summaryrefslogtreecommitdiff
path: root/Build/source/libs/luajit/LuaJIT-src/src
diff options
context:
space:
mode:
authorLuigi Scarso <luigi.scarso@gmail.com>2018-01-09 23:43:20 +0000
committerLuigi Scarso <luigi.scarso@gmail.com>2018-01-09 23:43:20 +0000
commitaa5d0e8bf3515c424ba0594a2d4e06ae31da7796 (patch)
treee56fd10d89732547ddb207c1998a1ded4360382e /Build/source/libs/luajit/LuaJIT-src/src
parentd33684eeca435784c95c62c7981dd02f7818b263 (diff)
Imported Luatex 1.06.3 (Lua 5.2.4)
git-svn-id: svn://tug.org/texlive/trunk@46269 c570f23f-e606-0410-a88d-b1316a301751
Diffstat (limited to 'Build/source/libs/luajit/LuaJIT-src/src')
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/Makefile26
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/Makefile.dep32
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.c4
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_asm.c13
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_fold.c4
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_lib.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_libbc.h17
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_peobj.c30
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/genlibbc.lua2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/genminilua.lua11
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/host/minilua.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/bc.lua2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/bcsave.lua16
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm.lua4
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips.lua55
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mipsel.lua2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_ppc.lua4
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x64.lua2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x86.lua11
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dump.lua29
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/p.lua5
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/v.lua4
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/zone.lua2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h35
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_aux.c65
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_base.c17
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_bit.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_debug.c10
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_ffi.c4
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_init.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_io.c13
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_jit.c18
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_math.c6
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_os.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_package.c45
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_string.c6
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lib_table.c22
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj.supp6
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_alloc.c257
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_api.c115
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_arch.h78
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_asm.c225
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_asm.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_asm_arm.h35
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_asm_mips.h722
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_asm_ppc.h37
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_asm_x86.h670
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_bc.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_bc.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_bcdump.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_bcread.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_bcwrite.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_buf.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_buf.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_carith.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_carith.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.c179
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.h26
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.c88
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.c6
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.c12
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_clib.c10
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_clib.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.c17
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.c84
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.h16
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_debug.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_debug.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_def.h15
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.c6
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.h3
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_emit_arm.h9
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_emit_mips.h120
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_emit_ppc.h8
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_emit_x86.h141
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_err.c46
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_err.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_errmsg.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ff.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.c107
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_frame.h39
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_func.c6
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_func.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_gc.c23
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_gc.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.c32
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ir.c165
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ir.h47
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_ircall.h54
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_iropt.h18
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_jit.h76
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_lex.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_lex.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_lib.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_lib.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_load.c5
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.c21
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_meta.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_meta.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_obj.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_obj.h10
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_opt_dce.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_opt_fold.c101
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_opt_loop.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_opt_mem.c6
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_opt_narrow.c46
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_opt_sink.c17
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_opt_split.c13
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_parse.c16
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_parse.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_profile.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_profile.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_record.c318
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_record.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_snap.c98
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_snap.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_state.c12
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_state.h4
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_str.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_str.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.c10
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt_num.c7
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_tab.c3
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_tab.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_target.h6
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm64.h246
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_target_mips.h133
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_target_ppc.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_target_x86.h25
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_trace.c81
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_trace.h3
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_traceerr.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_udata.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_udata.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_vm.h6
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.h2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_vmmath.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/ljamalg.c2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lua.h14
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/luaconf.h10
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/luajit.c123
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/luajit.h8
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lualib.h4
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/msvcbuild.bat16
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/vm_arm.dasc2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/vm_arm64.dasc287
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/vm_mips.dasc49
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/vm_ppc.dasc2
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/vm_x64.dasc49
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/vm_x86.dasc81
165 files changed, 4409 insertions, 1561 deletions
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/Makefile b/Build/source/libs/luajit/LuaJIT-src/src/Makefile
index 5139ccd2774..34c5e974a43 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/Makefile
+++ b/Build/source/libs/luajit/LuaJIT-src/src/Makefile
@@ -7,7 +7,7 @@
# Also works with MinGW and Cygwin on Windows.
# Please check msvcbuild.bat for building with MSVC on Windows.
#
-# Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+# Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
##############################################################################
MAJVER= 2
@@ -110,6 +110,9 @@ XCFLAGS+= -DLUAJIT_ENABLE_LUA52COMPAT
#XCFLAGS+= -DLUAJIT_NUMMODE=1
#XCFLAGS+= -DLUAJIT_NUMMODE=2
#
+# Enable GC64 mode for x64.
+#XCFLAGS+= -DLUAJIT_ENABLE_GC64
+#
##############################################################################
##############################################################################
@@ -121,8 +124,8 @@ XCFLAGS+= -DLUAJIT_ENABLE_LUA52COMPAT
#
# Use the system provided memory allocator (realloc) instead of the
# bundled memory allocator. This is slower, but sometimes helpful for
-# debugging. This option cannot be enabled on x64, since realloc usually
-# doesn't return addresses in the right address range.
+# debugging. This option cannot be enabled on x64 without GC64, since
+# realloc usually doesn't return addresses in the right address range.
# OTOH this option is mandatory for Valgrind's memcheck tool on x64 and
# the only way to get useful results from it for all other architectures.
#XCFLAGS+= -DLUAJIT_USE_SYSMALLOC
@@ -166,10 +169,6 @@ else
HOST_SYS= Windows
HOST_MSYS= cygwin
endif
- # Use Clang for OSX host.
- ifeq (Darwin,$(HOST_SYS))
- DEFAULT_CC= clang
- endif
endif
##############################################################################
@@ -210,7 +209,7 @@ TARGET_CC= $(STATIC_CC)
TARGET_STCC= $(STATIC_CC)
TARGET_DYNCC= $(DYNAMIC_CC)
TARGET_LD= $(CROSS)$(CC)
-TARGET_AR= $(CROSS)ar rcus
+TARGET_AR= $(CROSS)ar rcus 2>/dev/null
TARGET_STRIP= $(CROSS)strip
TARGET_LIBPATH= $(or $(PREFIX),/usr/local)/$(or $(MULTILIB),lib)
@@ -243,6 +242,9 @@ ifneq (,$(findstring LJ_TARGET_ARM ,$(TARGET_TESTARCH)))
TARGET_LJARCH= arm
else
ifneq (,$(findstring LJ_TARGET_ARM64 ,$(TARGET_TESTARCH)))
+ ifneq (,$(findstring __AARCH64EB__ ,$(TARGET_TESTARCH)))
+ TARGET_ARCH= -D__AARCH64EB__=1
+ endif
TARGET_LJARCH= arm64
else
ifneq (,$(findstring LJ_TARGET_PPC ,$(TARGET_TESTARCH)))
@@ -257,7 +259,11 @@ ifneq (,$(findstring LJ_TARGET_MIPS ,$(TARGET_TESTARCH)))
ifneq (,$(findstring MIPSEL ,$(TARGET_TESTARCH)))
TARGET_ARCH= -D__MIPSEL__=1
endif
- TARGET_LJARCH= mips
+ ifneq (,$(findstring LJ_TARGET_MIPS64 ,$(TARGET_TESTARCH)))
+ TARGET_LJARCH= mips64
+ else
+ TARGET_LJARCH= mips
+ endif
else
$(error Unsupported target architecture)
endif
@@ -310,7 +316,6 @@ ifeq (Darwin,$(TARGET_SYS))
export MACOSX_DEPLOYMENT_TARGET=10.4
endif
TARGET_STRIP+= -x
- TARGET_AR+= 2>/dev/null
TARGET_XSHLDFLAGS= -dynamiclib -single_module -undefined dynamic_lookup -fPIC
TARGET_DYNXLDOPTS=
TARGET_XSHLDFLAGS+= -install_name $(TARGET_DYLIBPATH) -compatibility_version $(MAJVER).$(MINVER) -current_version $(MAJVER).$(MINVER).$(RELVER)
@@ -321,7 +326,6 @@ ifeq (Darwin,$(TARGET_SYS))
else
ifeq (iOS,$(TARGET_SYS))
TARGET_STRIP+= -x
- TARGET_AR+= 2>/dev/null
TARGET_XSHLDFLAGS= -dynamiclib -single_module -undefined dynamic_lookup -fPIC
TARGET_DYNXLDOPTS=
TARGET_XSHLDFLAGS+= -install_name $(TARGET_DYLIBPATH) -compatibility_version $(MAJVER).$(MINVER) -current_version $(MAJVER).$(MINVER).$(RELVER)
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/Makefile.dep b/Build/source/libs/luajit/LuaJIT-src/src/Makefile.dep
index 7fc5ade4e18..44d66f198f6 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/Makefile.dep
+++ b/Build/source/libs/luajit/LuaJIT-src/src/Makefile.dep
@@ -3,8 +3,8 @@ lib_aux.o: lib_aux.c lua.h luaconf.h lauxlib.h lj_obj.h lj_def.h \
lj_dispatch.h lj_bc.h lj_traceerr.h lj_lib.h lj_alloc.h
lib_base.o: lib_base.c lua.h luaconf.h lauxlib.h lualib.h lj_obj.h \
lj_def.h lj_arch.h lj_gc.h lj_err.h lj_errmsg.h lj_debug.h lj_str.h \
- lj_tab.h lj_meta.h lj_state.h lj_ctype.h lj_cconv.h lj_bc.h lj_ff.h \
- lj_ffdef.h lj_dispatch.h lj_jit.h lj_ir.h lj_char.h lj_strscan.h \
+ lj_tab.h lj_meta.h lj_state.h lj_frame.h lj_bc.h lj_ctype.h lj_cconv.h \
+ lj_ff.h lj_ffdef.h lj_dispatch.h lj_jit.h lj_ir.h lj_char.h lj_strscan.h \
lj_strfmt.h lj_lib.h lj_libdef.h
lbitlib.o: lbitlib.c lua.h luaconf.h lauxlib.h lualib.h
lib_bit.o: lib_bit.c lua.h luaconf.h lauxlib.h lualib.h lj_obj.h lj_def.h \
@@ -164,7 +164,7 @@ lj_opt_sink.o: lj_opt_sink.c lj_obj.h lua.h luaconf.h lj_def.h lj_arch.h \
lj_ir.h lj_jit.h lj_iropt.h lj_target.h lj_target_*.h
lj_opt_split.o: lj_opt_split.c lj_obj.h lua.h luaconf.h lj_def.h \
lj_arch.h lj_err.h lj_errmsg.h lj_buf.h lj_gc.h lj_str.h lj_ir.h \
- lj_jit.h lj_ircall.h lj_iropt.h lj_vm.h
+ lj_jit.h lj_ircall.h lj_iropt.h lj_dispatch.h lj_bc.h lj_vm.h
lj_parse.o: lj_parse.c lj_obj.h lua.h luaconf.h lj_def.h lj_arch.h \
lj_gc.h lj_err.h lj_errmsg.h lj_debug.h lj_buf.h lj_str.h lj_tab.h \
lj_func.h lj_state.h lj_bc.h lj_ctype.h lj_strfmt.h lj_lex.h lj_parse.h \
@@ -216,19 +216,19 @@ ljamalg.o: ljamalg.c lua.h luaconf.h lauxlib.h lj_gc.c lj_obj.h lj_def.h \
lj_func.c lj_udata.c lj_meta.c lj_strscan.h lj_lib.h lj_debug.c \
lj_state.c lj_lex.h lj_alloc.h luajit.h lj_dispatch.c lj_ccallback.h \
lj_profile.h lj_vmevent.c lj_vmevent.h lj_vmmath.c lj_strscan.c \
- lj_strfmt.c lj_api.c lj_profile.c lj_lex.c lualib.h lj_parse.h \
- lj_parse.c lj_bcread.c lj_bcdump.h lj_bcwrite.c lj_load.c lj_ctype.c \
- lj_cdata.c lj_cconv.h lj_cconv.c lj_ccall.c lj_ccall.h lj_ccallback.c \
- lj_target.h lj_target_*.h lj_mcode.h lj_carith.c lj_carith.h lj_clib.c \
- lj_clib.h lj_cparse.c lj_cparse.h lj_lib.c lj_ir.c lj_ircall.h \
- lj_iropt.h lj_opt_mem.c lj_opt_fold.c lj_folddef.h lj_opt_narrow.c \
- lj_opt_dce.c lj_opt_loop.c lj_snap.h lj_opt_split.c lj_opt_sink.c \
- lj_mcode.c lj_snap.c lj_record.c lj_record.h lj_ffrecord.h lj_crecord.c \
- lj_crecord.h lj_ffrecord.c lj_recdef.h lj_asm.c lj_asm.h lj_emit_*.h \
- lj_asm_*.h lj_trace.c lj_gdbjit.h lj_gdbjit.c lj_alloc.c lib_aux.c \
- lib_base.c lj_libdef.h lib_math.c lib_string.c lib_table.c lib_io.c \
- lib_os.c lib_package.c lib_debug.c lib_bit.c lib_jit.c lib_ffi.c \
- lib_init.c
+ lj_strfmt.c lj_strfmt_num.c lj_api.c lj_profile.c lj_lex.c lualib.h \
+ lj_parse.h lj_parse.c lj_bcread.c lj_bcdump.h lj_bcwrite.c lj_load.c \
+ lj_ctype.c lj_cdata.c lj_cconv.h lj_cconv.c lj_ccall.c lj_ccall.h \
+ lj_ccallback.c lj_target.h lj_target_*.h lj_mcode.h lj_carith.c \
+ lj_carith.h lj_clib.c lj_clib.h lj_cparse.c lj_cparse.h lj_lib.c lj_ir.c \
+ lj_ircall.h lj_iropt.h lj_opt_mem.c lj_opt_fold.c lj_folddef.h \
+ lj_opt_narrow.c lj_opt_dce.c lj_opt_loop.c lj_snap.h lj_opt_split.c \
+ lj_opt_sink.c lj_mcode.c lj_snap.c lj_record.c lj_record.h lj_ffrecord.h \
+ lj_crecord.c lj_crecord.h lj_ffrecord.c lj_recdef.h lj_asm.c lj_asm.h \
+ lj_emit_*.h lj_asm_*.h lj_trace.c lj_gdbjit.h lj_gdbjit.c lj_alloc.c \
+ lib_aux.c lib_base.c lj_libdef.h lib_math.c lib_string.c lib_table.c \
+ lib_io.c lib_os.c lib_package.c lib_debug.c lib_bit.c lib_jit.c \
+ lib_ffi.c lib_init.c
luajit.o: luajit.c lua.h luaconf.h lauxlib.h lualib.h luajit.h lj_arch.h
host/buildvm.o: host/buildvm.c host/buildvm.h lj_def.h lua.h luaconf.h \
lj_arch.h lj_obj.h lj_def.h lj_arch.h lj_gc.h lj_obj.h lj_bc.h lj_ir.h \
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.c b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.c
index 3feae49f6fd..98a7a57a40c 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.c
@@ -1,6 +1,6 @@
/*
** LuaJIT VM builder.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** This is a tool to build the hand-tuned assembler code required for
** LuaJIT's bytecode interpreter. It supports a variety of output formats
@@ -110,7 +110,7 @@ static const char *sym_decorate(BuildCtx *ctx,
if (p) {
#if LJ_TARGET_X86ORX64
if (!LJ_64 && (ctx->mode == BUILD_coffasm || ctx->mode == BUILD_peobj))
- name[0] = '@';
+ name[0] = name[1] == 'R' ? '_' : '@'; /* Just for _RtlUnwind@16. */
else
*p = '\0';
#elif LJ_TARGET_PPC && !LJ_TARGET_OSX && !LJ_TARGET_CONSOLE
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.h b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.h
index 90d21b15f65..b90428dc48f 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm.h
@@ -1,6 +1,6 @@
/*
** LuaJIT VM builder.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _BUILDVM_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_asm.c b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_asm.c
index 28419c077a4..ffd14903c64 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_asm.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_asm.c
@@ -1,6 +1,6 @@
/*
** LuaJIT VM builder: Assembler source code emitter.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "buildvm.h"
@@ -93,10 +93,14 @@ static void emit_asm_words(BuildCtx *ctx, uint8_t *p, int n)
{
int i;
for (i = 0; i < n; i += 4) {
+ uint32_t ins = *(uint32_t *)(p+i);
+#if LJ_TARGET_ARM64 && LJ_BE
+ ins = lj_bswap(ins); /* ARM64 instructions are always little-endian. */
+#endif
if ((i & 15) == 0)
- fprintf(ctx->fp, "\t.long 0x%08x", *(uint32_t *)(p+i));
+ fprintf(ctx->fp, "\t.long 0x%08x", ins);
else
- fprintf(ctx->fp, ",0x%08x", *(uint32_t *)(p+i));
+ fprintf(ctx->fp, ",0x%08x", ins);
if ((i & 15) == 12) putc('\n', ctx->fp);
}
if ((n & 15) != 0) putc('\n', ctx->fp);
@@ -214,7 +218,8 @@ static void emit_asm_label(BuildCtx *ctx, const char *name, int size, int isfunc
case BUILD_machasm:
fprintf(ctx->fp,
"\n\t.private_extern %s\n"
- "%s:\n", name, name);
+ "\t.no_dead_strip %s\n"
+ "%s:\n", name, name, name);
break;
default:
break;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_fold.c b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_fold.c
index eec1af1cbac..d579f4d416a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_fold.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_fold.c
@@ -1,6 +1,6 @@
/*
** LuaJIT VM builder: IR folding hash table generator.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "buildvm.h"
@@ -9,7 +9,7 @@
/* Context for the folding hash table generator. */
static int lineno;
-static int funcidx;
+static uint32_t funcidx;
static uint32_t foldkeys[BUILD_MAX_FOLD];
static uint32_t nkeys;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_lib.c b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_lib.c
index 1adc3cbd60a..2956fdb6cd0 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_lib.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_lib.c
@@ -1,6 +1,6 @@
/*
** LuaJIT VM builder: library definition compiler.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "buildvm.h"
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_libbc.h b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_libbc.h
index 45f8f8cbdf2..b2600bd5905 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_libbc.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_libbc.h
@@ -15,7 +15,12 @@ static const uint8_t libbc_code[] = {
8,2,0,0,88,3,23,128,59,3,2,0,43,4,0,0,64,4,2,0,76,3,2,0,88,3,18,128,16,1,14,
0,41,3,1,0,3,3,1,0,88,3,14,128,3,1,2,0,88,3,12,128,59,3,1,0,22,4,1,1,18,5,2,
0,41,6,1,0,77,4,4,128,23,8,1,7,59,9,7,0,64,9,8,0,79,4,252,127,43,4,0,0,64,4,
-2,0,76,3,2,0,75,0,1,0,0,2,0
+2,0,76,3,2,0,75,0,1,0,0,2,0,5,12,0,0,0,35,16,0,12,0,16,1,14,0,16,2,14,0,16,
+3,14,0,11,4,0,0,88,5,1,128,18,4,0,0,16,4,12,0,3,1,2,0,88,5,24,128,33,5,1,3,
+0,2,3,0,88,6,4,128,2,3,1,0,88,6,2,128,4,4,0,0,88,6,9,128,18,6,1,0,18,7,2,0,
+41,8,1,0,77,6,4,128,32,10,5,9,59,11,9,0,64,11,10,4,79,6,252,127,88,6,8,128,
+18,6,2,0,18,7,1,0,41,8,255,255,77,6,4,128,32,10,5,9,59,11,9,0,64,11,10,4,79,
+6,252,127,76,4,2,0,0
#else
0,1,2,0,0,1,2,24,1,0,0,76,1,2,0,241,135,158,166,3,220,203,178,130,4,0,1,2,0,
0,1,2,24,1,0,0,76,1,2,0,243,244,148,165,20,198,190,199,252,3,0,1,2,0,0,0,3,
@@ -28,7 +33,12 @@ static const uint8_t libbc_code[] = {
8,2,0,0,88,3,23,128,59,3,2,0,43,4,0,0,64,4,2,0,76,3,2,0,88,3,18,128,16,1,14,
0,41,3,1,0,3,3,1,0,88,3,14,128,3,1,2,0,88,3,12,128,59,3,1,0,22,4,1,1,18,5,2,
0,41,6,1,0,77,4,4,128,23,8,1,7,59,9,7,0,64,9,8,0,79,4,252,127,43,4,0,0,64,4,
-2,0,76,3,2,0,75,0,1,0,0,2,0
+2,0,76,3,2,0,75,0,1,0,0,2,0,5,12,0,0,0,35,16,0,12,0,16,1,14,0,16,2,14,0,16,
+3,14,0,11,4,0,0,88,5,1,128,18,4,0,0,16,4,12,0,3,1,2,0,88,5,24,128,33,5,1,3,
+0,2,3,0,88,6,4,128,2,3,1,0,88,6,2,128,4,4,0,0,88,6,9,128,18,6,1,0,18,7,2,0,
+41,8,1,0,77,6,4,128,32,10,5,9,59,11,9,0,64,11,10,4,79,6,252,127,88,6,8,128,
+18,6,2,0,18,7,1,0,41,8,255,255,77,6,4,128,32,10,5,9,59,11,9,0,64,11,10,4,79,
+6,252,127,76,4,2,0,0
#endif
};
@@ -40,6 +50,7 @@ static const struct { const char *name; int ofs; } libbc_map[] = {
{"table_foreach",136},
{"table_getn",207},
{"table_remove",226},
-{NULL,355}
+{"table_move",355},
+{NULL,502}
};
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_peobj.c b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_peobj.c
index e8c927d8bf9..2eb2bb7bb33 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_peobj.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/buildvm_peobj.c
@@ -1,6 +1,6 @@
/*
** LuaJIT VM builder: PE object emitter.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Only used for building on Windows, since we cannot assume the presence
** of a suitable assembler. The host and target byte order must match.
@@ -109,6 +109,8 @@ enum {
#if LJ_TARGET_X64
PEOBJ_SECT_PDATA,
PEOBJ_SECT_XDATA,
+#elif LJ_TARGET_X86
+ PEOBJ_SECT_SXDATA,
#endif
PEOBJ_SECT_RDATA_Z,
PEOBJ_NSECTIONS
@@ -208,6 +210,13 @@ void emit_peobj(BuildCtx *ctx)
sofs += (pesect[PEOBJ_SECT_XDATA].nreloc = 1) * PEOBJ_RELOC_SIZE;
/* Flags: 40 = read, 30 = align4, 40 = initialized data. */
pesect[PEOBJ_SECT_XDATA].flags = 0x40300040;
+#elif LJ_TARGET_X86
+ memcpy(pesect[PEOBJ_SECT_SXDATA].name, ".sxdata", sizeof(".sxdata")-1);
+ pesect[PEOBJ_SECT_SXDATA].ofs = sofs;
+ sofs += (pesect[PEOBJ_SECT_SXDATA].size = 4);
+ pesect[PEOBJ_SECT_SXDATA].relocofs = sofs;
+ /* Flags: 40 = read, 30 = align4, 02 = lnk_info, 40 = initialized data. */
+ pesect[PEOBJ_SECT_SXDATA].flags = 0x40300240;
#endif
memcpy(pesect[PEOBJ_SECT_RDATA_Z].name, ".rdata$Z", sizeof(".rdata$Z")-1);
@@ -232,7 +241,7 @@ void emit_peobj(BuildCtx *ctx)
nrsym = ctx->nrelocsym;
pehdr.nsyms = 1+PEOBJ_NSECTIONS*2 + 1+ctx->nsym + nrsym;
#if LJ_TARGET_X64
- pehdr.nsyms += 1; /* Symbol for lj_err_unwind_win64. */
+ pehdr.nsyms += 1; /* Symbol for lj_err_unwind_win. */
#endif
/* Write PE object header and all sections. */
@@ -312,6 +321,19 @@ void emit_peobj(BuildCtx *ctx)
reloc.type = PEOBJ_RELOC_ADDR32NB;
owrite(ctx, &reloc, PEOBJ_RELOC_SIZE);
}
+#elif LJ_TARGET_X86
+ /* Write .sxdata section. */
+ for (i = 0; i < nrsym; i++) {
+ if (!strcmp(ctx->relocsym[i], "_lj_err_unwind_win")) {
+ uint32_t symidx = 1+2+i;
+ owrite(ctx, &symidx, 4);
+ break;
+ }
+ }
+ if (i == nrsym) {
+ fprintf(stderr, "Error: extern lj_err_unwind_win not used\n");
+ exit(1);
+ }
#endif
/* Write .rdata$Z section. */
@@ -333,8 +355,10 @@ void emit_peobj(BuildCtx *ctx)
#if LJ_TARGET_X64
emit_peobj_sym_sect(ctx, pesect, PEOBJ_SECT_PDATA);
emit_peobj_sym_sect(ctx, pesect, PEOBJ_SECT_XDATA);
- emit_peobj_sym(ctx, "lj_err_unwind_win64", 0,
+ emit_peobj_sym(ctx, "lj_err_unwind_win", 0,
PEOBJ_SECT_UNDEF, PEOBJ_TYPE_FUNC, PEOBJ_SCL_EXTERN);
+#elif LJ_TARGET_X86
+ emit_peobj_sym_sect(ctx, pesect, PEOBJ_SECT_SXDATA);
#endif
emit_peobj_sym(ctx, ctx->beginsym, 0,
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/genlibbc.lua b/Build/source/libs/luajit/LuaJIT-src/src/host/genlibbc.lua
index 7a17a3d06b8..6f5a05cce51 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/genlibbc.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/genlibbc.lua
@@ -2,7 +2,7 @@
-- Lua script to dump the bytecode of the library functions written in Lua.
-- The resulting 'buildvm_libbc.h' is used for the build process of LuaJIT.
----------------------------------------------------------------------------
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/genminilua.lua b/Build/source/libs/luajit/LuaJIT-src/src/host/genminilua.lua
index 0a4690c1d2f..50feff014d0 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/genminilua.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/genminilua.lua
@@ -2,7 +2,7 @@
-- Lua script to generate a customized, minified version of Lua.
-- The resulting 'minilua' is used for the build process of LuaJIT.
----------------------------------------------------------------------------
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
@@ -157,11 +157,11 @@ local function merge_includes(src)
if includes[name] then return "" end
includes[name] = true
local fp = assert(io.open(LUA_SOURCE..name, "r"))
- local src = fp:read("*a")
+ local inc = fp:read("*a")
assert(fp:close())
- src = gsub(src, "#ifndef%s+%w+_h\n#define%s+%w+_h\n", "")
- src = gsub(src, "#endif%s*$", "")
- return merge_includes(src)
+ inc = gsub(inc, "#ifndef%s+%w+_h\n#define%s+%w+_h\n", "")
+ inc = gsub(inc, "#endif%s*$", "")
+ return merge_includes(inc)
end)
end
@@ -300,6 +300,7 @@ local function strip_unused3(src)
src = gsub(src, "if%([^\n]*hookmask[^\n]*&&\n[^\n]*%b{}\n", "")
src = gsub(src, "(twoto%b()%()", "%1(size_t)")
src = gsub(src, "i<sizenode", "i<(int)sizenode")
+ src = gsub(src, "cast%(unsigned int,key%-1%)", "cast(unsigned int,key)-1")
return gsub(src, "\n\n+", "\n")
end
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/host/minilua.c b/Build/source/libs/luajit/LuaJIT-src/src/host/minilua.c
index aee192a710c..79150286cfc 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/host/minilua.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/host/minilua.c
@@ -1606,7 +1606,7 @@ luaC_barriert(L,t,key);
return gval(mp);
}
static const TValue*luaH_getnum(Table*t,int key){
-if(cast(unsigned int,key-1)<cast(unsigned int,t->sizearray))
+if(cast(unsigned int,key)-1<cast(unsigned int,t->sizearray))
return&t->array[key-1];
else{
lua_Number nk=cast_num(key);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/bc.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/bc.lua
index a8cb8496e43..193cf01f939 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/bc.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/bc.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT bytecode listing module.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/bcsave.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/bcsave.lua
index d0968b18760..c17c88e0ff1 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/bcsave.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/bcsave.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT module to save/list bytecode.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
@@ -63,8 +63,8 @@ local map_type = {
}
local map_arch = {
- x86 = true, x64 = true, arm = true, arm64 = true, ppc = true,
- mips = true, mipsel = true,
+ x86 = true, x64 = true, arm = true, arm64 = true, arm64be = true,
+ ppc = true, mips = true, mipsel = true,
}
local map_os = {
@@ -125,12 +125,12 @@ extern "C"
#ifdef _WIN32
__declspec(dllexport)
#endif
-const char %s%s[] = {
+const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname))
else
fp:write(string.format([[
#define %s%s_SIZE %d
-static const char %s%s[] = {
+static const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname, #s, LJBC_PREFIX, ctx.modname))
end
local t, n, m = {}, 0, 0
@@ -200,7 +200,7 @@ typedef struct {
]]
local symname = LJBC_PREFIX..ctx.modname
local is64, isbe = false, false
- if ctx.arch == "x64" or ctx.arch == "arm64" then
+ if ctx.arch == "x64" or ctx.arch == "arm64" or ctx.arch == "arm64be" then
is64 = true
elseif ctx.arch == "ppc" or ctx.arch == "mips" then
isbe = true
@@ -237,9 +237,9 @@ typedef struct {
hdr.eendian = isbe and 2 or 1
hdr.eversion = 1
hdr.type = f16(1)
- hdr.machine = f16(({ x86=3, x64=62, arm=40, arm64=183, ppc=20, mips=8, mipsel=8 })[ctx.arch])
+ hdr.machine = f16(({ x86=3, x64=62, arm=40, arm64=183, arm64be=183, ppc=20, mips=8, mipsel=8 })[ctx.arch])
if ctx.arch == "mips" or ctx.arch == "mipsel" then
- hdr.flags = 0x50001006
+ hdr.flags = f32(0x50001006)
end
hdr.version = f32(1)
hdr.shofs = fofs(ffi.offsetof(o, "sect"))
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm.lua
index 1296d816be0..c2dd776991f 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT ARM disassembler module.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
@@ -12,7 +12,7 @@
local type = type
local sub, byte, format = string.sub, string.byte, string.format
-local match, gmatch, gsub = string.match, string.gmatch, string.gsub
+local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, ror, tohex = bit.band, bit.bor, bit.ror, bit.tohex
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips.lua
index 2bf8b389c36..a12b8e62f3b 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS disassembler module.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
@@ -11,8 +11,8 @@
------------------------------------------------------------------------------
local type = type
-local sub, byte, format = string.sub, string.byte, string.format
-local match, gmatch, gsub = string.match, string.gmatch, string.gsub
+local byte, format = string.byte, string.format
+local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, tohex = bit.band, bit.bor, bit.tohex
@@ -34,15 +34,17 @@ local map_special = {
"jrS", "jalrD1S", "movzDST", "movnDST",
"syscallY", "breakY", false, "sync",
"mfhiD", "mthiS", "mfloD", "mtloS",
- false, false, false, false,
+ "dsllvDST", false, "dsrlvDST", "dsravDST",
"multST", "multuST", "divST", "divuST",
- false, false, false, false,
+ "dmultST", "dmultuST", "ddivST", "ddivuST",
"addDST", "addu|moveDST0", "subDST", "subu|neguDS0T",
- "andDST", "orDST", "xorDST", "nor|notDST0",
+ "andDST", "or|moveDST0", "xorDST", "nor|notDST0",
false, false, "sltDST", "sltuDST",
- false, false, false, false,
+ "daddDST", "dadduDST", "dsubDST", "dsubuDST",
"tgeSTZ", "tgeuSTZ", "tltSTZ", "tltuSTZ",
- "teqSTZ", false, "tneSTZ",
+ "teqSTZ", false, "tneSTZ", false,
+ "dsllDTA", false, "dsrlDTA", "dsraDTA",
+ "dsll32DTA", false, "dsrl32DTA", "dsra32DTA",
}
local map_special2 = {
@@ -60,11 +62,17 @@ local map_bshfl = {
[24] = "sehDT",
}
+local map_dbshfl = {
+ shift = 6, mask = 31,
+ [2] = "dsbhDT",
+ [5] = "dshdDT",
+}
+
local map_special3 = {
shift = 0, mask = 63,
- [0] = "extTSAK", [4] = "insTSAL",
- [32] = map_bshfl,
- [59] = "rdhwrTD",
+ [0] = "extTSAK", [1] = "dextmTSAP", [3] = "dextTSAK",
+ [4] = "insTSAL", [6] = "dinsuTSEQ", [7] = "dinsTSAL",
+ [32] = map_bshfl, [36] = map_dbshfl, [59] = "rdhwrTD",
}
local map_regimm = {
@@ -178,8 +186,8 @@ local map_cop1bc = {
local map_cop1 = {
shift = 21, mask = 31,
- [0] = "mfc1TG", false, "cfc1TG", "mfhc1TG",
- "mtc1TG", false, "ctc1TG", "mthc1TG",
+ [0] = "mfc1TG", "dmfc1TG", "cfc1TG", "mfhc1TG",
+ "mtc1TG", "dmtc1TG", "ctc1TG", "mthc1TG",
map_cop1bc, false, false, false,
false, false, false, false,
map_cop1s, map_cop1d, false, false,
@@ -213,16 +221,16 @@ local map_pri = {
"andiTSU", "ori|liTS0U", "xoriTSU", "luiTU",
map_cop0, map_cop1, false, map_cop1x,
"beql|beqzlST0B", "bnel|bnezlST0B", "blezlSB", "bgtzlSB",
- false, false, false, false,
- map_special2, false, false, map_special3,
+ "daddiTSI", "daddiuTSI", false, false,
+ map_special2, "jalxJ", false, map_special3,
"lbTSO", "lhTSO", "lwlTSO", "lwTSO",
"lbuTSO", "lhuTSO", "lwrTSO", false,
"sbTSO", "shTSO", "swlTSO", "swTSO",
false, false, "swrTSO", "cacheNSO",
"llTSO", "lwc1HSO", "lwc2TSO", "prefNSO",
- false, "ldc1HSO", "ldc2TSO", false,
+ false, "ldc1HSO", "ldc2TSO", "ldTSO",
"scTSO", "swc1HSO", "swc2TSO", false,
- false, "sdc1HSO", "sdc2TSO", false,
+ false, "sdc1HSO", "sdc2TSO", "sdTSO",
}
------------------------------------------------------------------------------
@@ -306,6 +314,8 @@ local function disass_ins(ctx)
x = "f"..band(rshift(op, 21), 31)
elseif p == "A" then
x = band(rshift(op, 6), 31)
+ elseif p == "E" then
+ x = band(rshift(op, 6), 31) + 32
elseif p == "M" then
x = band(rshift(op, 11), 31)
elseif p == "N" then
@@ -315,8 +325,12 @@ local function disass_ins(ctx)
if x == 0 then x = nil end
elseif p == "K" then
x = band(rshift(op, 11), 31) + 1
+ elseif p == "P" then
+ x = band(rshift(op, 11), 31) + 33
elseif p == "L" then
x = band(rshift(op, 11), 31) - last + 1
+ elseif p == "Q" then
+ x = band(rshift(op, 11), 31) - last + 33
elseif p == "I" then
x = arshift(lshift(op, 16), 16)
elseif p == "U" then
@@ -330,11 +344,12 @@ local function disass_ins(ctx)
elseif p == "B" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 16), 16)*4 + 4
ctx.rel = x
- x = "0x"..tohex(x)
+ x = format("0x%08x", x)
elseif p == "J" then
- x = band(ctx.addr + ctx.pos, 0xf0000000) + band(op, 0x03ffffff)*4
+ local a = ctx.addr + ctx.pos
+ x = a - band(a, 0x0fffffff) + band(op, 0x03ffffff)*4
ctx.rel = x
- x = "0x"..tohex(x)
+ x = format("0x%08x", x)
elseif p == "V" then
x = band(rshift(op, 8), 7)
if x == 0 then x = nil end
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mipsel.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mipsel.lua
index 3f2f6efc98f..f69b11f01fd 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mipsel.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mipsel.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT MIPSEL disassembler wrapper module.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the little-endian functions from the
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_ppc.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_ppc.lua
index 30f51ecdc14..2aeb1b29244 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_ppc.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_ppc.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT PPC disassembler module.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
@@ -13,7 +13,7 @@
------------------------------------------------------------------------------
local type = type
-local sub, byte, format = string.sub, string.byte, string.format
+local byte, format = string.byte, string.format
local match, gmatch, gsub = string.match, string.gmatch, string.gsub
local concat = table.concat
local bit = require("bit")
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x64.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x64.lua
index 077b1b127ca..d5714ee1f70 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x64.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x64.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT x64 disassembler wrapper module.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the 64 bit functions from the combined
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x86.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x86.lua
index f8a21ff30ec..4371233d2b4 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x86.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_x86.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT x86/x64 disassembler module.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
@@ -244,6 +244,7 @@ nil,"||psrlvVSXrvm","||psravdXrvm","||psllvVSXrvm",
[0xde] = "||aesdecXrvm", [0xdf] = "||aesdeclastXrvm",
--Fx
[0xf0] = "|||crc32TrBmt",[0xf1] = "|||crc32TrVmt",
+[0xf7] = "| sarxVrmv| shlxVrmv| shrxVrmv",
},
["3a"] = { -- [66] 0f 3a xx
@@ -273,6 +274,8 @@ nil,nil,nil,nil,
[0x60] = "||pcmpestrmXrmu",[0x61] = "||pcmpestriXrmu",
[0x62] = "||pcmpistrmXrmu",[0x63] = "||pcmpistriXrmu",
[0xdf] = "||aeskeygenassistXrmu",
+--Fx
+[0xf0] = "||| rorxVrmu",
},
}
@@ -414,8 +417,8 @@ local function putop(ctx, text, operands)
(ctx.rexx and "x" or "")..(ctx.rexb and "b" or "")..
(ctx.vexl and "l" or "")
if ctx.vexv and ctx.vexv ~= 0 then t = t.."v"..ctx.vexv end
- if t ~= "" then text = ctx.rex.."."..t.." "..text
- elseif ctx.rex == "vex" then text = "v"..text end
+ if t ~= "" then text = ctx.rex.."."..t.." "..gsub(text, "^ ", "")
+ elseif ctx.rex == "vex" then text = gsub("v"..text, "^v ", "") end
ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
ctx.rex = false; ctx.vexl = false; ctx.vexv = false
end
@@ -815,7 +818,7 @@ map_act = {
m = b%32; b = (b-m)/32
local nb = b%2; b = (b-nb)/2
if nb == 0 then ctx.rexb = true end
- local nx = b%2; b = (b-nx)/2
+ local nx = b%2
if nx == 0 then ctx.rexx = true end
b = byte(ctx.code, pos, pos)
if not b then return incomplete(ctx) end
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dump.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dump.lua
index 9d8330e1838..2bea652bf81 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/dump.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dump.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT compiler dump module.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
@@ -63,9 +63,9 @@ local traceinfo, traceir, tracek = jutil.traceinfo, jutil.traceir, jutil.tracek
local tracemc, tracesnap = jutil.tracemc, jutil.tracesnap
local traceexitstub, ircalladdr = jutil.traceexitstub, jutil.ircalladdr
local bit = require("bit")
-local band, shl, shr, tohex = bit.band, bit.lshift, bit.rshift, bit.tohex
+local band, shr, tohex = bit.band, bit.rshift, bit.tohex
local sub, gsub, format = string.sub, string.gsub, string.format
-local byte, char, rep = string.byte, string.char, string.rep
+local byte, rep = string.byte, string.rep
local type, tostring = type, tostring
local stdout, stderr = io.stdout, io.stderr
@@ -85,7 +85,7 @@ local nexitsym = 0
local function fillsymtab_tr(tr, nexit)
local t = {}
symtabmt.__index = t
- if jit.arch == "mips" or jit.arch == "mipsel" then
+ if jit.arch:sub(1, 4) == "mips" then
t[traceexitstub(tr, 0)] = "exit"
return
end
@@ -213,7 +213,7 @@ local colortype_ansi = {
"\027[35m%s\027[m",
}
-local function colorize_text(s, t)
+local function colorize_text(s)
return s
end
@@ -310,15 +310,17 @@ local function fmtfunc(func, pc)
end
end
-local function formatk(tr, idx)
+local function formatk(tr, idx, sn)
local k, t, slot = tracek(tr, idx)
local tn = type(k)
local s
if tn == "number" then
- if k == 2^52+2^51 then
+ if band(sn or 0, 0x30000) ~= 0 then
+ s = band(sn, 0x20000) ~= 0 and "contpc" or "ftsz"
+ elseif k == 2^52+2^51 then
s = "bias"
else
- s = format("%+.14g", k)
+ s = format(0 < k and k < 0x1p-1026 and "%+a" or "%+.14g", k)
end
elseif tn == "string" then
s = format(#k > 20 and '"%.20s"~' or '"%s"', gsub(k, "%c", ctlsub))
@@ -331,11 +333,13 @@ local function formatk(tr, idx)
s = format("userdata:%p", k)
else
s = format("[%p]", k)
- if s == "[0x00000000]" then s = "NULL" end
+ if s == "[NULL]" then s = "NULL" end
end
elseif t == 21 then -- int64_t
s = sub(tostring(k), 1, -3)
if sub(s, 1, 1) ~= "-" then s = "+"..s end
+ elseif sn == 0x1057fff then -- SNAP(1, SNAP_FRAME | SNAP_NORESTORE, REF_NIL)
+ return "----" -- Special case for LJ_FR2 slot 1.
else
s = tostring(k) -- For primitives.
end
@@ -354,7 +358,7 @@ local function printsnap(tr, snap)
n = n + 1
local ref = band(sn, 0xffff) - 0x8000 -- REF_BIAS
if ref < 0 then
- out:write(formatk(tr, ref))
+ out:write(formatk(tr, ref, sn))
elseif band(sn, 0x80000) ~= 0 then -- SNAP_SOFTFPNUM
out:write(colorize(format("%04d/%04d", ref, ref+1), 14))
else
@@ -552,7 +556,7 @@ local function dump_trace(what, tr, func, pc, otr, oex)
if what == "start" then
if dumpmode.H then out:write('<pre class="ljdump">\n') end
out:write("---- TRACE ", tr, " ", what)
- if otr then out:write(" ", otr, "/", oex) end
+ if otr then out:write(" ", otr, "/", oex == -1 and "stitch" or oex) end
out:write(" ", fmtfunc(func, pc), "\n")
elseif what == "stop" or what == "abort" then
out:write("---- TRACE ", tr, " ", what)
@@ -651,7 +655,8 @@ end
local function dumpon(opt, outfile)
if active then dumpoff() end
- local colormode = os.getenv("COLORTERM") and "A" or "T"
+ local term = os.getenv("TERM")
+ local colormode = (term and term:match("color") or os.getenv("COLORTERM")) and "A" or "T"
if opt then
opt = gsub(opt, "[TAH]", function(mode) colormode = mode; return ""; end)
end
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/p.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/p.lua
index 5323728b1df..7be105863d3 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/p.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/p.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT profiler.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
@@ -120,7 +120,7 @@ end
-- Show top N list.
local function prof_top(count1, count2, samples, indent)
local t, n = {}, 0
- for k, v in pairs(count1) do
+ for k in pairs(count1) do
n = n + 1
t[n] = k
end
@@ -156,6 +156,7 @@ local function prof_annotate(count1, samples)
ms = math.max(ms, v)
if pct >= prof_min then
local file, line = k:match("^(.*):(%d+)$")
+ if not file then file = k; line = 0 end
local fl = files[file]
if not fl then fl = {}; files[file] = fl; files[#files+1] = file end
line = tonumber(line)
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/v.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/v.lua
index 60c8b05aa4e..934de9859c7 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/v.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/v.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- Verbose mode of the LuaJIT compiler.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
@@ -99,7 +99,7 @@ end
local function dump_trace(what, tr, func, pc, otr, oex)
if what == "start" then
startloc = fmtfunc(func, pc)
- startex = otr and "("..otr.."/"..oex..") " or ""
+ startex = otr and "("..otr.."/"..(oex == -1 and "stitch" or oex)..") " or ""
else
if what == "abort" then
local loc = fmtfunc(func, pc)
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/zone.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/zone.lua
index f5f9656af63..fa702c4e989 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/jit/zone.lua
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/zone.lua
@@ -1,7 +1,7 @@
----------------------------------------------------------------------------
-- LuaJIT profiler zones.
--
--- Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h b/Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h
index ef0fd54dacc..dcb5d72346f 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h
@@ -15,9 +15,6 @@
#include "lua.h"
-#define luaL_getn(L,i) ((int)lua_objlen(L, i))
-#define luaL_setn(L,i,j) ((void)0) /* no op! */
-
/* extra error code for `luaL_load' */
#define LUA_ERRFILE (LUA_ERRERR+1)
@@ -58,6 +55,10 @@ LUALIB_API int (luaL_error) (lua_State *L, const char *fmt, ...);
LUALIB_API int (luaL_checkoption) (lua_State *L, int narg, const char *def,
const char *const lst[]);
+/* pre-defined references */
+#define LUA_NOREF (-2)
+#define LUA_REFNIL (-1)
+
LUALIB_API int (luaL_ref) (lua_State *L, int t);
LUALIB_API void (luaL_unref) (lua_State *L, int t, int ref);
@@ -84,6 +85,11 @@ LUALIB_API int (luaL_loadbufferx) (lua_State *L, const char *buff, size_t sz,
const char *name, const char *mode);
LUALIB_API void luaL_traceback (lua_State *L, lua_State *L1, const char *msg,
int level);
+LUALIB_API void (luaL_setfuncs) (lua_State *L, const luaL_Reg *l, int nup);
+LUALIB_API void (luaL_pushmodule) (lua_State *L, const char *modname,
+ int sizehint);
+LUALIB_API void *(luaL_testudata) (lua_State *L, int ud, const char *tname);
+LUALIB_API void (luaL_setmetatable) (lua_State *L, const char *tname);
@@ -111,7 +117,6 @@ typedef struct luaL_Stream {
-
/*
** ===============================================================
** some useful macros
@@ -139,6 +144,11 @@ typedef struct luaL_Stream {
#define luaL_opt(L,f,n,d) (lua_isnoneornil(L,(n)) ? (d) : f(L,(n)))
+/* From Lua 5.2. */
+#define luaL_newlibtable(L, l) \
+ lua_createtable(L, 0, sizeof(l)/sizeof((l)[0]) - 1)
+#define luaL_newlib(L, l) (luaL_newlibtable(L, l), luaL_setfuncs(L, l, 0))
+
/*
** {======================================================
** Generic Buffer manipulation
@@ -175,21 +185,4 @@ LUALIB_API int (RESERVED_load_aux_JIT) (lua_State *L, int status, int envarg);
/* }====================================================== */
-
-/* compatibility with ref system */
-
-/* pre-defined references */
-#define LUA_NOREF (-2)
-#define LUA_REFNIL (-1)
-
-#define lua_ref(L,lock) ((lock) ? luaL_ref(L, LUA_REGISTRYINDEX) : \
- (lua_pushstring(L, "unlocked references are obsolete"), lua_error(L), 0))
-
-#define lua_unref(L,ref) luaL_unref(L, LUA_REGISTRYINDEX, (ref))
-
-#define lua_getref(L,ref) lua_rawgeti(L, LUA_REGISTRYINDEX, (ref))
-
-
-#define luaL_reg luaL_Reg
-
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_aux.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_aux.c
index 0e61d951a24..c40565c340d 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_aux.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_aux.c
@@ -1,6 +1,6 @@
/*
** Auxiliary library for the Lua/C API.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major parts taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -107,38 +107,36 @@ LUALIB_API const char *luaL_findtable(lua_State *L, int idx,
static int libsize(const luaL_Reg *l)
{
int size = 0;
- for (; l->name; l++) size++;
+ for (; l && l->name; l++) size++;
return size;
}
+LUALIB_API void luaL_pushmodule(lua_State *L, const char *modname, int sizehint)
+{
+ luaL_findtable(L, LUA_REGISTRYINDEX, "_LOADED", 16);
+ lua_getfield(L, -1, modname);
+ if (!lua_istable(L, -1)) {
+ lua_pop(L, 1);
+ if (luaL_findtable(L, LUA_GLOBALSINDEX, modname, sizehint) != NULL)
+ lj_err_callerv(L, LJ_ERR_BADMODN, modname);
+ lua_pushvalue(L, -1);
+ lua_setfield(L, -3, modname); /* _LOADED[modname] = new table. */
+ }
+ lua_remove(L, -2); /* Remove _LOADED table. */
+}
+
LUALIB_API void luaL_openlib(lua_State *L, const char *libname,
const luaL_Reg *l, int nup)
{
lj_lib_checkfpu(L);
if (libname) {
- int size = libsize(l);
- /* check whether lib already exists */
- luaL_findtable(L, LUA_REGISTRYINDEX, "_LOADED", 16);
- lua_getfield(L, -1, libname); /* get _LOADED[libname] */
- if (!lua_istable(L, -1)) { /* not found? */
- lua_pop(L, 1); /* remove previous result */
- /* try global variable (and create one if it does not exist) */
- if (luaL_findtable(L, LUA_GLOBALSINDEX, libname, size) != NULL)
- lj_err_callerv(L, LJ_ERR_BADMODN, libname);
- lua_pushvalue(L, -1);
- lua_setfield(L, -3, libname); /* _LOADED[libname] = new table */
- }
- lua_remove(L, -2); /* remove _LOADED table */
- lua_insert(L, -(nup+1)); /* move library table to below upvalues */
- }
- for (; l->name; l++) {
- int i;
- for (i = 0; i < nup; i++) /* copy upvalues to the top */
- lua_pushvalue(L, -nup);
- lua_pushcclosure(L, l->func, nup);
- lua_setfield(L, -(nup+2), l->name);
+ luaL_pushmodule(L, libname, libsize(l));
+ lua_insert(L, -(nup + 1)); /* Move module table below upvalues. */
}
- lua_pop(L, nup); /* remove upvalues */
+ if (l)
+ luaL_setfuncs(L, l, nup);
+ else
+ lua_pop(L, nup); /* Remove upvalues. */
}
LUALIB_API void luaL_register(lua_State *L, const char *libname,
@@ -147,6 +145,19 @@ LUALIB_API void luaL_register(lua_State *L, const char *libname,
luaL_openlib(L, libname, l, 0);
}
+LUALIB_API void luaL_setfuncs(lua_State *L, const luaL_Reg *l, int nup)
+{
+ luaL_checkstack(L, nup, "too many upvalues");
+ for (; l->name; l++) {
+ int i;
+ for (i = 0; i < nup; i++) /* Copy upvalues to the top. */
+ lua_pushvalue(L, -nup);
+ lua_pushcclosure(L, l->func, nup);
+ lua_setfield(L, -(nup + 2), l->name);
+ }
+ lua_pop(L, nup); /* Remove upvalues. */
+}
+
LUALIB_API const char *luaL_gsub(lua_State *L, const char *s,
const char *p, const char *r)
{
@@ -302,7 +313,7 @@ static int panic(lua_State *L)
#ifdef LUAJIT_USE_SYSMALLOC
-#if LJ_64 && !defined(LUAJIT_USE_VALGRIND)
+#if LJ_64 && !LJ_GC64 && !defined(LUAJIT_USE_VALGRIND)
#error "Must use builtin allocator for 64 bit target"
#endif
@@ -334,7 +345,7 @@ LUALIB_API lua_State *luaL_newstate(void)
lua_State *L;
void *ud = lj_alloc_create();
if (ud == NULL) return NULL;
-#if LJ_64
+#if LJ_64 && !LJ_GC64
L = lj_state_newstate(lj_alloc_f, ud);
#else
L = lua_newstate(lj_alloc_f, ud);
@@ -343,7 +354,7 @@ LUALIB_API lua_State *luaL_newstate(void)
return L;
}
-#if LJ_64
+#if LJ_64 && !LJ_GC64
LUA_API lua_State *lua_newstate(lua_Alloc f, void *ud)
{
UNUSED(f); UNUSED(ud);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_base.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_base.c
index 6107bde0066..3a757870aa3 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_base.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_base.c
@@ -1,6 +1,6 @@
/*
** Base and coroutine library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2011 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -23,6 +23,7 @@
#include "lj_tab.h"
#include "lj_meta.h"
#include "lj_state.h"
+#include "lj_frame.h"
#if LJ_HASFFI
#include "lj_ctype.h"
#include "lj_cconv.h"
@@ -345,7 +346,7 @@ LJLIB_ASM_(xpcall) LJLIB_REC(.)
static int load_aux(lua_State *L, int status, int envarg)
{
- if (status == 0) {
+ if (status == LUA_OK) {
if (tvistab(L->base+envarg-1)) {
GCfunc *fn = funcV(L->top-1);
GCtab *t = tabV(L->base+envarg-1);
@@ -418,7 +419,7 @@ LJLIB_CF(dofile)
GCstr *fname = lj_lib_optstr(L, 1);
setnilV(L->top);
L->top = L->base+1;
- if (luaL_loadfile(L, fname ? strdata(fname) : NULL) != 0)
+ if (luaL_loadfile(L, fname ? strdata(fname) : NULL) != LUA_OK)
lua_error(L);
lua_call(L, 0, LUA_MULTRET);
return (int)(L->top - L->base) - 1;
@@ -536,7 +537,7 @@ LJLIB_CF(coroutine_status)
co = threadV(L->base);
if (co == L) s = "running";
else if (co->status == LUA_YIELD) s = "suspended";
- else if (co->status != 0) s = "dead";
+ else if (co->status != LUA_OK) s = "dead";
else if (co->base > tvref(co->stack)+1+LJ_FR2) s = "normal";
else if (co->top == co->base) s = "dead";
else s = "suspended";
@@ -557,6 +558,12 @@ LJLIB_CF(coroutine_running)
#endif
}
+LJLIB_CF(coroutine_isyieldable)
+{
+ setboolV(L->top++, cframe_canyield(L->cframe));
+ return 1;
+}
+
LJLIB_CF(coroutine_create)
{
lua_State *L1;
@@ -576,7 +583,7 @@ LJLIB_ASM(coroutine_yield)
static int ffh_resume(lua_State *L, lua_State *co, int wrap)
{
if (co->cframe != NULL || co->status > LUA_YIELD ||
- (co->status == 0 && co->top == co->base)) {
+ (co->status == LUA_OK && co->top == co->base)) {
ErrMsg em = co->cframe ? LJ_ERR_CORUN : LJ_ERR_CODEAD;
if (wrap) lj_err_caller(L, em);
setboolV(L->base-1-LJ_FR2, 0);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_bit.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_bit.c
index 1344d24d9fa..c979a44839c 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_bit.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_bit.c
@@ -1,6 +1,6 @@
/*
** Bit manipulation library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lib_bit_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_debug.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_debug.c
index cda7137efc0..f112b5bc870 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_debug.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_debug.c
@@ -1,6 +1,6 @@
/*
** Debug library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -283,13 +283,13 @@ LJLIB_CF(debug_setuservalue)
/* ------------------------------------------------------------------------ */
-static const char KEY_HOOK = 'h';
+#define KEY_HOOK ((void *)0x3004)
static void hookf(lua_State *L, lua_Debug *ar)
{
static const char *const hooknames[] =
{"call", "return", "line", "count", "tail return"};
- lua_pushlightuserdata(L, (void *)&KEY_HOOK);
+ lua_pushlightuserdata(L, KEY_HOOK);
lua_rawget(L, LUA_REGISTRYINDEX);
if (lua_isfunction(L, -1)) {
lua_pushstring(L, hooknames[(int)ar->event]);
@@ -334,7 +334,7 @@ LJLIB_CF(debug_sethook)
count = luaL_optint(L, arg+3, 0);
func = hookf; mask = makemask(smask, count);
}
- lua_pushlightuserdata(L, (void *)&KEY_HOOK);
+ lua_pushlightuserdata(L, KEY_HOOK);
lua_pushvalue(L, arg+1);
lua_rawset(L, LUA_REGISTRYINDEX);
lua_sethook(L, func, mask, count);
@@ -349,7 +349,7 @@ LJLIB_CF(debug_gethook)
if (hook != NULL && hook != hookf) { /* external hook? */
lua_pushliteral(L, "external hook");
} else {
- lua_pushlightuserdata(L, (void *)&KEY_HOOK);
+ lua_pushlightuserdata(L, KEY_HOOK);
lua_rawget(L, LUA_REGISTRYINDEX); /* get hook */
}
lua_pushstring(L, unmakemask(mask, buff));
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_ffi.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_ffi.c
index 1feee2157f9..136e98e8963 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_ffi.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_ffi.c
@@ -1,6 +1,6 @@
/*
** FFI library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lib_ffi_c
@@ -829,7 +829,7 @@ static GCtab *ffi_finalizer(lua_State *L)
settabV(L, L->top++, t);
setgcref(t->metatable, obj2gco(t));
setstrV(L, lj_tab_setstr(L, t, lj_str_newlit(L, "__mode")),
- lj_str_newlit(L, "K"));
+ lj_str_newlit(L, "k"));
t->nomm = (uint8_t)(~(1u<<MM_mode));
return t;
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_init.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_init.c
index bbb99ec0efd..405471dac0e 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_init.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_init.c
@@ -1,6 +1,6 @@
/*
** Library initialization.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major parts taken verbatim from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_io.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_io.c
index 31f0ea97521..9763ed466fe 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_io.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_io.c
@@ -1,6 +1,6 @@
/*
** I/O library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2011 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -203,13 +203,12 @@ static int io_file_read(lua_State *L, FILE *fp, int start)
for (n = start; nargs-- && ok; n++) {
if (tvisstr(L->base+n)) {
const char *p = strVdata(L->base+n);
- if (p[0] != '*')
- lj_err_arg(L, n+1, LJ_ERR_INVOPT);
- if (p[1] == 'n')
+ if (p[0] == '*') p++;
+ if (p[0] == 'n')
ok = io_file_readnum(L, fp);
- else if ((p[1] & ~0x20) == 'L')
- ok = io_file_readline(L, fp, (p[1] == 'l'));
- else if (p[1] == 'a')
+ else if ((p[0] & ~0x20) == 'L')
+ ok = io_file_readline(L, fp, (p[0] == 'l'));
+ else if (p[0] == 'a')
io_file_readall(L, fp);
else
lj_err_arg(L, n+1, LJ_ERR_INVFMT);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_jit.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_jit.c
index 402ca35c7eb..22ca0a1a249 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_jit.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_jit.c
@@ -1,6 +1,6 @@
/*
** JIT library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lib_jit_c
@@ -204,6 +204,7 @@ LJLIB_CF(jit_util_funcinfo)
lua_setfield(L, -2, "source");
lj_debug_pushloc(L, pt, pc);
lua_setfield(L, -2, "loc");
+ setprotoV(L, lj_tab_setstr(L, t, lj_str_newlit(L, "proto")), pt);
} else {
GCfunc *fn = funcV(L->base);
GCtab *t;
@@ -668,6 +669,11 @@ static uint32_t jit_cpudetect(lua_State *L)
if (fam >= 0x00000f00) /* K8, K10. */
flags |= JIT_F_PREFER_IMUL;
}
+ if (vendor[0] >= 7) {
+ uint32_t xfeatures[4];
+ lj_vm_cpuid(7, xfeatures);
+ flags |= ((xfeatures[1] >> 8)&1) * JIT_F_BMI2;
+ }
#endif
}
/* Check for required instruction set support on x86 (unnecessary on x64). */
@@ -710,15 +716,19 @@ static uint32_t jit_cpudetect(lua_State *L)
#if LJ_HASJIT
/* Compile-time MIPS CPU detection. */
#if LJ_ARCH_VERSION >= 20
- flags |= JIT_F_MIPS32R2;
+ flags |= JIT_F_MIPSXXR2;
#endif
/* Runtime MIPS CPU detection. */
#if defined(__GNUC__)
- if (!(flags & JIT_F_MIPS32R2)) {
+ if (!(flags & JIT_F_MIPSXXR2)) {
int x;
+#ifdef __mips16
+ x = 0; /* Runtime detection is difficult. Ensure optimal -march flags. */
+#else
/* On MIPS32R1 rotr is treated as srl. rotr r2,r2,1 -> srl r2,r2,1. */
__asm__("li $2, 1\n\t.long 0x00221042\n\tmove %0, $2" : "=r"(x) : : "$2");
- if (x) flags |= JIT_F_MIPS32R2; /* Either 0x80000000 (R2) or 0 (R1). */
+#endif
+ if (x) flags |= JIT_F_MIPSXXR2; /* Either 0x80000000 (R2) or 0 (R1). */
}
#endif
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_math.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_math.c
index f90dd331c44..ef9dda2da66 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_math.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_math.c
@@ -1,6 +1,6 @@
/*
** Math library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include <math.h>
@@ -221,10 +221,6 @@ LUALIB_API int luaopen_math(lua_State *L)
rs = (RandomState *)lua_newuserdata(L, sizeof(RandomState));
rs->valid = 0; /* Use lazy initialization to save some time on startup. */
LJ_LIB_REG(L, LUA_MATHLIBNAME, math);
-#if defined(LUA_COMPAT_MOD) && !LJ_52
- lua_getfield(L, -1, "fmod");
- lua_setfield(L, -2, "mod");
-#endif
return 1;
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_os.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_os.c
index 941c2a5db0b..9e78d49ac3c 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_os.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_os.c
@@ -1,6 +1,6 @@
/*
** OS library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_package.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_package.c
index ed56156e583..18adf09177f 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_package.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_package.c
@@ -1,6 +1,6 @@
/*
** Package library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2012 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -193,8 +193,7 @@ static void **ll_register(lua_State *L, const char *path)
lua_pop(L, 1);
plib = (void **)lua_newuserdata(L, sizeof(void *));
*plib = NULL;
- luaL_getmetatable(L, "_LOADLIB");
- lua_setmetatable(L, -2);
+ luaL_setmetatable(L, "_LOADLIB");
lua_pushfstring(L, "LOADLIB: %s", path);
lua_pushvalue(L, -2);
lua_settable(L, LUA_REGISTRYINDEX);
@@ -362,9 +361,9 @@ static int lj_cf_package_loader_c(lua_State *L)
return 1; /* library loaded successfully */
}
-#define LUA_POF "luaopen_"
-#define LUA_OFSEP "_"
-#define POF LUA_POF
+#define LUA_POF "luaopen_"
+#define LUA_OFSEP "_"
+#define POF LUA_POF
static const char *mkfuncname (lua_State *L, const char *modname) {
const char *funcname;
@@ -376,7 +375,6 @@ static const char *mkfuncname (lua_State *L, const char *modname) {
return funcname;
}
-
int loader_C_luatex (lua_State *L, const char *name, const char *filename) {
const char *funcname;
funcname = mkfuncname(L, name);
@@ -418,7 +416,6 @@ int loader_Call_luatex (lua_State *L, const char *name, const char *filename) {
return 1; /* library loaded successfully */
}
-
static int lj_cf_package_loader_preload(lua_State *L)
{
const char *name = luaL_checkstring(L, 1);
@@ -437,8 +434,7 @@ static int lj_cf_package_loader_preload(lua_State *L)
/* ------------------------------------------------------------------------ */
-static const int sentinel_ = 0;
-#define sentinel ((void *)&sentinel_)
+#define sentinel ((void *)0x4004)
static int lj_cf_package_require(lua_State *L)
{
@@ -528,29 +524,19 @@ static void modinit(lua_State *L, const char *modname)
static int lj_cf_package_module(lua_State *L)
{
const char *modname = luaL_checkstring(L, 1);
- int loaded = lua_gettop(L) + 1; /* index of _LOADED table */
- lua_getfield(L, LUA_REGISTRYINDEX, "_LOADED");
- lua_getfield(L, loaded, modname); /* get _LOADED[modname] */
- if (!lua_istable(L, -1)) { /* not found? */
- lua_pop(L, 1); /* remove previous result */
- /* try global variable (and create one if it does not exist) */
- if (luaL_findtable(L, LUA_GLOBALSINDEX, modname, 1) != NULL)
- lj_err_callerv(L, LJ_ERR_BADMODN, modname);
- lua_pushvalue(L, -1);
- lua_setfield(L, loaded, modname); /* _LOADED[modname] = new table */
- }
- /* check whether table already has a _NAME field */
+ int lastarg = (int)(L->top - L->base);
+ luaL_pushmodule(L, modname, 1);
lua_getfield(L, -1, "_NAME");
- if (!lua_isnil(L, -1)) { /* is table an initialized module? */
+ if (!lua_isnil(L, -1)) { /* Module already initialized? */
lua_pop(L, 1);
- } else { /* no; initialize it */
+ } else {
lua_pop(L, 1);
modinit(L, modname);
}
lua_pushvalue(L, -1);
setfenv(L);
- dooptions(L, loaded - 1);
- return 0;
+ dooptions(L, lastarg);
+ return LJ_52;
}
static int lj_cf_package_seeall(lua_State *L)
@@ -621,13 +607,16 @@ LUALIB_API int luaopen_package(lua_State *L)
lj_lib_pushcf(L, lj_cf_package_unloadlib, 1);
lua_setfield(L, -2, "__gc");
luaL_register(L, LUA_LOADLIBNAME, package_lib);
- lua_pushvalue(L, -1);
- lua_replace(L, LUA_ENVIRONINDEX);
+ lua_copy(L, -1, LUA_ENVIRONINDEX);
lua_createtable(L, sizeof(package_loaders)/sizeof(package_loaders[0])-1, 0);
for (i = 0; package_loaders[i] != NULL; i++) {
lj_lib_pushcf(L, package_loaders[i], 1);
lua_rawseti(L, -2, i+1);
}
+#if LJ_52
+ lua_pushvalue(L, -1);
+ lua_setfield(L, -3, "searchers");
+#endif
lua_setfield(L, -2, "loaders");
lua_getfield(L, LUA_REGISTRYINDEX, "LUA_NOENV");
noenv = lua_toboolean(L, -1);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_string.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_string.c
index 46b54a5e584..76b0730aca3 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_string.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_string.c
@@ -1,6 +1,6 @@
/*
** String library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -737,10 +737,6 @@ LUALIB_API int luaopen_string(lua_State *L)
GCtab *mt;
global_State *g;
LJ_LIB_REG(L, LUA_STRLIBNAME, string);
-#if defined(LUA_COMPAT_GFIND) && !LJ_52
- lua_getfield(L, -1, "gmatch");
- lua_setfield(L, -2, "gfind");
-#endif
mt = lj_tab_new(L, 0, 1);
/* NOBARRIER: basemt is a GC root. */
g = G(L);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lib_table.c b/Build/source/libs/luajit/LuaJIT-src/src/lib_table.c
index f9a3693dfcb..0450f1f684b 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lib_table.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lib_table.c
@@ -1,6 +1,6 @@
/*
** Table library.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -129,6 +129,26 @@ LJLIB_LUA(table_remove) /*
end
*/
+LJLIB_LUA(table_move) /*
+ function(a1, f, e, t, a2)
+ CHECK_tab(a1)
+ CHECK_int(f)
+ CHECK_int(e)
+ CHECK_int(t)
+ if a2 == nil then a2 = a1 end
+ CHECK_tab(a2)
+ if e >= f then
+ local d = t - f
+ if t > e or t <= f or a2 ~= a1 then
+ for i=f,e do a2[i+d] = a1[i] end
+ else
+ for i=e,f,-1 do a2[i+d] = a1[i] end
+ end
+ end
+ return a2
+ end
+*/
+
LJLIB_CF(table_concat) LJLIB_REC(.)
{
GCtab *t = lj_lib_checktab(L, 1);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj.supp b/Build/source/libs/luajit/LuaJIT-src/src/lj.supp
index acb9e789d08..217f7c8904f 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj.supp
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj.supp
@@ -27,15 +27,15 @@
{
Optimized string compare
Memcheck:Addr4
- fun:lj_str_fastcmp
+ fun:str_fastcmp
}
{
Optimized string compare
Memcheck:Addr1
- fun:lj_str_fastcmp
+ fun:str_fastcmp
}
{
Optimized string compare
Memcheck:Cond
- fun:lj_str_fastcmp
+ fun:str_fastcmp
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_alloc.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_alloc.c
index ddd50cae4f1..95d15d046ad 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_alloc.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_alloc.c
@@ -72,13 +72,56 @@
#define IS_DIRECT_BIT (SIZE_T_ONE)
+
+/* Determine system-specific block allocation method. */
#if LJ_TARGET_WINDOWS
#define WIN32_LEAN_AND_MEAN
#include <windows.h>
+#define LJ_ALLOC_VIRTUALALLOC 1
+
#if LJ_64 && !LJ_GC64
+#define LJ_ALLOC_NTAVM 1
+#endif
+
+#else
+
+#include <errno.h>
+/* If this include fails, then rebuild with: -DLUAJIT_USE_SYSMALLOC */
+#include <sys/mman.h>
+
+#define LJ_ALLOC_MMAP 1
+
+#if LJ_64
+
+#define LJ_ALLOC_MMAP_PROBE 1
+
+#if LJ_GC64
+#define LJ_ALLOC_MBITS 47 /* 128 TB in LJ_GC64 mode. */
+#elif LJ_TARGET_X64 && LJ_HASJIT
+/* Due to limitations in the x64 compiler backend. */
+#define LJ_ALLOC_MBITS 31 /* 2 GB on x64 with !LJ_GC64. */
+#else
+#define LJ_ALLOC_MBITS 32 /* 4 GB on other archs with !LJ_GC64. */
+#endif
+
+#endif
+
+#if LJ_64 && !LJ_GC64 && defined(MAP_32BIT)
+#define LJ_ALLOC_MMAP32 1
+#endif
+
+#if LJ_TARGET_LINUX
+#define LJ_ALLOC_MREMAP 1
+#endif
+
+#endif
+
+#if LJ_ALLOC_VIRTUALALLOC
+
+#if LJ_ALLOC_NTAVM
/* Undocumented, but hey, that's what we all love so much about Windows. */
typedef long (*PNTAVM)(HANDLE handle, void **addr, ULONG zbits,
size_t *size, ULONG alloctype, ULONG prot);
@@ -89,14 +132,15 @@ static PNTAVM ntavm;
*/
#define NTAVM_ZEROBITS 1
-static void INIT_MMAP(void)
+static void init_mmap(void)
{
ntavm = (PNTAVM)GetProcAddress(GetModuleHandleA("ntdll.dll"),
"NtAllocateVirtualMemory");
}
+#define INIT_MMAP() init_mmap()
/* Win64 32 bit MMAP via NtAllocateVirtualMemory. */
-static LJ_AINLINE void *CALL_MMAP(size_t size)
+static void *CALL_MMAP(size_t size)
{
DWORD olderr = GetLastError();
void *ptr = NULL;
@@ -107,7 +151,7 @@ static LJ_AINLINE void *CALL_MMAP(size_t size)
}
/* For direct MMAP, use MEM_TOP_DOWN to minimize interference */
-static LJ_AINLINE void *DIRECT_MMAP(size_t size)
+static void *DIRECT_MMAP(size_t size)
{
DWORD olderr = GetLastError();
void *ptr = NULL;
@@ -119,10 +163,8 @@ static LJ_AINLINE void *DIRECT_MMAP(size_t size)
#else
-#define INIT_MMAP() ((void)0)
-
/* Win32 MMAP via VirtualAlloc */
-static LJ_AINLINE void *CALL_MMAP(size_t size)
+static void *CALL_MMAP(size_t size)
{
DWORD olderr = GetLastError();
void *ptr = VirtualAlloc(0, size, MEM_RESERVE|MEM_COMMIT, PAGE_READWRITE);
@@ -131,7 +173,7 @@ static LJ_AINLINE void *CALL_MMAP(size_t size)
}
/* For direct MMAP, use MEM_TOP_DOWN to minimize interference */
-static LJ_AINLINE void *DIRECT_MMAP(size_t size)
+static void *DIRECT_MMAP(size_t size)
{
DWORD olderr = GetLastError();
void *ptr = VirtualAlloc(0, size, MEM_RESERVE|MEM_COMMIT|MEM_TOP_DOWN,
@@ -143,7 +185,7 @@ static LJ_AINLINE void *DIRECT_MMAP(size_t size)
#endif
/* This function supports releasing coalesed segments */
-static LJ_AINLINE int CALL_MUNMAP(void *ptr, size_t size)
+static int CALL_MUNMAP(void *ptr, size_t size)
{
DWORD olderr = GetLastError();
MEMORY_BASIC_INFORMATION minfo;
@@ -163,10 +205,7 @@ static LJ_AINLINE int CALL_MUNMAP(void *ptr, size_t size)
return 0;
}
-#else
-
-#include <errno.h>
-#include <sys/mman.h>
+#elif LJ_ALLOC_MMAP
#define MMAP_PROT (PROT_READ|PROT_WRITE)
#if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
@@ -174,107 +213,151 @@ static LJ_AINLINE int CALL_MUNMAP(void *ptr, size_t size)
#endif
#define MMAP_FLAGS (MAP_PRIVATE|MAP_ANONYMOUS)
-#if LJ_64 && !LJ_GC64
-/* 64 bit mode with 32 bit pointers needs special support for allocating
-** memory in the lower 2GB.
-*/
-
-#if defined(MAP_32BIT)
+#if LJ_ALLOC_MMAP_PROBE
-#if defined(__sun__)
-#define MMAP_REGION_START ((uintptr_t)0x1000)
+#ifdef MAP_TRYFIXED
+#define MMAP_FLAGS_PROBE (MMAP_FLAGS|MAP_TRYFIXED)
#else
-/* Actually this only gives us max. 1GB in current Linux kernels. */
-#define MMAP_REGION_START ((uintptr_t)0)
+#define MMAP_FLAGS_PROBE MMAP_FLAGS
#endif
-static LJ_AINLINE void *CALL_MMAP(size_t size)
-{
- int olderr = errno;
- void *ptr = mmap((void *)MMAP_REGION_START, size, MMAP_PROT, MAP_32BIT|MMAP_FLAGS, -1, 0);
- errno = olderr;
- return ptr;
-}
+#define LJ_ALLOC_MMAP_PROBE_MAX 30
+#define LJ_ALLOC_MMAP_PROBE_LINEAR 5
-#elif LJ_TARGET_OSX || LJ_TARGET_PS4 || defined(__FreeBSD__) || defined(__FreeBSD_kernel__) || defined(__NetBSD__) || defined(__OpenBSD__) || defined(__DragonFly__) || defined(__sun__) || defined(__CYGWIN__)
+#define LJ_ALLOC_MMAP_PROBE_LOWER ((uintptr_t)0x4000)
-/* OSX and FreeBSD mmap() use a naive first-fit linear search.
-** That's perfect for us. Except that -pagezero_size must be set for OSX,
-** otherwise the lower 4GB are blocked. And the 32GB RLIMIT_DATA needs
-** to be reduced to 250MB on FreeBSD.
+/* No point in a giant ifdef mess. Just try to open /dev/urandom.
+** It doesn't really matter if this fails, since we get some ASLR bits from
+** every unsuitable allocation, too. And we prefer linear allocation, anyway.
*/
-#if LJ_TARGET_OSX || defined(__DragonFly__)
-#define MMAP_REGION_START ((uintptr_t)0x10000)
-#elif LJ_TARGET_PS4
-#define MMAP_REGION_START ((uintptr_t)0x4000)
-#else
-#define MMAP_REGION_START ((uintptr_t)0x10000000)
-#endif
-#define MMAP_REGION_END ((uintptr_t)0x80000000)
+#include <fcntl.h>
+#include <unistd.h>
-#if (defined(__FreeBSD__) || defined(__FreeBSD_kernel__)) && !LJ_TARGET_PS4
-#include <sys/resource.h>
-#endif
+static uintptr_t mmap_probe_seed(void)
+{
+ uintptr_t val;
+ int fd = open("/dev/urandom", O_RDONLY);
+ if (fd != -1) {
+ int ok = ((size_t)read(fd, &val, sizeof(val)) == sizeof(val));
+ (void)close(fd);
+ if (ok) return val;
+ }
+ return 1; /* Punt. */
+}
-static LJ_AINLINE void *CALL_MMAP(size_t size)
+static void *mmap_probe(size_t size)
{
- int olderr = errno;
/* Hint for next allocation. Doesn't need to be thread-safe. */
- static uintptr_t alloc_hint = MMAP_REGION_START;
- int retry = 0;
-#if (defined(__FreeBSD__) || defined(__FreeBSD_kernel__)) && !LJ_TARGET_PS4
- static int rlimit_modified = 0;
- if (LJ_UNLIKELY(rlimit_modified == 0)) {
- struct rlimit rlim;
- rlim.rlim_cur = rlim.rlim_max = MMAP_REGION_START;
- setrlimit(RLIMIT_DATA, &rlim); /* Ignore result. May fail below. */
- rlimit_modified = 1;
- }
-#endif
- for (;;) {
- void *p = mmap((void *)alloc_hint, size, MMAP_PROT, MMAP_FLAGS, -1, 0);
- if ((uintptr_t)p >= MMAP_REGION_START &&
- (uintptr_t)p + size < MMAP_REGION_END) {
- alloc_hint = (uintptr_t)p + size;
+ static uintptr_t hint_addr = 0;
+ static uintptr_t hint_prng = 0;
+ int olderr = errno;
+ int retry;
+ for (retry = 0; retry < LJ_ALLOC_MMAP_PROBE_MAX; retry++) {
+ void *p = mmap((void *)hint_addr, size, MMAP_PROT, MMAP_FLAGS_PROBE, -1, 0);
+ uintptr_t addr = (uintptr_t)p;
+ if ((addr >> LJ_ALLOC_MBITS) == 0 && addr >= LJ_ALLOC_MMAP_PROBE_LOWER) {
+ /* We got a suitable address. Bump the hint address. */
+ hint_addr = addr + size;
errno = olderr;
return p;
}
- if (p != CMFAIL) munmap(p, size);
-#if defined(__sun__) || defined(__DragonFly__)
- alloc_hint += 0x1000000; /* Need near-exhaustive linear scan. */
- if (alloc_hint + size < MMAP_REGION_END) continue;
-#endif
- if (retry) break;
- retry = 1;
- alloc_hint = MMAP_REGION_START;
+ if (p != MFAIL) {
+ munmap(p, size);
+ } else if (errno == ENOMEM) {
+ return MFAIL;
+ }
+ if (hint_addr) {
+ /* First, try linear probing. */
+ if (retry < LJ_ALLOC_MMAP_PROBE_LINEAR) {
+ hint_addr += 0x1000000;
+ if (((hint_addr + size) >> LJ_ALLOC_MBITS) != 0)
+ hint_addr = 0;
+ continue;
+ } else if (retry == LJ_ALLOC_MMAP_PROBE_LINEAR) {
+ /* Next, try a no-hint probe to get back an ASLR address. */
+ hint_addr = 0;
+ continue;
+ }
+ }
+ /* Finally, try pseudo-random probing. */
+ if (LJ_UNLIKELY(hint_prng == 0)) {
+ hint_prng = mmap_probe_seed();
+ }
+ /* The unsuitable address we got has some ASLR PRNG bits. */
+ hint_addr ^= addr & ~((uintptr_t)(LJ_PAGESIZE-1));
+ do { /* The PRNG itself is very weak, but see above. */
+ hint_prng = hint_prng * 1103515245 + 12345;
+ hint_addr ^= hint_prng * (uintptr_t)LJ_PAGESIZE;
+ hint_addr &= (((uintptr_t)1 << LJ_ALLOC_MBITS)-1);
+ } while (hint_addr < LJ_ALLOC_MMAP_PROBE_LOWER);
}
errno = olderr;
- return CMFAIL;
+ return MFAIL;
}
+#endif
+
+#if LJ_ALLOC_MMAP32
+
+#if defined(__sun__)
+#define LJ_ALLOC_MMAP32_START ((uintptr_t)0x1000)
#else
+#define LJ_ALLOC_MMAP32_START ((uintptr_t)0)
+#endif
-#error "NYI: need an equivalent of MAP_32BIT for this 64 bit OS"
+static void *mmap_map32(size_t size)
+{
+#if LJ_ALLOC_MMAP_PROBE
+ static int fallback = 0;
+ if (fallback)
+ return mmap_probe(size);
+#endif
+ {
+ int olderr = errno;
+ void *ptr = mmap((void *)LJ_ALLOC_MMAP32_START, size, MMAP_PROT, MAP_32BIT|MMAP_FLAGS, -1, 0);
+ errno = olderr;
+ /* This only allows 1GB on Linux. So fallback to probing to get 2GB. */
+#if LJ_ALLOC_MMAP_PROBE
+ if (ptr == MFAIL) {
+ fallback = 1;
+ return mmap_probe(size);
+ }
+#endif
+ return ptr;
+ }
+}
#endif
+#if LJ_ALLOC_MMAP32
+#define CALL_MMAP(size) mmap_map32(size)
+#elif LJ_ALLOC_MMAP_PROBE
+#define CALL_MMAP(size) mmap_probe(size)
#else
-
-/* 32 bit mode and GC64 mode is easy. */
-static LJ_AINLINE void *CALL_MMAP(size_t size)
+static void *CALL_MMAP(size_t size)
{
int olderr = errno;
void *ptr = mmap(NULL, size, MMAP_PROT, MMAP_FLAGS, -1, 0);
errno = olderr;
return ptr;
}
-
#endif
-#define INIT_MMAP() ((void)0)
-#define DIRECT_MMAP(s) CALL_MMAP(s)
+#if (defined(__FreeBSD__) || defined(__FreeBSD_kernel__)) && !LJ_TARGET_PS4
+
+#include <sys/resource.h>
+
+static void init_mmap(void)
+{
+ struct rlimit rlim;
+ rlim.rlim_cur = rlim.rlim_max = 0x10000;
+ setrlimit(RLIMIT_DATA, &rlim); /* Ignore result. May fail later. */
+}
+#define INIT_MMAP() init_mmap()
+
+#endif
-static LJ_AINLINE int CALL_MUNMAP(void *ptr, size_t size)
+static int CALL_MUNMAP(void *ptr, size_t size)
{
int olderr = errno;
int ret = munmap(ptr, size);
@@ -282,10 +365,9 @@ static LJ_AINLINE int CALL_MUNMAP(void *ptr, size_t size)
return ret;
}
-#if LJ_TARGET_LINUX
+#if LJ_ALLOC_MREMAP
/* Need to define _GNU_SOURCE to get the mremap prototype. */
-static LJ_AINLINE void *CALL_MREMAP_(void *ptr, size_t osz, size_t nsz,
- int flags)
+static void *CALL_MREMAP_(void *ptr, size_t osz, size_t nsz, int flags)
{
int olderr = errno;
ptr = mremap(ptr, osz, nsz, flags);
@@ -305,6 +387,15 @@ static LJ_AINLINE void *CALL_MREMAP_(void *ptr, size_t osz, size_t nsz,
#endif
+
+#ifndef INIT_MMAP
+#define INIT_MMAP() ((void)0)
+#endif
+
+#ifndef DIRECT_MMAP
+#define DIRECT_MMAP(s) CALL_MMAP(s)
+#endif
+
#ifndef CALL_MREMAP
#define CALL_MREMAP(addr, osz, nsz, mv) ((void)osz, MFAIL)
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_api.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_api.c
index 0fbb9256343..d17a5754d40 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_api.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_api.c
@@ -1,6 +1,6 @@
/*
** Public Lua/C API.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -112,6 +112,13 @@ LUA_API void lua_xmove(lua_State *from, lua_State *to, int n)
from->top = f;
}
+LUA_API const lua_Number *lua_version(lua_State *L)
+{
+ static const lua_Number version = LUA_VERSION_NUM;
+ UNUSED(L);
+ return &version;
+}
+
/* -- Stack manipulation -------------------------------------------------- */
LUA_API int lua_gettop(lua_State *L)
@@ -152,30 +159,40 @@ LUA_API void lua_insert(lua_State *L, int idx)
copyTV(L, p, L->top);
}
-LUA_API void lua_replace(lua_State *L, int idx)
+static void copy_slot(lua_State *L, TValue *f, int idx)
{
- api_checknelems(L, 1);
if (idx == LUA_GLOBALSINDEX) {
- api_check(L, tvistab(L->top-1));
+ api_check(L, tvistab(f));
/* NOBARRIER: A thread (i.e. L) is never black. */
- setgcref(L->env, obj2gco(tabV(L->top-1)));
+ setgcref(L->env, obj2gco(tabV(f)));
} else if (idx == LUA_ENVIRONINDEX) {
GCfunc *fn = curr_func(L);
if (fn->c.gct != ~LJ_TFUNC)
lj_err_msg(L, LJ_ERR_NOENV);
- api_check(L, tvistab(L->top-1));
- setgcref(fn->c.env, obj2gco(tabV(L->top-1)));
- lj_gc_barrier(L, fn, L->top-1);
+ api_check(L, tvistab(f));
+ setgcref(fn->c.env, obj2gco(tabV(f)));
+ lj_gc_barrier(L, fn, f);
} else {
TValue *o = index2adr(L, idx);
api_checkvalidindex(L, o);
- copyTV(L, o, L->top-1);
+ copyTV(L, o, f);
if (idx < LUA_GLOBALSINDEX) /* Need a barrier for upvalues. */
- lj_gc_barrier(L, curr_func(L), L->top-1);
+ lj_gc_barrier(L, curr_func(L), f);
}
+}
+
+LUA_API void lua_replace(lua_State *L, int idx)
+{
+ api_checknelems(L, 1);
+ copy_slot(L, L->top - 1, idx);
L->top--;
}
+LUA_API void lua_copy(lua_State *L, int fromidx, int toidx)
+{
+ copy_slot(L, index2adr(L, fromidx), toidx);
+}
+
LUA_API void lua_pushvalue(lua_State *L, int idx)
{
copyTV(L, L->top, index2adr(L, idx));
@@ -325,6 +342,22 @@ LUA_API lua_Number lua_tonumber(lua_State *L, int idx)
return 0;
}
+LUA_API lua_Number lua_tonumberx(lua_State *L, int idx, int *ok)
+{
+ cTValue *o = index2adr(L, idx);
+ TValue tmp;
+ if (LJ_LIKELY(tvisnumber(o))) {
+ if (ok) *ok = 1;
+ return numberVnum(o);
+ } else if (tvisstr(o) && lj_strscan_num(strV(o), &tmp)) {
+ if (ok) *ok = 1;
+ return numV(&tmp);
+ } else {
+ if (ok) *ok = 0;
+ return 0;
+ }
+}
+
LUALIB_API lua_Number luaL_checknumber(lua_State *L, int idx)
{
cTValue *o = index2adr(L, idx);
@@ -362,7 +395,7 @@ LUA_API lua_Integer lua_tointeger(lua_State *L, int idx)
if (!(tvisstr(o) && lj_strscan_number(strV(o), &tmp)))
return 0;
if (tvisint(&tmp))
- return (lua_Integer)intV(&tmp);
+ return intV(&tmp);
n = numV(&tmp);
}
#if LJ_64
@@ -372,6 +405,35 @@ LUA_API lua_Integer lua_tointeger(lua_State *L, int idx)
#endif
}
+LUA_API lua_Integer lua_tointegerx(lua_State *L, int idx, int *ok)
+{
+ cTValue *o = index2adr(L, idx);
+ TValue tmp;
+ lua_Number n;
+ if (LJ_LIKELY(tvisint(o))) {
+ if (ok) *ok = 1;
+ return intV(o);
+ } else if (LJ_LIKELY(tvisnum(o))) {
+ n = numV(o);
+ } else {
+ if (!(tvisstr(o) && lj_strscan_number(strV(o), &tmp))) {
+ if (ok) *ok = 0;
+ return 0;
+ }
+ if (tvisint(&tmp)) {
+ if (ok) *ok = 1;
+ return intV(&tmp);
+ }
+ n = numV(&tmp);
+ }
+ if (ok) *ok = 1;
+#if LJ_64
+ return (lua_Integer)n;
+#else
+ return lj_num2int(n);
+#endif
+}
+
LUALIB_API lua_Integer luaL_checkinteger(lua_State *L, int idx)
{
cTValue *o = index2adr(L, idx);
@@ -858,7 +920,7 @@ LUA_API void lua_upvaluejoin(lua_State *L, int idx1, int n1, int idx2, int n2)
lj_gc_objbarrier(L, fn1, gcref(fn1->l.uvptr[n1]));
}
-LUALIB_API void *luaL_checkudata(lua_State *L, int idx, const char *tname)
+LUALIB_API void *luaL_testudata(lua_State *L, int idx, const char *tname)
{
cTValue *o = index2adr(L, idx);
if (tvisudata(o)) {
@@ -867,8 +929,14 @@ LUALIB_API void *luaL_checkudata(lua_State *L, int idx, const char *tname)
if (tv && tvistab(tv) && tabV(tv) == tabref(ud->metatable))
return uddata(ud);
}
- lj_err_argtype(L, idx, tname);
- return NULL; /* unreachable */
+ return NULL; /* value is not a userdata with a metatable */
+}
+
+LUALIB_API void *luaL_checkudata(lua_State *L, int idx, const char *tname)
+{
+ void *p = luaL_testudata(L, idx, tname);
+ if (!p) lj_err_argtype(L, idx, tname);
+ return p;
}
/* -- Object setters ------------------------------------------------------ */
@@ -977,6 +1045,12 @@ LUA_API int lua_setmetatable(lua_State *L, int idx)
return 1;
}
+LUALIB_API void luaL_setmetatable(lua_State *L, const char *tname)
+{
+ lua_getfield(L, LUA_REGISTRYINDEX, tname);
+ lua_setmetatable(L, -2);
+}
+
LUA_API int lua_setfenv(lua_State *L, int idx)
{
cTValue *o = index2adr(L, idx);
@@ -1032,7 +1106,7 @@ static TValue *api_call_base(lua_State *L, int nargs)
LUA_API void lua_call(lua_State *L, int nargs, int nresults)
{
- api_check(L, L->status == 0 || L->status == LUA_ERRERR);
+ api_check(L, L->status == LUA_OK || L->status == LUA_ERRERR);
api_checknelems(L, nargs+1);
lj_vm_call(L, api_call_base(L, nargs), nresults+1);
}
@@ -1043,7 +1117,7 @@ LUA_API int lua_pcall(lua_State *L, int nargs, int nresults, int errfunc)
uint8_t oldh = hook_save(g);
ptrdiff_t ef;
int status;
- api_check(L, L->status == 0 || L->status == LUA_ERRERR);
+ api_check(L, L->status == LUA_OK || L->status == LUA_ERRERR);
api_checknelems(L, nargs+1);
if (errfunc == 0) {
ef = 0;
@@ -1075,7 +1149,7 @@ LUA_API int lua_cpcall(lua_State *L, lua_CFunction func, void *ud)
global_State *g = G(L);
uint8_t oldh = hook_save(g);
int status;
- api_check(L, L->status == 0 || L->status == LUA_ERRERR);
+ api_check(L, L->status == LUA_OK || L->status == LUA_ERRERR);
status = lj_vm_cpcall(L, func, ud, cpcall);
if (status) hook_restore(g, oldh);
return status;
@@ -1096,6 +1170,11 @@ LUALIB_API int luaL_callmeta(lua_State *L, int idx, const char *field)
/* -- Coroutine yield and resume ------------------------------------------ */
+LUA_API int lua_isyieldable(lua_State *L)
+{
+ return cframe_canyield(L->cframe);
+}
+
LUA_API int lua_yield(lua_State *L, int nresults)
{
void *cf = L->cframe;
@@ -1140,7 +1219,7 @@ LUA_API int lua_resume(lua_State *L, int nargs)
{
if (L->cframe == NULL && L->status <= LUA_YIELD)
return lj_vm_resume(L,
- L->status == 0 ? api_call_base(L, nargs) : L->top - nargs,
+ L->status == LUA_OK ? api_call_base(L, nargs) : L->top - nargs,
0, 0);
L->top = L->base;
setstrV(L, L->top, lj_err_str(L, LJ_ERR_COSUSP));
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_arch.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_arch.h
index d95848fe5cf..e8ad844ff1d 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_arch.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_arch.h
@@ -1,6 +1,6 @@
/*
** Target architecture selection.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_ARCH_H
@@ -25,6 +25,10 @@
#define LUAJIT_ARCH_ppc 5
#define LUAJIT_ARCH_MIPS 6
#define LUAJIT_ARCH_mips 6
+#define LUAJIT_ARCH_MIPS32 6
+#define LUAJIT_ARCH_mips32 6
+#define LUAJIT_ARCH_MIPS64 7
+#define LUAJIT_ARCH_mips64 7
/* Target OS. */
#define LUAJIT_OS_OTHER 0
@@ -47,8 +51,10 @@
#define LUAJIT_TARGET LUAJIT_ARCH_ARM64
#elif defined(__ppc__) || defined(__ppc) || defined(__PPC__) || defined(__PPC) || defined(__powerpc__) || defined(__powerpc) || defined(__POWERPC__) || defined(__POWERPC) || defined(_M_PPC)
#define LUAJIT_TARGET LUAJIT_ARCH_PPC
+#elif defined(__mips64__) || defined(__mips64) || defined(__MIPS64__) || defined(__MIPS64)
+#define LUAJIT_TARGET LUAJIT_ARCH_MIPS64
#elif defined(__mips__) || defined(__mips) || defined(__MIPS__) || defined(__MIPS)
-#define LUAJIT_TARGET LUAJIT_ARCH_MIPS
+#define LUAJIT_TARGET LUAJIT_ARCH_MIPS32
#else
#error "No support for this architecture (yet)"
#endif
@@ -68,7 +74,10 @@
defined(__NetBSD__) || defined(__OpenBSD__) || \
defined(__DragonFly__)) && !defined(__ORBIS__)
#define LUAJIT_OS LUAJIT_OS_BSD
-#elif (defined(__sun__) && defined(__svr4__)) || defined(__CYGWIN__)
+#elif (defined(__sun__) && defined(__svr4__)) || defined(__HAIKU__)
+#define LUAJIT_OS LUAJIT_OS_POSIX
+#elif defined(__CYGWIN__)
+#define LJ_TARGET_CYGWIN 1
#define LUAJIT_OS LUAJIT_OS_POSIX
#else
#define LUAJIT_OS LUAJIT_OS_OTHER
@@ -137,7 +146,7 @@
#define LJ_ARCH_NAME "x86"
#define LJ_ARCH_BITS 32
#define LJ_ARCH_ENDIAN LUAJIT_LE
-#if LJ_TARGET_WINDOWS || __CYGWIN__
+#if LJ_TARGET_WINDOWS || LJ_TARGET_CYGWIN
#define LJ_ABI_WIN 1
#else
#define LJ_ABI_WIN 0
@@ -155,7 +164,7 @@
#define LJ_ARCH_NAME "x64"
#define LJ_ARCH_BITS 64
#define LJ_ARCH_ENDIAN LUAJIT_LE
-#if LJ_TARGET_WINDOWS || __CYGWIN__
+#if LJ_TARGET_WINDOWS || LJ_TARGET_CYGWIN
#define LJ_ABI_WIN 1
#else
#define LJ_ABI_WIN 0
@@ -206,9 +215,14 @@
#elif LUAJIT_TARGET == LUAJIT_ARCH_ARM64
-#define LJ_ARCH_NAME "arm64"
#define LJ_ARCH_BITS 64
+#if defined(__AARCH64EB__)
+#define LJ_ARCH_NAME "arm64be"
+#define LJ_ARCH_ENDIAN LUAJIT_BE
+#else
+#define LJ_ARCH_NAME "arm64"
#define LJ_ARCH_ENDIAN LUAJIT_LE
+#endif
#define LJ_TARGET_ARM64 1
#define LJ_TARGET_EHRETREG 0
#define LJ_TARGET_JUMPRANGE 27 /* +-2^27 = +-128MB */
@@ -217,7 +231,6 @@
#define LJ_TARGET_UNIFYROT 2 /* Want only IR_BROR. */
#define LJ_TARGET_GC64 1
#define LJ_ARCH_NUMMODE LJ_NUMMODE_DUAL
-#define LJ_ARCH_NOJIT 1 /* NYI */
#define LJ_ARCH_VERSION 80
@@ -286,13 +299,21 @@
#define LJ_ARCH_XENON 1
#endif
-#elif LUAJIT_TARGET == LUAJIT_ARCH_MIPS
+#elif LUAJIT_TARGET == LUAJIT_ARCH_MIPS32 || LUAJIT_TARGET == LUAJIT_ARCH_MIPS64
#if defined(__MIPSEL__) || defined(__MIPSEL) || defined(_MIPSEL)
+#if LUAJIT_TARGET == LUAJIT_ARCH_MIPS32
#define LJ_ARCH_NAME "mipsel"
+#else
+#define LJ_ARCH_NAME "mips64el"
+#endif
#define LJ_ARCH_ENDIAN LUAJIT_LE
#else
+#if LUAJIT_TARGET == LUAJIT_ARCH_MIPS32
#define LJ_ARCH_NAME "mips"
+#else
+#define LJ_ARCH_NAME "mips64"
+#endif
#define LJ_ARCH_ENDIAN LUAJIT_BE
#endif
@@ -304,11 +325,6 @@
#endif
#endif
-/* Temporarily disable features until the code has been merged. */
-#if !defined(LUAJIT_NO_UNWIND) && __GNU_COMPACT_EH__
-#define LUAJIT_NO_UNWIND 1
-#endif
-
#if !defined(LJ_ABI_SOFTFP)
#ifdef __mips_soft_float
#define LJ_ABI_SOFTFP 1
@@ -317,7 +333,17 @@
#endif
#endif
+#if LUAJIT_TARGET == LUAJIT_ARCH_MIPS32
#define LJ_ARCH_BITS 32
+#define LJ_TARGET_MIPS32 1
+#else
+#if LJ_ABI_SOFTFP || !LJ_ARCH_HASFPU
+#define LJ_ARCH_NOJIT 1 /* NYI */
+#endif
+#define LJ_ARCH_BITS 64
+#define LJ_TARGET_MIPS64 1
+#define LJ_TARGET_GC64 1
+#endif
#define LJ_TARGET_MIPS 1
#define LJ_TARGET_EHRETREG 4
#define LJ_TARGET_JUMPRANGE 27 /* 2*2^27 = 256MB-aligned region */
@@ -326,7 +352,7 @@
#define LJ_TARGET_UNIFYROT 2 /* Want only IR_BROR. */
#define LJ_ARCH_NUMMODE LJ_NUMMODE_DUAL
-#if _MIPS_ARCH_MIPS32R2
+#if _MIPS_ARCH_MIPS32R2 || _MIPS_ARCH_MIPS64R2
#define LJ_ARCH_VERSION 20
#else
#define LJ_ARCH_VERSION 10
@@ -350,13 +376,13 @@
#if __GNUC__ < 4
#error "Need at least GCC 4.0 or newer"
#endif
-#elif LJ_TARGET_ARM || LJ_TARGET_PPC
+#elif LJ_TARGET_ARM || LJ_TARGET_PPC
#if (__GNUC__ < 4) || ((__GNUC__ == 4) && __GNUC_MINOR__ < 2)
#error "Need at least GCC 4.2 or newer"
#endif
#elif LJ_TARGET_ARM64
#if __clang__
-#if (__clang_major__ < 3) || ((__clang_major__ == 3) && __clang_minor__ < 5)
+#if ((__clang_major__ < 3) || ((__clang_major__ == 3) && __clang_minor__ < 5)) && !defined(__NX_TOOLCHAIN_MAJOR__)
#error "Need at least Clang 3.5 or newer"
#endif
#else
@@ -388,9 +414,6 @@
#error "Only ARM EABI or iOS 3.0+ ABI is supported"
#endif
#elif LJ_TARGET_ARM64
-#if defined(__AARCH64EB__)
-#error "No support for big-endian ARM64"
-#endif
#if defined(_ILP32)
#error "No support for ILP32 model on ARM64"
#endif
@@ -407,9 +430,13 @@
#ifdef __NO_FPRS__
#error "No support for PPC/e500 anymore (use LuaJIT 2.0)"
#endif
-#elif LJ_TARGET_MIPS
-#if defined(_LP64)
-#error "No support for MIPS64"
+#elif LJ_TARGET_MIPS32
+#if !((defined(_MIPS_SIM_ABI32) && _MIPS_SIM == _MIPS_SIM_ABI32) || (defined(_ABIO32) && _MIPS_SIM == _ABIO32))
+#error "Only o32 ABI supported for MIPS32"
+#endif
+#elif LJ_TARGET_MIPS64
+#if !((defined(_MIPS_SIM_ABI64) && _MIPS_SIM == _MIPS_SIM_ABI64) || (defined(_ABI64) && _MIPS_SIM == _ABI64))
+#error "Only n64 ABI supported for MIPS64"
#endif
#endif
#endif
@@ -450,7 +477,7 @@
#endif
/* Disable or enable the JIT compiler. */
-#if defined(LUAJIT_DISABLE_JIT) || defined(LJ_ARCH_NOJIT) || defined(LJ_OS_NOJIT) || LJ_FR2 || LJ_GC64
+#if defined(LUAJIT_DISABLE_JIT) || defined(LJ_ARCH_NOJIT) || defined(LJ_OS_NOJIT)
#define LJ_HASJIT 0
#else
#define LJ_HASJIT 1
@@ -521,6 +548,11 @@
#define LJ_NO_SYSTEM 1
#endif
+#if !defined(LUAJIT_NO_UNWIND) && __GNU_COMPACT_EH__
+/* NYI: no support for compact unwind specification, yet. */
+#define LUAJIT_NO_UNWIND 1
+#endif
+
#if defined(LUAJIT_NO_UNWIND) || defined(__symbian__) || LJ_TARGET_IOS || LJ_TARGET_PS3 || LJ_TARGET_PS4
#define LJ_NO_UNWIND 1
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm.c
index 93f6bcd6c96..c2cf5a95a5b 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm.c
@@ -1,6 +1,6 @@
/*
** IR assembler (SSA IR -> machine code).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_asm_c
@@ -91,7 +91,7 @@ typedef struct ASMState {
MCode *realign; /* Realign loop if not NULL. */
#ifdef RID_NUM_KREF
- int32_t krefk[RID_NUM_KREF];
+ intptr_t krefk[RID_NUM_KREF];
#endif
IRRef1 phireg[RID_MAX]; /* PHI register references. */
uint16_t parentmap[LJ_MAX_JSLOTS]; /* Parent instruction to RegSP map. */
@@ -144,7 +144,7 @@ static LJ_AINLINE void checkmclim(ASMState *as)
#define ra_krefreg(ref) ((Reg)(RID_MIN_KREF + (Reg)(ref)))
#define ra_krefk(as, ref) (as->krefk[(ref)])
-static LJ_AINLINE void ra_setkref(ASMState *as, Reg r, int32_t k)
+static LJ_AINLINE void ra_setkref(ASMState *as, Reg r, intptr_t k)
{
IRRef ref = (IRRef)(r - RID_MIN_KREF);
as->krefk[ref] = k;
@@ -171,6 +171,8 @@ IRFLDEF(FLOFS)
#include "lj_emit_x86.h"
#elif LJ_TARGET_ARM
#include "lj_emit_arm.h"
+#elif LJ_TARGET_ARM64
+#include "lj_emit_arm64.h"
#elif LJ_TARGET_PPC
#include "lj_emit_ppc.h"
#elif LJ_TARGET_MIPS
@@ -322,7 +324,11 @@ static Reg ra_rematk(ASMState *as, IRRef ref)
lua_assert(!rset_test(as->freeset, r));
ra_free(as, r);
ra_modified(as, r);
+#if LJ_64
+ emit_loadu64(as, r, ra_krefk(as, ref));
+#else
emit_loadi(as, r, ra_krefk(as, ref));
+#endif
return r;
}
ir = IR(ref);
@@ -334,7 +340,7 @@ static Reg ra_rematk(ASMState *as, IRRef ref)
RA_DBGX((as, "remat $i $r", ir, r));
#if !LJ_SOFTFP
if (ir->o == IR_KNUM) {
- emit_loadn(as, r, ir_knum(ir));
+ emit_loadk64(as, r, ir);
} else
#endif
if (emit_canremat(REF_BASE) && ir->o == IR_BASE) {
@@ -346,6 +352,12 @@ static Reg ra_rematk(ASMState *as, IRRef ref)
#if LJ_64
} else if (ir->o == IR_KINT64) {
emit_loadu64(as, r, ir_kint64(ir)->u64);
+#if LJ_GC64
+ } else if (ir->o == IR_KGC) {
+ emit_loadu64(as, r, (uintptr_t)ir_kgc(ir));
+ } else if (ir->o == IR_KPTR || ir->o == IR_KKPTR) {
+ emit_loadu64(as, r, (uintptr_t)ir_kptr(ir));
+#endif
#endif
} else {
lua_assert(ir->o == IR_KINT || ir->o == IR_KGC ||
@@ -518,7 +530,7 @@ static void ra_evictk(ASMState *as)
#ifdef RID_NUM_KREF
/* Allocate a register for a constant. */
-static Reg ra_allock(ASMState *as, int32_t k, RegSet allow)
+static Reg ra_allock(ASMState *as, intptr_t k, RegSet allow)
{
/* First try to find a register which already holds the same constant. */
RegSet pick, work = ~as->freeset & RSET_GPR;
@@ -527,9 +539,31 @@ static Reg ra_allock(ASMState *as, int32_t k, RegSet allow)
IRRef ref;
r = rset_pickbot(work);
ref = regcost_ref(as->cost[r]);
+#if LJ_64
+ if (ref < ASMREF_L) {
+ if (ra_iskref(ref)) {
+ if (k == ra_krefk(as, ref))
+ return r;
+ } else {
+ IRIns *ir = IR(ref);
+ if ((ir->o == IR_KINT64 && k == (int64_t)ir_kint64(ir)->u64) ||
+#if LJ_GC64
+ (ir->o == IR_KINT && k == ir->i) ||
+ (ir->o == IR_KGC && k == (intptr_t)ir_kgc(ir)) ||
+ ((ir->o == IR_KPTR || ir->o == IR_KKPTR) &&
+ k == (intptr_t)ir_kptr(ir))
+#else
+ (ir->o != IR_KINT64 && k == ir->i)
+#endif
+ )
+ return r;
+ }
+ }
+#else
if (ref < ASMREF_L &&
k == (ra_iskref(ref) ? ra_krefk(as, ref) : IR(ref)->i))
return r;
+#endif
rset_clear(work, r);
}
pick = as->freeset & allow;
@@ -549,7 +583,7 @@ static Reg ra_allock(ASMState *as, int32_t k, RegSet allow)
}
/* Allocate a specific register for a constant. */
-static void ra_allockreg(ASMState *as, int32_t k, Reg r)
+static void ra_allockreg(ASMState *as, intptr_t k, Reg r)
{
Reg kr = ra_allock(as, k, RID2RSET(r));
if (kr != r) {
@@ -619,10 +653,20 @@ static Reg ra_alloc1(ASMState *as, IRRef ref, RegSet allow)
return r;
}
+/* Add a register rename to the IR. */
+static void ra_addrename(ASMState *as, Reg down, IRRef ref, SnapNo snapno)
+{
+ IRRef ren;
+ lj_ir_set(as->J, IRT(IR_RENAME, IRT_NIL), ref, snapno);
+ ren = tref_ref(lj_ir_emit(as->J));
+ as->J->cur.ir[ren].r = (uint8_t)down;
+ as->J->cur.ir[ren].s = SPS_NONE;
+}
+
/* Rename register allocation and emit move. */
static void ra_rename(ASMState *as, Reg down, Reg up)
{
- IRRef ren, ref = regcost_ref(as->cost[up] = as->cost[down]);
+ IRRef ref = regcost_ref(as->cost[up] = as->cost[down]);
IRIns *ir = IR(ref);
ir->r = (uint8_t)up;
as->cost[down] = 0;
@@ -635,11 +679,7 @@ static void ra_rename(ASMState *as, Reg down, Reg up)
RA_DBGX((as, "rename $f $r $r", regcost_ref(as->cost[up]), down, up));
emit_movrr(as, ir, down, up); /* Backwards codegen needs inverse move. */
if (!ra_hasspill(IR(ref)->s)) { /* Add the rename to the IR. */
- lj_ir_set(as->J, IRT(IR_RENAME, IRT_NIL), ref, as->snapno);
- ren = tref_ref(lj_ir_emit(as->J));
- as->ir = as->T->ir; /* The IR may have been reallocated. */
- IR(ren)->r = (uint8_t)down;
- IR(ren)->s = SPS_NONE;
+ ra_addrename(as, down, ref, as->snapno);
}
}
@@ -689,17 +729,21 @@ static void ra_left(ASMState *as, Reg dest, IRRef lref)
if (ra_noreg(left)) {
if (irref_isk(lref)) {
if (ir->o == IR_KNUM) {
- cTValue *tv = ir_knum(ir);
/* FP remat needs a load except for +0. Still better than eviction. */
- if (tvispzero(tv) || !(as->freeset & RSET_FPR)) {
- emit_loadn(as, dest, tv);
+ if (tvispzero(ir_knum(ir)) || !(as->freeset & RSET_FPR)) {
+ emit_loadk64(as, dest, ir);
return;
}
#if LJ_64
} else if (ir->o == IR_KINT64) {
- emit_loadu64(as, dest, ir_kint64(ir)->u64);
+ emit_loadk64(as, dest, ir);
+ return;
+#if LJ_GC64
+ } else if (ir->o == IR_KGC || ir->o == IR_KPTR || ir->o == IR_KKPTR) {
+ emit_loadk64(as, dest, ir);
return;
#endif
+#endif
} else if (ir->o != IR_KPRI) {
lua_assert(ir->o == IR_KINT || ir->o == IR_KGC ||
ir->o == IR_KPTR || ir->o == IR_KKPTR || ir->o == IR_KNULL);
@@ -941,7 +985,7 @@ static void asm_snap_prep(ASMState *as)
} else {
/* Process any renames above the highwater mark. */
for (; as->snaprename < as->T->nins; as->snaprename++) {
- IRIns *ir = IR(as->snaprename);
+ IRIns *ir = &as->T->ir[as->snaprename];
if (asm_snap_checkrename(as, ir->op1))
ir->op2 = REF_BIAS-1; /* Kill rename. */
}
@@ -1055,7 +1099,7 @@ static void asm_bufhdr(ASMState *as, IRIns *ir)
}
} else {
Reg tmp = ra_scratch(as, rset_exclude(RSET_GPR, sb));
- /* Passing ir isn't strictly correct, but it's an IRT_P32, too. */
+ /* Passing ir isn't strictly correct, but it's an IRT_PGC, too. */
emit_storeofs(as, ir, tmp, sb, offsetof(SBuf, p));
emit_loadofs(as, ir, tmp, sb, offsetof(SBuf, b));
}
@@ -1472,12 +1516,7 @@ static void asm_phi_fixup(ASMState *as)
irt_clearmark(ir->t);
/* Left PHI gained a spill slot before the loop? */
if (ra_hasspill(ir->s)) {
- IRRef ren;
- lj_ir_set(as->J, IRT(IR_RENAME, IRT_NIL), lref, as->loopsnapno);
- ren = tref_ref(lj_ir_emit(as->J));
- as->ir = as->T->ir; /* The IR may have been reallocated. */
- IR(ren)->r = (uint8_t)r;
- IR(ren)->s = SPS_NONE;
+ ra_addrename(as, r, lref, as->loopsnapno);
}
}
rset_clear(work, r);
@@ -1552,6 +1591,8 @@ static void asm_loop(ASMState *as)
#include "lj_asm_x86.h"
#elif LJ_TARGET_ARM
#include "lj_asm_arm.h"
+#elif LJ_TARGET_ARM64
+#include "lj_asm_arm64.h"
#elif LJ_TARGET_PPC
#include "lj_asm_ppc.h"
#elif LJ_TARGET_MIPS
@@ -1888,7 +1929,7 @@ static BCReg asm_baseslot(ASMState *as, SnapShot *snap, int *gotframe)
SnapEntry sn = map[n-1];
if ((sn & SNAP_FRAME)) {
*gotframe = 1;
- return snap_slot(sn);
+ return snap_slot(sn) - LJ_FR2;
}
}
return 0;
@@ -1908,16 +1949,20 @@ static void asm_tail_link(ASMState *as)
if (as->T->link == 0) {
/* Setup fixed registers for exit to interpreter. */
- const BCIns *pc = snap_pc(as->T->snapmap[snap->mapofs + snap->nent]);
+ const BCIns *pc = snap_pc(&as->T->snapmap[snap->mapofs + snap->nent]);
int32_t mres;
if (bc_op(*pc) == BC_JLOOP) { /* NYI: find a better way to do this. */
BCIns *retpc = &traceref(as->J, bc_d(*pc))->startins;
if (bc_isret(bc_op(*retpc)))
pc = retpc;
}
+#if LJ_GC64
+ emit_loadu64(as, RID_LPC, u64ptr(pc));
+#else
ra_allockreg(as, i32ptr(J2GG(as->J)->dispatch), RID_DISPATCH);
ra_allockreg(as, i32ptr(pc), RID_LPC);
- mres = (int32_t)(snap->nslots - baseslot);
+#endif
+ mres = (int32_t)(snap->nslots - baseslot - LJ_FR2);
switch (bc_op(*pc)) {
case BC_CALLM: case BC_CALLMT:
mres -= (int32_t)(1 + LJ_FR2 + bc_a(*pc) + bc_c(*pc)); break;
@@ -1932,6 +1977,11 @@ static void asm_tail_link(ASMState *as)
}
emit_addptr(as, RID_BASE, 8*(int32_t)baseslot);
+ if (as->J->ktrace) { /* Patch ktrace slot with the final GCtrace pointer. */
+ setgcref(IR(as->J->ktrace)[LJ_GC64].gcr, obj2gco(as->J->curfinal));
+ IR(as->J->ktrace)->o = IR_KGC;
+ }
+
/* Sync the interpreter state with the on-trace state. */
asm_stack_restore(as, snap);
@@ -1957,17 +2007,22 @@ static void asm_setup_regsp(ASMState *as)
ra_setup(as);
/* Clear reg/sp for constants. */
- for (ir = IR(T->nk), lastir = IR(REF_BASE); ir < lastir; ir++)
+ for (ir = IR(T->nk), lastir = IR(REF_BASE); ir < lastir; ir++) {
ir->prev = REGSP_INIT;
+ if (irt_is64(ir->t) && ir->o != IR_KNULL) {
+#if LJ_GC64
+ ir->i = 0; /* Will become non-zero only for RIP-relative addresses. */
+#else
+ /* Make life easier for backends by putting address of constant in i. */
+ ir->i = (int32_t)(intptr_t)(ir+1);
+#endif
+ ir++;
+ }
+ }
/* REF_BASE is used for implicit references to the BASE register. */
lastir->prev = REGSP_HINT(RID_BASE);
- ir = IR(nins-1);
- if (ir->o == IR_RENAME) {
- do { ir--; nins--; } while (ir->o == IR_RENAME);
- T->nins = nins; /* Remove any renames left over from ASM restart. */
- }
as->snaprename = nins;
as->snapref = nins;
as->snapno = T->nsnap;
@@ -2150,7 +2205,10 @@ static void asm_setup_regsp(ASMState *as)
#endif
#if LJ_TARGET_X86ORX64
/* Non-constant shift counts need to be in RID_ECX on x86/x64. */
- case IR_BSHL: case IR_BSHR: case IR_BSAR: case IR_BROL: case IR_BROR:
+ case IR_BSHL: case IR_BSHR: case IR_BSAR:
+ if ((as->flags & JIT_F_BMI2)) /* Except if BMI2 is available. */
+ break;
+ case IR_BROL: case IR_BROR:
if (!irref_isk(ir->op2) && !ra_hashint(IR(ir->op2)->r)) {
IR(ir->op2)->r = REGSP_HINT(RID_ECX);
if (inloop)
@@ -2196,14 +2254,25 @@ void lj_asm_trace(jit_State *J, GCtrace *T)
ASMState *as = &as_;
MCode *origtop;
+ /* Remove nops/renames left over from ASM restart due to LJ_TRERR_MCODELM. */
+ {
+ IRRef nins = T->nins;
+ IRIns *ir = &T->ir[nins-1];
+ if (ir->o == IR_NOP || ir->o == IR_RENAME) {
+ do { ir--; nins--; } while (ir->o == IR_NOP || ir->o == IR_RENAME);
+ T->nins = nins;
+ }
+ }
+
/* Ensure an initialized instruction beyond the last one for HIOP checks. */
- J->cur.nins = lj_ir_nextins(J);
- J->cur.ir[J->cur.nins].o = IR_NOP;
+ /* This also allows one RENAME to be added without reallocating curfinal. */
+ as->orignins = lj_ir_nextins(J);
+ J->cur.ir[as->orignins].o = IR_NOP;
/* Setup initial state. Copy some fields to reduce indirections. */
as->J = J;
as->T = T;
- as->ir = T->ir;
+ J->curfinal = lj_trace_alloc(J->L, T); /* This copies the IR, too. */
as->flags = J->flags;
as->loopref = J->loopref;
as->realign = NULL;
@@ -2216,12 +2285,41 @@ void lj_asm_trace(jit_State *J, GCtrace *T)
as->mclim = as->mcbot + MCLIM_REDZONE;
asm_setup_target(as);
- do {
+ /*
+ ** This is a loop, because the MCode may have to be (re-)assembled
+ ** multiple times:
+ **
+ ** 1. as->realign is set (and the assembly aborted), if the arch-specific
+ ** backend wants the MCode to be aligned differently.
+ **
+ ** This is currently only the case on x86/x64, where small loops get
+ ** an aligned loop body plus a short branch. Not much effort is wasted,
+ ** because the abort happens very quickly and only once.
+ **
+ ** 2. The IR is immovable, since the MCode embeds pointers to various
+ ** constants inside the IR. But RENAMEs may need to be added to the IR
+ ** during assembly, which might grow and reallocate the IR. We check
+ ** at the end if the IR (in J->cur.ir) has actually grown, resize the
+ ** copy (in J->curfinal.ir) and try again.
+ **
+ ** 95% of all traces have zero RENAMEs, 3% have one RENAME, 1.5% have
+ ** 2 RENAMEs and only 0.5% have more than that. That's why we opt to
+ ** always have one spare slot in the IR (see above), which means we
+ ** have to redo the assembly for only ~2% of all traces.
+ **
+ ** Very, very rarely, this needs to be done repeatedly, since the
+ ** location of constants inside the IR (actually, reachability from
+ ** a global pointer) may affect register allocation and thus the
+ ** number of RENAMEs.
+ */
+ for (;;) {
as->mcp = as->mctop;
#ifdef LUA_USE_ASSERT
as->mcp_prev = as->mcp;
#endif
- as->curins = T->nins;
+ as->ir = J->curfinal->ir; /* Use the copied IR. */
+ as->curins = J->cur.nins = as->orignins;
+
RA_DBG_START();
RA_DBGX((as, "===== STOP ====="));
@@ -2249,22 +2347,40 @@ void lj_asm_trace(jit_State *J, GCtrace *T)
checkmclim(as);
asm_ir(as, ir);
}
- } while (as->realign); /* Retry in case the MCode needs to be realigned. */
- /* Emit head of trace. */
- RA_DBG_REF();
- checkmclim(as);
- if (as->gcsteps > 0) {
- as->curins = as->T->snap[0].ref;
- asm_snap_prep(as); /* The GC check is a guard. */
- asm_gc_check(as);
+ if (as->realign && J->curfinal->nins >= T->nins)
+ continue; /* Retry in case only the MCode needs to be realigned. */
+
+ /* Emit head of trace. */
+ RA_DBG_REF();
+ checkmclim(as);
+ if (as->gcsteps > 0) {
+ as->curins = as->T->snap[0].ref;
+ asm_snap_prep(as); /* The GC check is a guard. */
+ asm_gc_check(as);
+ as->curins = as->stopins;
+ }
+ ra_evictk(as);
+ if (as->parent)
+ asm_head_side(as);
+ else
+ asm_head_root(as);
+ asm_phi_fixup(as);
+
+ if (J->curfinal->nins >= T->nins) { /* IR didn't grow? */
+ lua_assert(J->curfinal->nk == T->nk);
+ memcpy(J->curfinal->ir + as->orignins, T->ir + as->orignins,
+ (T->nins - as->orignins) * sizeof(IRIns)); /* Copy RENAMEs. */
+ T->nins = J->curfinal->nins;
+ break; /* Done. */
+ }
+
+ /* Otherwise try again with a bigger IR. */
+ lj_trace_free(J2G(J), J->curfinal);
+ J->curfinal = NULL; /* In case lj_trace_alloc() OOMs. */
+ J->curfinal = lj_trace_alloc(J->L, T);
+ as->realign = NULL;
}
- ra_evictk(as);
- if (as->parent)
- asm_head_side(as);
- else
- asm_head_root(as);
- asm_phi_fixup(as);
RA_DBGX((as, "===== START ===="));
RA_DBG_FLUSH();
@@ -2277,6 +2393,9 @@ void lj_asm_trace(jit_State *J, GCtrace *T)
if (!as->loopref)
asm_tail_fixup(as, T->link); /* Note: this may change as->mctop! */
T->szmcode = (MSize)((char *)as->mctop - (char *)as->mcp);
+#if LJ_TARGET_MCODE_FIXUP
+ asm_mcode_fixup(T->mcode, T->szmcode);
+#endif
lj_mcode_sync(T->mcode, origtop);
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm.h
index 9f4654ec0a8..2819481b6de 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm.h
@@ -1,6 +1,6 @@
/*
** IR assembler (SSA IR -> machine code).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_ASM_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_arm.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_arm.h
index ff4068a37fa..37bfa40f2fa 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_arm.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_arm.h
@@ -1,6 +1,6 @@
/*
** ARM IR assembler (SSA IR -> machine code).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/* -- Register allocator extensions --------------------------------------- */
@@ -909,7 +909,6 @@ static void asm_hrefk(ASMState *as, IRIns *ir)
static void asm_uref(ASMState *as, IRIns *ir)
{
- /* NYI: Check that UREFO is still open and not aliasing a slot. */
Reg dest = ra_dest(as, ir, RSET_GPR);
if (irref_isk(ir->op1)) {
GCfunc *fn = ir_kfunc(IR(ir->op1));
@@ -998,22 +997,26 @@ static ARMIns asm_fxstoreins(IRIns *ir)
static void asm_fload(ASMState *as, IRIns *ir)
{
- Reg dest = ra_dest(as, ir, RSET_GPR);
- Reg idx = ra_alloc1(as, ir->op1, RSET_GPR);
- ARMIns ai = asm_fxloadins(ir);
- int32_t ofs;
- if (ir->op2 == IRFL_TAB_ARRAY) {
- ofs = asm_fuseabase(as, ir->op1);
- if (ofs) { /* Turn the t->array load into an add for colocated arrays. */
- emit_dn(as, ARMI_ADD|ARMI_K12|ofs, dest, idx);
- return;
+ if (ir->op1 == REF_NIL) {
+ lua_assert(!ra_used(ir)); /* We can end up here if DCE is turned off. */
+ } else {
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg idx = ra_alloc1(as, ir->op1, RSET_GPR);
+ ARMIns ai = asm_fxloadins(ir);
+ int32_t ofs;
+ if (ir->op2 == IRFL_TAB_ARRAY) {
+ ofs = asm_fuseabase(as, ir->op1);
+ if (ofs) { /* Turn the t->array load into an add for colocated arrays. */
+ emit_dn(as, ARMI_ADD|ARMI_K12|ofs, dest, idx);
+ return;
+ }
}
+ ofs = field_ofs[ir->op2];
+ if ((ai & 0x04000000))
+ emit_lso(as, ai, dest, idx, ofs);
+ else
+ emit_lsox(as, ai, dest, idx, ofs);
}
- ofs = field_ofs[ir->op2];
- if ((ai & 0x04000000))
- emit_lso(as, ai, dest, idx, ofs);
- else
- emit_lsox(as, ai, dest, idx, ofs);
}
static void asm_fstore(ASMState *as, IRIns *ir)
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_mips.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_mips.h
index 9b7f89b096a..affe7d89962 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_mips.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_mips.h
@@ -1,6 +1,6 @@
/*
** MIPS IR assembler (SSA IR -> machine code).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/* -- Register allocator extensions --------------------------------------- */
@@ -23,7 +23,7 @@ static Reg ra_alloc1z(ASMState *as, IRRef ref, RegSet allow)
{
Reg r = IR(ref)->r;
if (ra_noreg(r)) {
- if (!(allow & RSET_FPR) && irref_isk(ref) && IR(ref)->i == 0)
+ if (!(allow & RSET_FPR) && irref_isk(ref) && get_kval(IR(ref)) == 0)
return RID_ZERO;
r = ra_allocref(as, ref, allow);
} else {
@@ -166,9 +166,9 @@ static Reg asm_fuseahuref(ASMState *as, IRRef ref, int32_t *ofsp, RegSet allow)
} else if (ir->o == IR_UREFC) {
if (irref_isk(ir->op1)) {
GCfunc *fn = ir_kfunc(IR(ir->op1));
- int32_t ofs = i32ptr(&gcref(fn->l.uvptr[(ir->op2 >> 8)])->uv.tv);
- int32_t jgl = (intptr_t)J2G(as->J);
- if ((uint32_t)(ofs-jgl) < 65536) {
+ intptr_t ofs = (intptr_t)&gcref(fn->l.uvptr[(ir->op2 >> 8)])->uv.tv;
+ intptr_t jgl = (intptr_t)J2G(as->J);
+ if ((uintptr_t)(ofs-jgl) < 65536) {
*ofsp = ofs-jgl-32768;
return RID_JGL;
} else {
@@ -190,20 +190,21 @@ static void asm_fusexref(ASMState *as, MIPSIns mi, Reg rt, IRRef ref,
Reg base;
if (ra_noreg(ir->r) && canfuse(as, ir)) {
if (ir->o == IR_ADD) {
- int32_t ofs2;
- if (irref_isk(ir->op2) && (ofs2 = ofs + IR(ir->op2)->i, checki16(ofs2))) {
+ intptr_t ofs2;
+ if (irref_isk(ir->op2) && (ofs2 = ofs + get_kval(IR(ir->op2)),
+ checki16(ofs2))) {
ref = ir->op1;
- ofs = ofs2;
+ ofs = (int32_t)ofs2;
}
} else if (ir->o == IR_STRREF) {
- int32_t ofs2 = 65536;
+ intptr_t ofs2 = 65536;
lua_assert(ofs == 0);
ofs = (int32_t)sizeof(GCstr);
if (irref_isk(ir->op2)) {
- ofs2 = ofs + IR(ir->op2)->i;
+ ofs2 = ofs + get_kval(IR(ir->op2));
ref = ir->op1;
} else if (irref_isk(ir->op1)) {
- ofs2 = ofs + IR(ir->op1)->i;
+ ofs2 = ofs + get_kval(IR(ir->op1));
ref = ir->op2;
}
if (!checki16(ofs2)) {
@@ -211,7 +212,7 @@ static void asm_fusexref(ASMState *as, MIPSIns mi, Reg rt, IRRef ref,
Reg right, left = ra_alloc2(as, ir, allow);
right = (left >> 8); left &= 255;
emit_hsi(as, mi, rt, RID_TMP, ofs);
- emit_dst(as, MIPSI_ADDU, RID_TMP, left, right);
+ emit_dst(as, MIPSI_AADDU, RID_TMP, left, right);
return;
}
ofs = ofs2;
@@ -227,7 +228,7 @@ static void asm_fusexref(ASMState *as, MIPSIns mi, Reg rt, IRRef ref,
static void asm_gencall(ASMState *as, const CCallInfo *ci, IRRef *args)
{
uint32_t n, nargs = CCI_XNARGS(ci);
- int32_t ofs = 16;
+ int32_t ofs = LJ_32 ? 16 : 0;
#if LJ_SOFTFP
Reg gpr = REGARG_FIRSTGPR;
#else
@@ -249,15 +250,15 @@ static void asm_gencall(ASMState *as, const CCallInfo *ci, IRRef *args)
!(ci->flags & CCI_VARARG)) {
lua_assert(rset_test(as->freeset, fpr)); /* Already evicted. */
ra_leftov(as, fpr, ref);
- fpr += 2;
- gpr += irt_isnum(ir->t) ? 2 : 1;
+ fpr += LJ_32 ? 2 : 1;
+ gpr += (LJ_32 && irt_isnum(ir->t)) ? 2 : 1;
} else
#endif
{
-#if !LJ_SOFTFP
+#if LJ_32 && !LJ_SOFTFP
fpr = REGARG_LASTFPR+1;
#endif
- if (irt_isnum(ir->t)) gpr = (gpr+1) & ~1;
+ if (LJ_32 && irt_isnum(ir->t)) gpr = (gpr+1) & ~1;
if (gpr <= REGARG_LASTGPR) {
lua_assert(rset_test(as->freeset, gpr)); /* Already evicted. */
#if !LJ_SOFTFP
@@ -269,35 +270,55 @@ static void asm_gencall(ASMState *as, const CCallInfo *ci, IRRef *args)
r = ra_alloc1(as, ref, RSET_FPR);
as->freeset |= (of & RSET_RANGE(REGARG_FIRSTGPR, REGARG_LASTGPR+1));
if (irt_isnum(ir->t)) {
+#if LJ_32
emit_tg(as, MIPSI_MFC1, gpr+(LJ_BE?0:1), r+1);
emit_tg(as, MIPSI_MFC1, gpr+(LJ_BE?1:0), r);
lua_assert(rset_test(as->freeset, gpr+1)); /* Already evicted. */
gpr += 2;
+#else
+ emit_tg(as, MIPSI_DMFC1, gpr, r);
+ gpr++; fpr++;
+#endif
} else if (irt_isfloat(ir->t)) {
emit_tg(as, MIPSI_MFC1, gpr, r);
gpr++;
+#if LJ_64
+ fpr++;
+#endif
}
} else
#endif
{
ra_leftov(as, gpr, ref);
gpr++;
+#if LJ_64
+ fpr++;
+#endif
}
} else {
Reg r = ra_alloc1z(as, ref, !LJ_SOFTFP && irt_isfp(ir->t) ? RSET_FPR : RSET_GPR);
+#if LJ_32
if (irt_isnum(ir->t)) ofs = (ofs + 4) & ~4;
emit_spstore(as, ir, r, ofs);
ofs += irt_isnum(ir->t) ? 8 : 4;
+#else
+ emit_spstore(as, ir, r, ofs + ((LJ_BE && (LJ_SOFTFP || r < RID_MAX_GPR) && !irt_is64(ir->t)) ? 4 : 0));
+ ofs += 8;
+#endif
}
}
} else {
#if !LJ_SOFTFP
fpr = REGARG_LASTFPR+1;
#endif
- if (gpr <= REGARG_LASTGPR)
+ if (gpr <= REGARG_LASTGPR) {
gpr++;
- else
- ofs += 4;
+#if LJ_64
+ fpr++;
+#endif
+ } else {
+ ofs += LJ_32 ? 4 : 8;
+ }
}
checkmclim(as);
}
@@ -307,15 +328,19 @@ static void asm_gencall(ASMState *as, const CCallInfo *ci, IRRef *args)
static void asm_setupresult(ASMState *as, IRIns *ir, const CCallInfo *ci)
{
RegSet drop = RSET_SCRATCH;
+#if LJ_32
int hiop = ((ir+1)->o == IR_HIOP && !irt_isnil((ir+1)->t));
+#endif
#if !LJ_SOFTFP
if ((ci->flags & CCI_NOFPRCLOBBER))
drop &= ~RSET_FPR;
#endif
if (ra_hasreg(ir->r))
rset_clear(drop, ir->r); /* Dest reg handled below. */
+#if LJ_32
if (hiop && ra_hasreg((ir+1)->r))
rset_clear(drop, (ir+1)->r); /* Dest reg handled below. */
+#endif
ra_evictset(as, drop); /* Evictions must be performed first. */
if (ra_used(ir)) {
lua_assert(!irt_ispri(ir->t));
@@ -326,18 +351,28 @@ static void asm_setupresult(ASMState *as, IRIns *ir, const CCallInfo *ci)
if (ra_hasreg(dest)) {
ra_free(as, dest);
ra_modified(as, dest);
+#if LJ_32
emit_tg(as, MIPSI_MTC1, RID_RETHI, dest+1);
emit_tg(as, MIPSI_MTC1, RID_RETLO, dest);
+#else
+ emit_tg(as, MIPSI_DMTC1, RID_RET, dest);
+#endif
}
if (ofs) {
+#if LJ_32
emit_tsi(as, MIPSI_SW, RID_RETLO, RID_SP, ofs+(LJ_BE?4:0));
emit_tsi(as, MIPSI_SW, RID_RETHI, RID_SP, ofs+(LJ_BE?0:4));
+#else
+ emit_tsi(as, MIPSI_SD, RID_RET, RID_SP, ofs);
+#endif
}
} else {
ra_destreg(as, ir, RID_FPRET);
}
+#if LJ_32
} else if (hiop) {
ra_destpair(as, ir);
+#endif
} else {
ra_destreg(as, ir, RID_RET);
}
@@ -356,7 +391,7 @@ static void asm_callx(ASMState *as, IRIns *ir)
func = ir->op2; irf = IR(func);
if (irf->o == IR_CARG) { func = irf->op1; irf = IR(func); }
if (irref_isk(func)) { /* Call to constant address. */
- ci.func = (ASMFunction)(void *)(irf->i);
+ ci.func = (ASMFunction)(void *)get_kval(irf);
} else { /* Need specific register for indirect calls. */
Reg r = ra_alloc1(as, func, RID2RSET(RID_CFUNCADDR));
MCode *p = as->mcp;
@@ -399,8 +434,8 @@ static void asm_retf(ASMState *as, IRIns *ir)
emit_setgl(as, base, jit_base);
emit_addptr(as, base, -8*delta);
asm_guard(as, MIPSI_BNE, RID_TMP,
- ra_allock(as, i32ptr(pc), rset_exclude(RSET_GPR, base)));
- emit_tsi(as, MIPSI_LW, RID_TMP, base, -8);
+ ra_allock(as, igcptr(pc), rset_exclude(RSET_GPR, base)));
+ emit_tsi(as, MIPSI_AL, RID_TMP, base, -8);
}
/* -- Type conversions ---------------------------------------------------- */
@@ -435,10 +470,15 @@ static void asm_conv(ASMState *as, IRIns *ir)
#if !LJ_SOFTFP
int stfp = (st == IRT_NUM || st == IRT_FLOAT);
#endif
+#if LJ_64
+ int st64 = (st == IRT_I64 || st == IRT_U64 || st == IRT_P64);
+#endif
IRRef lref = ir->op1;
+#if LJ_32
lua_assert(!(irt_isint64(ir->t) ||
(st == IRT_I64 || st == IRT_U64))); /* Handled by SPLIT. */
-#if LJ_SOFTFP
+#endif
+#if LJ_32 && LJ_SOFTFP
/* FP conversions are handled by SPLIT. */
lua_assert(!irt_isfp(ir->t) && !(st == IRT_NUM || st == IRT_FLOAT));
/* Can't check for same types: SPLIT uses CONV int.int + BXOR for sfp NEG. */
@@ -453,26 +493,50 @@ static void asm_conv(ASMState *as, IRIns *ir)
/* y = (x ^ 0x8000000) + 2147483648.0 */
Reg left = ra_alloc1(as, lref, RSET_GPR);
Reg tmp = ra_scratch(as, rset_exclude(RSET_FPR, dest));
- emit_fgh(as, irt_isfloat(ir->t) ? MIPSI_ADD_S : MIPSI_ADD_D,
- dest, dest, tmp);
- emit_fg(as, irt_isfloat(ir->t) ? MIPSI_CVT_S_W : MIPSI_CVT_D_W,
- dest, dest);
if (irt_isfloat(ir->t))
- emit_lsptr(as, MIPSI_LWC1, (tmp & 31),
- (void *)lj_ir_k64_find(as->J, U64x(4f000000,4f000000)),
- RSET_GPR);
- else
- emit_lsptr(as, MIPSI_LDC1, (tmp & 31),
- (void *)lj_ir_k64_find(as->J, U64x(41e00000,00000000)),
- RSET_GPR);
+ emit_fg(as, MIPSI_CVT_S_D, dest, dest);
+ /* Must perform arithmetic with doubles to keep the precision. */
+ emit_fgh(as, MIPSI_ADD_D, dest, dest, tmp);
+ emit_fg(as, MIPSI_CVT_D_W, dest, dest);
+ emit_lsptr(as, MIPSI_LDC1, (tmp & 31),
+ (void *)&as->J->k64[LJ_K64_2P31], RSET_GPR);
emit_tg(as, MIPSI_MTC1, RID_TMP, dest);
emit_dst(as, MIPSI_XOR, RID_TMP, RID_TMP, left);
emit_ti(as, MIPSI_LUI, RID_TMP, 0x8000);
+#if LJ_64
+ } else if(st == IRT_U64) { /* U64 to FP conversion. */
+ /* if (x >= 1u<<63) y = (double)(int64_t)(x&(1u<<63)-1) + pow(2.0, 63) */
+ Reg left = ra_alloc1(as, lref, RSET_GPR);
+ Reg tmp = ra_scratch(as, rset_exclude(RSET_FPR, dest));
+ MCLabel l_end = emit_label(as);
+ if (irt_isfloat(ir->t)) {
+ emit_fgh(as, MIPSI_ADD_S, dest, dest, tmp);
+ emit_lsptr(as, MIPSI_LWC1, (tmp & 31), (void *)&as->J->k32[LJ_K32_2P63],
+ rset_exclude(RSET_GPR, left));
+ emit_fg(as, MIPSI_CVT_S_L, dest, dest);
+ } else {
+ emit_fgh(as, MIPSI_ADD_D, dest, dest, tmp);
+ emit_lsptr(as, MIPSI_LDC1, (tmp & 31), (void *)&as->J->k64[LJ_K64_2P63],
+ rset_exclude(RSET_GPR, left));
+ emit_fg(as, MIPSI_CVT_D_L, dest, dest);
+ }
+ emit_branch(as, MIPSI_BGEZ, left, RID_ZERO, l_end);
+ emit_tg(as, MIPSI_DMTC1, RID_TMP, dest);
+ emit_tsml(as, MIPSI_DEXTM, RID_TMP, left, 30, 0);
+#endif
} else { /* Integer to FP conversion. */
Reg left = ra_alloc1(as, lref, RSET_GPR);
+#if LJ_32
emit_fg(as, irt_isfloat(ir->t) ? MIPSI_CVT_S_W : MIPSI_CVT_D_W,
dest, dest);
emit_tg(as, MIPSI_MTC1, left, dest);
+#else
+ MIPSIns mi = irt_isfloat(ir->t) ?
+ (st64 ? MIPSI_CVT_S_L : MIPSI_CVT_S_W) :
+ (st64 ? MIPSI_CVT_D_L : MIPSI_CVT_D_W);
+ emit_fg(as, mi, dest, dest);
+ emit_tg(as, st64 ? MIPSI_DMTC1 : MIPSI_MTC1, left, dest);
+#endif
}
} else if (stfp) { /* FP to integer conversion. */
if (irt_isguard(ir->t)) {
@@ -483,7 +547,7 @@ static void asm_conv(ASMState *as, IRIns *ir)
Reg dest = ra_dest(as, ir, RSET_GPR);
Reg left = ra_alloc1(as, lref, RSET_FPR);
Reg tmp = ra_scratch(as, rset_exclude(RSET_FPR, left));
- if (irt_isu32(ir->t)) {
+ if (irt_isu32(ir->t)) { /* FP to U32 conversion. */
/* y = (int)floor(x - 2147483648.0) ^ 0x80000000 */
emit_dst(as, MIPSI_XOR, dest, dest, RID_TMP);
emit_ti(as, MIPSI_LUI, RID_TMP, 0x8000);
@@ -494,16 +558,54 @@ static void asm_conv(ASMState *as, IRIns *ir)
tmp, left, tmp);
if (st == IRT_FLOAT)
emit_lsptr(as, MIPSI_LWC1, (tmp & 31),
- (void *)lj_ir_k64_find(as->J, U64x(4f000000,4f000000)),
- RSET_GPR);
+ (void *)&as->J->k32[LJ_K32_2P31], RSET_GPR);
else
emit_lsptr(as, MIPSI_LDC1, (tmp & 31),
- (void *)lj_ir_k64_find(as->J, U64x(41e00000,00000000)),
- RSET_GPR);
+ (void *)&as->J->k64[LJ_K64_2P31], RSET_GPR);
+#if LJ_64
+ } else if (irt_isu64(ir->t)) { /* FP to U64 conversion. */
+ MCLabel l_end;
+ emit_tg(as, MIPSI_DMFC1, dest, tmp);
+ l_end = emit_label(as);
+ /* For inputs >= 2^63 add -2^64 and convert again. */
+ if (st == IRT_NUM) {
+ emit_fg(as, MIPSI_TRUNC_L_D, tmp, tmp);
+ emit_fgh(as, MIPSI_ADD_D, tmp, left, tmp);
+ emit_lsptr(as, MIPSI_LDC1, (tmp & 31),
+ (void *)&as->J->k64[LJ_K64_M2P64],
+ rset_exclude(RSET_GPR, dest));
+ emit_fg(as, MIPSI_TRUNC_L_D, tmp, left); /* Delay slot. */
+ emit_branch(as, MIPSI_BC1T, 0, 0, l_end);
+ emit_fgh(as, MIPSI_C_OLT_D, 0, left, tmp);
+ emit_lsptr(as, MIPSI_LDC1, (tmp & 31),
+ (void *)&as->J->k64[LJ_K64_2P63],
+ rset_exclude(RSET_GPR, dest));
+ } else {
+ emit_fg(as, MIPSI_TRUNC_L_S, tmp, tmp);
+ emit_fgh(as, MIPSI_ADD_S, tmp, left, tmp);
+ emit_lsptr(as, MIPSI_LWC1, (tmp & 31),
+ (void *)&as->J->k32[LJ_K32_M2P64],
+ rset_exclude(RSET_GPR, dest));
+ emit_fg(as, MIPSI_TRUNC_L_S, tmp, left); /* Delay slot. */
+ emit_branch(as, MIPSI_BC1T, 0, 0, l_end);
+ emit_fgh(as, MIPSI_C_OLT_S, 0, left, tmp);
+ emit_lsptr(as, MIPSI_LWC1, (tmp & 31),
+ (void *)&as->J->k32[LJ_K32_2P63],
+ rset_exclude(RSET_GPR, dest));
+ }
+#endif
} else {
+#if LJ_32
emit_tg(as, MIPSI_MFC1, dest, tmp);
emit_fg(as, st == IRT_FLOAT ? MIPSI_TRUNC_W_S : MIPSI_TRUNC_W_D,
tmp, left);
+#else
+ MIPSIns mi = irt_is64(ir->t) ?
+ (st == IRT_NUM ? MIPSI_TRUNC_L_D : MIPSI_TRUNC_L_S) :
+ (st == IRT_NUM ? MIPSI_TRUNC_W_D : MIPSI_TRUNC_W_S);
+ emit_tg(as, irt_is64(ir->t) ? MIPSI_DMFC1 : MIPSI_MFC1, dest, left);
+ emit_fg(as, mi, left, left);
+#endif
}
}
} else
@@ -514,7 +616,7 @@ static void asm_conv(ASMState *as, IRIns *ir)
Reg left = ra_alloc1(as, ir->op1, RSET_GPR);
lua_assert(irt_isint(ir->t) || irt_isu32(ir->t));
if ((ir->op2 & IRCONV_SEXT)) {
- if ((as->flags & JIT_F_MIPS32R2)) {
+ if (LJ_64 || (as->flags & JIT_F_MIPSXXR2)) {
emit_dst(as, st == IRT_I8 ? MIPSI_SEB : MIPSI_SEH, dest, 0, left);
} else {
uint32_t shift = st == IRT_I8 ? 24 : 16;
@@ -526,8 +628,35 @@ static void asm_conv(ASMState *as, IRIns *ir)
(int32_t)(st == IRT_U8 ? 0xff : 0xffff));
}
} else { /* 32/64 bit integer conversions. */
+#if LJ_32
/* Only need to handle 32/32 bit no-op (cast) on 32 bit archs. */
ra_leftov(as, dest, lref); /* Do nothing, but may need to move regs. */
+#else
+ if (irt_is64(ir->t)) {
+ if (st64) {
+ /* 64/64 bit no-op (cast)*/
+ ra_leftov(as, dest, lref);
+ } else {
+ Reg left = ra_alloc1(as, lref, RSET_GPR);
+ if ((ir->op2 & IRCONV_SEXT)) { /* 32 to 64 bit sign extension. */
+ emit_dta(as, MIPSI_SLL, dest, left, 0);
+ } else { /* 32 to 64 bit zero extension. */
+ emit_tsml(as, MIPSI_DEXT, dest, left, 31, 0);
+ }
+ }
+ } else {
+ if (st64) {
+ /* This is either a 32 bit reg/reg mov which zeroes the hiword
+ ** or a load of the loword from a 64 bit address.
+ */
+ Reg left = ra_alloc1(as, lref, RSET_GPR);
+ emit_tsml(as, MIPSI_DEXT, dest, left, 31, 0);
+ } else { /* 32/32 bit no-op (cast). */
+ /* Do nothing, but may need to move regs. */
+ ra_leftov(as, dest, lref);
+ }
+ }
+#endif
}
}
}
@@ -570,23 +699,50 @@ static void asm_strto(ASMState *as, IRIns *ir)
args[1] = ASMREF_TMP1; /* TValue *n */
asm_gencall(as, ci, args);
/* Store the result to the spill slot or temp slots. */
- emit_tsi(as, MIPSI_ADDIU, ra_releasetmp(as, ASMREF_TMP1),
+ emit_tsi(as, MIPSI_AADDIU, ra_releasetmp(as, ASMREF_TMP1),
RID_SP, ofs);
}
/* -- Memory references --------------------------------------------------- */
+#if LJ_64
+/* Store tagged value for ref at base+ofs. */
+static void asm_tvstore64(ASMState *as, Reg base, int32_t ofs, IRRef ref)
+{
+ RegSet allow = rset_exclude(RSET_GPR, base);
+ IRIns *ir = IR(ref);
+ lua_assert(irt_ispri(ir->t) || irt_isaddr(ir->t) || irt_isinteger(ir->t));
+ if (irref_isk(ref)) {
+ TValue k;
+ lj_ir_kvalue(as->J->L, &k, ir);
+ emit_tsi(as, MIPSI_SD, ra_allock(as, (int64_t)k.u64, allow), base, ofs);
+ } else {
+ Reg src = ra_alloc1(as, ref, allow);
+ Reg type = ra_allock(as, (int64_t)irt_toitype(ir->t) << 47,
+ rset_exclude(allow, src));
+ emit_tsi(as, MIPSI_SD, RID_TMP, base, ofs);
+ if (irt_isinteger(ir->t)) {
+ emit_dst(as, MIPSI_DADDU, RID_TMP, RID_TMP, type);
+ emit_tsml(as, MIPSI_DEXT, RID_TMP, src, 31, 0);
+ } else {
+ emit_dst(as, MIPSI_DADDU, RID_TMP, src, type);
+ }
+ }
+}
+#endif
+
/* Get pointer to TValue. */
static void asm_tvptr(ASMState *as, Reg dest, IRRef ref)
{
IRIns *ir = IR(ref);
if (irt_isnum(ir->t)) {
if (irref_isk(ref)) /* Use the number constant itself as a TValue. */
- ra_allockreg(as, i32ptr(ir_knum(ir)), dest);
+ ra_allockreg(as, igcptr(ir_knum(ir)), dest);
else /* Otherwise force a spill and use the spill slot. */
- emit_tsi(as, MIPSI_ADDIU, dest, RID_SP, ra_spill(as, ir));
+ emit_tsi(as, MIPSI_AADDIU, dest, RID_SP, ra_spill(as, ir));
} else {
/* Otherwise use g->tmptv to hold the TValue. */
+#if LJ_32
RegSet allow = rset_exclude(RSET_GPR, dest);
Reg type;
emit_tsi(as, MIPSI_ADDIU, dest, RID_JGL, (int32_t)(offsetof(global_State, tmptv)-32768));
@@ -599,6 +755,11 @@ static void asm_tvptr(ASMState *as, Reg dest, IRRef ref)
else
type = ra_allock(as, (int32_t)irt_toitype(ir->t), allow);
emit_setgl(as, type, tmptv.it);
+#else
+ asm_tvstore64(as, dest, 0, ref);
+ emit_tsi(as, MIPSI_DADDIU, dest, RID_JGL,
+ (int32_t)(offsetof(global_State, tmptv)-32768));
+#endif
}
}
@@ -613,13 +774,13 @@ static void asm_aref(ASMState *as, IRIns *ir)
ofs += 8*IR(ir->op2)->i;
if (checki16(ofs)) {
base = ra_alloc1(as, refa, RSET_GPR);
- emit_tsi(as, MIPSI_ADDIU, dest, base, ofs);
+ emit_tsi(as, MIPSI_AADDIU, dest, base, ofs);
return;
}
}
base = ra_alloc1(as, ir->op1, RSET_GPR);
idx = ra_alloc1(as, ir->op2, rset_exclude(RSET_GPR, base));
- emit_dst(as, MIPSI_ADDU, dest, RID_TMP, base);
+ emit_dst(as, MIPSI_AADDU, dest, RID_TMP, base);
emit_dta(as, MIPSI_SLL, RID_TMP, idx, 3);
}
@@ -640,19 +801,19 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
Reg key = RID_NONE, type = RID_NONE, tmpnum = RID_NONE, tmp1 = RID_TMP, tmp2;
IRRef refkey = ir->op2;
IRIns *irkey = IR(refkey);
+ int isk = irref_isk(refkey);
IRType1 kt = irkey->t;
uint32_t khash;
MCLabel l_end, l_loop, l_next;
rset_clear(allow, tab);
-#if LJ_SOFTFP
- if (!irref_isk(refkey)) {
+#if LJ_32 && LJ_SOFTFP
+ if (!isk) {
key = ra_alloc1(as, refkey, allow);
rset_clear(allow, key);
if (irkey[1].o == IR_HIOP) {
if (ra_hasreg((irkey+1)->r)) {
- tmpnum = (irkey+1)->r;
- type = RID_TMP;
+ type = tmpnum = (irkey+1)->r;
tmp1 = ra_scratch(as, allow);
rset_clear(allow, tmp1);
ra_noweak(as, tmpnum);
@@ -672,8 +833,10 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
} else if (!irt_ispri(kt)) {
key = ra_alloc1(as, refkey, allow);
rset_clear(allow, key);
+#if LJ_32
type = ra_allock(as, (int32_t)irt_toitype(irkey->t), allow);
rset_clear(allow, type);
+#endif
}
#endif
tmp2 = ra_scratch(as, allow);
@@ -687,9 +850,9 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
else if (destused)
emit_loada(as, dest, niltvg(J2G(as->J)));
/* Follow hash chain until the end. */
- emit_move(as, dest, tmp2);
+ emit_move(as, dest, tmp1);
l_loop = --as->mcp;
- emit_tsi(as, MIPSI_LW, tmp2, dest, (int32_t)offsetof(Node, next));
+ emit_tsi(as, MIPSI_AL, tmp1, dest, (int32_t)offsetof(Node, next));
l_next = emit_label(as);
/* Type and value comparison. */
@@ -701,38 +864,66 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
emit_branch(as, MIPSI_BC1T, 0, 0, l_end);
emit_fgh(as, MIPSI_C_EQ_D, 0, tmpnum, key);
*--as->mcp = MIPSI_NOP; /* Avoid NaN comparison overhead. */
- emit_branch(as, MIPSI_BEQ, tmp2, RID_ZERO, l_next);
- emit_tsi(as, MIPSI_SLTIU, tmp2, tmp2, (int32_t)LJ_TISNUM);
+ emit_branch(as, MIPSI_BEQ, tmp1, RID_ZERO, l_next);
+ emit_tsi(as, MIPSI_SLTIU, tmp1, tmp1, (int32_t)LJ_TISNUM);
+#if LJ_32
emit_hsi(as, MIPSI_LDC1, tmpnum, dest, (int32_t)offsetof(Node, key.n));
} else {
if (irt_ispri(kt)) {
- emit_branch(as, MIPSI_BEQ, tmp2, type, l_end);
+ emit_branch(as, MIPSI_BEQ, tmp1, type, l_end);
} else {
- emit_branch(as, MIPSI_BEQ, tmp1, key, l_end);
- emit_tsi(as, MIPSI_LW, tmp1, dest, (int32_t)offsetof(Node, key.gcr));
- emit_branch(as, MIPSI_BNE, tmp2, type, l_next);
+ emit_branch(as, MIPSI_BEQ, tmp2, key, l_end);
+ emit_tsi(as, MIPSI_LW, tmp2, dest, (int32_t)offsetof(Node, key.gcr));
+ emit_branch(as, MIPSI_BNE, tmp1, type, l_next);
+ }
+ }
+ emit_tsi(as, MIPSI_LW, tmp1, dest, (int32_t)offsetof(Node, key.it));
+ *l_loop = MIPSI_BNE | MIPSF_S(tmp1) | ((as->mcp-l_loop-1) & 0xffffu);
+#else
+ emit_dta(as, MIPSI_DSRA32, tmp1, tmp1, 15);
+ emit_tg(as, MIPSI_DMTC1, tmp1, tmpnum);
+ emit_tsi(as, MIPSI_LD, tmp1, dest, (int32_t)offsetof(Node, key.u64));
+ } else if (irt_isaddr(kt)) {
+ Reg refk = tmp2;
+ if (isk) {
+ int64_t k = ((int64_t)irt_toitype(irkey->t) << 47) | irkey[1].tv.u64;
+ refk = ra_allock(as, k, allow);
+ rset_clear(allow, refk);
}
+ emit_branch(as, MIPSI_BEQ, tmp1, refk, l_end);
+ emit_tsi(as, MIPSI_LD, tmp1, dest, offsetof(Node, key));
+ } else {
+ Reg pri = ra_allock(as, ~((int64_t)~irt_toitype(ir->t) << 47), allow);
+ rset_clear(allow, pri);
+ lua_assert(irt_ispri(kt) && !irt_isnil(kt));
+ emit_branch(as, MIPSI_BEQ, tmp1, pri, l_end);
+ emit_tsi(as, MIPSI_LD, tmp1, dest, offsetof(Node, key));
+ }
+ *l_loop = MIPSI_BNE | MIPSF_S(tmp1) | ((as->mcp-l_loop-1) & 0xffffu);
+ if (!isk && irt_isaddr(kt)) {
+ type = ra_allock(as, (int64_t)irt_toitype(kt) << 47, allow);
+ emit_dst(as, MIPSI_DADDU, tmp2, key, type);
+ rset_clear(allow, type);
}
- emit_tsi(as, MIPSI_LW, tmp2, dest, (int32_t)offsetof(Node, key.it));
- *l_loop = MIPSI_BNE | MIPSF_S(tmp2) | ((as->mcp-l_loop-1) & 0xffffu);
+#endif
/* Load main position relative to tab->node into dest. */
- khash = irref_isk(refkey) ? ir_khash(irkey) : 1;
+ khash = isk ? ir_khash(irkey) : 1;
if (khash == 0) {
- emit_tsi(as, MIPSI_LW, dest, tab, (int32_t)offsetof(GCtab, node));
+ emit_tsi(as, MIPSI_AL, dest, tab, (int32_t)offsetof(GCtab, node));
} else {
Reg tmphash = tmp1;
- if (irref_isk(refkey))
+ if (isk)
tmphash = ra_allock(as, khash, allow);
- emit_dst(as, MIPSI_ADDU, dest, dest, tmp1);
+ emit_dst(as, MIPSI_AADDU, dest, dest, tmp1);
lua_assert(sizeof(Node) == 24);
emit_dst(as, MIPSI_SUBU, tmp1, tmp2, tmp1);
emit_dta(as, MIPSI_SLL, tmp1, tmp1, 3);
emit_dta(as, MIPSI_SLL, tmp2, tmp1, 5);
emit_dst(as, MIPSI_AND, tmp1, tmp2, tmphash);
- emit_tsi(as, MIPSI_LW, dest, tab, (int32_t)offsetof(GCtab, node));
+ emit_tsi(as, MIPSI_AL, dest, tab, (int32_t)offsetof(GCtab, node));
emit_tsi(as, MIPSI_LW, tmp2, tab, (int32_t)offsetof(GCtab, hmask));
- if (irref_isk(refkey)) {
+ if (isk) {
/* Nothing to do. */
} else if (irt_isstr(kt)) {
emit_tsi(as, MIPSI_LW, tmp1, key, (int32_t)offsetof(GCstr, hash));
@@ -742,9 +933,10 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
emit_dst(as, MIPSI_XOR, tmp1, tmp1, tmp2);
emit_rotr(as, tmp1, tmp1, dest, (-HASH_ROT2-HASH_ROT1)&31);
emit_dst(as, MIPSI_SUBU, tmp2, tmp2, dest);
+#if LJ_32
if (LJ_SOFTFP ? (irkey[1].o == IR_HIOP) : irt_isnum(kt)) {
emit_dst(as, MIPSI_XOR, tmp2, tmp2, tmp1);
- if ((as->flags & JIT_F_MIPS32R2)) {
+ if ((as->flags & JIT_F_MIPSXXR2)) {
emit_dta(as, MIPSI_ROTR, dest, tmp1, (-HASH_ROT1)&31);
} else {
emit_dst(as, MIPSI_OR, dest, dest, tmp1);
@@ -764,6 +956,23 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
emit_rotr(as, dest, tmp1, tmp2, (-HASH_ROT1)&31);
emit_dst(as, MIPSI_ADDU, tmp1, key, ra_allock(as, HASH_BIAS, allow));
}
+#else
+ emit_dst(as, MIPSI_XOR, tmp2, tmp2, tmp1);
+ emit_dta(as, MIPSI_ROTR, dest, tmp1, (-HASH_ROT1)&31);
+ if (irt_isnum(kt)) {
+ emit_dst(as, MIPSI_ADDU, tmp1, tmp1, tmp1);
+ emit_dta(as, MIPSI_DSRA32, tmp1, tmp1, 0);
+ emit_dta(as, MIPSI_SLL, tmp2, LJ_SOFTFP ? key : tmp1, 0);
+#if !LJ_SOFTFP
+ emit_tg(as, MIPSI_DMFC1, tmp1, key);
+#endif
+ } else {
+ checkmclim(as);
+ emit_dta(as, MIPSI_DSRA32, tmp1, tmp1, 0);
+ emit_dta(as, MIPSI_SLL, tmp2, key, 0);
+ emit_dst(as, MIPSI_DADDU, tmp1, key, type);
+ }
+#endif
}
}
}
@@ -776,17 +985,24 @@ static void asm_hrefk(ASMState *as, IRIns *ir)
int32_t kofs = ofs + (int32_t)offsetof(Node, key);
Reg dest = (ra_used(ir)||ofs > 32736) ? ra_dest(as, ir, RSET_GPR) : RID_NONE;
Reg node = ra_alloc1(as, ir->op1, RSET_GPR);
- Reg key = RID_NONE, type = RID_TMP, idx = node;
RegSet allow = rset_exclude(RSET_GPR, node);
+ Reg idx = node;
+#if LJ_32
+ Reg key = RID_NONE, type = RID_TMP;
int32_t lo, hi;
+#else
+ Reg key = ra_scratch(as, allow);
+ int64_t k;
+#endif
lua_assert(ofs % sizeof(Node) == 0);
if (ofs > 32736) {
idx = dest;
rset_clear(allow, dest);
kofs = (int32_t)offsetof(Node, key);
} else if (ra_hasreg(dest)) {
- emit_tsi(as, MIPSI_ADDIU, dest, node, ofs);
+ emit_tsi(as, MIPSI_AADDIU, dest, node, ofs);
}
+#if LJ_32
if (!irt_ispri(irkey->t)) {
key = ra_scratch(as, allow);
rset_clear(allow, key);
@@ -805,30 +1021,41 @@ nolo:
asm_guard(as, MIPSI_BNE, type, hi ? ra_allock(as, hi, allow) : RID_ZERO);
if (ra_hasreg(key)) emit_tsi(as, MIPSI_LW, key, idx, kofs+(LJ_BE?4:0));
emit_tsi(as, MIPSI_LW, type, idx, kofs+(LJ_BE?0:4));
+#else
+ if (irt_ispri(irkey->t)) {
+ lua_assert(!irt_isnil(irkey->t));
+ k = ~((int64_t)~irt_toitype(irkey->t) << 47);
+ } else if (irt_isnum(irkey->t)) {
+ k = (int64_t)ir_knum(irkey)->u64;
+ } else {
+ k = ((int64_t)irt_toitype(irkey->t) << 47) | (int64_t)ir_kgc(irkey);
+ }
+ asm_guard(as, MIPSI_BNE, key, ra_allock(as, k, allow));
+ emit_tsi(as, MIPSI_LD, key, idx, kofs);
+#endif
if (ofs > 32736)
- emit_tsi(as, MIPSI_ADDU, dest, node, ra_allock(as, ofs, allow));
+ emit_tsi(as, MIPSI_AADDU, dest, node, ra_allock(as, ofs, allow));
}
static void asm_uref(ASMState *as, IRIns *ir)
{
- /* NYI: Check that UREFO is still open and not aliasing a slot. */
Reg dest = ra_dest(as, ir, RSET_GPR);
if (irref_isk(ir->op1)) {
GCfunc *fn = ir_kfunc(IR(ir->op1));
MRef *v = &gcref(fn->l.uvptr[(ir->op2 >> 8)])->uv.v;
- emit_lsptr(as, MIPSI_LW, dest, v, RSET_GPR);
+ emit_lsptr(as, MIPSI_AL, dest, v, RSET_GPR);
} else {
Reg uv = ra_scratch(as, RSET_GPR);
Reg func = ra_alloc1(as, ir->op1, RSET_GPR);
if (ir->o == IR_UREFC) {
asm_guard(as, MIPSI_BEQ, RID_TMP, RID_ZERO);
- emit_tsi(as, MIPSI_ADDIU, dest, uv, (int32_t)offsetof(GCupval, tv));
+ emit_tsi(as, MIPSI_AADDIU, dest, uv, (int32_t)offsetof(GCupval, tv));
emit_tsi(as, MIPSI_LBU, RID_TMP, uv, (int32_t)offsetof(GCupval, closed));
} else {
- emit_tsi(as, MIPSI_LW, dest, uv, (int32_t)offsetof(GCupval, v));
+ emit_tsi(as, MIPSI_AL, dest, uv, (int32_t)offsetof(GCupval, v));
}
- emit_tsi(as, MIPSI_LW, uv, func,
- (int32_t)offsetof(GCfuncL, uvptr) + 4*(int32_t)(ir->op2 >> 8));
+ emit_tsi(as, MIPSI_AL, uv, func, (int32_t)offsetof(GCfuncL, uvptr) +
+ (int32_t)sizeof(MRef) * (int32_t)(ir->op2 >> 8));
}
}
@@ -840,6 +1067,7 @@ static void asm_fref(ASMState *as, IRIns *ir)
static void asm_strref(ASMState *as, IRIns *ir)
{
+#if LJ_32
Reg dest = ra_dest(as, ir, RSET_GPR);
IRRef ref = ir->op2, refk = ir->op1;
int32_t ofs = (int32_t)sizeof(GCstr);
@@ -871,6 +1099,20 @@ static void asm_strref(ASMState *as, IRIns *ir)
else
emit_dst(as, MIPSI_ADDU, dest, r,
ra_allock(as, ofs, rset_exclude(RSET_GPR, r)));
+#else
+ RegSet allow = RSET_GPR;
+ Reg dest = ra_dest(as, ir, allow);
+ Reg base = ra_alloc1(as, ir->op1, allow);
+ IRIns *irr = IR(ir->op2);
+ int32_t ofs = sizeof(GCstr);
+ rset_clear(allow, base);
+ if (irref_isk(ir->op2) && checki16(ofs + irr->i)) {
+ emit_tsi(as, MIPSI_DADDIU, dest, base, ofs + irr->i);
+ } else {
+ emit_tsi(as, MIPSI_DADDIU, dest, dest, ofs);
+ emit_dst(as, MIPSI_DADDU, dest, base, ra_alloc1(as, ir->op2, allow));
+ }
+#endif
}
/* -- Loads and stores ---------------------------------------------------- */
@@ -884,7 +1126,7 @@ static MIPSIns asm_fxloadins(IRIns *ir)
case IRT_U16: return MIPSI_LHU;
case IRT_NUM: lua_assert(!LJ_SOFTFP); return MIPSI_LDC1;
case IRT_FLOAT: if (!LJ_SOFTFP) return MIPSI_LWC1;
- default: return MIPSI_LW;
+ default: return (LJ_64 && irt_is64(ir->t)) ? MIPSI_LD : MIPSI_LW;
}
}
@@ -895,24 +1137,30 @@ static MIPSIns asm_fxstoreins(IRIns *ir)
case IRT_I16: case IRT_U16: return MIPSI_SH;
case IRT_NUM: lua_assert(!LJ_SOFTFP); return MIPSI_SDC1;
case IRT_FLOAT: if (!LJ_SOFTFP) return MIPSI_SWC1;
- default: return MIPSI_SW;
+ default: return (LJ_64 && irt_is64(ir->t)) ? MIPSI_SD : MIPSI_SW;
}
}
static void asm_fload(ASMState *as, IRIns *ir)
{
Reg dest = ra_dest(as, ir, RSET_GPR);
- Reg idx = ra_alloc1(as, ir->op1, RSET_GPR);
MIPSIns mi = asm_fxloadins(ir);
+ Reg idx;
int32_t ofs;
- if (ir->op2 == IRFL_TAB_ARRAY) {
- ofs = asm_fuseabase(as, ir->op1);
- if (ofs) { /* Turn the t->array load into an add for colocated arrays. */
- emit_tsi(as, MIPSI_ADDIU, dest, idx, ofs);
- return;
+ if (ir->op1 == REF_NIL) {
+ idx = RID_JGL;
+ ofs = (ir->op2 << 2) - 32768 - GG_OFS(g);
+ } else {
+ idx = ra_alloc1(as, ir->op1, RSET_GPR);
+ if (ir->op2 == IRFL_TAB_ARRAY) {
+ ofs = asm_fuseabase(as, ir->op1);
+ if (ofs) { /* Turn the t->array load into an add for colocated arrays. */
+ emit_tsi(as, MIPSI_AADDIU, dest, idx, ofs);
+ return;
+ }
}
+ ofs = field_ofs[ir->op2];
}
- ofs = field_ofs[ir->op2];
lua_assert(!irt_isfp(ir->t));
emit_tsi(as, mi, dest, idx, ofs);
}
@@ -952,36 +1200,59 @@ static void asm_xstore_(ASMState *as, IRIns *ir, int32_t ofs)
static void asm_ahuvload(ASMState *as, IRIns *ir)
{
- int hiop = (LJ_SOFTFP && (ir+1)->o == IR_HIOP);
- IRType t = hiop ? IRT_NUM : irt_type(ir->t);
+ int hiop = (LJ_32 && LJ_SOFTFP && (ir+1)->o == IR_HIOP);
Reg dest = RID_NONE, type = RID_TMP, idx;
RegSet allow = RSET_GPR;
int32_t ofs = 0;
- if (hiop && ra_used(ir+1)) {
- type = ra_dest(as, ir+1, allow);
- rset_clear(allow, type);
+ IRType1 t = ir->t;
+ if (hiop) {
+ t.irt = IRT_NUM;
+ if (ra_used(ir+1)) {
+ type = ra_dest(as, ir+1, allow);
+ rset_clear(allow, type);
+ }
}
if (ra_used(ir)) {
lua_assert((LJ_SOFTFP ? 0 : irt_isnum(ir->t)) ||
irt_isint(ir->t) || irt_isaddr(ir->t));
- dest = ra_dest(as, ir, (!LJ_SOFTFP && t == IRT_NUM) ? RSET_FPR : allow);
+ dest = ra_dest(as, ir, (!LJ_SOFTFP && irt_isnum(t)) ? RSET_FPR : allow);
rset_clear(allow, dest);
+#if LJ_64
+ if (irt_isaddr(t))
+ emit_tsml(as, MIPSI_DEXTM, dest, dest, 14, 0);
+ else if (irt_isint(t))
+ emit_dta(as, MIPSI_SLL, dest, dest, 0);
+#endif
}
idx = asm_fuseahuref(as, ir->op1, &ofs, allow);
rset_clear(allow, idx);
- if (t == IRT_NUM) {
+ if (irt_isnum(t)) {
asm_guard(as, MIPSI_BEQ, RID_TMP, RID_ZERO);
emit_tsi(as, MIPSI_SLTIU, RID_TMP, type, (int32_t)LJ_TISNUM);
} else {
- asm_guard(as, MIPSI_BNE, type, ra_allock(as, irt_toitype_(t), allow));
+ asm_guard(as, MIPSI_BNE, type,
+ ra_allock(as, (int32_t)irt_toitype(t), allow));
}
+#if LJ_32
if (ra_hasreg(dest)) {
- if (!LJ_SOFTFP && t == IRT_NUM)
+ if (!LJ_SOFTFP && irt_isnum(t))
emit_hsi(as, MIPSI_LDC1, dest, idx, ofs);
else
emit_tsi(as, MIPSI_LW, dest, idx, ofs+(LJ_BE?4:0));
}
emit_tsi(as, MIPSI_LW, type, idx, ofs+(LJ_BE?0:4));
+#else
+ if (ra_hasreg(dest)) {
+ if (!LJ_SOFTFP && irt_isnum(t)) {
+ emit_hsi(as, MIPSI_LDC1, dest, idx, ofs);
+ dest = type;
+ }
+ } else {
+ dest = type;
+ }
+ emit_dta(as, MIPSI_DSRA32, type, dest, 15);
+ emit_tsi(as, MIPSI_LD, dest, idx, ofs);
+#endif
}
static void asm_ahustore(ASMState *as, IRIns *ir)
@@ -993,103 +1264,159 @@ static void asm_ahustore(ASMState *as, IRIns *ir)
return;
if (!LJ_SOFTFP && irt_isnum(ir->t)) {
src = ra_alloc1(as, ir->op2, RSET_FPR);
+ idx = asm_fuseahuref(as, ir->op1, &ofs, allow);
+ emit_hsi(as, MIPSI_SDC1, src, idx, ofs);
} else {
- int hiop = (LJ_SOFTFP && (ir+1)->o == IR_HIOP);
+#if LJ_32
if (!irt_ispri(ir->t)) {
src = ra_alloc1(as, ir->op2, allow);
rset_clear(allow, src);
}
- if (hiop)
+ if (LJ_SOFTFP && (ir+1)->o == IR_HIOP)
type = ra_alloc1(as, (ir+1)->op2, allow);
else
type = ra_allock(as, (int32_t)irt_toitype(ir->t), allow);
rset_clear(allow, type);
- }
- idx = asm_fuseahuref(as, ir->op1, &ofs, allow);
- if (!LJ_SOFTFP && irt_isnum(ir->t)) {
- emit_hsi(as, MIPSI_SDC1, src, idx, ofs);
- } else {
+ idx = asm_fuseahuref(as, ir->op1, &ofs, allow);
if (ra_hasreg(src))
emit_tsi(as, MIPSI_SW, src, idx, ofs+(LJ_BE?4:0));
emit_tsi(as, MIPSI_SW, type, idx, ofs+(LJ_BE?0:4));
+#else
+ Reg tmp = RID_TMP;
+ if (irt_ispri(ir->t)) {
+ tmp = ra_allock(as, ~((int64_t)~irt_toitype(ir->t) << 47), allow);
+ rset_clear(allow, tmp);
+ } else {
+ src = ra_alloc1(as, ir->op2, allow);
+ rset_clear(allow, src);
+ type = ra_allock(as, (int64_t)irt_toitype(ir->t) << 47, allow);
+ rset_clear(allow, type);
+ }
+ idx = asm_fuseahuref(as, ir->op1, &ofs, allow);
+ emit_tsi(as, MIPSI_SD, tmp, idx, ofs);
+ if (ra_hasreg(src)) {
+ if (irt_isinteger(ir->t)) {
+ emit_dst(as, MIPSI_DADDU, tmp, tmp, type);
+ emit_tsml(as, MIPSI_DEXT, tmp, src, 31, 0);
+ } else {
+ emit_dst(as, MIPSI_DADDU, tmp, src, type);
+ }
+ }
+#endif
}
}
static void asm_sload(ASMState *as, IRIns *ir)
{
- int32_t ofs = 8*((int32_t)ir->op1-1) + ((ir->op2 & IRSLOAD_FRAME) ? 4 : 0);
- int hiop = (LJ_SOFTFP && (ir+1)->o == IR_HIOP);
- IRType t = hiop ? IRT_NUM : irt_type(ir->t);
Reg dest = RID_NONE, type = RID_NONE, base;
RegSet allow = RSET_GPR;
+ IRType1 t = ir->t;
+#if LJ_32
+ int32_t ofs = 8*((int32_t)ir->op1-1) + ((ir->op2 & IRSLOAD_FRAME) ? 4 : 0);
+ int hiop = (LJ_32 && LJ_SOFTFP && (ir+1)->o == IR_HIOP);
+ if (hiop)
+ t.irt = IRT_NUM;
+#else
+ int32_t ofs = 8*((int32_t)ir->op1-2);
+#endif
lua_assert(!(ir->op2 & IRSLOAD_PARENT)); /* Handled by asm_head_side(). */
lua_assert(irt_isguard(ir->t) || !(ir->op2 & IRSLOAD_TYPECHECK));
-#if LJ_SOFTFP
+#if LJ_32 && LJ_SOFTFP
lua_assert(!(ir->op2 & IRSLOAD_CONVERT)); /* Handled by LJ_SOFTFP SPLIT. */
if (hiop && ra_used(ir+1)) {
type = ra_dest(as, ir+1, allow);
rset_clear(allow, type);
}
#else
- if ((ir->op2 & IRSLOAD_CONVERT) && irt_isguard(ir->t) && t == IRT_INT) {
+ if ((ir->op2 & IRSLOAD_CONVERT) && irt_isguard(t) && irt_isint(t)) {
dest = ra_scratch(as, RSET_FPR);
asm_tointg(as, ir, dest);
- t = IRT_NUM; /* Continue with a regular number type check. */
+ t.irt = IRT_NUM; /* Continue with a regular number type check. */
} else
#endif
if (ra_used(ir)) {
lua_assert((LJ_SOFTFP ? 0 : irt_isnum(ir->t)) ||
irt_isint(ir->t) || irt_isaddr(ir->t));
- dest = ra_dest(as, ir, (!LJ_SOFTFP && t == IRT_NUM) ? RSET_FPR : allow);
+ dest = ra_dest(as, ir, (!LJ_SOFTFP && irt_isnum(t)) ? RSET_FPR : allow);
rset_clear(allow, dest);
base = ra_alloc1(as, REF_BASE, allow);
rset_clear(allow, base);
if (!LJ_SOFTFP && (ir->op2 & IRSLOAD_CONVERT)) {
- if (t == IRT_INT) {
+ if (irt_isint(t)) {
Reg tmp = ra_scratch(as, RSET_FPR);
emit_tg(as, MIPSI_MFC1, dest, tmp);
emit_fg(as, MIPSI_TRUNC_W_D, tmp, tmp);
dest = tmp;
- t = IRT_NUM; /* Check for original type. */
+ t.irt = IRT_NUM; /* Check for original type. */
} else {
Reg tmp = ra_scratch(as, RSET_GPR);
emit_fg(as, MIPSI_CVT_D_W, dest, dest);
emit_tg(as, MIPSI_MTC1, tmp, dest);
dest = tmp;
- t = IRT_INT; /* Check for original type. */
+ t.irt = IRT_INT; /* Check for original type. */
}
}
+#if LJ_64
+ else if (irt_isaddr(t)) {
+ /* Clear type from pointers. */
+ emit_tsml(as, MIPSI_DEXTM, dest, dest, 14, 0);
+ } else if (irt_isint(t) && (ir->op2 & IRSLOAD_TYPECHECK)) {
+ /* Sign-extend integers. */
+ emit_dta(as, MIPSI_SLL, dest, dest, 0);
+ }
+#endif
goto dotypecheck;
}
base = ra_alloc1(as, REF_BASE, allow);
rset_clear(allow, base);
dotypecheck:
+#if LJ_32
if ((ir->op2 & IRSLOAD_TYPECHECK)) {
- if (ra_noreg(type)) {
- if (ofs < 256 && ra_hasreg(dest) && (dest & 1) == 0 &&
- rset_test((as->freeset & allow), dest+1)) {
- type = dest+1;
- ra_modified(as, type);
- } else {
- type = RID_TMP;
- }
- }
- if (t == IRT_NUM) {
+ if (ra_noreg(type))
+ type = RID_TMP;
+ if (irt_isnum(t)) {
asm_guard(as, MIPSI_BEQ, RID_TMP, RID_ZERO);
emit_tsi(as, MIPSI_SLTIU, RID_TMP, type, (int32_t)LJ_TISNUM);
} else {
- Reg ktype = ra_allock(as, irt_toitype_(t), allow);
+ Reg ktype = ra_allock(as, irt_toitype(t), allow);
asm_guard(as, MIPSI_BNE, type, ktype);
}
}
if (ra_hasreg(dest)) {
- if (!LJ_SOFTFP && t == IRT_NUM)
+ if (!LJ_SOFTFP && irt_isnum(t))
emit_hsi(as, MIPSI_LDC1, dest, base, ofs);
else
emit_tsi(as, MIPSI_LW, dest, base, ofs ^ (LJ_BE?4:0));
}
if (ra_hasreg(type))
emit_tsi(as, MIPSI_LW, type, base, ofs ^ (LJ_BE?0:4));
+#else
+ if ((ir->op2 & IRSLOAD_TYPECHECK)) {
+ type = dest < RID_MAX_GPR ? dest : RID_TMP;
+ if (irt_ispri(t)) {
+ asm_guard(as, MIPSI_BNE, type,
+ ra_allock(as, ~((int64_t)~irt_toitype(t) << 47) , allow));
+ } else {
+ if (irt_isnum(t)) {
+ asm_guard(as, MIPSI_BEQ, RID_TMP, RID_ZERO);
+ emit_tsi(as, MIPSI_SLTIU, RID_TMP, RID_TMP, (int32_t)LJ_TISNUM);
+ if (ra_hasreg(dest))
+ emit_hsi(as, MIPSI_LDC1, dest, base, ofs);
+ } else {
+ asm_guard(as, MIPSI_BNE, RID_TMP,
+ ra_allock(as, (int32_t)irt_toitype(t), allow));
+ }
+ emit_dta(as, MIPSI_DSRA32, RID_TMP, type, 15);
+ }
+ emit_tsi(as, MIPSI_LD, type, base, ofs);
+ } else if (ra_hasreg(dest)) {
+ if (irt_isnum(t))
+ emit_hsi(as, MIPSI_LDC1, dest, base, ofs);
+ else
+ emit_tsi(as, irt_isint(t) ? MIPSI_LW : MIPSI_LD, dest, base,
+ ofs ^ ((LJ_BE && irt_isint(t)) ? 4 : 0));
+ }
+#endif
}
/* -- Allocations --------------------------------------------------------- */
@@ -1116,8 +1443,8 @@ static void asm_cnew(ASMState *as, IRIns *ir)
/* Initialize immutable cdata object. */
if (ir->o == IR_CNEWI) {
RegSet allow = (RSET_GPR & ~RSET_SCRATCH);
+#if LJ_32
int32_t ofs = sizeof(GCcdata);
- lua_assert(sz == 4 || sz == 8);
if (sz == 8) {
ofs += 4;
lua_assert((ir+1)->o == IR_HIOP);
@@ -1130,6 +1457,11 @@ static void asm_cnew(ASMState *as, IRIns *ir)
if (ofs == sizeof(GCcdata)) break;
ofs -= 4; if (LJ_BE) ir++; else ir--;
}
+#else
+ emit_tsi(as, MIPSI_SD, ra_alloc1(as, ir->op2, allow),
+ RID_RET, sizeof(GCcdata));
+#endif
+ lua_assert(sz == 4 || sz == 8);
} else if (ir->op2 != REF_NIL) { /* Create VLA/VLS/aligned cdata. */
ci = &lj_ir_callinfo[IRCALL_lj_cdata_newv];
args[0] = ASMREF_L; /* lua_State *L */
@@ -1164,7 +1496,7 @@ static void asm_tbar(ASMState *as, IRIns *ir)
Reg mark = ra_scratch(as, rset_exclude(RSET_GPR, tab));
Reg link = RID_TMP;
MCLabel l_end = emit_label(as);
- emit_tsi(as, MIPSI_SW, link, tab, (int32_t)offsetof(GCtab, gclist));
+ emit_tsi(as, MIPSI_AS, link, tab, (int32_t)offsetof(GCtab, gclist));
emit_tsi(as, MIPSI_SB, mark, tab, (int32_t)offsetof(GCtab, marked));
emit_setgl(as, tab, gc.grayagain);
emit_getgl(as, link, gc.grayagain);
@@ -1187,7 +1519,7 @@ static void asm_obar(ASMState *as, IRIns *ir)
args[0] = ASMREF_TMP1; /* global_State *g */
args[1] = ir->op1; /* TValue *tv */
asm_gencall(as, ci, args);
- emit_tsi(as, MIPSI_ADDIU, ra_releasetmp(as, ASMREF_TMP1), RID_JGL, -32768);
+ emit_tsi(as, MIPSI_AADDIU, ra_releasetmp(as, ASMREF_TMP1), RID_JGL, -32768);
obj = IR(ir->op1)->r;
tmp = ra_scratch(as, rset_exclude(RSET_GPR, obj));
emit_branch(as, MIPSI_BEQ, RID_TMP, RID_ZERO, l_end);
@@ -1233,8 +1565,9 @@ static void asm_fpmath(ASMState *as, IRIns *ir)
static void asm_add(ASMState *as, IRIns *ir)
{
+ IRType1 t = ir->t;
#if !LJ_SOFTFP
- if (irt_isnum(ir->t)) {
+ if (irt_isnum(t)) {
asm_fparith(as, ir, MIPSI_ADD_D);
} else
#endif
@@ -1242,14 +1575,16 @@ static void asm_add(ASMState *as, IRIns *ir)
Reg dest = ra_dest(as, ir, RSET_GPR);
Reg right, left = ra_hintalloc(as, ir->op1, dest, RSET_GPR);
if (irref_isk(ir->op2)) {
- int32_t k = IR(ir->op2)->i;
+ intptr_t k = get_kval(IR(ir->op2));
if (checki16(k)) {
- emit_tsi(as, MIPSI_ADDIU, dest, left, k);
+ emit_tsi(as, (LJ_64 && irt_is64(t)) ? MIPSI_DADDIU : MIPSI_ADDIU, dest,
+ left, k);
return;
}
}
right = ra_alloc1(as, ir->op2, rset_exclude(RSET_GPR, left));
- emit_dst(as, MIPSI_ADDU, dest, left, right);
+ emit_dst(as, (LJ_64 && irt_is64(t)) ? MIPSI_DADDU : MIPSI_ADDU, dest,
+ left, right);
}
}
@@ -1264,7 +1599,8 @@ static void asm_sub(ASMState *as, IRIns *ir)
Reg dest = ra_dest(as, ir, RSET_GPR);
Reg right, left = ra_alloc2(as, ir, RSET_GPR);
right = (left >> 8); left &= 255;
- emit_dst(as, MIPSI_SUBU, dest, left, right);
+ emit_dst(as, (LJ_64 && irt_is64(ir->t)) ? MIPSI_DSUBU : MIPSI_SUBU, dest,
+ left, right);
}
}
@@ -1279,13 +1615,49 @@ static void asm_mul(ASMState *as, IRIns *ir)
Reg dest = ra_dest(as, ir, RSET_GPR);
Reg right, left = ra_alloc2(as, ir, RSET_GPR);
right = (left >> 8); left &= 255;
- emit_dst(as, MIPSI_MUL, dest, left, right);
+ if (LJ_64 && irt_is64(ir->t)) {
+ emit_dst(as, MIPSI_MFLO, dest, 0, 0);
+ emit_dst(as, MIPSI_DMULT, 0, left, right);
+ } else {
+ emit_dst(as, MIPSI_MUL, dest, left, right);
+ }
}
}
-#define asm_div(as, ir) asm_fparith(as, ir, MIPSI_DIV_D)
-#define asm_mod(as, ir) asm_callid(as, ir, IRCALL_lj_vm_modi)
-#define asm_pow(as, ir) asm_callid(as, ir, IRCALL_lj_vm_powi)
+static void asm_mod(ASMState *as, IRIns *ir)
+{
+#if LJ_64 && LJ_HASFFI
+ if (!irt_isint(ir->t))
+ asm_callid(as, ir, irt_isi64(ir->t) ? IRCALL_lj_carith_modi64 :
+ IRCALL_lj_carith_modu64);
+ else
+#endif
+ asm_callid(as, ir, IRCALL_lj_vm_modi);
+}
+
+#if !LJ_SOFTFP
+static void asm_pow(ASMState *as, IRIns *ir)
+{
+#if LJ_64 && LJ_HASFFI
+ if (!irt_isnum(ir->t))
+ asm_callid(as, ir, irt_isi64(ir->t) ? IRCALL_lj_carith_powi64 :
+ IRCALL_lj_carith_powu64);
+ else
+#endif
+ asm_callid(as, ir, IRCALL_lj_vm_powi);
+}
+
+static void asm_div(ASMState *as, IRIns *ir)
+{
+#if LJ_64 && LJ_HASFFI
+ if (!irt_isnum(ir->t))
+ asm_callid(as, ir, irt_isi64(ir->t) ? IRCALL_lj_carith_divi64 :
+ IRCALL_lj_carith_divu64);
+ else
+#endif
+ asm_fparith(as, ir, MIPSI_DIV_D);
+}
+#endif
static void asm_neg(ASMState *as, IRIns *ir)
{
@@ -1297,7 +1669,8 @@ static void asm_neg(ASMState *as, IRIns *ir)
{
Reg dest = ra_dest(as, ir, RSET_GPR);
Reg left = ra_hintalloc(as, ir->op1, dest, RSET_GPR);
- emit_dst(as, MIPSI_SUBU, dest, RID_ZERO, left);
+ emit_dst(as, (LJ_64 && irt_is64(ir->t)) ? MIPSI_DSUBU : MIPSI_SUBU, dest,
+ RID_ZERO, left);
}
}
@@ -1308,6 +1681,7 @@ static void asm_neg(ASMState *as, IRIns *ir)
static void asm_arithov(ASMState *as, IRIns *ir)
{
Reg right, left, tmp, dest = ra_dest(as, ir, RSET_GPR);
+ lua_assert(!irt_is64(ir->t));
if (irref_isk(ir->op2)) {
int k = IR(ir->op2)->i;
if (ir->o == IR_SUBOV) k = -k;
@@ -1355,7 +1729,7 @@ static void asm_mulov(ASMState *as, IRIns *ir)
emit_dst(as, MIPSI_MULT, 0, left, right);
}
-#if LJ_HASFFI
+#if LJ_32 && LJ_HASFFI
static void asm_add64(ASMState *as, IRIns *ir)
{
Reg dest = ra_dest(as, ir, RSET_GPR);
@@ -1457,7 +1831,8 @@ static void asm_bswap(ASMState *as, IRIns *ir)
{
Reg dest = ra_dest(as, ir, RSET_GPR);
Reg left = ra_alloc1(as, ir->op1, RSET_GPR);
- if ((as->flags & JIT_F_MIPS32R2)) {
+#if LJ_32
+ if ((as->flags & JIT_F_MIPSXXR2)) {
emit_dta(as, MIPSI_ROTR, dest, RID_TMP, 16);
emit_dst(as, MIPSI_WSBH, RID_TMP, 0, left);
} else {
@@ -1472,6 +1847,15 @@ static void asm_bswap(ASMState *as, IRIns *ir)
emit_dta(as, MIPSI_SRL, tmp, left, 24);
emit_dta(as, MIPSI_SLL, RID_TMP, left, 24);
}
+#else
+ if (irt_is64(ir->t)) {
+ emit_dst(as, MIPSI_DSHD, dest, 0, RID_TMP);
+ emit_dst(as, MIPSI_DSBH, RID_TMP, 0, left);
+ } else {
+ emit_dta(as, MIPSI_ROTR, dest, RID_TMP, 16);
+ emit_dst(as, MIPSI_WSBH, RID_TMP, 0, left);
+ }
+#endif
}
static void asm_bitop(ASMState *as, IRIns *ir, MIPSIns mi, MIPSIns mik)
@@ -1479,7 +1863,7 @@ static void asm_bitop(ASMState *as, IRIns *ir, MIPSIns mi, MIPSIns mik)
Reg dest = ra_dest(as, ir, RSET_GPR);
Reg right, left = ra_hintalloc(as, ir->op1, dest, RSET_GPR);
if (irref_isk(ir->op2)) {
- int32_t k = IR(ir->op2)->i;
+ intptr_t k = get_kval(IR(ir->op2));
if (checku16(k)) {
emit_tsi(as, mik, dest, left, k);
return;
@@ -1497,11 +1881,14 @@ static void asm_bitshift(ASMState *as, IRIns *ir, MIPSIns mi, MIPSIns mik)
{
Reg dest = ra_dest(as, ir, RSET_GPR);
if (irref_isk(ir->op2)) { /* Constant shifts. */
- uint32_t shift = (uint32_t)(IR(ir->op2)->i & 31);
- emit_dta(as, mik, dest, ra_hintalloc(as, ir->op1, dest, RSET_GPR), shift);
+ uint32_t shift = (uint32_t)IR(ir->op2)->i;
+ if (LJ_64 && irt_is64(ir->t)) mik |= (shift & 32) ? MIPSI_D32 : MIPSI_D;
+ emit_dta(as, mik, dest, ra_hintalloc(as, ir->op1, dest, RSET_GPR),
+ (shift & 31));
} else {
Reg right, left = ra_alloc2(as, ir, RSET_GPR);
right = (left >> 8); left &= 255;
+ if (LJ_64 && irt_is64(ir->t)) mi |= MIPSI_DV;
emit_dst(as, mi, dest, right, left); /* Shift amount is in rs. */
}
}
@@ -1513,7 +1900,7 @@ static void asm_bitshift(ASMState *as, IRIns *ir, MIPSIns mi, MIPSIns mik)
static void asm_bror(ASMState *as, IRIns *ir)
{
- if ((as->flags & JIT_F_MIPS32R2)) {
+ if (LJ_64 || (as->flags & JIT_F_MIPSXXR2)) {
asm_bitshift(as, ir, MIPSI_ROTRV, MIPSI_ROTR);
} else {
Reg dest = ra_dest(as, ir, RSET_GPR);
@@ -1532,7 +1919,7 @@ static void asm_bror(ASMState *as, IRIns *ir)
}
}
-#if LJ_SOFTFP
+#if LJ_32 && LJ_SOFTFP
static void asm_sfpmin_max(ASMState *as, IRIns *ir)
{
CCallInfo ci = lj_ir_callinfo[(IROp)ir->o == IR_MIN ? IRCALL_lj_vm_sfmin : IRCALL_lj_vm_sfmax];
@@ -1581,7 +1968,7 @@ static void asm_min_max(ASMState *as, IRIns *ir, int ismax)
/* -- Comparisons --------------------------------------------------------- */
-#if LJ_SOFTFP
+#if LJ_32 && LJ_SOFTFP
/* SFP comparisons. */
static void asm_sfpcomp(ASMState *as, IRIns *ir)
{
@@ -1654,13 +2041,13 @@ static void asm_comp(ASMState *as, IRIns *ir)
} else {
Reg right, left = ra_alloc1(as, ir->op1, RSET_GPR);
if (op == IR_ABC) op = IR_UGT;
- if ((op&4) == 0 && irref_isk(ir->op2) && IR(ir->op2)->i == 0) {
+ if ((op&4) == 0 && irref_isk(ir->op2) && get_kval(IR(ir->op2)) == 0) {
MIPSIns mi = (op&2) ? ((op&1) ? MIPSI_BLEZ : MIPSI_BGTZ) :
((op&1) ? MIPSI_BLTZ : MIPSI_BGEZ);
asm_guard(as, mi, left, 0);
} else {
if (irref_isk(ir->op2)) {
- int32_t k = IR(ir->op2)->i;
+ intptr_t k = get_kval(IR(ir->op2));
if ((op&2)) k++;
if (checki16(k)) {
asm_guard(as, (op&1) ? MIPSI_BNE : MIPSI_BEQ, RID_TMP, RID_ZERO);
@@ -1679,7 +2066,8 @@ static void asm_comp(ASMState *as, IRIns *ir)
static void asm_equal(ASMState *as, IRIns *ir)
{
- Reg right, left = ra_alloc2(as, ir, (!LJ_SOFTFP && irt_isnum(ir->t)) ? RSET_FPR : RSET_GPR);
+ Reg right, left = ra_alloc2(as, ir, (!LJ_SOFTFP && irt_isnum(ir->t)) ?
+ RSET_FPR : RSET_GPR);
right = (left >> 8); left &= 255;
if (!LJ_SOFTFP && irt_isnum(ir->t)) {
asm_guard(as, (ir->o & 1) ? MIPSI_BC1T : MIPSI_BC1F, 0, 0);
@@ -1689,7 +2077,7 @@ static void asm_equal(ASMState *as, IRIns *ir)
}
}
-#if LJ_HASFFI
+#if LJ_32 && LJ_HASFFI
/* 64 bit integer comparisons. */
static void asm_comp64(ASMState *as, IRIns *ir)
{
@@ -1731,7 +2119,7 @@ static void asm_comp64eq(ASMState *as, IRIns *ir)
/* Hiword op of a split 64 bit op. Previous op must be the loword op. */
static void asm_hiop(ASMState *as, IRIns *ir)
{
-#if LJ_HASFFI || LJ_SOFTFP
+#if LJ_32 && (LJ_HASFFI || LJ_SOFTFP)
/* HIOP is marked as a store because it needs its own DCE logic. */
int uselo = ra_used(ir-1), usehi = ra_used(ir); /* Loword/hiword used? */
if (LJ_UNLIKELY(!(as->flags & JIT_F_OPT_DCE))) uselo = usehi = 1;
@@ -1835,36 +2223,42 @@ static void asm_stack_check(ASMState *as, BCReg topslot,
Reg tmp, pbase = irp ? (ra_hasreg(irp->r) ? irp->r : RID_TMP) : RID_BASE;
ExitNo oldsnap = as->snapno;
rset_clear(allow, pbase);
+#if LJ_32
tmp = allow ? rset_pickbot(allow) :
(pbase == RID_RETHI ? RID_RETLO : RID_RETHI);
+#else
+ tmp = allow ? rset_pickbot(allow) : RID_RET;
+#endif
as->snapno = exitno;
asm_guard(as, MIPSI_BNE, RID_TMP, RID_ZERO);
as->snapno = oldsnap;
if (allow == RSET_EMPTY) /* Restore temp. register. */
- emit_tsi(as, MIPSI_LW, tmp, RID_SP, 0);
+ emit_tsi(as, MIPSI_AL, tmp, RID_SP, 0);
else
ra_modified(as, tmp);
emit_tsi(as, MIPSI_SLTIU, RID_TMP, RID_TMP, (int32_t)(8*topslot));
- emit_dst(as, MIPSI_SUBU, RID_TMP, tmp, pbase);
- emit_tsi(as, MIPSI_LW, tmp, tmp, offsetof(lua_State, maxstack));
+ emit_dst(as, MIPSI_ASUBU, RID_TMP, tmp, pbase);
+ emit_tsi(as, MIPSI_AL, tmp, tmp, offsetof(lua_State, maxstack));
if (pbase == RID_TMP)
emit_getgl(as, RID_TMP, jit_base);
emit_getgl(as, tmp, cur_L);
if (allow == RSET_EMPTY) /* Spill temp. register. */
- emit_tsi(as, MIPSI_SW, tmp, RID_SP, 0);
+ emit_tsi(as, MIPSI_AS, tmp, RID_SP, 0);
}
/* Restore Lua stack from on-trace state. */
static void asm_stack_restore(ASMState *as, SnapShot *snap)
{
SnapEntry *map = &as->T->snapmap[snap->mapofs];
- SnapEntry *flinks = &as->T->snapmap[snap_nextofs(as->T, snap)-1];
+#if LJ_32 || defined(LUA_USE_ASSERT)
+ SnapEntry *flinks = &as->T->snapmap[snap_nextofs(as->T, snap)-1-LJ_FR2];
+#endif
MSize n, nent = snap->nent;
/* Store the value of all modified slots to the Lua stack. */
for (n = 0; n < nent; n++) {
SnapEntry sn = map[n];
BCReg s = snap_slot(sn);
- int32_t ofs = 8*((int32_t)s-1);
+ int32_t ofs = 8*((int32_t)s-1-LJ_FR2);
IRRef ref = snap_ref(sn);
IRIns *ir = IR(ref);
if ((sn & SNAP_NORESTORE))
@@ -1884,8 +2278,9 @@ static void asm_stack_restore(ASMState *as, SnapShot *snap)
emit_hsi(as, MIPSI_SDC1, src, RID_BASE, ofs);
#endif
} else {
- Reg type;
+#if LJ_32
RegSet allow = rset_exclude(RSET_GPR, RID_BASE);
+ Reg type;
lua_assert(irt_ispri(ir->t) || irt_isaddr(ir->t) || irt_isinteger(ir->t));
if (!irt_ispri(ir->t)) {
Reg src = ra_alloc1(as, ref, allow);
@@ -1903,6 +2298,9 @@ static void asm_stack_restore(ASMState *as, SnapShot *snap)
type = ra_allock(as, (int32_t)irt_toitype(ir->t), allow);
}
emit_tsi(as, MIPSI_SW, type, RID_BASE, ofs+(LJ_BE?0:4));
+#else
+ asm_tvstore64(as, RID_BASE, ofs, ref);
+#endif
}
checkmclim(as);
}
@@ -1926,7 +2324,7 @@ static void asm_gc_check(ASMState *as)
args[0] = ASMREF_TMP1; /* global_State *g */
args[1] = ASMREF_TMP2; /* MSize steps */
asm_gencall(as, ci, args);
- emit_tsi(as, MIPSI_ADDIU, ra_releasetmp(as, ASMREF_TMP1), RID_JGL, -32768);
+ emit_tsi(as, MIPSI_AADDIU, ra_releasetmp(as, ASMREF_TMP1), RID_JGL, -32768);
tmp = ra_releasetmp(as, ASMREF_TMP2);
emit_loadi(as, tmp, as->gcsteps);
/* Jump around GC step if GC total < GC threshold. */
@@ -2001,7 +2399,7 @@ static void asm_tail_fixup(ASMState *as, TraceNo lnk)
MCode *target = lnk ? traceref(as->J,lnk)->mcode : (MCode *)lj_vm_exit_interp;
int32_t spadj = as->T->spadjust;
MCode *p = as->mctop-1;
- *p = spadj ? (MIPSI_ADDIU|MIPSF_T(RID_SP)|MIPSF_S(RID_SP)|spadj) : MIPSI_NOP;
+ *p = spadj ? (MIPSI_AADDIU|MIPSF_T(RID_SP)|MIPSF_S(RID_SP)|spadj) : MIPSI_NOP;
p[-1] = MIPSI_J|(((uintptr_t)target>>2)&0x03ffffffu);
}
@@ -2019,9 +2417,14 @@ static Reg asm_setup_call_slots(ASMState *as, IRIns *ir, const CCallInfo *ci)
{
IRRef args[CCI_NARGS_MAX*2];
uint32_t i, nargs = CCI_XNARGS(ci);
+#if LJ_32
int nslots = 4, ngpr = REGARG_NUMGPR, nfpr = REGARG_NUMFPR;
+#else
+ int nslots = 0, ngpr = REGARG_NUMGPR;
+#endif
asm_collectargs(as, ir, ci, args);
for (i = 0; i < nargs; i++) {
+#if LJ_32
if (!LJ_SOFTFP && args[i] && irt_isfp(IR(args[i])->t) &&
nfpr > 0 && !(ci->flags & CCI_VARARG)) {
nfpr--;
@@ -2034,6 +2437,9 @@ static Reg asm_setup_call_slots(ASMState *as, IRIns *ir, const CCallInfo *ci)
nfpr = 0;
if (ngpr > 0) ngpr--; else nslots++;
}
+#else
+ if (ngpr > 0) ngpr--; else nslots += 2;
+#endif
}
if (nslots > as->evenspill) /* Leave room for args in stack slots. */
as->evenspill = nslots;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_ppc.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_ppc.h
index 0ebed40fdaa..6daa861b91a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_ppc.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_ppc.h
@@ -1,6 +1,6 @@
/*
** PPC IR assembler (SSA IR -> machine code).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/* -- Register allocator extensions --------------------------------------- */
@@ -393,8 +393,7 @@ static void asm_tointg(ASMState *as, IRIns *ir, Reg left)
emit_asi(as, PPCI_XORIS, RID_TMP, dest, 0x8000);
emit_tai(as, PPCI_LWZ, dest, RID_SP, SPOFS_TMPLO);
emit_lsptr(as, PPCI_LFS, (fbias & 31),
- (void *)lj_ir_k64_find(as->J, U64x(59800004,59800000)),
- RSET_GPR);
+ (void *)&as->J->k32[LJ_K32_2P52_2P31], RSET_GPR);
emit_fai(as, PPCI_STFD, tmp, RID_SP, SPOFS_TMP);
emit_fb(as, PPCI_FCTIWZ, tmp, left);
}
@@ -433,13 +432,11 @@ static void asm_conv(ASMState *as, IRIns *ir)
Reg left = ra_alloc1(as, lref, allow);
Reg hibias = ra_allock(as, 0x43300000, rset_clear(allow, left));
Reg fbias = ra_scratch(as, rset_exclude(RSET_FPR, dest));
- const float *kbias;
if (irt_isfloat(ir->t)) emit_fb(as, PPCI_FRSP, dest, dest);
emit_fab(as, PPCI_FSUB, dest, dest, fbias);
emit_fai(as, PPCI_LFD, dest, RID_SP, SPOFS_TMP);
- kbias = (const float *)lj_ir_k64_find(as->J, U64x(59800004,59800000));
- if (st == IRT_U32) kbias++;
- emit_lsptr(as, PPCI_LFS, (fbias & 31), (void *)kbias,
+ emit_lsptr(as, PPCI_LFS, (fbias & 31),
+ &as->J->k32[st == IRT_U32 ? LJ_K32_2P52 : LJ_K32_2P52_2P31],
rset_clear(allow, hibias));
emit_tai(as, PPCI_STW, st == IRT_U32 ? left : RID_TMP,
RID_SP, SPOFS_TMPLO);
@@ -472,8 +469,7 @@ static void asm_conv(ASMState *as, IRIns *ir)
emit_fb(as, PPCI_FCTIWZ, tmp, tmp);
emit_fab(as, PPCI_FSUB, tmp, left, tmp);
emit_lsptr(as, PPCI_LFS, (tmp & 31),
- (void *)lj_ir_k64_find(as->J, U64x(4f000000,00000000)),
- RSET_GPR);
+ (void *)&as->J->k32[LJ_K32_2P31], RSET_GPR);
} else {
emit_tai(as, PPCI_LWZ, dest, RID_SP, SPOFS_TMPLO);
emit_fai(as, PPCI_STFD, tmp, RID_SP, SPOFS_TMP);
@@ -717,7 +713,6 @@ static void asm_hrefk(ASMState *as, IRIns *ir)
static void asm_uref(ASMState *as, IRIns *ir)
{
- /* NYI: Check that UREFO is still open and not aliasing a slot. */
Reg dest = ra_dest(as, ir, RSET_GPR);
if (irref_isk(ir->op1)) {
GCfunc *fn = ir_kfunc(IR(ir->op1));
@@ -809,17 +804,23 @@ static PPCIns asm_fxstoreins(IRIns *ir)
static void asm_fload(ASMState *as, IRIns *ir)
{
Reg dest = ra_dest(as, ir, RSET_GPR);
- Reg idx = ra_alloc1(as, ir->op1, RSET_GPR);
PPCIns pi = asm_fxloadins(ir);
+ Reg idx;
int32_t ofs;
- if (ir->op2 == IRFL_TAB_ARRAY) {
- ofs = asm_fuseabase(as, ir->op1);
- if (ofs) { /* Turn the t->array load into an add for colocated arrays. */
- emit_tai(as, PPCI_ADDI, dest, idx, ofs);
- return;
+ if (ir->op1 == REF_NIL) {
+ idx = RID_JGL;
+ ofs = (ir->op2 << 2) - 32768;
+ } else {
+ idx = ra_alloc1(as, ir->op1, RSET_GPR);
+ if (ir->op2 == IRFL_TAB_ARRAY) {
+ ofs = asm_fuseabase(as, ir->op1);
+ if (ofs) { /* Turn the t->array load into an add for colocated arrays. */
+ emit_tai(as, PPCI_ADDI, dest, idx, ofs);
+ return;
+ }
}
+ ofs = field_ofs[ir->op2];
}
- ofs = field_ofs[ir->op2];
lua_assert(!irt_isi8(ir->t));
emit_tai(as, pi, dest, idx, ofs);
}
@@ -975,7 +976,7 @@ static void asm_sload(ASMState *as, IRIns *ir)
emit_fab(as, PPCI_FSUB, dest, dest, fbias);
emit_fai(as, PPCI_LFD, dest, RID_SP, SPOFS_TMP);
emit_lsptr(as, PPCI_LFS, (fbias & 31),
- (void *)lj_ir_k64_find(as->J, U64x(59800004,59800000)),
+ (void *)&as->J->k32[LJ_K32_2P52_2P31],
rset_clear(allow, hibias));
emit_tai(as, PPCI_STW, tmp, RID_SP, SPOFS_TMPLO);
emit_tai(as, PPCI_STW, hibias, RID_SP, SPOFS_TMPHI);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_x86.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_x86.h
index 512e05346e6..3e189b1d04e 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_x86.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_x86.h
@@ -1,6 +1,6 @@
/*
** x86/x64 IR assembler (SSA IR -> machine code).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/* -- Guard handling ------------------------------------------------------ */
@@ -21,12 +21,14 @@ static MCode *asm_exitstub_gen(ASMState *as, ExitNo group)
}
/* Push the high byte of the exitno for each exit stub group. */
*mxp++ = XI_PUSHi8; *mxp++ = (MCode)((group*EXITSTUBS_PER_GROUP)>>8);
+#if !LJ_GC64
/* Store DISPATCH at original stack slot 0. Account for the two push ops. */
*mxp++ = XI_MOVmi;
*mxp++ = MODRM(XM_OFS8, 0, RID_ESP);
*mxp++ = MODRM(XM_SCALE1, RID_ESP, RID_ESP);
*mxp++ = 2*sizeof(void *);
*(int32_t *)mxp = ptr2addr(J2GG(as->J)->dispatch); mxp += 4;
+#endif
/* Jump to exit handler which fills in the ExitState. */
*mxp++ = XI_JMP; mxp += 4;
*((int32_t *)(mxp-4)) = jmprel(mxp, (MCode *)(void *)lj_vm_exit_handler);
@@ -62,10 +64,14 @@ static void asm_guardcc(ASMState *as, int cc)
target = p;
cc ^= 1;
if (as->realign) {
+ if (LJ_GC64 && LJ_UNLIKELY(as->mrm.base == RID_RIP))
+ as->mrm.ofs += 2; /* Fixup RIP offset for pending fused load. */
emit_sjcc(as, cc, target);
return;
}
}
+ if (LJ_GC64 && LJ_UNLIKELY(as->mrm.base == RID_RIP))
+ as->mrm.ofs += 6; /* Fixup RIP offset for pending fused load. */
emit_jcc(as, cc, target);
}
@@ -79,6 +85,15 @@ static int asm_isk32(ASMState *as, IRRef ref, int32_t *k)
{
if (irref_isk(ref)) {
IRIns *ir = IR(ref);
+#if LJ_GC64
+ if (ir->o == IR_KNULL || !irt_is64(ir->t)) {
+ *k = ir->i;
+ return 1;
+ } else if (checki32((int64_t)ir_k64(ir)->u64)) {
+ *k = (int32_t)ir_k64(ir)->u64;
+ return 1;
+ }
+#else
if (ir->o != IR_KINT64) {
*k = ir->i;
return 1;
@@ -86,6 +101,7 @@ static int asm_isk32(ASMState *as, IRRef ref, int32_t *k)
*k = (int32_t)ir_kint64(ir)->u64;
return 1;
}
+#endif
}
return 0;
}
@@ -185,9 +201,19 @@ static void asm_fuseahuref(ASMState *as, IRRef ref, RegSet allow)
if (irref_isk(ir->op1)) {
GCfunc *fn = ir_kfunc(IR(ir->op1));
GCupval *uv = &gcref(fn->l.uvptr[(ir->op2 >> 8)])->uv;
+#if LJ_GC64
+ int64_t ofs = dispofs(as, &uv->tv);
+ if (checki32(ofs) && checki32(ofs+4)) {
+ as->mrm.ofs = (int32_t)ofs;
+ as->mrm.base = RID_DISPATCH;
+ as->mrm.idx = RID_NONE;
+ return;
+ }
+#else
as->mrm.ofs = ptr2addr(&uv->tv);
as->mrm.base = as->mrm.idx = RID_NONE;
return;
+#endif
}
break;
default:
@@ -205,14 +231,40 @@ static void asm_fuseahuref(ASMState *as, IRRef ref, RegSet allow)
static void asm_fusefref(ASMState *as, IRIns *ir, RegSet allow)
{
lua_assert(ir->o == IR_FLOAD || ir->o == IR_FREF);
- as->mrm.ofs = field_ofs[ir->op2];
as->mrm.idx = RID_NONE;
+ if (ir->op1 == REF_NIL) {
+#if LJ_GC64
+ as->mrm.ofs = (int32_t)(ir->op2 << 2) - GG_OFS(dispatch);
+ as->mrm.base = RID_DISPATCH;
+#else
+ as->mrm.ofs = (int32_t)(ir->op2 << 2) + ptr2addr(J2GG(as->J));
+ as->mrm.base = RID_NONE;
+#endif
+ return;
+ }
+ as->mrm.ofs = field_ofs[ir->op2];
if (irref_isk(ir->op1)) {
- as->mrm.ofs += IR(ir->op1)->i;
+ IRIns *op1 = IR(ir->op1);
+#if LJ_GC64
+ if (ir->op1 == REF_NIL) {
+ as->mrm.ofs -= GG_OFS(dispatch);
+ as->mrm.base = RID_DISPATCH;
+ return;
+ } else if (op1->o == IR_KPTR || op1->o == IR_KKPTR) {
+ intptr_t ofs = dispofs(as, ir_kptr(op1));
+ if (checki32(as->mrm.ofs + ofs)) {
+ as->mrm.ofs += (int32_t)ofs;
+ as->mrm.base = RID_DISPATCH;
+ return;
+ }
+ }
+#else
+ as->mrm.ofs += op1->i;
as->mrm.base = RID_NONE;
- } else {
- as->mrm.base = (uint8_t)ra_alloc1(as, ir->op1, allow);
+ return;
+#endif
}
+ as->mrm.base = (uint8_t)ra_alloc1(as, ir->op1, allow);
}
/* Fuse string reference into memory operand. */
@@ -223,7 +275,7 @@ static void asm_fusestrref(ASMState *as, IRIns *ir, RegSet allow)
as->mrm.base = as->mrm.idx = RID_NONE;
as->mrm.scale = XM_SCALE1;
as->mrm.ofs = sizeof(GCstr);
- if (irref_isk(ir->op1)) {
+ if (!LJ_GC64 && irref_isk(ir->op1)) {
as->mrm.ofs += IR(ir->op1)->i;
} else {
Reg r = ra_alloc1(as, ir->op1, allow);
@@ -255,10 +307,20 @@ static void asm_fusexref(ASMState *as, IRRef ref, RegSet allow)
IRIns *ir = IR(ref);
as->mrm.idx = RID_NONE;
if (ir->o == IR_KPTR || ir->o == IR_KKPTR) {
+#if LJ_GC64
+ intptr_t ofs = dispofs(as, ir_kptr(ir));
+ if (checki32(ofs)) {
+ as->mrm.ofs = (int32_t)ofs;
+ as->mrm.base = RID_DISPATCH;
+ return;
+ }
+ } if (0) {
+#else
as->mrm.ofs = ir->i;
as->mrm.base = RID_NONE;
} else if (ir->o == IR_STRREF) {
asm_fusestrref(as, ir, allow);
+#endif
} else {
as->mrm.ofs = 0;
if (canfuse(as, ir) && ir->o == IR_ADD && ra_noreg(ir->r)) {
@@ -301,7 +363,45 @@ static void asm_fusexref(ASMState *as, IRRef ref, RegSet allow)
}
}
-/* Fuse load into memory operand. */
+/* Fuse load of 64 bit IR constant into memory operand. */
+static Reg asm_fuseloadk64(ASMState *as, IRIns *ir)
+{
+ const uint64_t *k = &ir_k64(ir)->u64;
+ if (!LJ_GC64 || checki32((intptr_t)k)) {
+ as->mrm.ofs = ptr2addr(k);
+ as->mrm.base = RID_NONE;
+#if LJ_GC64
+ } else if (checki32(dispofs(as, k))) {
+ as->mrm.ofs = (int32_t)dispofs(as, k);
+ as->mrm.base = RID_DISPATCH;
+ } else if (checki32(mcpofs(as, k)) && checki32(mcpofs(as, k+1)) &&
+ checki32(mctopofs(as, k)) && checki32(mctopofs(as, k+1))) {
+ as->mrm.ofs = (int32_t)mcpofs(as, k);
+ as->mrm.base = RID_RIP;
+ } else {
+ if (ir->i) {
+ lua_assert(*k == *(uint64_t*)(as->mctop - ir->i));
+ } else {
+ while ((uintptr_t)as->mcbot & 7) *as->mcbot++ = XI_INT3;
+ *(uint64_t*)as->mcbot = *k;
+ ir->i = (int32_t)(as->mctop - as->mcbot);
+ as->mcbot += 8;
+ as->mclim = as->mcbot + MCLIM_REDZONE;
+ }
+ as->mrm.ofs = (int32_t)mcpofs(as, as->mctop - ir->i);
+ as->mrm.base = RID_RIP;
+#endif
+ }
+ as->mrm.idx = RID_NONE;
+ return RID_MRM;
+}
+
+/* Fuse load into memory operand.
+**
+** Important caveat: this may emit RIP-relative loads! So don't place any
+** code emitters between this function and the use of its result.
+** The only permitted exception is asm_guardcc().
+*/
static Reg asm_fuseload(ASMState *as, IRRef ref, RegSet allow)
{
IRIns *ir = IR(ref);
@@ -320,26 +420,35 @@ static Reg asm_fuseload(ASMState *as, IRRef ref, RegSet allow)
if (ir->o == IR_KNUM) {
RegSet avail = as->freeset & ~as->modset & RSET_FPR;
lua_assert(allow != RSET_EMPTY);
- if (!(avail & (avail-1))) { /* Fuse if less than two regs available. */
- as->mrm.ofs = ptr2addr(ir_knum(ir));
- as->mrm.base = as->mrm.idx = RID_NONE;
- return RID_MRM;
- }
- } else if (ir->o == IR_KINT64) {
+ if (!(avail & (avail-1))) /* Fuse if less than two regs available. */
+ return asm_fuseloadk64(as, ir);
+ } else if (ref == REF_BASE || ir->o == IR_KINT64) {
RegSet avail = as->freeset & ~as->modset & RSET_GPR;
lua_assert(allow != RSET_EMPTY);
if (!(avail & (avail-1))) { /* Fuse if less than two regs available. */
- as->mrm.ofs = ptr2addr(ir_kint64(ir));
- as->mrm.base = as->mrm.idx = RID_NONE;
- return RID_MRM;
+ if (ref == REF_BASE) {
+#if LJ_GC64
+ as->mrm.ofs = (int32_t)dispofs(as, &J2G(as->J)->jit_base);
+ as->mrm.base = RID_DISPATCH;
+#else
+ as->mrm.ofs = ptr2addr(&J2G(as->J)->jit_base);
+ as->mrm.base = RID_NONE;
+#endif
+ as->mrm.idx = RID_NONE;
+ return RID_MRM;
+ } else {
+ return asm_fuseloadk64(as, ir);
+ }
}
} else if (mayfuse(as, ref)) {
RegSet xallow = (allow & RSET_GPR) ? allow : RSET_GPR;
if (ir->o == IR_SLOAD) {
if (!(ir->op2 & (IRSLOAD_PARENT|IRSLOAD_CONVERT)) &&
- noconflict(as, ref, IR_RETF, 0)) {
+ noconflict(as, ref, IR_RETF, 0) &&
+ !(LJ_GC64 && irt_isaddr(ir->t))) {
as->mrm.base = (uint8_t)ra_alloc1(as, REF_BASE, xallow);
- as->mrm.ofs = 8*((int32_t)ir->op1-1) + ((ir->op2&IRSLOAD_FRAME)?4:0);
+ as->mrm.ofs = 8*((int32_t)ir->op1-1-LJ_FR2) +
+ (!LJ_FR2 && (ir->op2 & IRSLOAD_FRAME) ? 4 : 0);
as->mrm.idx = RID_NONE;
return RID_MRM;
}
@@ -351,7 +460,8 @@ static Reg asm_fuseload(ASMState *as, IRRef ref, RegSet allow)
return RID_MRM;
}
} else if (ir->o == IR_ALOAD || ir->o == IR_HLOAD || ir->o == IR_ULOAD) {
- if (noconflict(as, ref, ir->o + IRDELTA_L2S, 0)) {
+ if (noconflict(as, ref, ir->o + IRDELTA_L2S, 0) &&
+ !(LJ_GC64 && irt_isaddr(ir->t))) {
asm_fuseahuref(as, ir->op1, xallow);
return RID_MRM;
}
@@ -364,12 +474,16 @@ static Reg asm_fuseload(ASMState *as, IRRef ref, RegSet allow)
asm_fusexref(as, ir->op1, xallow);
return RID_MRM;
}
- } else if (ir->o == IR_VLOAD) {
+ } else if (ir->o == IR_VLOAD && !(LJ_GC64 && irt_isaddr(ir->t))) {
asm_fuseahuref(as, ir->op1, xallow);
return RID_MRM;
}
}
- if (!(as->freeset & allow) && !irref_isk(ref) &&
+ if (ir->o == IR_FLOAD && ir->op1 == REF_NIL) {
+ asm_fusefref(as, ir, RSET_EMPTY);
+ return RID_MRM;
+ }
+ if (!(as->freeset & allow) && !emit_canremat(ref) &&
(allow == RSET_EMPTY || ra_hasspill(ir->s) || iscrossref(as, ref)))
goto fusespill;
return ra_allocref(as, ref, allow);
@@ -485,8 +599,8 @@ static void asm_gencall(ASMState *as, const CCallInfo *ci, IRRef *args)
if (r) { /* Argument is in a register. */
if (r < RID_MAX_GPR && ref < ASMREF_TMP1) {
#if LJ_64
- if (ir->o == IR_KINT64)
- emit_loadu64(as, r, ir_kint64(ir)->u64);
+ if (LJ_GC64 ? !(ir->o == IR_KINT || ir->o == IR_KNULL) : ir->o == IR_KINT64)
+ emit_loadu64(as, r, ir_k64(ir)->u64);
else
#endif
emit_loadi(as, r, ir->i);
@@ -642,6 +756,9 @@ static void asm_callx(ASMState *as, IRIns *ir)
static void asm_retf(ASMState *as, IRIns *ir)
{
Reg base = ra_alloc1(as, REF_BASE, RSET_GPR);
+#if LJ_FR2
+ Reg rpc = ra_scratch(as, rset_exclude(RSET_GPR, base));
+#endif
void *pc = ir_kptr(IR(ir->op2));
int32_t delta = 1+LJ_FR2+bc_a(*((const BCIns *)pc - 1));
as->topslot -= (BCReg)delta;
@@ -650,7 +767,12 @@ static void asm_retf(ASMState *as, IRIns *ir)
emit_setgl(as, base, jit_base);
emit_addptr(as, base, -8*delta);
asm_guardcc(as, CC_NE);
+#if LJ_FR2
+ emit_rmro(as, XO_CMP, rpc|REX_GC64, base, -8);
+ emit_loadu64(as, rpc, u64ptr(pc));
+#else
emit_gmroi(as, XG_ARITHi(XOg_CMP), base, -4, ptr2addr(pc));
+#endif
}
/* -- Type conversions ---------------------------------------------------- */
@@ -674,8 +796,9 @@ static void asm_tobit(ASMState *as, IRIns *ir)
Reg tmp = ra_noreg(IR(ir->op1)->r) ?
ra_alloc1(as, ir->op1, RSET_FPR) :
ra_scratch(as, RSET_FPR);
- Reg right = asm_fuseload(as, ir->op2, rset_exclude(RSET_FPR, tmp));
+ Reg right;
emit_rr(as, XO_MOVDto, tmp, dest);
+ right = asm_fuseload(as, ir->op2, rset_exclude(RSET_FPR, tmp));
emit_mrm(as, XO_ADDSD, tmp, right);
ra_left(as, tmp, ir->op1);
}
@@ -696,13 +819,13 @@ static void asm_conv(ASMState *as, IRIns *ir)
if (left == dest) return; /* Avoid the XO_XORPS. */
} else if (LJ_32 && st == IRT_U32) { /* U32 to FP conversion on x86. */
/* number = (2^52+2^51 .. u32) - (2^52+2^51) */
- cTValue *k = lj_ir_k64_find(as->J, U64x(43380000,00000000));
+ cTValue *k = &as->J->k64[LJ_K64_TOBIT];
Reg bias = ra_scratch(as, rset_exclude(RSET_FPR, dest));
if (irt_isfloat(ir->t))
emit_rr(as, XO_CVTSD2SS, dest, dest);
emit_rr(as, XO_SUBSD, dest, bias); /* Subtract 2^52+2^51 bias. */
emit_rr(as, XO_XORPS, dest, bias); /* Merge bias and integer. */
- emit_loadn(as, bias, k);
+ emit_rma(as, XO_MOVSD, bias, k);
emit_mrm(as, XO_MOVD, dest, asm_fuseload(as, lref, RSET_GPR));
return;
} else { /* Integer to FP conversion. */
@@ -711,7 +834,7 @@ static void asm_conv(ASMState *as, IRIns *ir)
asm_fuseloadm(as, lref, RSET_GPR, st64);
if (LJ_64 && st == IRT_U64) {
MCLabel l_end = emit_label(as);
- const void *k = lj_ir_k64_find(as->J, U64x(43f00000,00000000));
+ cTValue *k = &as->J->k64[LJ_K64_2P64];
emit_rma(as, XO_ADDSD, dest, k); /* Add 2^64 to compensate. */
emit_sjcc(as, CC_NS, l_end);
emit_rr(as, XO_TEST, left|REX_64, left); /* Check if u64 >= 2^63. */
@@ -738,23 +861,20 @@ static void asm_conv(ASMState *as, IRIns *ir)
emit_gri(as, XG_ARITHi(XOg_ADD), dest, (int32_t)0x80000000);
emit_rr(as, op, dest|REX_64, tmp);
if (st == IRT_NUM)
- emit_rma(as, XO_ADDSD, tmp, lj_ir_k64_find(as->J,
- LJ_64 ? U64x(c3f00000,00000000) : U64x(c1e00000,00000000)));
+ emit_rma(as, XO_ADDSD, tmp, &as->J->k64[LJ_K64_M2P64_31]);
else
- emit_rma(as, XO_ADDSS, tmp, lj_ir_k64_find(as->J,
- LJ_64 ? U64x(00000000,df800000) : U64x(00000000,cf000000)));
+ emit_rma(as, XO_ADDSS, tmp, &as->J->k32[LJ_K32_M2P64_31]);
emit_sjcc(as, CC_NS, l_end);
emit_rr(as, XO_TEST, dest|REX_64, dest); /* Check if dest negative. */
emit_rr(as, op, dest|REX_64, tmp);
ra_left(as, tmp, lref);
} else {
- Reg left = asm_fuseload(as, lref, RSET_FPR);
if (LJ_64 && irt_isu32(ir->t))
emit_rr(as, XO_MOV, dest, dest); /* Zero hiword. */
emit_mrm(as, op,
dest|((LJ_64 &&
(irt_is64(ir->t) || irt_isu32(ir->t))) ? REX_64 : 0),
- left);
+ asm_fuseload(as, lref, RSET_FPR));
}
}
} else if (st >= IRT_I8 && st <= IRT_U16) { /* Extend to 32 bit integer. */
@@ -828,8 +948,7 @@ static void asm_conv_fp_int64(ASMState *as, IRIns *ir)
if (((ir-1)->op2 & IRCONV_SRCMASK) == IRT_U64) {
/* For inputs in [2^63,2^64-1] add 2^64 to compensate. */
MCLabel l_end = emit_label(as);
- emit_rma(as, XO_FADDq, XOg_FADDq,
- lj_ir_k64_find(as->J, U64x(43f00000,00000000)));
+ emit_rma(as, XO_FADDq, XOg_FADDq, &as->J->k64[LJ_K64_2P64]);
emit_sjcc(as, CC_NS, l_end);
emit_rr(as, XO_TEST, hi, hi); /* Check if u64 >= 2^63. */
} else {
@@ -869,8 +988,7 @@ static void asm_conv_int64_fp(ASMState *as, IRIns *ir)
emit_rmro(as, XO_FISTTPq, XOg_FISTTPq, RID_ESP, 0);
else
emit_rmro(as, XO_FISTPq, XOg_FISTPq, RID_ESP, 0);
- emit_rma(as, XO_FADDq, XOg_FADDq,
- lj_ir_k64_find(as->J, U64x(c3f00000,00000000)));
+ emit_rma(as, XO_FADDq, XOg_FADDq, &as->J->k64[LJ_K64_M2P64]);
emit_sjcc(as, CC_NS, l_pop);
emit_rr(as, XO_TEST, hi, hi); /* Check if out-of-range (2^63). */
}
@@ -934,6 +1052,25 @@ static void asm_tvptr(ASMState *as, Reg dest, IRRef ref)
emit_rmro(as, XO_LEA, dest|REX_64, RID_ESP, ra_spill(as, ir));
} else {
/* Otherwise use g->tmptv to hold the TValue. */
+#if LJ_GC64
+ if (irref_isk(ref)) {
+ TValue k;
+ lj_ir_kvalue(as->J->L, &k, ir);
+ emit_movmroi(as, dest, 4, k.u32.hi);
+ emit_movmroi(as, dest, 0, k.u32.lo);
+ } else {
+ /* TODO: 64 bit store + 32 bit load-modify-store is suboptimal. */
+ Reg src = ra_alloc1(as, ref, rset_exclude(RSET_GPR, dest));
+ if (irt_is64(ir->t)) {
+ emit_u32(as, irt_toitype(ir->t) << 15);
+ emit_rmro(as, XO_ARITHi, XOg_OR, dest, 4);
+ } else {
+ /* Currently, no caller passes integers that might end up here. */
+ emit_movmroi(as, dest, 4, (irt_toitype(ir->t) << 15));
+ }
+ emit_movtomro(as, REX_64IR(ir, src), dest, 0);
+ }
+#else
if (!irref_isk(ref)) {
Reg src = ra_alloc1(as, ref, rset_exclude(RSET_GPR, dest));
emit_movtomro(as, REX_64IR(ir, src), dest, 0);
@@ -942,6 +1079,7 @@ static void asm_tvptr(ASMState *as, Reg dest, IRRef ref)
}
if (!(LJ_64 && irt_islightud(ir->t)))
emit_movmroi(as, dest, 4, irt_toitype(ir->t));
+#endif
emit_loada(as, dest, &J2G(as->J)->tmptv);
}
}
@@ -951,9 +1089,9 @@ static void asm_aref(ASMState *as, IRIns *ir)
Reg dest = ra_dest(as, ir, RSET_GPR);
asm_fusearef(as, ir, RSET_GPR);
if (!(as->mrm.idx == RID_NONE && as->mrm.ofs == 0))
- emit_mrm(as, XO_LEA, dest, RID_MRM);
+ emit_mrm(as, XO_LEA, dest|REX_GC64, RID_MRM);
else if (as->mrm.base != dest)
- emit_rr(as, XO_MOV, dest, as->mrm.base);
+ emit_rr(as, XO_MOV, dest|REX_GC64, as->mrm.base);
}
/* Inlined hash lookup. Specialized for key type and for const keys.
@@ -980,7 +1118,7 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
if (!isk) {
rset_clear(allow, tab);
key = ra_alloc1(as, ir->op2, irt_isnum(kt) ? RSET_FPR : allow);
- if (!irt_isstr(kt))
+ if (LJ_GC64 || !irt_isstr(kt))
tmp = ra_scratch(as, rset_exclude(allow, key));
}
@@ -993,8 +1131,8 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
/* Follow hash chain until the end. */
l_loop = emit_sjcc_label(as, CC_NZ);
- emit_rr(as, XO_TEST, dest, dest);
- emit_rmro(as, XO_MOV, dest, dest, offsetof(Node, next));
+ emit_rr(as, XO_TEST, dest|REX_GC64, dest);
+ emit_rmro(as, XO_MOV, dest|REX_GC64, dest, offsetof(Node, next));
l_next = emit_label(as);
/* Type and value comparison. */
@@ -1015,7 +1153,7 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
emit_rmro(as, XO_UCOMISD, key, dest, offsetof(Node, key.n));
emit_sjcc(as, CC_AE, l_next);
/* The type check avoids NaN penalties and complaints from Valgrind. */
-#if LJ_64
+#if LJ_64 && !LJ_GC64
emit_u32(as, LJ_TISNUM);
emit_rmro(as, XO_ARITHi, XOg_CMP, dest, offsetof(Node, key.it));
#else
@@ -1023,10 +1161,28 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
emit_rmro(as, XO_ARITHi8, XOg_CMP, dest, offsetof(Node, key.it));
#endif
}
-#if LJ_64
+#if LJ_64 && !LJ_GC64
} else if (irt_islightud(kt)) {
emit_rmro(as, XO_CMP, key|REX_64, dest, offsetof(Node, key.u64));
#endif
+#if LJ_GC64
+ } else if (irt_isaddr(kt)) {
+ if (isk) {
+ TValue k;
+ k.u64 = ((uint64_t)irt_toitype(irkey->t) << 47) | irkey[1].tv.u64;
+ emit_gmroi(as, XG_ARITHi(XOg_CMP), dest, offsetof(Node, key.u32.lo),
+ k.u32.lo);
+ emit_sjcc(as, CC_NE, l_next);
+ emit_gmroi(as, XG_ARITHi(XOg_CMP), dest, offsetof(Node, key.u32.hi),
+ k.u32.hi);
+ } else {
+ emit_rmro(as, XO_CMP, tmp|REX_64, dest, offsetof(Node, key.u64));
+ }
+ } else {
+ lua_assert(irt_ispri(kt) && !irt_isnil(kt));
+ emit_u32(as, (irt_toitype(kt)<<15)|0x7fff);
+ emit_rmro(as, XO_ARITHi, XOg_CMP, dest, offsetof(Node, key.it));
+#else
} else {
if (!irt_ispri(kt)) {
lua_assert(irt_isaddr(kt));
@@ -1040,16 +1196,23 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
lua_assert(!irt_isnil(kt));
emit_i8(as, irt_toitype(kt));
emit_rmro(as, XO_ARITHi8, XOg_CMP, dest, offsetof(Node, key.it));
+#endif
}
emit_sfixup(as, l_loop);
checkmclim(as);
+#if LJ_GC64
+ if (!isk && irt_isaddr(kt)) {
+ emit_rr(as, XO_OR, tmp|REX_64, key);
+ emit_loadu64(as, tmp, (uint64_t)irt_toitype(kt) << 47);
+ }
+#endif
/* Load main position relative to tab->node into dest. */
khash = isk ? ir_khash(irkey) : 1;
if (khash == 0) {
- emit_rmro(as, XO_MOV, dest, tab, offsetof(GCtab, node));
+ emit_rmro(as, XO_MOV, dest|REX_GC64, tab, offsetof(GCtab, node));
} else {
- emit_rmro(as, XO_ARITH(XOg_ADD), dest, tab, offsetof(GCtab, node));
+ emit_rmro(as, XO_ARITH(XOg_ADD), dest|REX_GC64, tab, offsetof(GCtab,node));
if ((as->flags & JIT_F_PREFER_IMUL)) {
emit_i8(as, sizeof(Node));
emit_rr(as, XO_IMULi8, dest, dest);
@@ -1084,7 +1247,19 @@ static void asm_href(ASMState *as, IRIns *ir, IROp merge)
#endif
} else {
emit_rr(as, XO_MOV, tmp, key);
+#if LJ_GC64
+ checkmclim(as);
+ emit_gri(as, XG_ARITHi(XOg_XOR), dest, irt_toitype(kt) << 15);
+ if ((as->flags & JIT_F_BMI2)) {
+ emit_i8(as, 32);
+ emit_mrm(as, XV_RORX|VEX_64, dest, key);
+ } else {
+ emit_shifti(as, XOg_SHR|REX_64, dest, 32);
+ emit_rr(as, XO_MOV, dest|REX_64, key|REX_64);
+ }
+#else
emit_rmro(as, XO_LEA, dest, key, HASH_BIAS);
+#endif
}
}
}
@@ -1104,11 +1279,11 @@ static void asm_hrefk(ASMState *as, IRIns *ir)
if (ra_hasreg(dest)) {
if (ofs != 0) {
if (dest == node && !(as->flags & JIT_F_LEA_AGU))
- emit_gri(as, XG_ARITHi(XOg_ADD), dest, ofs);
+ emit_gri(as, XG_ARITHi(XOg_ADD), dest|REX_GC64, ofs);
else
- emit_rmro(as, XO_LEA, dest, node, ofs);
+ emit_rmro(as, XO_LEA, dest|REX_GC64, node, ofs);
} else if (dest != node) {
- emit_rr(as, XO_MOV, dest, node);
+ emit_rr(as, XO_MOV, dest|REX_GC64, node);
}
}
asm_guardcc(as, CC_NE);
@@ -1120,13 +1295,24 @@ static void asm_hrefk(ASMState *as, IRIns *ir)
lua_assert(irt_isnum(irkey->t) || irt_isgcv(irkey->t));
/* Assumes -0.0 is already canonicalized to +0.0. */
emit_loadu64(as, key, irt_isnum(irkey->t) ? ir_knum(irkey)->u64 :
+#if LJ_GC64
+ ((uint64_t)irt_toitype(irkey->t) << 47) |
+ (uint64_t)ir_kgc(irkey));
+#else
((uint64_t)irt_toitype(irkey->t) << 32) |
(uint64_t)(uint32_t)ptr2addr(ir_kgc(irkey)));
+#endif
} else {
lua_assert(!irt_isnil(irkey->t));
+#if LJ_GC64
+ emit_i32(as, (irt_toitype(irkey->t)<<15)|0x7fff);
+ emit_rmro(as, XO_ARITHi, XOg_CMP, node,
+ ofs + (int32_t)offsetof(Node, key.it));
+#else
emit_i8(as, irt_toitype(irkey->t));
emit_rmro(as, XO_ARITHi8, XOg_CMP, node,
ofs + (int32_t)offsetof(Node, key.it));
+#endif
}
#else
l_exit = emit_label(as);
@@ -1157,25 +1343,25 @@ static void asm_hrefk(ASMState *as, IRIns *ir)
static void asm_uref(ASMState *as, IRIns *ir)
{
- /* NYI: Check that UREFO is still open and not aliasing a slot. */
Reg dest = ra_dest(as, ir, RSET_GPR);
if (irref_isk(ir->op1)) {
GCfunc *fn = ir_kfunc(IR(ir->op1));
MRef *v = &gcref(fn->l.uvptr[(ir->op2 >> 8)])->uv.v;
- emit_rma(as, XO_MOV, dest, v);
+ emit_rma(as, XO_MOV, dest|REX_GC64, v);
} else {
Reg uv = ra_scratch(as, RSET_GPR);
Reg func = ra_alloc1(as, ir->op1, RSET_GPR);
if (ir->o == IR_UREFC) {
- emit_rmro(as, XO_LEA, dest, uv, offsetof(GCupval, tv));
+ emit_rmro(as, XO_LEA, dest|REX_GC64, uv, offsetof(GCupval, tv));
asm_guardcc(as, CC_NE);
emit_i8(as, 1);
emit_rmro(as, XO_ARITHib, XOg_CMP, uv, offsetof(GCupval, closed));
} else {
- emit_rmro(as, XO_MOV, dest, uv, offsetof(GCupval, v));
+ emit_rmro(as, XO_MOV, dest|REX_GC64, uv, offsetof(GCupval, v));
}
- emit_rmro(as, XO_MOV, uv, func,
- (int32_t)offsetof(GCfuncL, uvptr) + 4*(int32_t)(ir->op2 >> 8));
+ emit_rmro(as, XO_MOV, uv|REX_GC64, func,
+ (int32_t)offsetof(GCfuncL, uvptr) +
+ (int32_t)sizeof(MRef) * (int32_t)(ir->op2 >> 8));
}
}
@@ -1193,9 +1379,9 @@ static void asm_strref(ASMState *as, IRIns *ir)
if (as->mrm.base == RID_NONE)
emit_loadi(as, dest, as->mrm.ofs);
else if (as->mrm.base == dest && as->mrm.idx == RID_NONE)
- emit_gri(as, XG_ARITHi(XOg_ADD), dest, as->mrm.ofs);
+ emit_gri(as, XG_ARITHi(XOg_ADD), dest|REX_GC64, as->mrm.ofs);
else
- emit_mrm(as, XO_LEA, dest, RID_MRM);
+ emit_mrm(as, XO_LEA, dest|REX_GC64, RID_MRM);
}
/* -- Loads and stores ---------------------------------------------------- */
@@ -1264,7 +1450,7 @@ static void asm_fxstore(ASMState *as, IRIns *ir)
case IRT_I16: case IRT_U16: xo = XO_MOVtow; break;
case IRT_NUM: xo = XO_MOVSDto; break;
case IRT_FLOAT: xo = XO_MOVSSto; break;
-#if LJ_64
+#if LJ_64 && !LJ_GC64
case IRT_LIGHTUD: lua_assert(0); /* NYI: mask 64 bit lightuserdata. */
#endif
default:
@@ -1296,7 +1482,7 @@ static void asm_fxstore(ASMState *as, IRIns *ir)
#define asm_fstore(as, ir) asm_fxstore(as, ir)
#define asm_xstore(as, ir) asm_fxstore(as, ir)
-#if LJ_64
+#if LJ_64 && !LJ_GC64
static Reg asm_load_lightud64(ASMState *as, IRIns *ir, int typecheck)
{
if (ra_used(ir) || typecheck) {
@@ -1318,9 +1504,12 @@ static Reg asm_load_lightud64(ASMState *as, IRIns *ir, int typecheck)
static void asm_ahuvload(ASMState *as, IRIns *ir)
{
+#if LJ_GC64
+ Reg tmp = RID_NONE;
+#endif
lua_assert(irt_isnum(ir->t) || irt_ispri(ir->t) || irt_isaddr(ir->t) ||
(LJ_DUALNUM && irt_isint(ir->t)));
-#if LJ_64
+#if LJ_64 && !LJ_GC64
if (irt_islightud(ir->t)) {
Reg dest = asm_load_lightud64(as, ir, 1);
if (ra_hasreg(dest)) {
@@ -1334,20 +1523,64 @@ static void asm_ahuvload(ASMState *as, IRIns *ir)
RegSet allow = irt_isnum(ir->t) ? RSET_FPR : RSET_GPR;
Reg dest = ra_dest(as, ir, allow);
asm_fuseahuref(as, ir->op1, RSET_GPR);
+#if LJ_GC64
+ if (irt_isaddr(ir->t)) {
+ emit_shifti(as, XOg_SHR|REX_64, dest, 17);
+ asm_guardcc(as, CC_NE);
+ emit_i8(as, irt_toitype(ir->t));
+ emit_rr(as, XO_ARITHi8, XOg_CMP, dest);
+ emit_i8(as, XI_O16);
+ if ((as->flags & JIT_F_BMI2)) {
+ emit_i8(as, 47);
+ emit_mrm(as, XV_RORX|VEX_64, dest, RID_MRM);
+ } else {
+ emit_shifti(as, XOg_ROR|REX_64, dest, 47);
+ emit_mrm(as, XO_MOV, dest|REX_64, RID_MRM);
+ }
+ return;
+ } else
+#endif
emit_mrm(as, dest < RID_MAX_GPR ? XO_MOV : XO_MOVSD, dest, RID_MRM);
} else {
- asm_fuseahuref(as, ir->op1, RSET_GPR);
+ RegSet gpr = RSET_GPR;
+#if LJ_GC64
+ if (irt_isaddr(ir->t)) {
+ tmp = ra_scratch(as, RSET_GPR);
+ gpr = rset_exclude(gpr, tmp);
+ }
+#endif
+ asm_fuseahuref(as, ir->op1, gpr);
}
/* Always do the type check, even if the load result is unused. */
as->mrm.ofs += 4;
asm_guardcc(as, irt_isnum(ir->t) ? CC_AE : CC_NE);
if (LJ_64 && irt_type(ir->t) >= IRT_NUM) {
lua_assert(irt_isinteger(ir->t) || irt_isnum(ir->t));
+#if LJ_GC64
+ emit_u32(as, LJ_TISNUM << 15);
+#else
emit_u32(as, LJ_TISNUM);
+#endif
+ emit_mrm(as, XO_ARITHi, XOg_CMP, RID_MRM);
+#if LJ_GC64
+ } else if (irt_isaddr(ir->t)) {
+ as->mrm.ofs -= 4;
+ emit_i8(as, irt_toitype(ir->t));
+ emit_mrm(as, XO_ARITHi8, XOg_CMP, tmp);
+ emit_shifti(as, XOg_SAR|REX_64, tmp, 47);
+ emit_mrm(as, XO_MOV, tmp|REX_64, RID_MRM);
+ } else if (irt_isnil(ir->t)) {
+ as->mrm.ofs -= 4;
+ emit_i8(as, -1);
+ emit_mrm(as, XO_ARITHi8, XOg_CMP|REX_64, RID_MRM);
+ } else {
+ emit_u32(as, (irt_toitype(ir->t) << 15) | 0x7fff);
emit_mrm(as, XO_ARITHi, XOg_CMP, RID_MRM);
+#else
} else {
emit_i8(as, irt_toitype(ir->t));
emit_mrm(as, XO_ARITHi8, XOg_CMP, RID_MRM);
+#endif
}
}
@@ -1359,12 +1592,28 @@ static void asm_ahustore(ASMState *as, IRIns *ir)
Reg src = ra_alloc1(as, ir->op2, RSET_FPR);
asm_fuseahuref(as, ir->op1, RSET_GPR);
emit_mrm(as, XO_MOVSDto, src, RID_MRM);
-#if LJ_64
+#if LJ_64 && !LJ_GC64
} else if (irt_islightud(ir->t)) {
Reg src = ra_alloc1(as, ir->op2, RSET_GPR);
asm_fuseahuref(as, ir->op1, rset_exclude(RSET_GPR, src));
emit_mrm(as, XO_MOVto, src|REX_64, RID_MRM);
#endif
+#if LJ_GC64
+ } else if (irref_isk(ir->op2)) {
+ TValue k;
+ lj_ir_kvalue(as->J->L, &k, IR(ir->op2));
+ asm_fuseahuref(as, ir->op1, RSET_GPR);
+ if (tvisnil(&k)) {
+ emit_i32(as, -1);
+ emit_mrm(as, XO_MOVmi, REX_64, RID_MRM);
+ } else {
+ emit_u32(as, k.u32.lo);
+ emit_mrm(as, XO_MOVmi, 0, RID_MRM);
+ as->mrm.ofs += 4;
+ emit_u32(as, k.u32.hi);
+ emit_mrm(as, XO_MOVmi, 0, RID_MRM);
+ }
+#endif
} else {
IRIns *irr = IR(ir->op2);
RegSet allow = RSET_GPR;
@@ -1375,6 +1624,17 @@ static void asm_ahustore(ASMState *as, IRIns *ir)
}
asm_fuseahuref(as, ir->op1, allow);
if (ra_hasreg(src)) {
+#if LJ_GC64
+ if (!(LJ_DUALNUM && irt_isinteger(ir->t))) {
+ /* TODO: 64 bit store + 32 bit load-modify-store is suboptimal. */
+ as->mrm.ofs += 4;
+ emit_u32(as, irt_toitype(ir->t) << 15);
+ emit_mrm(as, XO_ARITHi, XOg_OR, RID_MRM);
+ as->mrm.ofs -= 4;
+ emit_mrm(as, XO_MOVto, src|REX_64, RID_MRM);
+ return;
+ }
+#endif
emit_mrm(as, XO_MOVto, src, RID_MRM);
} else if (!irt_ispri(irr->t)) {
lua_assert(irt_isaddr(ir->t) || (LJ_DUALNUM && irt_isinteger(ir->t)));
@@ -1382,14 +1642,20 @@ static void asm_ahustore(ASMState *as, IRIns *ir)
emit_mrm(as, XO_MOVmi, 0, RID_MRM);
}
as->mrm.ofs += 4;
+#if LJ_GC64
+ lua_assert(LJ_DUALNUM && irt_isinteger(ir->t));
+ emit_i32(as, LJ_TNUMX << 15);
+#else
emit_i32(as, (int32_t)irt_toitype(ir->t));
+#endif
emit_mrm(as, XO_MOVmi, 0, RID_MRM);
}
}
static void asm_sload(ASMState *as, IRIns *ir)
{
- int32_t ofs = 8*((int32_t)ir->op1-1) + ((ir->op2 & IRSLOAD_FRAME) ? 4 : 0);
+ int32_t ofs = 8*((int32_t)ir->op1-1-LJ_FR2) +
+ (!LJ_FR2 && (ir->op2 & IRSLOAD_FRAME) ? 4 : 0);
IRType1 t = ir->t;
Reg base;
lua_assert(!(ir->op2 & IRSLOAD_PARENT)); /* Handled by asm_head_side(). */
@@ -1402,7 +1668,7 @@ static void asm_sload(ASMState *as, IRIns *ir)
base = ra_alloc1(as, REF_BASE, RSET_GPR);
emit_rmro(as, XO_MOVSD, left, base, ofs);
t.irt = IRT_NUM; /* Continue with a regular number type check. */
-#if LJ_64
+#if LJ_64 && !LJ_GC64
} else if (irt_islightud(t)) {
Reg dest = asm_load_lightud64(as, ir, (ir->op2 & IRSLOAD_TYPECHECK));
if (ra_hasreg(dest)) {
@@ -1420,6 +1686,36 @@ static void asm_sload(ASMState *as, IRIns *ir)
t.irt = irt_isint(t) ? IRT_NUM : IRT_INT; /* Check for original type. */
emit_rmro(as, irt_isint(t) ? XO_CVTSI2SD : XO_CVTTSD2SI, dest, base, ofs);
} else {
+#if LJ_GC64
+ if (irt_isaddr(t)) {
+ /* LJ_GC64 type check + tag removal without BMI2 and with BMI2:
+ **
+ ** mov r64, [addr] rorx r64, [addr], 47
+ ** ror r64, 47
+ ** cmp r16, itype cmp r16, itype
+ ** jne ->exit jne ->exit
+ ** shr r64, 16 shr r64, 16
+ */
+ emit_shifti(as, XOg_SHR|REX_64, dest, 17);
+ if ((ir->op2 & IRSLOAD_TYPECHECK)) {
+ asm_guardcc(as, CC_NE);
+ emit_i8(as, irt_toitype(t));
+ emit_rr(as, XO_ARITHi8, XOg_CMP, dest);
+ emit_i8(as, XI_O16);
+ }
+ if ((as->flags & JIT_F_BMI2)) {
+ emit_i8(as, 47);
+ emit_rmro(as, XV_RORX|VEX_64, dest, base, ofs);
+ } else {
+ if ((ir->op2 & IRSLOAD_TYPECHECK))
+ emit_shifti(as, XOg_ROR|REX_64, dest, 47);
+ else
+ emit_shifti(as, XOg_SHL|REX_64, dest, 17);
+ emit_rmro(as, XO_MOV, dest|REX_64, base, ofs);
+ }
+ return;
+ } else
+#endif
emit_rmro(as, irt_isnum(t) ? XO_MOVSD : XO_MOV, dest, base, ofs);
}
} else {
@@ -1432,11 +1728,42 @@ static void asm_sload(ASMState *as, IRIns *ir)
asm_guardcc(as, irt_isnum(t) ? CC_AE : CC_NE);
if (LJ_64 && irt_type(t) >= IRT_NUM) {
lua_assert(irt_isinteger(t) || irt_isnum(t));
+#if LJ_GC64
+ emit_u32(as, LJ_TISNUM << 15);
+#else
emit_u32(as, LJ_TISNUM);
+#endif
+ emit_rmro(as, XO_ARITHi, XOg_CMP, base, ofs+4);
+#if LJ_GC64
+ } else if (irt_isnil(t)) {
+ /* LJ_GC64 type check for nil:
+ **
+ ** cmp qword [addr], -1
+ ** jne ->exit
+ */
+ emit_i8(as, -1);
+ emit_rmro(as, XO_ARITHi8, XOg_CMP|REX_64, base, ofs);
+ } else if (irt_ispri(t)) {
+ emit_u32(as, (irt_toitype(t) << 15) | 0x7fff);
emit_rmro(as, XO_ARITHi, XOg_CMP, base, ofs+4);
} else {
+ /* LJ_GC64 type check only:
+ **
+ ** mov r64, [addr]
+ ** sar r64, 47
+ ** cmp r32, itype
+ ** jne ->exit
+ */
+ Reg tmp = ra_scratch(as, rset_exclude(RSET_GPR, base));
+ emit_i8(as, irt_toitype(t));
+ emit_rr(as, XO_ARITHi8, XOg_CMP, tmp);
+ emit_shifti(as, XOg_SAR|REX_64, tmp, 47);
+ emit_rmro(as, XO_MOV, tmp|REX_64, base, ofs+4);
+#else
+ } else {
emit_i8(as, irt_toitype(t));
emit_rmro(as, XO_ARITHi8, XOg_CMP, base, ofs+4);
+#endif
}
}
}
@@ -1464,8 +1791,9 @@ static void asm_cnew(ASMState *as, IRIns *ir)
Reg r64 = sz == 8 ? REX_64 : 0;
if (irref_isk(ir->op2)) {
IRIns *irk = IR(ir->op2);
- uint64_t k = irk->o == IR_KINT64 ? ir_k64(irk)->u64 :
- (uint64_t)(uint32_t)irk->i;
+ uint64_t k = (irk->o == IR_KINT64 ||
+ (LJ_GC64 && (irk->o == IR_KPTR || irk->o == IR_KKPTR))) ?
+ ir_k64(irk)->u64 : (uint64_t)(uint32_t)irk->i;
if (sz == 4 || checki32((int64_t)k)) {
emit_i32(as, (int32_t)k);
emit_rmro(as, XO_MOVmi, r64, RID_RET, sizeof(GCcdata));
@@ -1530,7 +1858,7 @@ static void asm_tbar(ASMState *as, IRIns *ir)
Reg tab = ra_alloc1(as, ir->op1, RSET_GPR);
Reg tmp = ra_scratch(as, rset_exclude(RSET_GPR, tab));
MCLabel l_end = emit_label(as);
- emit_movtomro(as, tmp, tab, offsetof(GCtab, gclist));
+ emit_movtomro(as, tmp|REX_GC64, tab, offsetof(GCtab, gclist));
emit_setgl(as, tab, gc.grayagain);
emit_getgl(as, tmp, gc.grayagain);
emit_i8(as, ~LJ_GC_BLACK);
@@ -1956,7 +2284,7 @@ static void asm_bswap(ASMState *as, IRIns *ir)
#define asm_bor(as, ir) asm_intarith(as, ir, XOg_OR)
#define asm_bxor(as, ir) asm_intarith(as, ir, XOg_XOR)
-static void asm_bitshift(ASMState *as, IRIns *ir, x86Shift xs)
+static void asm_bitshift(ASMState *as, IRIns *ir, x86Shift xs, x86Op xv)
{
IRRef rref = ir->op2;
IRIns *irr = IR(rref);
@@ -1965,11 +2293,27 @@ static void asm_bitshift(ASMState *as, IRIns *ir, x86Shift xs)
int shift;
dest = ra_dest(as, ir, RSET_GPR);
shift = irr->i & (irt_is64(ir->t) ? 63 : 31);
+ if (!xv && shift && (as->flags & JIT_F_BMI2)) {
+ Reg left = asm_fuseloadm(as, ir->op1, RSET_GPR, irt_is64(ir->t));
+ if (left != dest) { /* BMI2 rotate right by constant. */
+ emit_i8(as, xs == XOg_ROL ? -shift : shift);
+ emit_mrm(as, VEX_64IR(ir, XV_RORX), dest, left);
+ return;
+ }
+ }
switch (shift) {
case 0: break;
case 1: emit_rr(as, XO_SHIFT1, REX_64IR(ir, xs), dest); break;
default: emit_shifti(as, REX_64IR(ir, xs), dest, shift); break;
}
+ } else if ((as->flags & JIT_F_BMI2) && xv) { /* BMI2 variable shifts. */
+ Reg left, right;
+ dest = ra_dest(as, ir, RSET_GPR);
+ right = ra_alloc1(as, rref, RSET_GPR);
+ left = asm_fuseloadm(as, ir->op1, rset_exclude(RSET_GPR, right),
+ irt_is64(ir->t));
+ emit_mrm(as, VEX_64IR(ir, xv) ^ (right << 19), dest, left);
+ return;
} else { /* Variable shifts implicitly use register cl (i.e. ecx). */
Reg right;
dest = ra_dest(as, ir, rset_exclude(RSET_GPR, RID_ECX));
@@ -1995,11 +2339,11 @@ static void asm_bitshift(ASMState *as, IRIns *ir, x86Shift xs)
*/
}
-#define asm_bshl(as, ir) asm_bitshift(as, ir, XOg_SHL)
-#define asm_bshr(as, ir) asm_bitshift(as, ir, XOg_SHR)
-#define asm_bsar(as, ir) asm_bitshift(as, ir, XOg_SAR)
-#define asm_brol(as, ir) asm_bitshift(as, ir, XOg_ROL)
-#define asm_bror(as, ir) asm_bitshift(as, ir, XOg_ROR)
+#define asm_bshl(as, ir) asm_bitshift(as, ir, XOg_SHL, XV_SHLX)
+#define asm_bshr(as, ir) asm_bitshift(as, ir, XOg_SHR, XV_SHRX)
+#define asm_bsar(as, ir) asm_bitshift(as, ir, XOg_SAR, XV_SARX)
+#define asm_brol(as, ir) asm_bitshift(as, ir, XOg_ROL, 0)
+#define asm_bror(as, ir) asm_bitshift(as, ir, XOg_ROR, 0)
/* -- Comparisons --------------------------------------------------------- */
@@ -2050,7 +2394,6 @@ static void asm_comp(ASMState *as, IRIns *ir)
cc ^= (VCC_PS|(5<<4)); /* A <-> B, AE <-> BE, PS <-> none */
}
left = ra_alloc1(as, lref, RSET_FPR);
- right = asm_fuseload(as, rref, rset_exclude(RSET_FPR, left));
l_around = emit_label(as);
asm_guardcc(as, cc >> 4);
if (cc & VCC_P) { /* Extra CC_P branch required? */
@@ -2067,6 +2410,7 @@ static void asm_comp(ASMState *as, IRIns *ir)
emit_jcc(as, CC_P, as->mcp);
}
}
+ right = asm_fuseload(as, rref, rset_exclude(RSET_FPR, left));
emit_mrm(as, XO_UCOMISD, left, right);
} else {
IRRef lref = ir->op1, rref = ir->op2;
@@ -2343,13 +2687,18 @@ static void asm_stack_check(ASMState *as, BCReg topslot,
emit_rmro(as, XO_MOV, r|REX_64, RID_ESP, 0);
else
ra_modified(as, r);
- emit_gri(as, XG_ARITHi(XOg_CMP), r, (int32_t)(8*topslot));
+ emit_gri(as, XG_ARITHi(XOg_CMP), r|REX_GC64, (int32_t)(8*topslot));
if (ra_hasreg(pbase) && pbase != r)
- emit_rr(as, XO_ARITH(XOg_SUB), r, pbase);
+ emit_rr(as, XO_ARITH(XOg_SUB), r|REX_GC64, pbase);
else
+#if LJ_GC64
+ emit_rmro(as, XO_ARITH(XOg_SUB), r|REX_64, RID_DISPATCH,
+ (int32_t)dispofs(as, &J2G(as->J)->jit_base));
+#else
emit_rmro(as, XO_ARITH(XOg_SUB), r, RID_NONE,
ptr2addr(&J2G(as->J)->jit_base));
- emit_rmro(as, XO_MOV, r, r, offsetof(lua_State, maxstack));
+#endif
+ emit_rmro(as, XO_MOV, r|REX_GC64, r, offsetof(lua_State, maxstack));
emit_getgl(as, r, cur_L);
if (allow == RSET_EMPTY) /* Spill temp. register. */
emit_rmro(as, XO_MOVto, r|REX_64, RID_ESP, 0);
@@ -2359,13 +2708,15 @@ static void asm_stack_check(ASMState *as, BCReg topslot,
static void asm_stack_restore(ASMState *as, SnapShot *snap)
{
SnapEntry *map = &as->T->snapmap[snap->mapofs];
- SnapEntry *flinks = &as->T->snapmap[snap_nextofs(as->T, snap)-1];
+#if !LJ_FR2 || defined(LUA_USE_ASSERT)
+ SnapEntry *flinks = &as->T->snapmap[snap_nextofs(as->T, snap)-1-LJ_FR2];
+#endif
MSize n, nent = snap->nent;
/* Store the value of all modified slots to the Lua stack. */
for (n = 0; n < nent; n++) {
SnapEntry sn = map[n];
BCReg s = snap_slot(sn);
- int32_t ofs = 8*((int32_t)s-1);
+ int32_t ofs = 8*((int32_t)s-1-LJ_FR2);
IRRef ref = snap_ref(sn);
IRIns *ir = IR(ref);
if ((sn & SNAP_NORESTORE))
@@ -2378,16 +2729,44 @@ static void asm_stack_restore(ASMState *as, SnapShot *snap)
(LJ_DUALNUM && irt_isinteger(ir->t)));
if (!irref_isk(ref)) {
Reg src = ra_alloc1(as, ref, rset_exclude(RSET_GPR, RID_BASE));
+#if LJ_GC64
+ if (irt_is64(ir->t)) {
+ /* TODO: 64 bit store + 32 bit load-modify-store is suboptimal. */
+ emit_u32(as, irt_toitype(ir->t) << 15);
+ emit_rmro(as, XO_ARITHi, XOg_OR, RID_BASE, ofs+4);
+ } else if (LJ_DUALNUM && irt_isinteger(ir->t)) {
+ emit_movmroi(as, RID_BASE, ofs+4, LJ_TISNUM << 15);
+ } else {
+ emit_movmroi(as, RID_BASE, ofs+4, (irt_toitype(ir->t)<<15)|0x7fff);
+ }
+#endif
emit_movtomro(as, REX_64IR(ir, src), RID_BASE, ofs);
+#if LJ_GC64
+ } else {
+ TValue k;
+ lj_ir_kvalue(as->J->L, &k, ir);
+ if (tvisnil(&k)) {
+ emit_i32(as, -1);
+ emit_rmro(as, XO_MOVmi, REX_64, RID_BASE, ofs);
+ } else {
+ emit_movmroi(as, RID_BASE, ofs+4, k.u32.hi);
+ emit_movmroi(as, RID_BASE, ofs, k.u32.lo);
+ }
+#else
} else if (!irt_ispri(ir->t)) {
emit_movmroi(as, RID_BASE, ofs, ir->i);
+#endif
}
if ((sn & (SNAP_CONT|SNAP_FRAME))) {
+#if !LJ_FR2
if (s != 0) /* Do not overwrite link to previous frame. */
emit_movmroi(as, RID_BASE, ofs+4, (int32_t)(*flinks--));
+#endif
+#if !LJ_GC64
} else {
if (!(LJ_64 && irt_islightud(ir->t)))
emit_movmroi(as, RID_BASE, ofs+4, irt_toitype(ir->t));
+#endif
}
}
checkmclim(as);
@@ -2413,11 +2792,15 @@ static void asm_gc_check(ASMState *as)
args[1] = ASMREF_TMP2; /* MSize steps */
asm_gencall(as, ci, args);
tmp = ra_releasetmp(as, ASMREF_TMP1);
+#if LJ_GC64
+ emit_rmro(as, XO_LEA, tmp|REX_64, RID_DISPATCH, GG_DISP2G);
+#else
emit_loada(as, tmp, J2G(as->J));
+#endif
emit_loadi(as, ra_releasetmp(as, ASMREF_TMP2), as->gcsteps);
/* Jump around GC step if GC total < GC threshold. */
emit_sjcc(as, CC_B, l_end);
- emit_opgl(as, XO_ARITH(XOg_CMP), tmp, gc.threshold);
+ emit_opgl(as, XO_ARITH(XOg_CMP), tmp|REX_GC64, gc.threshold);
emit_getgl(as, tmp, gc.total);
as->gcsteps = 0;
checkmclim(as);
@@ -2482,7 +2865,7 @@ static void asm_head_root_base(ASMState *as)
if (rset_test(as->modset, r) || irt_ismarked(ir->t))
ir->r = RID_INIT; /* No inheritance for modified BASE register. */
if (r != RID_BASE)
- emit_rr(as, XO_MOV, r, RID_BASE);
+ emit_rr(as, XO_MOV, r|REX_GC64, RID_BASE);
}
}
@@ -2498,8 +2881,9 @@ static RegSet asm_head_side_base(ASMState *as, IRIns *irp, RegSet allow)
if (irp->r == r) {
rset_clear(allow, r); /* Mark same BASE register as coalesced. */
} else if (ra_hasreg(irp->r) && rset_test(as->freeset, irp->r)) {
+ /* Move from coalesced parent reg. */
rset_clear(allow, irp->r);
- emit_rr(as, XO_MOV, r, irp->r); /* Move from coalesced parent reg. */
+ emit_rr(as, XO_MOV, r|REX_GC64, irp->r);
} else {
emit_getgl(as, r, jit_base); /* Otherwise reload BASE. */
}
@@ -2600,10 +2984,111 @@ static Reg asm_setup_call_slots(ASMState *as, IRIns *ir, const CCallInfo *ci)
static void asm_setup_target(ASMState *as)
{
asm_exitstub_setup(as, as->T->nsnap);
+ as->mrm.base = 0;
}
/* -- Trace patching ------------------------------------------------------ */
+static const uint8_t map_op1[256] = {
+0x92,0x92,0x92,0x92,0x52,0x45,0x51,0x51,0x92,0x92,0x92,0x92,0x52,0x45,0x51,0x20,
+0x92,0x92,0x92,0x92,0x52,0x45,0x51,0x51,0x92,0x92,0x92,0x92,0x52,0x45,0x51,0x51,
+0x92,0x92,0x92,0x92,0x52,0x45,0x10,0x51,0x92,0x92,0x92,0x92,0x52,0x45,0x10,0x51,
+0x92,0x92,0x92,0x92,0x52,0x45,0x10,0x51,0x92,0x92,0x92,0x92,0x52,0x45,0x10,0x51,
+#if LJ_64
+0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x10,0x14,0x14,0x14,0x14,0x14,0x14,0x14,0x14,
+#else
+0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,
+#endif
+0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,
+0x51,0x51,0x92,0x92,0x10,0x10,0x12,0x11,0x45,0x86,0x52,0x93,0x51,0x51,0x51,0x51,
+0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,
+0x93,0x86,0x93,0x93,0x92,0x92,0x92,0x92,0x92,0x92,0x92,0x92,0x92,0x92,0x92,0x92,
+0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x51,0x47,0x51,0x51,0x51,0x51,0x51,
+#if LJ_64
+0x59,0x59,0x59,0x59,0x51,0x51,0x51,0x51,0x52,0x45,0x51,0x51,0x51,0x51,0x51,0x51,
+#else
+0x55,0x55,0x55,0x55,0x51,0x51,0x51,0x51,0x52,0x45,0x51,0x51,0x51,0x51,0x51,0x51,
+#endif
+0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x05,0x05,0x05,0x05,0x05,0x05,0x05,0x05,
+0x93,0x93,0x53,0x51,0x70,0x71,0x93,0x86,0x54,0x51,0x53,0x51,0x51,0x52,0x51,0x51,
+0x92,0x92,0x92,0x92,0x52,0x52,0x51,0x51,0x92,0x92,0x92,0x92,0x92,0x92,0x92,0x92,
+0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x45,0x45,0x47,0x52,0x51,0x51,0x51,0x51,
+0x10,0x51,0x10,0x10,0x51,0x51,0x63,0x66,0x51,0x51,0x51,0x51,0x51,0x51,0x92,0x92
+};
+
+static const uint8_t map_op2[256] = {
+0x93,0x93,0x93,0x93,0x52,0x52,0x52,0x52,0x52,0x52,0x51,0x52,0x51,0x93,0x52,0x94,
+0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,
+0x53,0x53,0x53,0x53,0x53,0x53,0x53,0x53,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,
+0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x34,0x51,0x35,0x51,0x51,0x51,0x51,0x51,
+0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,
+0x53,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,
+0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,
+0x94,0x54,0x54,0x54,0x93,0x93,0x93,0x52,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,
+0x46,0x46,0x46,0x46,0x46,0x46,0x46,0x46,0x46,0x46,0x46,0x46,0x46,0x46,0x46,0x46,
+0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,
+0x52,0x52,0x52,0x93,0x94,0x93,0x51,0x51,0x52,0x52,0x52,0x93,0x94,0x93,0x93,0x93,
+0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x94,0x93,0x93,0x93,0x93,0x93,
+0x93,0x93,0x94,0x93,0x94,0x94,0x94,0x93,0x52,0x52,0x52,0x52,0x52,0x52,0x52,0x52,
+0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,
+0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,
+0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x93,0x52
+};
+
+static uint32_t asm_x86_inslen(const uint8_t* p)
+{
+ uint32_t result = 0;
+ uint32_t prefixes = 0;
+ uint32_t x = map_op1[*p];
+ for (;;) {
+ switch (x >> 4) {
+ case 0: return result + x + (prefixes & 4);
+ case 1: prefixes |= x; x = map_op1[*++p]; result++; break;
+ case 2: x = map_op2[*++p]; break;
+ case 3: p++; goto mrm;
+ case 4: result -= (prefixes & 2); /* fallthrough */
+ case 5: return result + (x & 15);
+ case 6: /* Group 3. */
+ if (p[1] & 0x38) x = 2;
+ else if ((prefixes & 2) && (x == 0x66)) x = 4;
+ goto mrm;
+ case 7: /* VEX c4/c5. */
+ if (LJ_32 && p[1] < 0xc0) {
+ x = 2;
+ goto mrm;
+ }
+ if (x == 0x70) {
+ x = *++p & 0x1f;
+ result++;
+ if (x >= 2) {
+ p += 2;
+ result += 2;
+ goto mrm;
+ }
+ }
+ p++;
+ result++;
+ x = map_op2[*++p];
+ break;
+ case 8: result -= (prefixes & 2); /* fallthrough */
+ case 9: mrm: /* ModR/M and possibly SIB. */
+ result += (x & 15);
+ x = *++p;
+ switch (x >> 6) {
+ case 0: if ((x & 7) == 5) return result + 4; break;
+ case 1: result++; break;
+ case 2: result += 4; break;
+ case 3: return result;
+ }
+ if ((x & 7) == 4) {
+ result++;
+ if (x < 0x40 && (p[1] & 7) == 5) result += 4;
+ }
+ return result;
+ }
+ }
+}
+
/* Patch exit jumps of existing machine code to a new target. */
void lj_asm_patchexit(jit_State *J, GCtrace *T, ExitNo exitno, MCode *target)
{
@@ -2612,22 +3097,23 @@ void lj_asm_patchexit(jit_State *J, GCtrace *T, ExitNo exitno, MCode *target)
MSize len = T->szmcode;
MCode *px = exitstub_addr(J, exitno) - 6;
MCode *pe = p+len-6;
- uint32_t stateaddr = u32ptr(&J2G(J)->vmstate);
+#if LJ_GC64
+ uint32_t statei = (uint32_t)(GG_OFS(g.vmstate) - GG_OFS(dispatch));
+#else
+ uint32_t statei = u32ptr(&J2G(J)->vmstate);
+#endif
if (len > 5 && p[len-5] == XI_JMP && p+len-6 + *(int32_t *)(p+len-4) == px)
*(int32_t *)(p+len-4) = jmprel(p+len, target);
/* Do not patch parent exit for a stack check. Skip beyond vmstate update. */
- for (; p < pe; p++)
- if (*(uint32_t *)(p+(LJ_64 ? 3 : 2)) == stateaddr && p[0] == XI_MOVmi) {
- p += LJ_64 ? 11 : 10;
+ for (; p < pe; p += asm_x86_inslen(p)) {
+ intptr_t ofs = LJ_GC64 ? (p[0] & 0xf0) == 0x40 : LJ_64;
+ if (*(uint32_t *)(p+2+ofs) == statei && p[ofs+LJ_GC64-LJ_64] == XI_MOVmi)
break;
- }
+ }
lua_assert(p < pe);
- for (; p < pe; p++) {
- if ((*(uint16_t *)p & 0xf0ff) == 0x800f && p + *(int32_t *)(p+2) == px) {
+ for (; p < pe; p += asm_x86_inslen(p))
+ if ((*(uint16_t *)p & 0xf0ff) == 0x800f && p + *(int32_t *)(p+2) == px)
*(int32_t *)(p+2) = jmprel(p+6, target);
- p += 5;
- }
- }
lj_mcode_sync(T->mcode, T->mcode + T->szmcode);
lj_mcode_patch(J, mcarea, 1);
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_bc.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_bc.c
index f038769c220..a597692cab4 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_bc.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_bc.c
@@ -1,6 +1,6 @@
/*
** Bytecode instruction modes.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_bc_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_bc.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_bc.h
index 78826068693..69a45f281e9 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_bc.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_bc.h
@@ -1,6 +1,6 @@
/*
** Bytecode instruction format.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_BC_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_bcdump.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_bcdump.h
index 3752f151249..fdfc6ec0c6e 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_bcdump.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_bcdump.h
@@ -1,6 +1,6 @@
/*
** Bytecode dump definitions.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_BCDUMP_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_bcread.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_bcread.c
index 455c9d8242d..48c5e7c7f5a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_bcread.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_bcread.c
@@ -1,6 +1,6 @@
/*
** Bytecode reader.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_bcread_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_bcwrite.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_bcwrite.c
index 4728305932a..5e05caeaf5d 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_bcwrite.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_bcwrite.c
@@ -1,6 +1,6 @@
/*
** Bytecode writer.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_bcwrite_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_buf.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_buf.c
index 6251b075af2..0dfe7f98075 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_buf.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_buf.c
@@ -1,6 +1,6 @@
/*
** Buffer handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_buf_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_buf.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_buf.h
index fb632ee231c..a4051694445 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_buf.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_buf.h
@@ -1,6 +1,6 @@
/*
** Buffer handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_BUF_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_carith.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_carith.c
index 6102dcaa776..218abd260f4 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_carith.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_carith.c
@@ -1,6 +1,6 @@
/*
** C data arithmetic.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "lj_obj.h"
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_carith.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_carith.h
index b3bdff8a761..67d976bf0c0 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_carith.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_carith.h
@@ -1,6 +1,6 @@
/*
** C data arithmetic.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_CARITH_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.c
index eb7c1ee0d58..5c252e5b683 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.c
@@ -1,6 +1,6 @@
/*
** FFI C call handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "lj_obj.h"
@@ -301,7 +301,7 @@
unsigned int cl = ccall_classify_struct(cts, ctr); \
if ((cl & 4)) { /* Combine float HFA from separate registers. */ \
CTSize i = (cl >> 8) - 1; \
- do { ((uint32_t *)dp)[i] = cc->fpr[i].u32; } while (i--); \
+ do { ((uint32_t *)dp)[i] = cc->fpr[i].lo; } while (i--); \
} else { \
if (cl > 1) sp = (uint8_t *)&cc->fpr[0]; \
memcpy(dp, sp, ctr->size); \
@@ -331,7 +331,7 @@
#define CCALL_HANDLE_COMPLEXARG \
/* Pass complex by value in separate (!) FPRs or on stack. */ \
- isfp = ctr->size == 2*sizeof(float) ? 2 : 1;
+ isfp = sz == 2*sizeof(float) ? 2 : 1;
#define CCALL_HANDLE_REGARG \
if (LJ_TARGET_IOS && isva) { \
@@ -359,6 +359,13 @@
} \
}
+#if LJ_BE
+#define CCALL_HANDLE_RET \
+ if (ctype_isfp(ctr->info) && ctr->size == sizeof(float)) \
+ sp = (uint8_t *)&cc->fpr[0].f;
+#endif
+
+
#elif LJ_TARGET_PPC
/* -- PPC calling conventions --------------------------------------------- */
@@ -407,8 +414,8 @@
if (ctype_isfp(ctr->info) && ctr->size == sizeof(float)) \
ctr = ctype_get(cts, CTID_DOUBLE); /* FPRs always hold doubles. */
-#elif LJ_TARGET_MIPS
-/* -- MIPS calling conventions -------------------------------------------- */
+#elif LJ_TARGET_MIPS32
+/* -- MIPS o32 calling conventions ---------------------------------------- */
#define CCALL_HANDLE_STRUCTRET \
cc->retref = 1; /* Return all structs by reference. */ \
@@ -483,6 +490,78 @@
sp = (uint8_t *)&cc->fpr[0].f;
#endif
+#elif LJ_TARGET_MIPS64
+/* -- MIPS n64 calling conventions ---------------------------------------- */
+
+#define CCALL_HANDLE_STRUCTRET \
+ cc->retref = !(sz <= 16); \
+ if (cc->retref) cc->gpr[ngpr++] = (GPRArg)dp;
+
+#define CCALL_HANDLE_STRUCTRET2 \
+ ccall_copy_struct(cc, ctr, dp, sp, ccall_classify_struct(cts, ctr, ct));
+
+#define CCALL_HANDLE_COMPLEXRET \
+ /* Complex values are returned in 1 or 2 FPRs. */ \
+ cc->retref = 0;
+
+#if LJ_ABI_SOFTFP /* MIPS64 soft-float */
+
+#define CCALL_HANDLE_COMPLEXRET2 \
+ if (ctr->size == 2*sizeof(float)) { /* Copy complex float from GPRs. */ \
+ ((intptr_t *)dp)[0] = cc->gpr[0]; \
+ } else { /* Copy complex double from GPRs. */ \
+ ((intptr_t *)dp)[0] = cc->gpr[0]; \
+ ((intptr_t *)dp)[1] = cc->gpr[1]; \
+ }
+
+#define CCALL_HANDLE_COMPLEXARG \
+ /* Pass complex by value in 2 or 4 GPRs. */
+
+/* Position of soft-float 'float' return value depends on endianess. */
+#define CCALL_HANDLE_RET \
+ if (ctype_isfp(ctr->info) && ctr->size == sizeof(float)) \
+ sp = (uint8_t *)cc->gpr + LJ_ENDIAN_SELECT(0, 4);
+
+#else /* MIPS64 hard-float */
+
+#define CCALL_HANDLE_COMPLEXRET2 \
+ if (ctr->size == 2*sizeof(float)) { /* Copy complex float from FPRs. */ \
+ ((float *)dp)[0] = cc->fpr[0].f; \
+ ((float *)dp)[1] = cc->fpr[1].f; \
+ } else { /* Copy complex double from FPRs. */ \
+ ((double *)dp)[0] = cc->fpr[0].d; \
+ ((double *)dp)[1] = cc->fpr[1].d; \
+ }
+
+#define CCALL_HANDLE_COMPLEXARG \
+ if (sz == 2*sizeof(float)) { \
+ isfp = 2; \
+ if (ngpr < maxgpr) \
+ sz *= 2; \
+ }
+
+#define CCALL_HANDLE_RET \
+ if (ctype_isfp(ctr->info) && ctr->size == sizeof(float)) \
+ sp = (uint8_t *)&cc->fpr[0].f;
+
+#endif
+
+#define CCALL_HANDLE_STRUCTARG \
+ /* Pass all structs by value in registers and/or on the stack. */
+
+#define CCALL_HANDLE_REGARG \
+ if (ngpr < maxgpr) { \
+ dp = &cc->gpr[ngpr]; \
+ if (ngpr + n > maxgpr) { \
+ nsp += ngpr + n - maxgpr; /* Assumes contiguous gpr/stack fields. */ \
+ if (nsp > CCALL_MAXSTACK) goto err_nyi; /* Too many arguments. */ \
+ ngpr = maxgpr; \
+ } else { \
+ ngpr += n; \
+ } \
+ goto done; \
+ }
+
#else
#error "Missing calling convention definitions for this architecture"
#endif
@@ -722,6 +801,78 @@ noth: /* Not a homogeneous float/double aggregate. */
#endif
+/* -- MIPS64 ABI struct classification ---------------------------- */
+
+#if LJ_TARGET_MIPS64
+
+#define FTYPE_FLOAT 1
+#define FTYPE_DOUBLE 2
+
+/* Classify FP fields (max. 2) and their types. */
+static unsigned int ccall_classify_struct(CTState *cts, CType *ct, CType *ctf)
+{
+ int n = 0, ft = 0;
+ if ((ctf->info & CTF_VARARG) || (ct->info & CTF_UNION))
+ goto noth;
+ while (ct->sib) {
+ CType *sct;
+ ct = ctype_get(cts, ct->sib);
+ if (n == 2) {
+ goto noth;
+ } else if (ctype_isfield(ct->info)) {
+ sct = ctype_rawchild(cts, ct);
+ if (ctype_isfp(sct->info)) {
+ ft |= (sct->size == 4 ? FTYPE_FLOAT : FTYPE_DOUBLE) << 2*n;
+ n++;
+ } else {
+ goto noth;
+ }
+ } else if (ctype_isbitfield(ct->info) ||
+ ctype_isxattrib(ct->info, CTA_SUBTYPE)) {
+ goto noth;
+ }
+ }
+ if (n <= 2)
+ return ft;
+noth: /* Not a homogeneous float/double aggregate. */
+ return 0; /* Struct is in GPRs. */
+}
+
+void ccall_copy_struct(CCallState *cc, CType *ctr, void *dp, void *sp, int ft)
+{
+ if (LJ_ABI_SOFTFP ? ft :
+ ((ft & 3) == FTYPE_FLOAT || (ft >> 2) == FTYPE_FLOAT)) {
+ int i, ofs = 0;
+ for (i = 0; ft != 0; i++, ft >>= 2) {
+ if ((ft & 3) == FTYPE_FLOAT) {
+#if LJ_ABI_SOFTFP
+ /* The 2nd FP struct result is in CARG1 (gpr[2]) and not CRET2. */
+ memcpy((uint8_t *)dp + ofs,
+ (uint8_t *)&cc->gpr[2*i] + LJ_ENDIAN_SELECT(0, 4), 4);
+#else
+ *(float *)((uint8_t *)dp + ofs) = cc->fpr[i].f;
+#endif
+ ofs += 4;
+ } else {
+ ofs = (ofs + 7) & ~7; /* 64 bit alignment. */
+#if LJ_ABI_SOFTFP
+ *(intptr_t *)((uint8_t *)dp + ofs) = cc->gpr[2*i];
+#else
+ *(double *)((uint8_t *)dp + ofs) = cc->fpr[i].d;
+#endif
+ ofs += 8;
+ }
+ }
+ } else {
+#if !LJ_ABI_SOFTFP
+ if (ft) sp = (uint8_t *)&cc->fpr[0];
+#endif
+ memcpy(dp, sp, ctr->size);
+ }
+}
+
+#endif
+
/* -- Common C call handling ---------------------------------------------- */
/* Infer the destination CTypeID for a vararg argument. */
@@ -889,6 +1040,19 @@ static int ccall_set_args(lua_State *L, CTState *cts, CType *ct,
*(int32_t *)dp = d->size == 1 ? (int32_t)*(int8_t *)dp :
(int32_t)*(int16_t *)dp;
}
+#if LJ_TARGET_ARM64 && LJ_BE
+ if (isfp && d->size == sizeof(float))
+ ((float *)dp)[1] = ((float *)dp)[0]; /* Floats occupy high slot. */
+#endif
+#if LJ_TARGET_MIPS64 || (LJ_TARGET_ARM64 && LJ_BE)
+ if ((ctype_isinteger_or_bool(d->info) || ctype_isenum(d->info)
+#if LJ_TARGET_MIPS64
+ || (isfp && nsp == 0)
+#endif
+ ) && d->size <= 4) {
+ *(int64_t *)dp = (int64_t)*(int32_t *)dp; /* Sign-extend to 64 bit. */
+ }
+#endif
#if LJ_TARGET_X64 && LJ_ABI_WIN
if (isva) { /* Windows/x64 mirrors varargs in both register sets. */
if (nfpr == ngpr)
@@ -904,7 +1068,7 @@ static int ccall_set_args(lua_State *L, CTState *cts, CType *ct,
cc->fpr[nfpr-1].d[0] = cc->fpr[nfpr-2].d[1]; /* Split complex double. */
cc->fpr[nfpr-2].d[1] = 0;
}
-#elif LJ_TARGET_ARM64
+#elif LJ_TARGET_ARM64 || (LJ_TARGET_MIPS64 && !LJ_ABI_SOFTFP)
if (isfp == 2 && (uint8_t *)dp < (uint8_t *)cc->stack) {
/* Split float HFA or complex float into separate registers. */
CTSize i = (sz >> 2) - 1;
@@ -951,7 +1115,8 @@ static int ccall_get_results(lua_State *L, CTState *cts, CType *ct,
CCALL_HANDLE_COMPLEXRET2
return 1; /* One GC step. */
}
- if (LJ_BE && ctype_isinteger_or_bool(ctr->info) && ctr->size < CTSIZE_PTR)
+ if (LJ_BE && ctr->size < CTSIZE_PTR &&
+ (ctype_isinteger_or_bool(ctr->info) || ctype_isenum(ctr->info)))
sp += (CTSIZE_PTR - ctr->size);
#if CCALL_NUM_FPR
if (ctype_isfp(ctr->info) || ctype_isvector(ctr->info))
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.h
index a77afae3f87..59f664817a2 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ccall.h
@@ -1,6 +1,6 @@
/*
** FFI C call handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_CCALL_H
@@ -79,8 +79,8 @@ typedef union FPRArg {
typedef intptr_t GPRArg;
typedef union FPRArg {
double d;
- float f;
- uint32_t u32;
+ struct { LJ_ENDIAN_LOHI(float f; , float g;) };
+ struct { LJ_ENDIAN_LOHI(uint32_t lo; , uint32_t hi;) };
} FPRArg;
#elif LJ_TARGET_PPC
@@ -95,11 +95,11 @@ typedef union FPRArg {
typedef intptr_t GPRArg;
typedef double FPRArg;
-#elif LJ_TARGET_MIPS
+#elif LJ_TARGET_MIPS32
#define CCALL_NARG_GPR 4
#define CCALL_NARG_FPR (LJ_ABI_SOFTFP ? 0 : 2)
-#define CCALL_NRET_GPR 2
+#define CCALL_NRET_GPR (LJ_ABI_SOFTFP ? 4 : 2)
#define CCALL_NRET_FPR (LJ_ABI_SOFTFP ? 0 : 2)
#define CCALL_SPS_EXTRA 7
#define CCALL_SPS_FREE 1
@@ -110,6 +110,22 @@ typedef union FPRArg {
struct { LJ_ENDIAN_LOHI(float f; , float g;) };
} FPRArg;
+#elif LJ_TARGET_MIPS64
+
+/* FP args are positional and overlay the GPR array. */
+#define CCALL_NARG_GPR 8
+#define CCALL_NARG_FPR 0
+#define CCALL_NRET_GPR 2
+#define CCALL_NRET_FPR (LJ_ABI_SOFTFP ? 0 : 2)
+#define CCALL_SPS_EXTRA 3
+#define CCALL_SPS_FREE 1
+
+typedef intptr_t GPRArg;
+typedef union FPRArg {
+ double d;
+ struct { LJ_ENDIAN_LOHI(float f; , float g;) };
+} FPRArg;
+
#else
#error "Missing calling convention definitions for this architecture"
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.c
index 43c29e41265..846827b119b 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.c
@@ -1,6 +1,6 @@
/*
** FFI C callback handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "lj_obj.h"
@@ -63,9 +63,13 @@ static MSize CALLBACK_OFS2SLOT(MSize ofs)
#define CALLBACK_MCODE_HEAD 24
-#elif LJ_TARGET_MIPS
+#elif LJ_TARGET_MIPS32
-#define CALLBACK_MCODE_HEAD 24
+#define CALLBACK_MCODE_HEAD 20
+
+#elif LJ_TARGET_MIPS64
+
+#define CALLBACK_MCODE_HEAD 52
#else
@@ -169,16 +173,16 @@ static void callback_mcode_init(global_State *g, uint32_t *page)
uint32_t *p = page;
void *target = (void *)lj_vm_ffi_callback;
MSize slot;
- *p++ = A64I_LDRLx | A64F_D(RID_X11) | A64F_S19(4);
- *p++ = A64I_LDRLx | A64F_D(RID_X10) | A64F_S19(5);
- *p++ = A64I_BR | A64F_N(RID_X11);
- *p++ = A64I_NOP;
+ *p++ = A64I_LE(A64I_LDRLx | A64F_D(RID_X11) | A64F_S19(4));
+ *p++ = A64I_LE(A64I_LDRLx | A64F_D(RID_X10) | A64F_S19(5));
+ *p++ = A64I_LE(A64I_BR | A64F_N(RID_X11));
+ *p++ = A64I_LE(A64I_NOP);
((void **)p)[0] = target;
((void **)p)[1] = g;
p += 4;
for (slot = 0; slot < CALLBACK_MAX_SLOT; slot++) {
- *p++ = A64I_MOVZw | A64F_D(RID_X9) | A64F_U16(slot);
- *p = A64I_B | A64F_S26((page-p) & 0x03ffffffu);
+ *p++ = A64I_LE(A64I_MOVZw | A64F_D(RID_X9) | A64F_U16(slot));
+ *p = A64I_LE(A64I_B | A64F_S26((page-p) & 0x03ffffffu));
p++;
}
lua_assert(p - page <= CALLBACK_MCODE_SIZE);
@@ -206,14 +210,27 @@ static void callback_mcode_init(global_State *g, uint32_t *page)
static void callback_mcode_init(global_State *g, uint32_t *page)
{
uint32_t *p = page;
- void *target = (void *)lj_vm_ffi_callback;
+ uintptr_t target = (uintptr_t)(void *)lj_vm_ffi_callback;
+ uintptr_t ug = (uintptr_t)(void *)g;
MSize slot;
- *p++ = MIPSI_SW | MIPSF_T(RID_R1)|MIPSF_S(RID_SP) | 0;
- *p++ = MIPSI_LUI | MIPSF_T(RID_R3) | (u32ptr(target) >> 16);
- *p++ = MIPSI_LUI | MIPSF_T(RID_R2) | (u32ptr(g) >> 16);
- *p++ = MIPSI_ORI | MIPSF_T(RID_R3)|MIPSF_S(RID_R3) |(u32ptr(target)&0xffff);
+#if LJ_TARGET_MIPS32
+ *p++ = MIPSI_LUI | MIPSF_T(RID_R3) | (target >> 16);
+ *p++ = MIPSI_LUI | MIPSF_T(RID_R2) | (ug >> 16);
+#else
+ *p++ = MIPSI_LUI | MIPSF_T(RID_R3) | (target >> 48);
+ *p++ = MIPSI_LUI | MIPSF_T(RID_R2) | (ug >> 48);
+ *p++ = MIPSI_ORI | MIPSF_T(RID_R3)|MIPSF_S(RID_R3) | ((target >> 32) & 0xffff);
+ *p++ = MIPSI_ORI | MIPSF_T(RID_R2)|MIPSF_S(RID_R2) | ((ug >> 32) & 0xffff);
+ *p++ = MIPSI_DSLL | MIPSF_D(RID_R3)|MIPSF_T(RID_R3) | MIPSF_A(16);
+ *p++ = MIPSI_DSLL | MIPSF_D(RID_R2)|MIPSF_T(RID_R2) | MIPSF_A(16);
+ *p++ = MIPSI_ORI | MIPSF_T(RID_R3)|MIPSF_S(RID_R3) | ((target >> 16) & 0xffff);
+ *p++ = MIPSI_ORI | MIPSF_T(RID_R2)|MIPSF_S(RID_R2) | ((ug >> 16) & 0xffff);
+ *p++ = MIPSI_DSLL | MIPSF_D(RID_R3)|MIPSF_T(RID_R3) | MIPSF_A(16);
+ *p++ = MIPSI_DSLL | MIPSF_D(RID_R2)|MIPSF_T(RID_R2) | MIPSF_A(16);
+#endif
+ *p++ = MIPSI_ORI | MIPSF_T(RID_R3)|MIPSF_S(RID_R3) | (target & 0xffff);
*p++ = MIPSI_JR | MIPSF_S(RID_R3);
- *p++ = MIPSI_ORI | MIPSF_T(RID_R2)|MIPSF_S(RID_R2) | (u32ptr(g)&0xffff);
+ *p++ = MIPSI_ORI | MIPSF_T(RID_R2)|MIPSF_S(RID_R2) | (ug & 0xffff);
for (slot = 0; slot < CALLBACK_MAX_SLOT; slot++) {
*p = MIPSI_B | ((page-p-1) & 0x0000ffffu);
p++;
@@ -425,7 +442,7 @@ void lj_ccallback_mcode_free(CTState *cts)
if (ctype_isfp(ctr->info) && ctr->size == sizeof(float)) \
*(double *)dp = *(float *)dp; /* FPRs always hold doubles. */
-#elif LJ_TARGET_MIPS
+#elif LJ_TARGET_MIPS32
#define CALLBACK_HANDLE_GPR \
if (n > 1) ngpr = (ngpr + 1u) & ~1u; /* Align to regpair. */ \
@@ -455,6 +472,29 @@ void lj_ccallback_mcode_free(CTState *cts)
if (ctype_isfp(ctr->info) && ctr->size == sizeof(float)) \
((float *)dp)[1] = *(float *)dp;
+#elif LJ_TARGET_MIPS64
+
+#if !LJ_ABI_SOFTFP /* MIPS64 hard-float */
+#define CALLBACK_HANDLE_REGARG \
+ if (ngpr + n <= maxgpr) { \
+ sp = isfp ? (void*) &cts->cb.fpr[ngpr] : (void*) &cts->cb.gpr[ngpr]; \
+ ngpr += n; \
+ goto done; \
+ }
+#else /* MIPS64 soft-float */
+#define CALLBACK_HANDLE_REGARG \
+ if (ngpr + n <= maxgpr) { \
+ UNUSED(isfp); \
+ sp = (void*) &cts->cb.gpr[ngpr]; \
+ ngpr += n; \
+ goto done; \
+ }
+#endif
+
+#define CALLBACK_HANDLE_RET \
+ if (ctype_isfp(ctr->info) && ctr->size == sizeof(float)) \
+ ((float *)dp)[1] = *(float *)dp;
+
#else
#error "Missing calling convention definitions for this architecture"
#endif
@@ -542,7 +582,11 @@ static void callback_conv_args(CTState *cts, lua_State *L)
nsp += n;
done:
- if (LJ_BE && cta->size < CTSIZE_PTR)
+ if (LJ_BE && cta->size < CTSIZE_PTR
+#if LJ_TARGET_MIPS64
+ && !(isfp && nsp)
+#endif
+ )
sp = (void *)((uint8_t *)sp + CTSIZE_PTR-cta->size);
gcsteps += lj_cconv_tv_ct(cts, cta, 0, o++, sp);
}
@@ -580,6 +624,10 @@ static void callback_conv_result(CTState *cts, lua_State *L, TValue *o)
if (ctype_isfp(ctr->info))
dp = (uint8_t *)&cts->cb.fpr[0];
#endif
+#if LJ_TARGET_ARM64 && LJ_BE
+ if (ctype_isfp(ctr->info) && ctr->size == sizeof(float))
+ dp = (uint8_t *)&cts->cb.fpr[0].f[1];
+#endif
lj_cconv_ct_tv(cts, ctr, dp, o, 0);
#ifdef CALLBACK_HANDLE_RET
CALLBACK_HANDLE_RET
@@ -593,6 +641,12 @@ static void callback_conv_result(CTState *cts, lua_State *L, TValue *o)
*(int32_t *)dp = ctr->size == 1 ? (int32_t)*(int8_t *)dp :
(int32_t)*(int16_t *)dp;
}
+#if LJ_TARGET_MIPS64 || (LJ_TARGET_ARM64 && LJ_BE)
+ /* Always sign-extend results to 64 bits. Even a soft-fp 'float'. */
+ if (ctr->size <= 4 &&
+ (LJ_ABI_SOFTFP || ctype_isinteger_or_bool(ctr->info)))
+ *(int64_t *)dp = (int64_t)*(int32_t *)dp;
+#endif
#if LJ_TARGET_X86
if (ctype_isfp(ctr->info))
cts->cb.gpr[2] = ctr->size == sizeof(float) ? 1 : 2;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.h
index 99c778d1a50..a8cdad38630 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ccallback.h
@@ -1,6 +1,6 @@
/*
** FFI C callback handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_CCALLBACK_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.c
index 81f3c871c57..13b8230dc83 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.c
@@ -1,6 +1,6 @@
/*
** C type conversions.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "lj_obj.h"
@@ -448,8 +448,10 @@ int lj_cconv_tv_bf(CTState *cts, CType *s, TValue *o, uint8_t *sp)
setintV(o, (int32_t)val);
}
} else {
+ uint32_t b = (val >> pos) & 1;
lua_assert(bsz == 1);
- setboolV(o, (val >> pos) & 1);
+ setboolV(o, b);
+ setboolV(&cts->g->tmptv2, b); /* Remember for trace recorder. */
}
return 0; /* No GC step needed. */
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.h
index 35e72cbb00a..0a0b66c909a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_cconv.h
@@ -1,6 +1,6 @@
/*
** C type conversions.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_CCONV_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.c
index 05e27dad1ab..68e16d76fc9 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.c
@@ -1,6 +1,6 @@
/*
** C data management.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "lj_obj.h"
@@ -93,11 +93,13 @@ void lj_cdata_setfin(lua_State *L, GCcdata *cd, GCobj *obj, uint32_t it)
setcdataV(L, &tmp, cd);
lj_gc_anybarriert(L, t);
tv = lj_tab_set(L, t, &tmp);
- setgcV(L, tv, obj, it);
- if (!tvisnil(tv))
- cd->marked |= LJ_GC_CDATA_FIN;
- else
+ if (it == LJ_TNIL) {
+ setnilV(tv);
cd->marked &= ~LJ_GC_CDATA_FIN;
+ } else {
+ setgcV(L, tv, obj, it);
+ cd->marked |= LJ_GC_CDATA_FIN;
+ }
}
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.h
index ee5b740dbd7..5bb0f5dca24 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_cdata.h
@@ -1,6 +1,6 @@
/*
** C data management.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_CDATA_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_clib.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_clib.c
index 922dbfaec8c..614265903ac 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_clib.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_clib.c
@@ -1,6 +1,6 @@
/*
** FFI C library loader.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "lj_obj.h"
@@ -40,7 +40,7 @@ LJ_NORET LJ_NOINLINE static void clib_error_(lua_State *L)
#define clib_error(L, fmt, name) clib_error_(L)
-#if defined(__CYGWIN__)
+#if LJ_TARGET_CYGWIN
#define CLIB_SOPREFIX "cyg"
#else
#define CLIB_SOPREFIX "lib"
@@ -48,7 +48,7 @@ LJ_NORET LJ_NOINLINE static void clib_error_(lua_State *L)
#if LJ_TARGET_OSX
#define CLIB_SOEXT "%s.dylib"
-#elif defined(__CYGWIN__)
+#elif LJ_TARGET_CYGWIN
#define CLIB_SOEXT "%s.dll"
#else
#define CLIB_SOEXT "%s.so"
@@ -57,14 +57,14 @@ LJ_NORET LJ_NOINLINE static void clib_error_(lua_State *L)
static const char *clib_extname(lua_State *L, const char *name)
{
if (!strchr(name, '/')
-#ifdef __CYGWIN__
+#if LJ_TARGET_CYGWIN
&& !strchr(name, '\\')
#endif
) {
if (!strchr(name, '.')) {
name = lj_strfmt_pushf(L, CLIB_SOEXT, name);
L->top--;
-#ifdef __CYGWIN__
+#if LJ_TARGET_CYGWIN
} else {
return name;
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_clib.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_clib.h
index ce3ab85a8bb..fcc9dac5920 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_clib.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_clib.h
@@ -1,6 +1,6 @@
/*
** FFI C library loader.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_CLIB_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.c
index 7ec4a5eebfa..83cfd1128b3 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.c
@@ -1,6 +1,6 @@
/*
** C declaration parser.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "lj_obj.h"
@@ -297,13 +297,17 @@ static CPToken cp_next_(CPState *cp)
else return '/';
break;
case '|':
- if (cp_get(cp) != '|') return '|'; cp_get(cp); return CTOK_OROR;
+ if (cp_get(cp) != '|') return '|';
+ cp_get(cp); return CTOK_OROR;
case '&':
- if (cp_get(cp) != '&') return '&'; cp_get(cp); return CTOK_ANDAND;
+ if (cp_get(cp) != '&') return '&';
+ cp_get(cp); return CTOK_ANDAND;
case '=':
- if (cp_get(cp) != '=') return '='; cp_get(cp); return CTOK_EQ;
+ if (cp_get(cp) != '=') return '=';
+ cp_get(cp); return CTOK_EQ;
case '!':
- if (cp_get(cp) != '=') return '!'; cp_get(cp); return CTOK_NE;
+ if (cp_get(cp) != '=') return '!';
+ cp_get(cp); return CTOK_NE;
case '<':
if (cp_get(cp) == '=') { cp_get(cp); return CTOK_LE; }
else if (cp->c == '<') { cp_get(cp); return CTOK_SHL; }
@@ -313,7 +317,8 @@ static CPToken cp_next_(CPState *cp)
else if (cp->c == '>') { cp_get(cp); return CTOK_SHR; }
return '>';
case '-':
- if (cp_get(cp) != '>') return '-'; cp_get(cp); return CTOK_DEREF;
+ if (cp_get(cp) != '>') return '-';
+ cp_get(cp); return CTOK_DEREF;
case '$':
return cp_param(cp);
case '\0': return CTOK_EOF;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.h
index bfd1a9bb4d0..bad1060bbc9 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_cparse.h
@@ -1,6 +1,6 @@
/*
** C declaration parser.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_CPARSE_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.c
index c0f7e3d7957..e32ae23e187 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.c
@@ -1,6 +1,6 @@
/*
** Trace recorder for C data operations.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_ffrecord_c
@@ -712,6 +712,19 @@ static TRef crec_reassoc_ofs(jit_State *J, TRef tr, ptrdiff_t *ofsp, MSize sz)
return tr;
}
+/* Tailcall to function. */
+static void crec_tailcall(jit_State *J, RecordFFData *rd, cTValue *tv)
+{
+ TRef kfunc = lj_ir_kfunc(J, funcV(tv));
+#if LJ_FR2
+ J->base[-2] = kfunc;
+ J->base[-1] = TREF_FRAME;
+#else
+ J->base[-1] = kfunc | TREF_FRAME;
+#endif
+ rd->nres = -1; /* Pending tailcall. */
+}
+
/* Record ctype __index/__newindex metamethods. */
static void crec_index_meta(jit_State *J, CTState *cts, CType *ct,
RecordFFData *rd)
@@ -721,8 +734,7 @@ static void crec_index_meta(jit_State *J, CTState *cts, CType *ct,
if (!tv)
lj_trace_err(J, LJ_TRERR_BADTYPE);
if (tvisfunc(tv)) {
- J->base[-1] = lj_ir_kfunc(J, funcV(tv)) | TREF_FRAME;
- rd->nres = -1; /* Pending tailcall. */
+ crec_tailcall(J, rd, tv);
} else if (rd->data == 0 && tvistab(tv) && tref_isstr(J->base[1])) {
/* Specialize to result of __index lookup. */
cTValue *o = lj_tab_get(J->L, tabV(tv), &rd->argv[1]);
@@ -739,6 +751,48 @@ static void crec_index_meta(jit_State *J, CTState *cts, CType *ct,
}
}
+/* Record bitfield load/store. */
+static void crec_index_bf(jit_State *J, RecordFFData *rd, TRef ptr, CTInfo info)
+{
+ IRType t = IRT_I8 + 2*lj_fls(ctype_bitcsz(info)) + ((info&CTF_UNSIGNED)?1:0);
+ TRef tr = emitir(IRT(IR_XLOAD, t), ptr, 0);
+ CTSize pos = ctype_bitpos(info), bsz = ctype_bitbsz(info), shift = 32 - bsz;
+ lua_assert(t <= IRT_U32); /* NYI: 64 bit bitfields. */
+ if (rd->data == 0) { /* __index metamethod. */
+ if ((info & CTF_BOOL)) {
+ tr = emitir(IRTI(IR_BAND), tr, lj_ir_kint(J, (int32_t)((1u << pos))));
+ /* Assume not equal to zero. Fixup and emit pending guard later. */
+ lj_ir_set(J, IRTGI(IR_NE), tr, lj_ir_kint(J, 0));
+ J->postproc = LJ_POST_FIXGUARD;
+ tr = TREF_TRUE;
+ } else if (!(info & CTF_UNSIGNED)) {
+ tr = emitir(IRTI(IR_BSHL), tr, lj_ir_kint(J, shift - pos));
+ tr = emitir(IRTI(IR_BSAR), tr, lj_ir_kint(J, shift));
+ } else {
+ lua_assert(bsz < 32); /* Full-size fields cannot end up here. */
+ tr = emitir(IRTI(IR_BSHR), tr, lj_ir_kint(J, pos));
+ tr = emitir(IRTI(IR_BAND), tr, lj_ir_kint(J, (int32_t)((1u << bsz)-1)));
+ /* We can omit the U32 to NUM conversion, since bsz < 32. */
+ }
+ J->base[0] = tr;
+ } else { /* __newindex metamethod. */
+ CTState *cts = ctype_ctsG(J2G(J));
+ CType *ct = ctype_get(cts,
+ (info & CTF_BOOL) ? CTID_BOOL :
+ (info & CTF_UNSIGNED) ? CTID_UINT32 : CTID_INT32);
+ int32_t mask = (int32_t)(((1u << bsz)-1) << pos);
+ TRef sp = crec_ct_tv(J, ct, 0, J->base[2], &rd->argv[2]);
+ sp = emitir(IRTI(IR_BSHL), sp, lj_ir_kint(J, pos));
+ /* Use of the target type avoids forwarding conversions. */
+ sp = emitir(IRT(IR_BAND, t), sp, lj_ir_kint(J, mask));
+ tr = emitir(IRT(IR_BAND, t), tr, lj_ir_kint(J, (int32_t)~mask));
+ tr = emitir(IRT(IR_BOR, t), tr, sp);
+ emitir(IRT(IR_XSTORE, t), ptr, tr);
+ rd->nres = 0;
+ J->needsnap = 1;
+ }
+}
+
void LJ_FASTCALL recff_cdata_index(jit_State *J, RecordFFData *rd)
{
TRef idx, ptr = J->base[0];
@@ -813,6 +867,7 @@ again:
CType *fct;
fct = lj_ctype_getfield(cts, ct, name, &fofs);
if (fct) {
+ ofs += (ptrdiff_t)fofs;
/* Always specialize to the field name. */
emitir(IRTG(IR_EQ, IRT_STR), idx, lj_ir_kstr(J, name));
if (ctype_isconstval(fct->info)) {
@@ -824,12 +879,14 @@ again:
J->base[0] = lj_ir_kint(J, (int32_t)fct->size);
return; /* Interpreter will throw for newindex. */
} else if (ctype_isbitfield(fct->info)) {
- lj_trace_err(J, LJ_TRERR_NYICONV);
+ if (ofs)
+ ptr = emitir(IRT(IR_ADD, IRT_PTR), ptr, lj_ir_kintp(J, ofs));
+ crec_index_bf(J, rd, ptr, fct->info);
+ return;
} else {
lua_assert(ctype_isfield(fct->info));
sid = ctype_cid(fct->info);
}
- ofs += (ptrdiff_t)fofs;
}
} else if (ctype_iscomplex(ct->info)) {
if (name->len == 2 &&
@@ -1119,20 +1176,20 @@ static void crec_snap_caller(jit_State *J)
lua_State *L = J->L;
TValue *base = L->base, *top = L->top;
const BCIns *pc = J->pc;
- TRef ftr = J->base[-1];
+ TRef ftr = J->base[-1-LJ_FR2];
ptrdiff_t delta;
if (!frame_islua(base-1) || J->framedepth <= 0)
lj_trace_err(J, LJ_TRERR_NYICALL);
J->pc = frame_pc(base-1); delta = 1+LJ_FR2+bc_a(J->pc[-1]);
L->top = base; L->base = base - delta;
- J->base[-1] = TREF_FALSE;
+ J->base[-1-LJ_FR2] = TREF_FALSE;
J->base -= delta; J->baseslot -= (BCReg)delta;
- J->maxslot = (BCReg)delta; J->framedepth--;
+ J->maxslot = (BCReg)delta-LJ_FR2; J->framedepth--;
lj_snap_add(J);
L->base = base; L->top = top;
J->framedepth++; J->maxslot = 1;
J->base += delta; J->baseslot += (BCReg)delta;
- J->base[-1] = ftr; J->pc = pc;
+ J->base[-1-LJ_FR2] = ftr; J->pc = pc;
}
/* Record function call. */
@@ -1224,8 +1281,7 @@ void LJ_FASTCALL recff_cdata_call(jit_State *J, RecordFFData *rd)
tv = lj_ctype_meta(cts, ctype_isptr(ct->info) ? ctype_cid(ct->info) : id, mm);
if (tv) {
if (tvisfunc(tv)) {
- J->base[-1] = lj_ir_kfunc(J, funcV(tv)) | TREF_FRAME;
- rd->nres = -1; /* Pending tailcall. */
+ crec_tailcall(J, rd, tv);
return;
}
} else if (mm == MM_new) {
@@ -1238,7 +1294,7 @@ void LJ_FASTCALL recff_cdata_call(jit_State *J, RecordFFData *rd)
static TRef crec_arith_int64(jit_State *J, TRef *sp, CType **s, MMS mm)
{
- if (ctype_isnum(s[0]->info) && ctype_isnum(s[1]->info)) {
+ if (sp[0] && sp[1] && ctype_isnum(s[0]->info) && ctype_isnum(s[1]->info)) {
IRType dt;
CTypeID id;
TRef tr;
@@ -1296,6 +1352,7 @@ static TRef crec_arith_ptr(jit_State *J, TRef *sp, CType **s, MMS mm)
{
CTState *cts = ctype_ctsG(J2G(J));
CType *ctp = s[0];
+ if (!(sp[0] && sp[1])) return 0;
if (ctype_isptr(ctp->info) || ctype_isrefarray(ctp->info)) {
if ((mm == MM_sub || mm == MM_eq || mm == MM_lt || mm == MM_le) &&
(ctype_isptr(s[1]->info) || ctype_isrefarray(s[1]->info))) {
@@ -1373,8 +1430,7 @@ static TRef crec_arith_meta(jit_State *J, TRef *sp, CType **s, CTState *cts,
}
if (tv) {
if (tvisfunc(tv)) {
- J->base[-1] = lj_ir_kfunc(J, funcV(tv)) | TREF_FRAME;
- rd->nres = -1; /* Pending tailcall. */
+ crec_tailcall(J, rd, tv);
return 0;
} /* NYI: non-function metamethods. */
} else if ((MMS)rd->data == MM_eq) { /* Fallback cdata pointer comparison. */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.h
index 7a058813191..c165def4755 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_crecord.h
@@ -1,6 +1,6 @@
/*
** Trace recorder for C data operations.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_CRECORD_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.c
index bc47cc5b927..0ea89c7486a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.c
@@ -1,6 +1,6 @@
/*
** C type management.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include "lj_obj.h"
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.h
index e9b426f0dc9..0c220a88866 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ctype.h
@@ -1,6 +1,6 @@
/*
** C type management.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_CTYPE_H
@@ -42,18 +42,18 @@ LJ_STATIC_ASSERT(((int)CT_STRUCT & (int)CT_ARRAY) == CT_STRUCT);
** ---------- info ------------
** |type flags... A cid | size | sib | next | name |
** +----------------------------+--------+-------+-------+-------+--
-** |NUM BFvcUL.. A | size | | type | |
-** |STRUCT ..vcU..V A | size | field | name? | name? |
-** |PTR ..vcR... A cid | size | | type | |
-** |ARRAY VCvc...V A cid | size | | type | |
-** |VOID ..vc.... A | size | | type | |
+** |NUM BFcvUL.. A | size | | type | |
+** |STRUCT ..cvU..V A | size | field | name? | name? |
+** |PTR ..cvR... A cid | size | | type | |
+** |ARRAY VCcv...V A cid | size | | type | |
+** |VOID ..cv.... A | size | | type | |
** |ENUM A cid | size | const | name? | name? |
** |FUNC ....VS.. cc cid | nargs | field | name? | name? |
** |TYPEDEF cid | | | name | name |
** |ATTRIB attrnum cid | attr | sib? | type? | |
** |FIELD cid | offset | field | | name? |
-** |BITFIELD B.vcU csz bsz pos | offset | field | | name? |
-** |CONSTVAL c cid | value | const | name | name |
+** |BITFIELD B.cvU csz bsz pos | offset | field | | name? |
+** |CONSTVAL c cid | value | const | name | name |
** |EXTERN cid | | sib? | name | name |
** |KW tok | size | | name | name |
** +----------------------------+--------+-------+-------+-------+--
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_debug.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_debug.c
index 153035a90b2..959dc289c7c 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_debug.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_debug.c
@@ -1,6 +1,6 @@
/*
** Debugging and introspection.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_debug_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_debug.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_debug.h
index 1028688d145..5917c00bc6f 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_debug.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_debug.h
@@ -1,6 +1,6 @@
/*
** Debugging and introspection.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_DEBUG_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_def.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_def.h
index 6b898838ef2..517921ddecf 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_def.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_def.h
@@ -1,6 +1,6 @@
/*
** LuaJIT common internal definitions.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_DEF_H
@@ -95,6 +95,9 @@ typedef unsigned int uintptr_t;
#define U64x(hi, lo) (((uint64_t)0x##hi << 32) + (uint64_t)0x##lo)
#define i32ptr(p) ((int32_t)(intptr_t)(void *)(p))
#define u32ptr(p) ((uint32_t)(intptr_t)(void *)(p))
+#define i64ptr(p) ((int64_t)(intptr_t)(void *)(p))
+#define u64ptr(p) ((uint64_t)(intptr_t)(void *)(p))
+#define igcptr(p) (LJ_GC64 ? i64ptr(p) : i32ptr(p))
#define checki8(x) ((x) == (int32_t)(int8_t)(x))
#define checku8(x) ((x) == (int32_t)(uint8_t)(x))
@@ -103,14 +106,8 @@ typedef unsigned int uintptr_t;
#define checki32(x) ((x) == (int32_t)(x))
#define checku32(x) ((x) == (uint32_t)(x))
#define checkptr32(x) ((uintptr_t)(x) == (uint32_t)(uintptr_t)(x))
-#define checkptr47(x) (((uint64_t)(x) >> 47) == 0)
-#if LJ_GC64
-#define checkptrGC(x) (checkptr47((x)))
-#elif LJ_64
-#define checkptrGC(x) (checkptr32((x)))
-#else
-#define checkptrGC(x) 1
-#endif
+#define checkptr47(x) (((uint64_t)(uintptr_t)(x) >> 47) == 0)
+#define checkptrGC(x) (LJ_GC64 ? checkptr47((x)) : LJ_64 ? checkptr32((x)) :1)
/* Every half-decent C compiler transforms this into a rotate instruction. */
#define lj_rol(x, n) (((x)<<(n)) | ((x)>>(-(int)(n)&(8*sizeof(x)-1))))
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.c
index 36b920add7f..5d6795f88e3 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.c
@@ -1,6 +1,6 @@
/*
** Instruction dispatch handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_dispatch_c
@@ -75,7 +75,7 @@ void lj_dispatch_init(GG_State *GG)
for (i = 0; i < GG_NUM_ASMFF; i++)
GG->bcff[i] = BCINS_AD(BC__MAX+i, 0, 0);
#if LJ_TARGET_MIPS
- memcpy(GG->got, dispatch_got, LJ_GOT__MAX*4);
+ memcpy(GG->got, dispatch_got, LJ_GOT__MAX*sizeof(ASMFunction *));
#endif
}
@@ -267,7 +267,7 @@ int luaJIT_setmode(lua_State *L, int idx, int mode)
case LUAJIT_MODE_FUNC:
case LUAJIT_MODE_ALLFUNC:
case LUAJIT_MODE_ALLSUBFUNC: {
- cTValue *tv = idx == 0 ? frame_prev(L->base-1) :
+ cTValue *tv = idx == 0 ? frame_prev(L->base-1)-LJ_FR2 :
idx > 0 ? L->base + (idx-1) : L->top + idx;
GCproto *pt;
if ((idx == 0 || tvisfunc(tv)) && isluafunc(&gcval(tv)->fn))
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.h
index 827080770da..5bda51a2137 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_dispatch.h
@@ -1,6 +1,6 @@
/*
** Instruction dispatch handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_DISPATCH_H
@@ -107,6 +107,7 @@ typedef struct GG_State {
#define J2G(J) (&J2GG(J)->g)
#define G2J(gl) (&G2GG(gl)->J)
#define L2J(L) (&L2GG(L)->J)
+#define GG_G2J (GG_OFS(J) - GG_OFS(g))
#define GG_G2DISP (GG_OFS(dispatch) - GG_OFS(g))
#define GG_DISP2G (GG_OFS(g) - GG_OFS(dispatch))
#define GG_DISP2J (GG_OFS(J) - GG_OFS(dispatch))
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_arm.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_arm.h
index 47fee5fcd85..dee8bdccd19 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_arm.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_arm.h
@@ -1,6 +1,6 @@
/*
** ARM instruction emitter.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/* -- Constant encoding --------------------------------------------------- */
@@ -207,7 +207,7 @@ static void emit_loadi(ASMState *as, Reg r, int32_t i)
#define emit_loada(as, r, addr) emit_loadi(as, (r), i32ptr((addr)))
-static Reg ra_allock(ASMState *as, int32_t k, RegSet allow);
+static Reg ra_allock(ASMState *as, intptr_t k, RegSet allow);
/* Get/set from constant pointer. */
static void emit_lsptr(ASMState *as, ARMIns ai, Reg r, void *p)
@@ -219,8 +219,9 @@ static void emit_lsptr(ASMState *as, ARMIns ai, Reg r, void *p)
#if !LJ_SOFTFP
/* Load a number constant into an FPR. */
-static void emit_loadn(ASMState *as, Reg r, cTValue *tv)
+static void emit_loadk64(ASMState *as, Reg r, IRIns *ir)
{
+ cTValue *tv = ir_knum(ir);
int32_t i;
if ((as->flags & JIT_F_VFPV3) && !tv->u32.lo) {
uint32_t hi = tv->u32.hi;
@@ -273,7 +274,7 @@ static void emit_call(ASMState *as, void *target)
ptrdiff_t delta = ((char *)target - (char *)p) - 8;
if ((((delta>>2) + 0x00800000) >> 24) == 0) {
if ((delta & 1))
- *p = ARMI_BLX | ((uint32_t)(delta>>2) & 0x00ffffffu) | ((delta&2) << 27);
+ *p = ARMI_BLX | ((uint32_t)(delta>>2) & 0x00ffffffu) | ((delta&2) << 23);
else
*p = ARMI_BL | ((uint32_t)(delta>>2) & 0x00ffffffu);
} else { /* Target out of range: need indirect call. But don't use R0-R3. */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_mips.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_mips.h
index fdebe94b49b..8a9ee24dce7 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_mips.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_mips.h
@@ -1,8 +1,30 @@
/*
** MIPS instruction emitter.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
+#if LJ_64
+static intptr_t get_k64val(IRIns *ir)
+{
+ if (ir->o == IR_KINT64) {
+ return (intptr_t)ir_kint64(ir)->u64;
+ } else if (ir->o == IR_KGC) {
+ return (intptr_t)ir_kgc(ir);
+ } else if (ir->o == IR_KPTR || ir->o == IR_KKPTR) {
+ return (intptr_t)ir_kptr(ir);
+ } else {
+ lua_assert(ir->o == IR_KINT || ir->o == IR_KNULL);
+ return ir->i; /* Sign-extended. */
+ }
+}
+#endif
+
+#if LJ_64
+#define get_kval(ir) get_k64val(ir)
+#else
+#define get_kval(ir) ((ir)->i)
+#endif
+
/* -- Emit basic instructions --------------------------------------------- */
static void emit_dst(ASMState *as, MIPSIns mi, Reg rd, Reg rs, Reg rt)
@@ -35,7 +57,7 @@ static void emit_fgh(ASMState *as, MIPSIns mi, Reg rf, Reg rg, Reg rh)
static void emit_rotr(ASMState *as, Reg dest, Reg src, Reg tmp, uint32_t shift)
{
- if ((as->flags & JIT_F_MIPS32R2)) {
+ if (LJ_64 || (as->flags & JIT_F_MIPSXXR2)) {
emit_dta(as, MIPSI_ROTR, dest, src, shift);
} else {
emit_dst(as, MIPSI_OR, dest, dest, tmp);
@@ -44,13 +66,21 @@ static void emit_rotr(ASMState *as, Reg dest, Reg src, Reg tmp, uint32_t shift)
}
}
+#if LJ_64
+static void emit_tsml(ASMState *as, MIPSIns mi, Reg rt, Reg rs, uint32_t msb,
+ uint32_t lsb)
+{
+ *--as->mcp = mi | MIPSF_T(rt) | MIPSF_S(rs) | MIPSF_M(msb) | MIPSF_L(lsb);
+}
+#endif
+
/* -- Emit loads/stores --------------------------------------------------- */
/* Prefer rematerialization of BASE/L from global_State over spills. */
#define emit_canremat(ref) ((ref) <= REF_BASE)
/* Try to find a one step delta relative to another constant. */
-static int emit_kdelta1(ASMState *as, Reg t, int32_t i)
+static int emit_kdelta1(ASMState *as, Reg t, intptr_t i)
{
RegSet work = ~as->freeset & RSET_GPR;
while (work) {
@@ -58,9 +88,10 @@ static int emit_kdelta1(ASMState *as, Reg t, int32_t i)
IRRef ref = regcost_ref(as->cost[r]);
lua_assert(r != t);
if (ref < ASMREF_L) {
- int32_t delta = i - (ra_iskref(ref) ? ra_krefk(as, ref) : IR(ref)->i);
+ intptr_t delta = (intptr_t)((uintptr_t)i -
+ (uintptr_t)(ra_iskref(ref) ? ra_krefk(as, ref) : get_kval(IR(ref))));
if (checki16(delta)) {
- emit_tsi(as, MIPSI_ADDIU, t, r, delta);
+ emit_tsi(as, MIPSI_AADDIU, t, r, delta);
return 1;
}
}
@@ -76,8 +107,8 @@ static void emit_loadi(ASMState *as, Reg r, int32_t i)
emit_ti(as, MIPSI_LI, r, i);
} else {
if ((i & 0xffff)) {
- int32_t jgl = i32ptr(J2G(as->J));
- if ((uint32_t)(i-jgl) < 65536) {
+ intptr_t jgl = (intptr_t)(void *)J2G(as->J);
+ if ((uintptr_t)(i-jgl) < 65536) {
emit_tsi(as, MIPSI_ADDIU, r, RID_JGL, i-jgl-32768);
return;
} else if (emit_kdelta1(as, r, i)) {
@@ -92,16 +123,48 @@ static void emit_loadi(ASMState *as, Reg r, int32_t i)
}
}
+#if LJ_64
+/* Load a 64 bit constant into a GPR. */
+static void emit_loadu64(ASMState *as, Reg r, uint64_t u64)
+{
+ if (checki32((int64_t)u64)) {
+ emit_loadi(as, r, (int32_t)u64);
+ } else {
+ uint64_t delta = u64 - (uint64_t)(void *)J2G(as->J);
+ if (delta < 65536) {
+ emit_tsi(as, MIPSI_DADDIU, r, RID_JGL, (int32_t)(delta-32768));
+ } else if (emit_kdelta1(as, r, (intptr_t)u64)) {
+ return;
+ } else {
+ if ((u64 & 0xffff)) {
+ emit_tsi(as, MIPSI_ORI, r, r, u64 & 0xffff);
+ }
+ if (((u64 >> 16) & 0xffff)) {
+ emit_dta(as, MIPSI_DSLL, r, r, 16);
+ emit_tsi(as, MIPSI_ORI, r, r, (u64 >> 16) & 0xffff);
+ emit_dta(as, MIPSI_DSLL, r, r, 16);
+ } else {
+ emit_dta(as, MIPSI_DSLL32, r, r, 0);
+ }
+ emit_loadi(as, r, (int32_t)(u64 >> 32));
+ }
+ /* TODO: There are probably more optimization opportunities. */
+ }
+}
+
+#define emit_loada(as, r, addr) emit_loadu64(as, (r), u64ptr((addr)))
+#else
#define emit_loada(as, r, addr) emit_loadi(as, (r), i32ptr((addr)))
+#endif
-static Reg ra_allock(ASMState *as, int32_t k, RegSet allow);
-static void ra_allockreg(ASMState *as, int32_t k, Reg r);
+static Reg ra_allock(ASMState *as, intptr_t k, RegSet allow);
+static void ra_allockreg(ASMState *as, intptr_t k, Reg r);
/* Get/set from constant pointer. */
static void emit_lsptr(ASMState *as, MIPSIns mi, Reg r, void *p, RegSet allow)
{
- int32_t jgl = i32ptr(J2G(as->J));
- int32_t i = i32ptr(p);
+ intptr_t jgl = (intptr_t)(J2G(as->J));
+ intptr_t i = (intptr_t)(p);
Reg base;
if ((uint32_t)(i-jgl) < 65536) {
i = i-jgl-32768;
@@ -112,8 +175,24 @@ static void emit_lsptr(ASMState *as, MIPSIns mi, Reg r, void *p, RegSet allow)
emit_tsi(as, mi, r, base, i);
}
-#define emit_loadn(as, r, tv) \
- emit_lsptr(as, MIPSI_LDC1, ((r) & 31), (void *)(tv), RSET_GPR)
+#if LJ_64
+static void emit_loadk64(ASMState *as, Reg r, IRIns *ir)
+{
+ const uint64_t *k = &ir_k64(ir)->u64;
+ Reg r64 = r;
+ if (rset_test(RSET_FPR, r)) {
+ r64 = RID_TMP;
+ emit_tg(as, MIPSI_DMTC1, r64, r);
+ }
+ if ((uint32_t)((intptr_t)k-(intptr_t)J2G(as->J)) < 65536)
+ emit_lsptr(as, MIPSI_LD, r64, (void *)k, 0);
+ else
+ emit_loadu64(as, r64, *k);
+}
+#else
+#define emit_loadk64(as, r, ir) \
+ emit_lsptr(as, MIPSI_LDC1, ((r) & 31), (void *)&ir_knum((ir))->u64, RSET_GPR)
+#endif
/* Get/set global_State fields. */
static void emit_lsglptr(ASMState *as, MIPSIns mi, Reg r, int32_t ofs)
@@ -122,9 +201,9 @@ static void emit_lsglptr(ASMState *as, MIPSIns mi, Reg r, int32_t ofs)
}
#define emit_getgl(as, r, field) \
- emit_lsglptr(as, MIPSI_LW, (r), (int32_t)offsetof(global_State, field))
+ emit_lsglptr(as, MIPSI_AL, (r), (int32_t)offsetof(global_State, field))
#define emit_setgl(as, r, field) \
- emit_lsglptr(as, MIPSI_SW, (r), (int32_t)offsetof(global_State, field))
+ emit_lsglptr(as, MIPSI_AS, (r), (int32_t)offsetof(global_State, field))
/* Trace number is determined from per-trace exit stubs. */
#define emit_setvmstate(as, i) UNUSED(i)
@@ -157,13 +236,14 @@ static void emit_call(ASMState *as, void *target, int needcfa)
MCode *p = as->mcp;
*--p = MIPSI_NOP;
if ((((uintptr_t)target ^ (uintptr_t)p) >> 28) == 0) {
- *--p = MIPSI_JAL | (((uintptr_t)target >>2) & 0x03ffffffu);
+ *--p = (((uintptr_t)target & 1) ? MIPSI_JALX : MIPSI_JAL) |
+ (((uintptr_t)target >>2) & 0x03ffffffu);
} else { /* Target out of range: need indirect call. */
*--p = MIPSI_JALR | MIPSF_S(RID_CFUNCADDR);
needcfa = 1;
}
as->mcp = p;
- if (needcfa) ra_allockreg(as, i32ptr(target), RID_CFUNCADDR);
+ if (needcfa) ra_allockreg(as, (intptr_t)target, RID_CFUNCADDR);
}
/* -- Emit generic operations --------------------------------------------- */
@@ -184,7 +264,7 @@ static void emit_movrr(ASMState *as, IRIns *ir, Reg dst, Reg src)
static void emit_loadofs(ASMState *as, IRIns *ir, Reg r, Reg base, int32_t ofs)
{
if (r < RID_MAX_GPR)
- emit_tsi(as, MIPSI_LW, r, base, ofs);
+ emit_tsi(as, irt_is64(ir->t) ? MIPSI_LD : MIPSI_LW, r, base, ofs);
else
emit_tsi(as, irt_isnum(ir->t) ? MIPSI_LDC1 : MIPSI_LWC1,
(r & 31), base, ofs);
@@ -194,7 +274,7 @@ static void emit_loadofs(ASMState *as, IRIns *ir, Reg r, Reg base, int32_t ofs)
static void emit_storeofs(ASMState *as, IRIns *ir, Reg r, Reg base, int32_t ofs)
{
if (r < RID_MAX_GPR)
- emit_tsi(as, MIPSI_SW, r, base, ofs);
+ emit_tsi(as, irt_is64(ir->t) ? MIPSI_SD : MIPSI_SW, r, base, ofs);
else
emit_tsi(as, irt_isnum(ir->t) ? MIPSI_SDC1 : MIPSI_SWC1,
(r&31), base, ofs);
@@ -205,7 +285,7 @@ static void emit_addptr(ASMState *as, Reg r, int32_t ofs)
{
if (ofs) {
lua_assert(checki16(ofs));
- emit_tsi(as, MIPSI_ADDIU, r, r, ofs);
+ emit_tsi(as, MIPSI_AADDIU, r, r, ofs);
}
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_ppc.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_ppc.h
index 4eb933ea549..21c3c2ace70 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_ppc.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_ppc.h
@@ -1,6 +1,6 @@
/*
** PPC instruction emitter.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/* -- Emit basic instructions --------------------------------------------- */
@@ -98,7 +98,7 @@ static void emit_loadi(ASMState *as, Reg r, int32_t i)
#define emit_loada(as, r, addr) emit_loadi(as, (r), i32ptr((addr)))
-static Reg ra_allock(ASMState *as, int32_t k, RegSet allow);
+static Reg ra_allock(ASMState *as, intptr_t k, RegSet allow);
/* Get/set from constant pointer. */
static void emit_lsptr(ASMState *as, PPCIns pi, Reg r, void *p, RegSet allow)
@@ -115,8 +115,8 @@ static void emit_lsptr(ASMState *as, PPCIns pi, Reg r, void *p, RegSet allow)
emit_tai(as, pi, r, base, i);
}
-#define emit_loadn(as, r, tv) \
- emit_lsptr(as, PPCI_LFD, ((r) & 31), (void *)(tv), RSET_GPR)
+#define emit_loadk64(as, r, ir) \
+ emit_lsptr(as, PPCI_LFD, ((r) & 31), (void *)&ir_knum((ir))->u64, RSET_GPR)
/* Get/set global_State fields. */
static void emit_lsglptr(ASMState *as, PPCIns pi, Reg r, int32_t ofs)
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_x86.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_x86.h
index caf30859ace..5207f9dada5 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_x86.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_x86.h
@@ -1,6 +1,6 @@
/*
** x86/x64 instruction emitter.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/* -- Emit basic instructions --------------------------------------------- */
@@ -13,10 +13,17 @@
if (rex != 0x40) *--(p) = rex; }
#define FORCE_REX 0x200
#define REX_64 (FORCE_REX|0x080000)
+#define VEX_64 0x800000
#else
#define REXRB(p, rr, rb) ((void)0)
#define FORCE_REX 0
#define REX_64 0
+#define VEX_64 0
+#endif
+#if LJ_GC64
+#define REX_GC64 REX_64
+#else
+#define REX_GC64 0
#endif
#define emit_i8(as, i) (*--as->mcp = (MCode)(i))
@@ -31,6 +38,13 @@ static LJ_AINLINE MCode *emit_op(x86Op xo, Reg rr, Reg rb, Reg rx,
MCode *p, int delta)
{
int n = (int8_t)xo;
+ if (n == -60) { /* VEX-encoded instruction */
+#if LJ_64
+ xo ^= (((rr>>1)&4)+((rx>>2)&2)+((rb>>3)&1))<<13;
+#endif
+ *(uint32_t *)(p+delta-5) = (uint32_t)xo;
+ return p+delta-5;
+ }
#if defined(__GNUC__)
if (__builtin_constant_p(xo) && n == -2)
p[delta-2] = (MCode)(xo >> 24);
@@ -85,26 +99,17 @@ static int32_t ptr2addr(const void *p)
#define ptr2addr(p) (i32ptr((p)))
#endif
-/* op r, [addr] */
-static void emit_rma(ASMState *as, x86Op xo, Reg rr, const void *addr)
-{
- MCode *p = as->mcp;
- *(int32_t *)(p-4) = ptr2addr(addr);
-#if LJ_64
- p[-5] = MODRM(XM_SCALE1, RID_ESP, RID_EBP);
- as->mcp = emit_opm(xo, XM_OFS0, rr, RID_ESP, p, -5);
-#else
- as->mcp = emit_opm(xo, XM_OFS0, rr, RID_EBP, p, -4);
-#endif
-}
-
/* op r, [base+ofs] */
static void emit_rmro(ASMState *as, x86Op xo, Reg rr, Reg rb, int32_t ofs)
{
MCode *p = as->mcp;
x86Mode mode;
if (ra_hasreg(rb)) {
- if (ofs == 0 && (rb&7) != RID_EBP) {
+ if (LJ_GC64 && rb == RID_RIP) {
+ mode = XM_OFS0;
+ p -= 4;
+ *(int32_t *)p = ofs;
+ } else if (ofs == 0 && (rb&7) != RID_EBP) {
mode = XM_OFS0;
} else if (checki8(ofs)) {
*--p = (MCode)ofs;
@@ -202,6 +207,11 @@ static void emit_mrm(ASMState *as, x86Op xo, Reg rr, Reg rb)
*--p = MODRM(XM_SCALE1, RID_ESP, RID_EBP);
rb = RID_ESP;
#endif
+ } else if (LJ_GC64 && rb == RID_RIP) {
+ lua_assert(as->mrm.idx == RID_NONE);
+ mode = XM_OFS0;
+ p -= 4;
+ *(int32_t *)p = as->mrm.ofs;
} else {
if (as->mrm.ofs == 0 && (rb&7) != RID_EBP) {
mode = XM_OFS0;
@@ -255,8 +265,8 @@ static void emit_movmroi(ASMState *as, Reg base, int32_t ofs, int32_t i)
/* Get/set global_State fields. */
#define emit_opgl(as, xo, r, field) \
emit_rma(as, (xo), (r), (void *)&J2G(as->J)->field)
-#define emit_getgl(as, r, field) emit_opgl(as, XO_MOV, (r), field)
-#define emit_setgl(as, r, field) emit_opgl(as, XO_MOVto, (r), field)
+#define emit_getgl(as, r, field) emit_opgl(as, XO_MOV, (r)|REX_GC64, field)
+#define emit_setgl(as, r, field) emit_opgl(as, XO_MOVto, (r)|REX_GC64, field)
#define emit_setvmstate(as, i) \
(emit_i32(as, i), emit_opgl(as, XO_MOVmi, 0, vmstate))
@@ -279,9 +289,21 @@ static void emit_loadi(ASMState *as, Reg r, int32_t i)
}
}
+#if LJ_GC64
+#define dispofs(as, k) \
+ ((intptr_t)((uintptr_t)(k) - (uintptr_t)J2GG(as->J)->dispatch))
+#define mcpofs(as, k) \
+ ((intptr_t)((uintptr_t)(k) - (uintptr_t)as->mcp))
+#define mctopofs(as, k) \
+ ((intptr_t)((uintptr_t)(k) - (uintptr_t)as->mctop))
+/* mov r, addr */
+#define emit_loada(as, r, addr) \
+ emit_loadu64(as, (r), (uintptr_t)(addr))
+#else
/* mov r, addr */
#define emit_loada(as, r, addr) \
emit_loadi(as, (r), ptr2addr((addr)))
+#endif
#if LJ_64
/* mov r, imm64 or shorter 32 bit extended load. */
@@ -293,6 +315,15 @@ static void emit_loadu64(ASMState *as, Reg r, uint64_t u64)
MCode *p = as->mcp;
*(int32_t *)(p-4) = (int32_t)u64;
as->mcp = emit_opm(XO_MOVmi, XM_REG, REX_64, r, p, -4);
+#if LJ_GC64
+ } else if (checki32(dispofs(as, u64))) {
+ emit_rmro(as, XO_LEA, r|REX_64, RID_DISPATCH, (int32_t)dispofs(as, u64));
+ } else if (checki32(mcpofs(as, u64)) && checki32(mctopofs(as, u64))) {
+ /* Since as->realign assumes the code size doesn't change, check
+ ** RIP-relative addressing reachability for both as->mcp and as->mctop.
+ */
+ emit_rmro(as, XO_LEA, r|REX_64, RID_RIP, (int32_t)mcpofs(as, u64));
+#endif
} else { /* Full-size 64 bit load. */
MCode *p = as->mcp;
*(uint64_t *)(p-8) = u64;
@@ -304,13 +335,71 @@ static void emit_loadu64(ASMState *as, Reg r, uint64_t u64)
}
#endif
-/* movsd r, [&tv->n] / xorps r, r */
-static void emit_loadn(ASMState *as, Reg r, cTValue *tv)
+/* op r, [addr] */
+static void emit_rma(ASMState *as, x86Op xo, Reg rr, const void *addr)
{
- if (tvispzero(tv)) /* Use xor only for +0. */
- emit_rr(as, XO_XORPS, r, r);
- else
- emit_rma(as, XO_MOVSD, r, &tv->n);
+#if LJ_GC64
+ if (checki32(dispofs(as, addr))) {
+ emit_rmro(as, xo, rr, RID_DISPATCH, (int32_t)dispofs(as, addr));
+ } else if (checki32(mcpofs(as, addr)) && checki32(mctopofs(as, addr))) {
+ emit_rmro(as, xo, rr, RID_RIP, (int32_t)mcpofs(as, addr));
+ } else if (!checki32((intptr_t)addr) && (xo == XO_MOV || xo == XO_MOVSD)) {
+ emit_rmro(as, xo, rr, rr, 0);
+ emit_loadu64(as, rr, (uintptr_t)addr);
+ } else
+#endif
+ {
+ MCode *p = as->mcp;
+ *(int32_t *)(p-4) = ptr2addr(addr);
+#if LJ_64
+ p[-5] = MODRM(XM_SCALE1, RID_ESP, RID_EBP);
+ as->mcp = emit_opm(xo, XM_OFS0, rr, RID_ESP, p, -5);
+#else
+ as->mcp = emit_opm(xo, XM_OFS0, rr, RID_EBP, p, -4);
+#endif
+ }
+}
+
+/* Load 64 bit IR constant into register. */
+static void emit_loadk64(ASMState *as, Reg r, IRIns *ir)
+{
+ Reg r64;
+ x86Op xo;
+ const uint64_t *k = &ir_k64(ir)->u64;
+ if (rset_test(RSET_FPR, r)) {
+ r64 = r;
+ xo = XO_MOVSD;
+ } else {
+ r64 = r | REX_64;
+ xo = XO_MOV;
+ }
+ if (*k == 0) {
+ emit_rr(as, rset_test(RSET_FPR, r) ? XO_XORPS : XO_ARITH(XOg_XOR), r, r);
+#if LJ_GC64
+ } else if (checki32((intptr_t)k) || checki32(dispofs(as, k)) ||
+ (checki32(mcpofs(as, k)) && checki32(mctopofs(as, k)))) {
+ emit_rma(as, xo, r64, k);
+ } else {
+ if (ir->i) {
+ lua_assert(*k == *(uint64_t*)(as->mctop - ir->i));
+ } else if (as->curins <= as->stopins && rset_test(RSET_GPR, r)) {
+ emit_loadu64(as, r, *k);
+ return;
+ } else {
+ /* If all else fails, add the FP constant at the MCode area bottom. */
+ while ((uintptr_t)as->mcbot & 7) *as->mcbot++ = XI_INT3;
+ *(uint64_t *)as->mcbot = *k;
+ ir->i = (int32_t)(as->mctop - as->mcbot);
+ as->mcbot += 8;
+ as->mclim = as->mcbot + MCLIM_REDZONE;
+ lj_mcode_commitbot(as->J, as->mcbot);
+ }
+ emit_rmro(as, xo, r64, RID_RIP, (int32_t)mcpofs(as, as->mctop - ir->i));
+#else
+ } else {
+ emit_rma(as, xo, r64, k);
+#endif
+ }
}
/* -- Emit control-flow instructions -------------------------------------- */
@@ -412,8 +501,10 @@ static void emit_call_(ASMState *as, MCode *target)
/* Use 64 bit operations to handle 64 bit IR types. */
#if LJ_64
#define REX_64IR(ir, r) ((r) + (irt_is64((ir)->t) ? REX_64 : 0))
+#define VEX_64IR(ir, r) ((r) + (irt_is64((ir)->t) ? VEX_64 : 0))
#else
#define REX_64IR(ir, r) (r)
+#define VEX_64IR(ir, r) (r)
#endif
/* Generic move between two regs. */
@@ -449,9 +540,9 @@ static void emit_addptr(ASMState *as, Reg r, int32_t ofs)
{
if (ofs) {
if ((as->flags & JIT_F_LEA_AGU))
- emit_rmro(as, XO_LEA, r, r, ofs);
+ emit_rmro(as, XO_LEA, r|REX_GC64, r, ofs);
else
- emit_gri(as, XG_ARITHi(XOg_ADD), r, ofs);
+ emit_gri(as, XG_ARITHi(XOg_ADD), r|REX_GC64, ofs);
}
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_err.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_err.c
index a847ca07db2..b6be357e791 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_err.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_err.c
@@ -1,6 +1,6 @@
/*
** Error handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_err_c
@@ -46,7 +46,8 @@
** the wrapper function feature. Lua errors thrown through C++ frames
** cannot be caught by C++ code and C++ destructors are not run.
**
-** EXT is the default on x64 systems, INT is the default on all other systems.
+** EXT is the default on x64 systems and on Windows, INT is the default on all
+** other systems.
**
** EXT can be manually enabled on POSIX systems using GCC and DWARF2 stack
** unwinding with -DLUAJIT_UNWIND_EXTERNAL. *All* C code must be compiled
@@ -55,7 +56,6 @@
** and all C libraries that have callbacks which may be used to call back
** into Lua. C++ code must *not* be compiled with -fno-exceptions.
**
-** EXT cannot be enabled on WIN32 since system exceptions use code-driven SEH.
** EXT is mandatory on WIN64 since the calling convention has an abundance
** of callee-saved registers (rbx, rbp, rsi, rdi, r12-r15, xmm6-xmm15).
** The POSIX/x64 interpreter only saves r12/r13 for INT (e.g. PS4).
@@ -63,7 +63,7 @@
#if defined(__GNUC__) && (LJ_TARGET_X64 || defined(LUAJIT_UNWIND_EXTERNAL)) && !LJ_NO_UNWIND
#define LJ_UNWIND_EXT 1
-#elif LJ_TARGET_X64 && LJ_TARGET_WINDOWS
+#elif LJ_TARGET_WINDOWS
#define LJ_UNWIND_EXT 1
#endif
@@ -384,7 +384,7 @@ static void err_raise_ext(int errcode)
#endif /* LJ_TARGET_ARM */
-#elif LJ_TARGET_X64 && LJ_ABI_WIN
+#elif LJ_ABI_WIN
/*
** Someone in Redmond owes me several days of my life. A lot of this is
@@ -402,6 +402,7 @@ static void err_raise_ext(int errcode)
#define WIN32_LEAN_AND_MEAN
#include <windows.h>
+#if LJ_TARGET_X64
/* Taken from: http://www.nynaeve.net/?p=99 */
typedef struct UndocumentedDispatcherContext {
ULONG64 ControlPc;
@@ -416,11 +417,14 @@ typedef struct UndocumentedDispatcherContext {
ULONG ScopeIndex;
ULONG Fill0;
} UndocumentedDispatcherContext;
+#else
+typedef void *UndocumentedDispatcherContext;
+#endif
/* Another wild guess. */
extern void __DestructExceptionObject(EXCEPTION_RECORD *rec, int nothrow);
-#ifdef MINGW_SDK_INIT
+#if LJ_TARGET_X64 && defined(MINGW_SDK_INIT)
/* Workaround for broken MinGW64 declaration. */
VOID RtlUnwindEx_FIXED(PVOID,PVOID,PVOID,PVOID,PVOID,PVOID) asm("RtlUnwindEx");
#define RtlUnwindEx RtlUnwindEx_FIXED
@@ -434,10 +438,15 @@ VOID RtlUnwindEx_FIXED(PVOID,PVOID,PVOID,PVOID,PVOID,PVOID) asm("RtlUnwindEx");
#define LJ_EXCODE_CHECK(cl) (((cl) ^ LJ_EXCODE) <= 0xff)
#define LJ_EXCODE_ERRCODE(cl) ((int)((cl) & 0xff))
-/* Win64 exception handler for interpreter frame. */
-LJ_FUNCA EXCEPTION_DISPOSITION lj_err_unwind_win64(EXCEPTION_RECORD *rec,
- void *cf, CONTEXT *ctx, UndocumentedDispatcherContext *dispatch)
+/* Windows exception handler for interpreter frame. */
+LJ_FUNCA int lj_err_unwind_win(EXCEPTION_RECORD *rec,
+ void *f, CONTEXT *ctx, UndocumentedDispatcherContext *dispatch)
{
+#if LJ_TARGET_X64
+ void *cf = f;
+#else
+ void *cf = (char *)f - CFRAME_OFS_SEH;
+#endif
lua_State *L = cframe_L(cf);
int errcode = LJ_EXCODE_CHECK(rec->ExceptionCode) ?
LJ_EXCODE_ERRCODE(rec->ExceptionCode) : LUA_ERRRUN;
@@ -455,8 +464,9 @@ LJ_FUNCA EXCEPTION_DISPOSITION lj_err_unwind_win64(EXCEPTION_RECORD *rec,
setstrV(L, L->top++, lj_err_str(L, LJ_ERR_ERRCPP));
} else if (!LJ_EXCODE_CHECK(rec->ExceptionCode)) {
/* Don't catch access violations etc. */
- return ExceptionContinueSearch;
+ return 1; /* ExceptionContinueSearch */
}
+#if LJ_TARGET_X64
/* Unwind the stack and call all handlers for all lower C frames
** (including ourselves) again with EH_UNWINDING set. Then set
** rsp = cf, rax = errcode and jump to the specified target.
@@ -466,9 +476,21 @@ LJ_FUNCA EXCEPTION_DISPOSITION lj_err_unwind_win64(EXCEPTION_RECORD *rec,
lj_vm_unwind_c_eh),
rec, (void *)(uintptr_t)errcode, ctx, dispatch->HistoryTable);
/* RtlUnwindEx should never return. */
+#else
+ UNUSED(ctx);
+ UNUSED(dispatch);
+ /* Call all handlers for all lower C frames (including ourselves) again
+ ** with EH_UNWINDING set. Then call the specified function, passing cf
+ ** and errcode.
+ */
+ lj_vm_rtlunwind(cf, (void *)rec,
+ (cframe_unwind_ff(cf2) && errcode != LUA_YIELD) ?
+ (void *)lj_vm_unwind_ff : (void *)lj_vm_unwind_c, errcode);
+ /* lj_vm_rtlunwind does not return. */
+#endif
}
}
- return ExceptionContinueSearch;
+ return 1; /* ExceptionContinueSearch */
}
/* Raise Windows exception. */
@@ -487,7 +509,7 @@ LJ_NOINLINE void LJ_FASTCALL lj_err_throw(lua_State *L, int errcode)
global_State *g = G(L);
lj_trace_abort(g);
setmref(g->jit_base, NULL);
- L->status = 0;
+ L->status = LUA_OK;
#if LJ_UNWIND_EXT
err_raise_ext(errcode);
/*
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_err.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_err.h
index 6c7576137b5..cba5fb7149e 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_err.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_err.h
@@ -1,6 +1,6 @@
/*
** Error handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_ERR_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_errmsg.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_errmsg.h
index 0ed8f4e7398..060a9f89750 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_errmsg.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_errmsg.h
@@ -1,6 +1,6 @@
/*
** VM error messages.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/* This file may be included multiple times with different ERRDEF macros. */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ff.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_ff.h
index 9af3b12d265..31d65a00dba 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ff.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ff.h
@@ -1,6 +1,6 @@
/*
** Fast function IDs.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_FF_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.c
index d05dc5d710d..dfdee2dbfe1 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.c
@@ -1,6 +1,6 @@
/*
** Fast function call recorder.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_ffrecord_c
@@ -102,42 +102,41 @@ static void recff_stitch(jit_State *J)
ASMFunction cont = lj_cont_stitch;
lua_State *L = J->L;
TValue *base = L->base;
+ BCReg nslot = J->maxslot + 1 + LJ_FR2;
+ TValue *nframe = base + 1 + LJ_FR2;
const BCIns *pc = frame_pc(base-1);
TValue *pframe = frame_prevl(base-1);
- TRef trcont;
- lua_assert(!LJ_FR2); /* TODO_FR2: handle frame shift. */
/* Move func + args up in Lua stack and insert continuation. */
- memmove(&base[1], &base[-1], sizeof(TValue)*(J->maxslot+1));
- setframe_ftsz(base+1, ((char *)(base+1) - (char *)pframe) + FRAME_CONT);
- setcont(base, cont);
+ memmove(&base[1], &base[-1-LJ_FR2], sizeof(TValue)*nslot);
+ setframe_ftsz(nframe, ((char *)nframe - (char *)pframe) + FRAME_CONT);
+ setcont(base-LJ_FR2, cont);
setframe_pc(base, pc);
- setnilV(base-1); /* Incorrect, but rec_check_slots() won't run anymore. */
- L->base += 2;
- L->top += 2;
+ setnilV(base-1-LJ_FR2); /* Incorrect, but rec_check_slots() won't run anymore. */
+ L->base += 2 + LJ_FR2;
+ L->top += 2 + LJ_FR2;
/* Ditto for the IR. */
- memmove(&J->base[1], &J->base[-1], sizeof(TRef)*(J->maxslot+1));
-#if LJ_64
- trcont = lj_ir_kptr(J, (void *)((int64_t)cont-(int64_t)lj_vm_asm_begin));
+ memmove(&J->base[1], &J->base[-1-LJ_FR2], sizeof(TRef)*nslot);
+#if LJ_FR2
+ J->base[2] = TREF_FRAME;
+ J->base[-1] = lj_ir_k64(J, IR_KNUM, u64ptr(contptr(cont)));
+ J->base[0] = lj_ir_k64(J, IR_KNUM, u64ptr(pc)) | TREF_CONT;
#else
- trcont = lj_ir_kptr(J, (void *)cont);
+ J->base[0] = lj_ir_kptr(J, contptr(cont)) | TREF_CONT;
#endif
- J->base[0] = trcont | TREF_CONT;
- J->ktracep = lj_ir_k64_reserve(J);
- lua_assert(irt_toitype_(IRT_P64) == LJ_TTRACE);
- J->base[-1] = emitir(IRT(IR_XLOAD, IRT_P64), lj_ir_kptr(J, &J->ktracep->gcr), 0);
- J->base += 2;
- J->baseslot += 2;
+ J->ktrace = tref_ref((J->base[-1-LJ_FR2] = lj_ir_ktrace(J)));
+ J->base += 2 + LJ_FR2;
+ J->baseslot += 2 + LJ_FR2;
J->framedepth++;
lj_record_stop(J, LJ_TRLINK_STITCH, 0);
/* Undo Lua stack changes. */
- memmove(&base[-1], &base[1], sizeof(TValue)*(J->maxslot+1));
+ memmove(&base[-1-LJ_FR2], &base[1], sizeof(TValue)*nslot);
setframe_pc(base-1, pc);
- L->base -= 2;
- L->top -= 2;
+ L->base -= 2 + LJ_FR2;
+ L->top -= 2 + LJ_FR2;
}
/* Fallback handler for fast functions that are not recorded (yet). */
@@ -179,7 +178,7 @@ static void LJ_FASTCALL recff_nyi(jit_State *J, RecordFFData *rd)
/* Emit BUFHDR for the global temporary buffer. */
static TRef recff_bufhdr(jit_State *J)
{
- return emitir(IRT(IR_BUFHDR, IRT_P32),
+ return emitir(IRT(IR_BUFHDR, IRT_PGC),
lj_ir_kptr(J, &J2G(J)->tmpbuf), IRBUFHDR_RESET);
}
@@ -229,7 +228,7 @@ static void LJ_FASTCALL recff_setmetatable(jit_State *J, RecordFFData *rd)
ix.tab = tr;
copyTV(J->L, &ix.tabv, &rd->argv[0]);
lj_record_mm_lookup(J, &ix, MM_metatable); /* Guard for no __metatable. */
- fref = emitir(IRT(IR_FREF, IRT_P32), tr, IRFL_TAB_META);
+ fref = emitir(IRT(IR_FREF, IRT_PGC), tr, IRFL_TAB_META);
mtref = tref_isnil(mt) ? lj_ir_knull(J, IRT_TAB) : mt;
emitir(IRT(IR_FSTORE, IRT_TAB), fref, mtref);
if (!tref_isnil(mt))
@@ -295,7 +294,7 @@ int32_t lj_ffrecord_select_mode(jit_State *J, TRef tr, TValue *tv)
if (strV(tv)->len == 1) {
emitir(IRTG(IR_EQ, IRT_STR), tr, lj_ir_kstr(J, strV(tv)));
} else {
- TRef trptr = emitir(IRT(IR_STRREF, IRT_P32), tr, lj_ir_kint(J, 0));
+ TRef trptr = emitir(IRT(IR_STRREF, IRT_PGC), tr, lj_ir_kint(J, 0));
TRef trchar = emitir(IRT(IR_XLOAD, IRT_U8), trptr, IRXLOAD_READONLY);
emitir(IRTG(IR_EQ, IRT_INT), trchar, lj_ir_kint(J, '#'));
}
@@ -380,10 +379,10 @@ static int recff_metacall(jit_State *J, RecordFFData *rd, MMS mm)
int errcode;
TValue argv0;
/* Temporarily insert metamethod below object. */
- J->base[1] = J->base[0];
+ J->base[1+LJ_FR2] = J->base[0];
J->base[0] = ix.mobj;
copyTV(J->L, &argv0, &rd->argv[0]);
- copyTV(J->L, &rd->argv[1], &rd->argv[0]);
+ copyTV(J->L, &rd->argv[1+LJ_FR2], &rd->argv[0]);
copyTV(J->L, &rd->argv[0], &ix.mobjv);
/* Need to protect lj_record_tailcall because it may throw. */
errcode = lj_vm_cpcall(J->L, NULL, J, recff_metacall_cp);
@@ -450,6 +449,10 @@ static void LJ_FASTCALL recff_xpairs(jit_State *J, RecordFFData *rd)
static void LJ_FASTCALL recff_pcall(jit_State *J, RecordFFData *rd)
{
if (J->maxslot >= 1) {
+#if LJ_FR2
+ /* Shift function arguments up. */
+ memmove(J->base + 1, J->base, sizeof(TRef) * J->maxslot);
+#endif
lj_record_call(J, 0, J->maxslot - 1);
rd->nres = -1; /* Pending call. */
} /* else: Interpreter will throw. */
@@ -469,13 +472,16 @@ static void LJ_FASTCALL recff_xpcall(jit_State *J, RecordFFData *rd)
TValue argv0, argv1;
TRef tmp;
int errcode;
- lua_assert(!LJ_FR2); /* TODO_FR2: handle different frame setup. */
/* Swap function and traceback. */
tmp = J->base[0]; J->base[0] = J->base[1]; J->base[1] = tmp;
copyTV(J->L, &argv0, &rd->argv[0]);
copyTV(J->L, &argv1, &rd->argv[1]);
copyTV(J->L, &rd->argv[0], &argv1);
copyTV(J->L, &rd->argv[1], &argv0);
+#if LJ_FR2
+ /* Shift function arguments up. */
+ memmove(J->base + 2, J->base + 1, sizeof(TRef) * (J->maxslot-1));
+#endif
/* Need to protect lj_record_call because it may throw. */
errcode = lj_vm_cpcall(J->L, NULL, J, recff_xpcall_cp);
/* Always undo Lua stack swap to avoid confusing the interpreter. */
@@ -504,7 +510,7 @@ static void LJ_FASTCALL recff_getfenv(jit_State *J, RecordFFData *rd)
static void LJ_FASTCALL recff_math_abs(jit_State *J, RecordFFData *rd)
{
TRef tr = lj_ir_tonum(J, J->base[0]);
- J->base[0] = emitir(IRTN(IR_ABS), tr, lj_ir_knum_abs(J));
+ J->base[0] = emitir(IRTN(IR_ABS), tr, lj_ir_ksimd(J, LJ_KSIMD_ABS));
UNUSED(rd);
}
@@ -613,10 +619,8 @@ static void LJ_FASTCALL recff_math_modf(jit_State *J, RecordFFData *rd)
static void LJ_FASTCALL recff_math_pow(jit_State *J, RecordFFData *rd)
{
- TRef tr = lj_ir_tonum(J, J->base[0]);
- if (!tref_isnumber_str(J->base[1]))
- lj_trace_err(J, LJ_TRERR_BADTYPE);
- J->base[0] = lj_opt_narrow_pow(J, tr, J->base[1], &rd->argv[1]);
+ J->base[0] = lj_opt_narrow_pow(J, J->base[0], J->base[1],
+ &rd->argv[0], &rd->argv[1]);
UNUSED(rd);
}
@@ -822,7 +826,7 @@ static void LJ_FASTCALL recff_string_range(jit_State *J, RecordFFData *rd)
/* Also handle empty range here, to avoid extra traces. */
TRef trptr, trslen = emitir(IRTI(IR_SUB), trend, trstart);
emitir(IRTGI(IR_GE), trslen, tr0);
- trptr = emitir(IRT(IR_STRREF, IRT_P32), trstr, trstart);
+ trptr = emitir(IRT(IR_STRREF, IRT_PGC), trstr, trstart);
J->base[0] = emitir(IRT(IR_SNEW, IRT_STR), trptr, trslen);
} else { /* Range underflow: return empty string. */
emitir(IRTGI(IR_LT), trend, trstart);
@@ -838,7 +842,7 @@ static void LJ_FASTCALL recff_string_range(jit_State *J, RecordFFData *rd)
rd->nres = len;
for (i = 0; i < len; i++) {
TRef tmp = emitir(IRTI(IR_ADD), trstart, lj_ir_kint(J, (int32_t)i));
- tmp = emitir(IRT(IR_STRREF, IRT_P32), trstr, tmp);
+ tmp = emitir(IRT(IR_STRREF, IRT_PGC), trstr, tmp);
J->base[i] = emitir(IRT(IR_XLOAD, IRT_U8), tmp, IRXLOAD_READONLY);
}
} else { /* Empty range or range underflow: return no results. */
@@ -860,7 +864,7 @@ static void LJ_FASTCALL recff_string_char(jit_State *J, RecordFFData *rd)
if (i > 1) { /* Concatenate the strings, if there's more than one. */
TRef hdr = recff_bufhdr(J), tr = hdr;
for (i = 0; J->base[i] != 0; i++)
- tr = emitir(IRT(IR_BUFPUT, IRT_P32), tr, J->base[i]);
+ tr = emitir(IRT(IR_BUFPUT, IRT_PGC), tr, J->base[i]);
J->base[0] = emitir(IRT(IR_BUFSTR, IRT_STR), tr, hdr);
}
UNUSED(rd);
@@ -877,14 +881,14 @@ static void LJ_FASTCALL recff_string_rep(jit_State *J, RecordFFData *rd)
emitir(IRTGI(vrep > 1 ? IR_GT : IR_LE), rep, lj_ir_kint(J, 1));
if (vrep > 1) {
TRef hdr2 = recff_bufhdr(J);
- TRef tr2 = emitir(IRT(IR_BUFPUT, IRT_P32), hdr2, sep);
- tr2 = emitir(IRT(IR_BUFPUT, IRT_P32), tr2, str);
+ TRef tr2 = emitir(IRT(IR_BUFPUT, IRT_PGC), hdr2, sep);
+ tr2 = emitir(IRT(IR_BUFPUT, IRT_PGC), tr2, str);
str2 = emitir(IRT(IR_BUFSTR, IRT_STR), tr2, hdr2);
}
}
tr = hdr = recff_bufhdr(J);
if (str2) {
- tr = emitir(IRT(IR_BUFPUT, IRT_P32), tr, str);
+ tr = emitir(IRT(IR_BUFPUT, IRT_PGC), tr, str);
str = str2;
rep = emitir(IRTI(IR_ADD), rep, lj_ir_kint(J, -1));
}
@@ -935,8 +939,8 @@ static void LJ_FASTCALL recff_string_find(jit_State *J, RecordFFData *rd)
if ((J->base[2] && tref_istruecond(J->base[3])) ||
(emitir(IRTG(IR_EQ, IRT_STR), trpat, lj_ir_kstr(J, pat)),
!lj_str_haspattern(pat))) { /* Search for fixed string. */
- TRef trsptr = emitir(IRT(IR_STRREF, IRT_P32), trstr, trstart);
- TRef trpptr = emitir(IRT(IR_STRREF, IRT_P32), trpat, tr0);
+ TRef trsptr = emitir(IRT(IR_STRREF, IRT_PGC), trstr, trstart);
+ TRef trpptr = emitir(IRT(IR_STRREF, IRT_PGC), trpat, tr0);
TRef trslen = emitir(IRTI(IR_SUB), trlen, trstart);
TRef trplen = emitir(IRTI(IR_FLOAD), trpat, IRFL_STR_LEN);
TRef tr = lj_ir_call(J, IRCALL_lj_str_find, trsptr, trpptr, trslen, trplen);
@@ -944,13 +948,13 @@ static void LJ_FASTCALL recff_string_find(jit_State *J, RecordFFData *rd)
if (lj_str_find(strdata(str)+(MSize)start, strdata(pat),
str->len-(MSize)start, pat->len)) {
TRef pos;
- emitir(IRTG(IR_NE, IRT_P32), tr, trp0);
- pos = emitir(IRTI(IR_SUB), tr, emitir(IRT(IR_STRREF, IRT_P32), trstr, tr0));
+ emitir(IRTG(IR_NE, IRT_PGC), tr, trp0);
+ pos = emitir(IRTI(IR_SUB), tr, emitir(IRT(IR_STRREF, IRT_PGC), trstr, tr0));
J->base[0] = emitir(IRTI(IR_ADD), pos, lj_ir_kint(J, 1));
J->base[1] = emitir(IRTI(IR_ADD), pos, trplen);
rd->nres = 2;
} else {
- emitir(IRTG(IR_EQ, IRT_P32), tr, trp0);
+ emitir(IRTG(IR_EQ, IRT_PGC), tr, trp0);
J->base[0] = TREF_NIL;
}
} else { /* Search for pattern. */
@@ -977,7 +981,7 @@ static void LJ_FASTCALL recff_string_format(jit_State *J, RecordFFData *rd)
IRCallID id;
switch (STRFMT_TYPE(sf)) {
case STRFMT_LIT:
- tr = emitir(IRT(IR_BUFPUT, IRT_P32), tr,
+ tr = emitir(IRT(IR_BUFPUT, IRT_PGC), tr,
lj_ir_kstr(J, lj_str_new(J->L, fs.str, fs.len)));
break;
case STRFMT_INT:
@@ -986,7 +990,7 @@ static void LJ_FASTCALL recff_string_format(jit_State *J, RecordFFData *rd)
if (!tref_isinteger(tra))
goto handle_num;
if (sf == STRFMT_INT) { /* Shortcut for plain %d. */
- tr = emitir(IRT(IR_BUFPUT, IRT_P32), tr,
+ tr = emitir(IRT(IR_BUFPUT, IRT_PGC), tr,
emitir(IRT(IR_TOSTR, IRT_STR), tra, IRTOSTR_INT));
} else {
#if LJ_HASFFI
@@ -1016,7 +1020,7 @@ static void LJ_FASTCALL recff_string_format(jit_State *J, RecordFFData *rd)
return;
}
if (sf == STRFMT_STR) /* Shortcut for plain %s. */
- tr = emitir(IRT(IR_BUFPUT, IRT_P32), tr, tra);
+ tr = emitir(IRT(IR_BUFPUT, IRT_PGC), tr, tra);
else if ((sf & STRFMT_T_QUOTED))
tr = lj_ir_call(J, IRCALL_lj_strfmt_putquoted, tr, tra);
else
@@ -1025,7 +1029,7 @@ static void LJ_FASTCALL recff_string_format(jit_State *J, RecordFFData *rd)
case STRFMT_CHAR:
tra = lj_opt_narrow_toint(J, tra);
if (sf == STRFMT_CHAR) /* Shortcut for plain %c. */
- tr = emitir(IRT(IR_BUFPUT, IRT_P32), tr,
+ tr = emitir(IRT(IR_BUFPUT, IRT_PGC), tr,
emitir(IRT(IR_TOSTR, IRT_STR), tra, IRTOSTR_CHAR));
else
tr = lj_ir_call(J, IRCALL_lj_strfmt_putfchar, tr, trsf, tra);
@@ -1110,8 +1114,13 @@ static TRef recff_io_fp(jit_State *J, TRef *udp, int32_t id)
{
TRef tr, ud, fp;
if (id) { /* io.func() */
+#if LJ_GC64
+ /* TODO: fix ARM32 asm_fload(), so we can use this for all archs. */
+ ud = lj_ir_ggfload(J, IRT_UDATA, GG_OFS(g.gcroot[id]));
+#else
tr = lj_ir_kptr(J, &J2G(J)->gcroot[id]);
ud = emitir(IRT(IR_XLOAD, IRT_UDATA), tr, 0);
+#endif
} else { /* fp:method() */
ud = J->base[0];
if (!tref_isudata(ud))
@@ -1133,7 +1142,7 @@ static void LJ_FASTCALL recff_io_write(jit_State *J, RecordFFData *rd)
ptrdiff_t i = rd->data == 0 ? 1 : 0;
for (; J->base[i]; i++) {
TRef str = lj_ir_tostr(J, J->base[i]);
- TRef buf = emitir(IRT(IR_STRREF, IRT_P32), str, zero);
+ TRef buf = emitir(IRT(IR_STRREF, IRT_PGC), str, zero);
TRef len = emitir(IRTI(IR_FLOAD), str, IRFL_STR_LEN);
if (tref_isk(len) && IR(tref_ref(len))->i == 1) {
IRIns *irs = IR(tref_ref(str));
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.h
index 7a188beb298..3b407450d58 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ffrecord.h
@@ -1,6 +1,6 @@
/*
** Fast function call recorder.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_FFRECORD_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_frame.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_frame.h
index fc0e281c2e0..19c49a4aef4 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_frame.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_frame.h
@@ -1,6 +1,6 @@
/*
** Stack frames.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_FRAME_H
@@ -116,6 +116,17 @@ enum { LJ_CONT_TAILCALL, LJ_CONT_FFI_CALLBACK }; /* Special continuations. */
/* These definitions must match with the arch-specific *.dasc files. */
#if LJ_TARGET_X86
+#if LJ_ABI_WIN
+#define CFRAME_OFS_ERRF (19*4)
+#define CFRAME_OFS_NRES (18*4)
+#define CFRAME_OFS_PREV (17*4)
+#define CFRAME_OFS_L (16*4)
+#define CFRAME_OFS_SEH (9*4)
+#define CFRAME_OFS_PC (6*4)
+#define CFRAME_OFS_MULTRES (5*4)
+#define CFRAME_SIZE (16*4)
+#define CFRAME_SHIFT_MULTRES 0
+#else
#define CFRAME_OFS_ERRF (15*4)
#define CFRAME_OFS_NRES (14*4)
#define CFRAME_OFS_PREV (13*4)
@@ -124,6 +135,7 @@ enum { LJ_CONT_TAILCALL, LJ_CONT_FFI_CALLBACK }; /* Special continuations. */
#define CFRAME_OFS_MULTRES (5*4)
#define CFRAME_SIZE (12*4)
#define CFRAME_SHIFT_MULTRES 0
+#endif
#elif LJ_TARGET_X64
#if LJ_ABI_WIN
#define CFRAME_OFS_PREV (13*8)
@@ -217,26 +229,41 @@ enum { LJ_CONT_TAILCALL, LJ_CONT_FFI_CALLBACK }; /* Special continuations. */
#define CFRAME_SIZE 272
#define CFRAME_SHIFT_MULTRES 3
#endif
-#elif LJ_TARGET_MIPS
+#elif LJ_TARGET_MIPS32
#if LJ_ARCH_HASFPU
#define CFRAME_OFS_ERRF 124
#define CFRAME_OFS_NRES 120
#define CFRAME_OFS_PREV 116
#define CFRAME_OFS_L 112
-#define CFRAME_OFS_PC 20
-#define CFRAME_OFS_MULTRES 16
#define CFRAME_SIZE 112
-#define CFRAME_SHIFT_MULTRES 3
#else
#define CFRAME_OFS_ERRF 76
#define CFRAME_OFS_NRES 72
#define CFRAME_OFS_PREV 68
#define CFRAME_OFS_L 64
+#define CFRAME_SIZE 64
+#endif
#define CFRAME_OFS_PC 20
#define CFRAME_OFS_MULTRES 16
-#define CFRAME_SIZE 64
#define CFRAME_SHIFT_MULTRES 3
+#elif LJ_TARGET_MIPS64
+#if LJ_ARCH_HASFPU
+#define CFRAME_OFS_ERRF 188
+#define CFRAME_OFS_NRES 184
+#define CFRAME_OFS_PREV 176
+#define CFRAME_OFS_L 168
+#define CFRAME_OFS_PC 160
+#define CFRAME_SIZE 192
+#else
+#define CFRAME_OFS_ERRF 124
+#define CFRAME_OFS_NRES 120
+#define CFRAME_OFS_PREV 112
+#define CFRAME_OFS_L 104
+#define CFRAME_OFS_PC 96
+#define CFRAME_SIZE 128
#endif
+#define CFRAME_OFS_MULTRES 0
+#define CFRAME_SHIFT_MULTRES 3
#else
#error "Missing CFRAME_* definitions for this architecture"
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_func.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_func.c
index 431a56d76df..639dad87684 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_func.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_func.c
@@ -1,6 +1,6 @@
/*
** Function handling (prototypes, functions and upvalues).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -140,7 +140,9 @@ GCfunc *lj_func_newL_empty(lua_State *L, GCproto *pt, GCtab *env)
/* NOBARRIER: The GCfunc is new (marked white). */
for (i = 0; i < nuv; i++) {
GCupval *uv = func_emptyuv(L);
- uv->dhash = (uint32_t)(uintptr_t)pt ^ ((uint32_t)proto_uv(pt)[i] << 24);
+ int32_t v = proto_uv(pt)[i];
+ uv->immutable = ((v / PROTO_UV_IMMUTABLE) & 1);
+ uv->dhash = (uint32_t)(uintptr_t)pt ^ (v << 24);
setgcref(fn->l.uvptr[i], obj2gco(uv));
}
fn->l.nupvalues = (uint8_t)nuv;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_func.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_func.h
index e0c3c555ed3..901751b9813 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_func.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_func.h
@@ -1,6 +1,6 @@
/*
** Function handling (prototypes, functions and upvalues).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_FUNC_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_gc.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_gc.c
index 53f1d974cf3..2aaf5b2c4f7 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_gc.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_gc.c
@@ -1,6 +1,6 @@
/*
** Garbage collector.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -169,12 +169,19 @@ static int gc_traverse_tab(global_State *g, GCtab *t)
while ((c = *modestr++)) {
if (c == 'k') weak |= LJ_GC_WEAKKEY;
else if (c == 'v') weak |= LJ_GC_WEAKVAL;
- else if (c == 'K') weak = (int)(~0u & ~LJ_GC_WEAKVAL);
}
- if (weak > 0) { /* Weak tables are cleared in the atomic phase. */
- t->marked = (uint8_t)((t->marked & ~LJ_GC_WEAK) | weak);
- setgcrefr(t->gclist, g->gc.weak);
- setgcref(g->gc.weak, obj2gco(t));
+ if (weak) { /* Weak tables are cleared in the atomic phase. */
+#if LJ_HASFFI
+ CTState *cts = ctype_ctsG(g);
+ if (cts && cts->finalizer == t) {
+ weak = (int)(~0u & ~LJ_GC_WEAKVAL);
+ } else
+#endif
+ {
+ t->marked = (uint8_t)((t->marked & ~LJ_GC_WEAK) | weak);
+ setgcrefr(t->gclist, g->gc.weak);
+ setgcref(g->gc.weak, obj2gco(t));
+ }
}
}
if (weak == LJ_GC_WEAK) /* Nothing to mark if both keys/values are weak. */
@@ -238,6 +245,8 @@ static void gc_traverse_trace(global_State *g, GCtrace *T)
IRIns *ir = &T->ir[ref];
if (ir->o == IR_KGC)
gc_markobj(g, ir_kgc(ir));
+ if (irt_is64(ir->t) && ir->o != IR_KNULL)
+ ref++;
}
if (T->link) gc_marktrace(g, T->link);
if (T->nextroot) gc_marktrace(g, T->nextroot);
@@ -308,7 +317,7 @@ static size_t propagatemark(global_State *g)
if (gc_traverse_tab(g, t) > 0)
black2gray(o); /* Keep weak tables gray. */
return sizeof(GCtab) + sizeof(TValue) * t->asize +
- sizeof(Node) * (t->hmask + 1);
+ (t->hmask ? sizeof(Node) * (t->hmask + 1) : 0);
} else if (LJ_LIKELY(gct == ~LJ_TFUNC)) {
GCfunc *fn = gco2func(o);
gc_traverse_func(g, fn);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_gc.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_gc.h
index eee09afb5a5..669bbe92403 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_gc.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_gc.h
@@ -1,6 +1,6 @@
/*
** Garbage collector.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_GC_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.c
index f2bd865e600..c219ffac0a8 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.c
@@ -1,6 +1,6 @@
/*
** Client for the GDB JIT API.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_gdbjit_c
@@ -296,6 +296,9 @@ enum {
#elif LJ_TARGET_ARM
DW_REG_SP = 13,
DW_REG_RA = 14,
+#elif LJ_TARGET_ARM64
+ DW_REG_SP = 31,
+ DW_REG_RA = 30,
#elif LJ_TARGET_PPC
DW_REG_SP = 1,
DW_REG_RA = 65,
@@ -374,6 +377,8 @@ static const ELFheader elfhdr_template = {
.machine = 62,
#elif LJ_TARGET_ARM
.machine = 40,
+#elif LJ_TARGET_ARM64
+ .machine = 183,
#elif LJ_TARGET_PPC
.machine = 20,
#elif LJ_TARGET_MIPS
@@ -563,6 +568,13 @@ static void LJ_FASTCALL gdbjit_ehframe(GDBJITctx *ctx)
int i;
for (i = 11; i >= 4; i--) { DB(DW_CFA_offset|i); DUV(2+(11-i)); }
}
+#elif LJ_TARGET_ARM64
+ {
+ int i;
+ DB(DW_CFA_offset|31); DUV(2);
+ for (i = 28; i >= 19; i--) { DB(DW_CFA_offset|i); DUV(3+(28-i)); }
+ for (i = 15; i >= 8; i--) { DB(DW_CFA_offset|32|i); DUV(28-i); }
+ }
#elif LJ_TARGET_PPC
{
int i;
@@ -719,6 +731,20 @@ static void gdbjit_buildobj(GDBJITctx *ctx)
/* -- Interface to GDB JIT API -------------------------------------------- */
+static int gdbjit_lock;
+
+static void gdbjit_lock_acquire()
+{
+ while (__sync_lock_test_and_set(&gdbjit_lock, 1)) {
+ /* Just spin; futexes or pthreads aren't worth the portability cost. */
+ }
+}
+
+static void gdbjit_lock_release()
+{
+ __sync_lock_release(&gdbjit_lock);
+}
+
/* Add new entry to GDB JIT symbol chain. */
static void gdbjit_newentry(lua_State *L, GDBJITctx *ctx)
{
@@ -730,6 +756,7 @@ static void gdbjit_newentry(lua_State *L, GDBJITctx *ctx)
ctx->T->gdbjit_entry = (void *)eo;
/* Link new entry to chain and register it. */
eo->entry.prev_entry = NULL;
+ gdbjit_lock_acquire();
eo->entry.next_entry = __jit_debug_descriptor.first_entry;
if (eo->entry.next_entry)
eo->entry.next_entry->prev_entry = &eo->entry;
@@ -739,6 +766,7 @@ static void gdbjit_newentry(lua_State *L, GDBJITctx *ctx)
__jit_debug_descriptor.relevant_entry = &eo->entry;
__jit_debug_descriptor.action_flag = GDBJIT_REGISTER;
__jit_debug_register_code();
+ gdbjit_lock_release();
}
/* Add debug info for newly compiled trace and notify GDB. */
@@ -770,6 +798,7 @@ void lj_gdbjit_deltrace(jit_State *J, GCtrace *T)
{
GDBJITentryobj *eo = (GDBJITentryobj *)T->gdbjit_entry;
if (eo) {
+ gdbjit_lock_acquire();
if (eo->entry.prev_entry)
eo->entry.prev_entry->next_entry = eo->entry.next_entry;
else
@@ -779,6 +808,7 @@ void lj_gdbjit_deltrace(jit_State *J, GCtrace *T)
__jit_debug_descriptor.relevant_entry = &eo->entry;
__jit_debug_descriptor.action_flag = GDBJIT_UNREGISTER;
__jit_debug_register_code();
+ gdbjit_lock_release();
lj_mem_free(J2G(J), eo, eo->sz);
}
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.h
index 1043bc69862..bbaa1568c44 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_gdbjit.h
@@ -1,6 +1,6 @@
/*
** Client for the GDB JIT API.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_GDBJIT_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ir.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_ir.c
index 63c9825475f..5baece67e63 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ir.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ir.c
@@ -1,6 +1,6 @@
/*
** SSA IR (Intermediate Representation) emitter.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_ir_c
@@ -91,7 +91,7 @@ static void lj_ir_growbot(jit_State *J)
IRIns *baseir = J->irbuf + J->irbotlim;
MSize szins = J->irtoplim - J->irbotlim;
lua_assert(szins != 0);
- lua_assert(J->cur.nk == J->irbotlim);
+ lua_assert(J->cur.nk == J->irbotlim || J->cur.nk-1 == J->irbotlim);
if (J->cur.nins + (szins >> 1) < J->irtoplim) {
/* More than half of the buffer is free on top: shift up by a quarter. */
MSize ofs = szins >> 2;
@@ -145,6 +145,16 @@ TRef lj_ir_call(jit_State *J, IRCallID id, ...)
return emitir(CCI_OPTYPE(ci), tr, id);
}
+/* Load field of type t from GG_State + offset. Must be 32 bit aligned. */
+LJ_FUNC TRef lj_ir_ggfload(jit_State *J, IRType t, uintptr_t ofs)
+{
+ lua_assert((ofs & 3) == 0);
+ ofs >>= 2;
+ lua_assert(ofs >= IRFL__MAX && ofs <= 0x3ff); /* 10 bit FOLD key limit. */
+ lj_ir_set(J, IRT(IR_FLOAD, t), REF_NIL, ofs);
+ return lj_opt_fold(J);
+}
+
/* -- Interning of constants ---------------------------------------------- */
/*
@@ -165,6 +175,24 @@ static LJ_AINLINE IRRef ir_nextk(jit_State *J)
return ref;
}
+/* Get ref of next 64 bit IR constant and optionally grow IR.
+** Note: this may invalidate all IRIns *!
+*/
+static LJ_AINLINE IRRef ir_nextk64(jit_State *J)
+{
+ IRRef ref = J->cur.nk - 2;
+ lua_assert(J->state != LJ_TRACE_ASM);
+ if (LJ_UNLIKELY(ref < J->irbotlim)) lj_ir_growbot(J);
+ J->cur.nk = ref;
+ return ref;
+}
+
+#if LJ_GC64
+#define ir_nextkgc ir_nextk64
+#else
+#define ir_nextkgc ir_nextk
+#endif
+
/* Intern int32_t constant. */
TRef LJ_FASTCALL lj_ir_kint(jit_State *J, int32_t k)
{
@@ -184,95 +212,21 @@ found:
return TREF(ref, IRT_INT);
}
-/* The MRef inside the KNUM/KINT64 IR instructions holds the address of the
-** 64 bit constant. The constants themselves are stored in a chained array
-** and shared across traces.
-**
-** Rationale for choosing this data structure:
-** - The address of the constants is embedded in the generated machine code
-** and must never move. A resizable array or hash table wouldn't work.
-** - Most apps need very few non-32 bit integer constants (less than a dozen).
-** - Linear search is hard to beat in terms of speed and low complexity.
-*/
-typedef struct K64Array {
- MRef next; /* Pointer to next list. */
- MSize numk; /* Number of used elements in this array. */
- TValue k[LJ_MIN_K64SZ]; /* Array of constants. */
-} K64Array;
-
-/* Free all chained arrays. */
-void lj_ir_k64_freeall(jit_State *J)
-{
- K64Array *k;
- for (k = mref(J->k64, K64Array); k; ) {
- K64Array *next = mref(k->next, K64Array);
- lj_mem_free(J2G(J), k, sizeof(K64Array));
- k = next;
- }
- setmref(J->k64, NULL);
-}
-
-/* Get new 64 bit constant slot. */
-static TValue *ir_k64_add(jit_State *J, K64Array *kp, uint64_t u64)
-{
- TValue *ntv;
- if (!(kp && kp->numk < LJ_MIN_K64SZ)) { /* Allocate a new array. */
- K64Array *kn = lj_mem_newt(J->L, sizeof(K64Array), K64Array);
- setmref(kn->next, NULL);
- kn->numk = 0;
- if (kp)
- setmref(kp->next, kn); /* Chain to the end of the list. */
- else
- setmref(J->k64, kn); /* Link first array. */
- kp = kn;
- }
- ntv = &kp->k[kp->numk++]; /* Add to current array. */
- ntv->u64 = u64;
- return ntv;
-}
-
-/* Find 64 bit constant in chained array or add it. */
-cTValue *lj_ir_k64_find(jit_State *J, uint64_t u64)
-{
- K64Array *k, *kp = NULL;
- MSize idx;
- /* Search for the constant in the whole chain of arrays. */
- for (k = mref(J->k64, K64Array); k; k = mref(k->next, K64Array)) {
- kp = k; /* Remember previous element in list. */
- for (idx = 0; idx < k->numk; idx++) { /* Search one array. */
- TValue *tv = &k->k[idx];
- if (tv->u64 == u64) /* Needed for +-0/NaN/absmask. */
- return tv;
- }
- }
- /* Otherwise add a new constant. */
- return ir_k64_add(J, kp, u64);
-}
-
-TValue *lj_ir_k64_reserve(jit_State *J)
-{
- K64Array *k, *kp = NULL;
- lj_ir_k64_find(J, 0); /* Intern dummy 0 to protect the reserved slot. */
- /* Find last K64Array, if any. */
- for (k = mref(J->k64, K64Array); k; k = mref(k->next, K64Array)) kp = k;
- return ir_k64_add(J, kp, 0); /* Set to 0. Final value is set later. */
-}
-
-/* Intern 64 bit constant, given by its address. */
-TRef lj_ir_k64(jit_State *J, IROp op, cTValue *tv)
+/* Intern 64 bit constant, given by its 64 bit pattern. */
+TRef lj_ir_k64(jit_State *J, IROp op, uint64_t u64)
{
IRIns *ir, *cir = J->cur.ir;
IRRef ref;
IRType t = op == IR_KNUM ? IRT_NUM : IRT_I64;
for (ref = J->chain[op]; ref; ref = cir[ref].prev)
- if (ir_k64(&cir[ref]) == tv)
+ if (ir_k64(&cir[ref])->u64 == u64)
goto found;
- ref = ir_nextk(J);
+ ref = ir_nextk64(J);
ir = IR(ref);
- lua_assert(checkptrGC(tv));
- setmref(ir->ptr, tv);
+ ir[1].tv.u64 = u64;
ir->t.irt = t;
ir->o = op;
+ ir->op12 = 0;
ir->prev = J->chain[op];
J->chain[op] = (IRRef1)ref;
found:
@@ -282,13 +236,13 @@ found:
/* Intern FP constant, given by its 64 bit pattern. */
TRef lj_ir_knum_u64(jit_State *J, uint64_t u64)
{
- return lj_ir_k64(J, IR_KNUM, lj_ir_k64_find(J, u64));
+ return lj_ir_k64(J, IR_KNUM, u64);
}
/* Intern 64 bit integer constant. */
TRef lj_ir_kint64(jit_State *J, uint64_t u64)
{
- return lj_ir_k64(J, IR_KINT64, lj_ir_k64_find(J, u64));
+ return lj_ir_k64(J, IR_KINT64, u64);
}
/* Check whether a number is int and return it. -0 is NOT considered an int. */
@@ -323,15 +277,15 @@ TRef lj_ir_kgc(jit_State *J, GCobj *o, IRType t)
{
IRIns *ir, *cir = J->cur.ir;
IRRef ref;
- lua_assert(!LJ_GC64); /* TODO_GC64: major changes required. */
lua_assert(!isdead(J2G(J), o));
for (ref = J->chain[IR_KGC]; ref; ref = cir[ref].prev)
if (ir_kgc(&cir[ref]) == o)
goto found;
- ref = ir_nextk(J);
+ ref = ir_nextkgc(J);
ir = IR(ref);
/* NOBARRIER: Current trace is a GC root. */
- setgcref(ir->gcr, o);
+ ir->op12 = 0;
+ setgcref(ir[LJ_GC64].gcr, o);
ir->t.irt = (uint8_t)t;
ir->o = IR_KGC;
ir->prev = J->chain[IR_KGC];
@@ -340,24 +294,44 @@ found:
return TREF(ref, t);
}
-/* Intern 32 bit pointer constant. */
+/* Allocate GCtrace constant placeholder (no interning). */
+TRef lj_ir_ktrace(jit_State *J)
+{
+ IRRef ref = ir_nextkgc(J);
+ IRIns *ir = IR(ref);
+ lua_assert(irt_toitype_(IRT_P64) == LJ_TTRACE);
+ ir->t.irt = IRT_P64;
+ ir->o = LJ_GC64 ? IR_KNUM : IR_KNULL; /* Not IR_KGC yet, but same size. */
+ ir->op12 = 0;
+ ir->prev = 0;
+ return TREF(ref, IRT_P64);
+}
+
+/* Intern pointer constant. */
TRef lj_ir_kptr_(jit_State *J, IROp op, void *ptr)
{
IRIns *ir, *cir = J->cur.ir;
IRRef ref;
- lua_assert((void *)(intptr_t)i32ptr(ptr) == ptr);
+#if LJ_64 && !LJ_GC64
+ lua_assert((void *)(uintptr_t)u32ptr(ptr) == ptr);
+#endif
for (ref = J->chain[op]; ref; ref = cir[ref].prev)
- if (mref(cir[ref].ptr, void) == ptr)
+ if (ir_kptr(&cir[ref]) == ptr)
goto found;
+#if LJ_GC64
+ ref = ir_nextk64(J);
+#else
ref = ir_nextk(J);
+#endif
ir = IR(ref);
- setmref(ir->ptr, ptr);
- ir->t.irt = IRT_P32;
+ ir->op12 = 0;
+ setmref(ir[LJ_GC64].ptr, ptr);
+ ir->t.irt = IRT_PGC;
ir->o = op;
ir->prev = J->chain[op];
J->chain[op] = (IRRef1)ref;
found:
- return TREF(ref, IRT_P32);
+ return TREF(ref, IRT_PGC);
}
/* Intern typed NULL constant. */
@@ -412,9 +386,8 @@ void lj_ir_kvalue(lua_State *L, TValue *tv, const IRIns *ir)
case IR_KPRI: setpriV(tv, irt_toitype(ir->t)); break;
case IR_KINT: setintV(tv, ir->i); break;
case IR_KGC: setgcV(L, tv, ir_kgc(ir), irt_toitype(ir->t)); break;
- case IR_KPTR: case IR_KKPTR: case IR_KNULL:
- setlightudV(tv, mref(ir->ptr, void));
- break;
+ case IR_KPTR: case IR_KKPTR: setlightudV(tv, ir_kptr(ir)); break;
+ case IR_KNULL: setlightudV(tv, NULL); break;
case IR_KNUM: setnumV(tv, ir_knum(ir)->n); break;
#if LJ_HASFFI
case IR_KINT64: {
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ir.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_ir.h
index cd8df59d39d..34c27853948 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ir.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ir.h
@@ -1,6 +1,6 @@
/*
** SSA IR (Intermediate Representation) format.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_IR_H
@@ -220,7 +220,7 @@ IRFLDEF(FLENUM)
/* SLOAD mode bits, stored in op2. */
#define IRSLOAD_PARENT 0x01 /* Coalesce with parent trace. */
-#define IRSLOAD_FRAME 0x02 /* Load hiword of frame. */
+#define IRSLOAD_FRAME 0x02 /* Load 32 bits of ftsz. */
#define IRSLOAD_TYPECHECK 0x04 /* Needs type check. */
#define IRSLOAD_CONVERT 0x08 /* Number to integer conversion. */
#define IRSLOAD_READONLY 0x10 /* Read-only, omit slot store. */
@@ -294,7 +294,9 @@ LJ_DATA const uint8_t lj_ir_mode[IR__MAX+1];
/* -- IR instruction types ------------------------------------------------ */
-/* Map of itypes to non-negative numbers. ORDER LJ_T.
+#define IRTSIZE_PGC (LJ_GC64 ? 8 : 4)
+
+/* Map of itypes to non-negative numbers and their sizes. ORDER LJ_T.
** LJ_TUPVAL/LJ_TTRACE never appear in a TValue. Use these itypes for
** IRT_P32 and IRT_P64, which never escape the IR.
** The various integers are only used in the IR and can only escape to
@@ -302,12 +304,13 @@ LJ_DATA const uint8_t lj_ir_mode[IR__MAX+1];
** contiguous and next to IRT_NUM (see the typerange macros below).
*/
#define IRTDEF(_) \
- _(NIL, 4) _(FALSE, 4) _(TRUE, 4) _(LIGHTUD, LJ_64 ? 8 : 4) _(STR, 4) \
- _(P32, 4) _(THREAD, 4) _(PROTO, 4) _(FUNC, 4) _(P64, 8) _(CDATA, 4) \
- _(TAB, 4) _(UDATA, 4) \
+ _(NIL, 4) _(FALSE, 4) _(TRUE, 4) _(LIGHTUD, LJ_64 ? 8 : 4) \
+ _(STR, IRTSIZE_PGC) _(P32, 4) _(THREAD, IRTSIZE_PGC) _(PROTO, IRTSIZE_PGC) \
+ _(FUNC, IRTSIZE_PGC) _(P64, 8) _(CDATA, IRTSIZE_PGC) _(TAB, IRTSIZE_PGC) \
+ _(UDATA, IRTSIZE_PGC) \
_(FLOAT, 4) _(NUM, 8) _(I8, 1) _(U8, 1) _(I16, 2) _(U16, 2) \
_(INT, 4) _(U32, 4) _(I64, 8) _(U64, 8) \
- _(SOFTFP, 4) /* There is room for 9 more types. */
+ _(SOFTFP, 4) /* There is room for 8 more types. */
/* IR result type and flags (8 bit). */
typedef enum {
@@ -318,9 +321,10 @@ IRTDEF(IRTENUM)
/* Native pointer type and the corresponding integer type. */
IRT_PTR = LJ_64 ? IRT_P64 : IRT_P32,
+ IRT_PGC = LJ_GC64 ? IRT_P64 : IRT_P32,
+ IRT_IGC = LJ_GC64 ? IRT_I64 : IRT_INT,
IRT_INTP = LJ_64 ? IRT_I64 : IRT_INT,
IRT_UINTP = LJ_64 ? IRT_U64 : IRT_U32,
- /* TODO_GC64: major changes required for all uses of IRT_P32. */
/* Additional flags. */
IRT_MARK = 0x20, /* Marker for misc. purposes. */
@@ -408,7 +412,7 @@ static LJ_AINLINE IRType itype2irt(const TValue *tv)
static LJ_AINLINE uint32_t irt_toitype_(IRType t)
{
- lua_assert(!LJ_64 || t != IRT_LIGHTUD);
+ lua_assert(!LJ_64 || LJ_GC64 || t != IRT_LIGHTUD);
if (LJ_DUALNUM && t > IRT_NUM) {
return LJ_TISNUM;
} else {
@@ -521,7 +525,9 @@ typedef uint32_t TRef;
** +-------+-------+---+---+---+---+
** | op1 | op2 | t | o | r | s |
** +-------+-------+---+---+---+---+
-** | op12/i/gco | ot | prev | (alternative fields in union)
+** | op12/i/gco32 | ot | prev | (alternative fields in union)
+** +-------+-------+---+---+---+---+
+** | TValue/gco64 | (2nd IR slot for 64 bit constants)
** +---------------+-------+-------+
** 32 16 16
**
@@ -549,22 +555,27 @@ typedef union IRIns {
)
};
int32_t i; /* 32 bit signed integer literal (overlaps op12). */
- GCRef gcr; /* GCobj constant (overlaps op12). */
- MRef ptr; /* Pointer constant (overlaps op12). */
+ GCRef gcr; /* GCobj constant (overlaps op12 or entire slot). */
+ MRef ptr; /* Pointer constant (overlaps op12 or entire slot). */
+ TValue tv; /* TValue constant (overlaps entire slot). */
} IRIns;
-/* TODO_GC64: major changes required. */
-#define ir_kgc(ir) check_exp((ir)->o == IR_KGC, gcref((ir)->gcr))
+#define ir_kgc(ir) check_exp((ir)->o == IR_KGC, gcref((ir)[LJ_GC64].gcr))
#define ir_kstr(ir) (gco2str(ir_kgc((ir))))
#define ir_ktab(ir) (gco2tab(ir_kgc((ir))))
#define ir_kfunc(ir) (gco2func(ir_kgc((ir))))
#define ir_kcdata(ir) (gco2cd(ir_kgc((ir))))
-#define ir_knum(ir) check_exp((ir)->o == IR_KNUM, mref((ir)->ptr, cTValue))
-#define ir_kint64(ir) check_exp((ir)->o == IR_KINT64, mref((ir)->ptr,cTValue))
+#define ir_knum(ir) check_exp((ir)->o == IR_KNUM, &(ir)[1].tv)
+#define ir_kint64(ir) check_exp((ir)->o == IR_KINT64, &(ir)[1].tv)
#define ir_k64(ir) \
- check_exp((ir)->o == IR_KNUM || (ir)->o == IR_KINT64, mref((ir)->ptr,cTValue))
+ check_exp((ir)->o == IR_KNUM || (ir)->o == IR_KINT64 || \
+ (LJ_GC64 && \
+ ((ir)->o == IR_KGC || \
+ (ir)->o == IR_KPTR || (ir)->o == IR_KKPTR)), \
+ &(ir)[1].tv)
#define ir_kptr(ir) \
- check_exp((ir)->o == IR_KPTR || (ir)->o == IR_KKPTR, mref((ir)->ptr, void))
+ check_exp((ir)->o == IR_KPTR || (ir)->o == IR_KKPTR, \
+ mref((ir)[LJ_GC64].ptr, void))
/* A store or any other op with a non-weak guard has a side-effect. */
static LJ_AINLINE int ir_sideeff(IRIns *ir)
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_ircall.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_ircall.h
index 144135a4a7f..973c36e6ec9 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_ircall.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_ircall.h
@@ -1,6 +1,6 @@
/*
** IR CALL* instruction definitions.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_IRCALL_H
@@ -78,13 +78,13 @@ typedef struct CCallInfo {
#define IRCALLCOND_SOFTFP_FFI(x) NULL
#endif
-#if LJ_SOFTFP && LJ_TARGET_MIPS
+#if LJ_SOFTFP && LJ_TARGET_MIPS32
#define IRCALLCOND_SOFTFP_MIPS(x) x
#else
#define IRCALLCOND_SOFTFP_MIPS(x) NULL
#endif
-#define LJ_NEED_FP64 (LJ_TARGET_ARM || LJ_TARGET_PPC || LJ_TARGET_MIPS)
+#define LJ_NEED_FP64 (LJ_TARGET_ARM || LJ_TARGET_PPC || LJ_TARGET_MIPS32)
#if LJ_HASFFI && (LJ_SOFTFP || LJ_NEED_FP64)
#define IRCALLCOND_FP64_FFI(x) x
@@ -104,12 +104,6 @@ typedef struct CCallInfo {
#define IRCALLCOND_FFI32(x) NULL
#endif
-#if LJ_TARGET_X86
-#define CCI_RANDFPR 0 /* Clang on OSX/x86 is overzealous. */
-#else
-#define CCI_RANDFPR CCI_NOFPRCLOBBER
-#endif
-
#if LJ_SOFTFP
#define XA_FP CCI_XA
#define XA2_FP (CCI_XA+CCI_XA)
@@ -129,40 +123,40 @@ typedef struct CCallInfo {
/* Function definitions for CALL* instructions. */
#define IRCALLDEF(_) \
_(ANY, lj_str_cmp, 2, FN, INT, CCI_NOFPRCLOBBER) \
- _(ANY, lj_str_find, 4, N, P32, 0) \
+ _(ANY, lj_str_find, 4, N, PGC, 0) \
_(ANY, lj_str_new, 3, S, STR, CCI_L) \
_(ANY, lj_strscan_num, 2, FN, INT, 0) \
_(ANY, lj_strfmt_int, 2, FN, STR, CCI_L) \
_(ANY, lj_strfmt_num, 2, FN, STR, CCI_L) \
_(ANY, lj_strfmt_char, 2, FN, STR, CCI_L) \
- _(ANY, lj_strfmt_putint, 2, FL, P32, 0) \
- _(ANY, lj_strfmt_putnum, 2, FL, P32, 0) \
- _(ANY, lj_strfmt_putquoted, 2, FL, P32, 0) \
- _(ANY, lj_strfmt_putfxint, 3, L, P32, XA_64) \
- _(ANY, lj_strfmt_putfnum_int, 3, L, P32, XA_FP) \
- _(ANY, lj_strfmt_putfnum_uint, 3, L, P32, XA_FP) \
- _(ANY, lj_strfmt_putfnum, 3, L, P32, XA_FP) \
- _(ANY, lj_strfmt_putfstr, 3, L, P32, 0) \
- _(ANY, lj_strfmt_putfchar, 3, L, P32, 0) \
- _(ANY, lj_buf_putmem, 3, S, P32, 0) \
- _(ANY, lj_buf_putstr, 2, FL, P32, 0) \
- _(ANY, lj_buf_putchar, 2, FL, P32, 0) \
- _(ANY, lj_buf_putstr_reverse, 2, FL, P32, 0) \
- _(ANY, lj_buf_putstr_lower, 2, FL, P32, 0) \
- _(ANY, lj_buf_putstr_upper, 2, FL, P32, 0) \
- _(ANY, lj_buf_putstr_rep, 3, L, P32, 0) \
- _(ANY, lj_buf_puttab, 5, L, P32, 0) \
+ _(ANY, lj_strfmt_putint, 2, FL, PGC, 0) \
+ _(ANY, lj_strfmt_putnum, 2, FL, PGC, 0) \
+ _(ANY, lj_strfmt_putquoted, 2, FL, PGC, 0) \
+ _(ANY, lj_strfmt_putfxint, 3, L, PGC, XA_64) \
+ _(ANY, lj_strfmt_putfnum_int, 3, L, PGC, XA_FP) \
+ _(ANY, lj_strfmt_putfnum_uint, 3, L, PGC, XA_FP) \
+ _(ANY, lj_strfmt_putfnum, 3, L, PGC, XA_FP) \
+ _(ANY, lj_strfmt_putfstr, 3, L, PGC, 0) \
+ _(ANY, lj_strfmt_putfchar, 3, L, PGC, 0) \
+ _(ANY, lj_buf_putmem, 3, S, PGC, 0) \
+ _(ANY, lj_buf_putstr, 2, FL, PGC, 0) \
+ _(ANY, lj_buf_putchar, 2, FL, PGC, 0) \
+ _(ANY, lj_buf_putstr_reverse, 2, FL, PGC, 0) \
+ _(ANY, lj_buf_putstr_lower, 2, FL, PGC, 0) \
+ _(ANY, lj_buf_putstr_upper, 2, FL, PGC, 0) \
+ _(ANY, lj_buf_putstr_rep, 3, L, PGC, 0) \
+ _(ANY, lj_buf_puttab, 5, L, PGC, 0) \
_(ANY, lj_buf_tostr, 1, FL, STR, 0) \
_(ANY, lj_tab_new_ah, 3, A, TAB, CCI_L) \
_(ANY, lj_tab_new1, 2, FS, TAB, CCI_L) \
_(ANY, lj_tab_dup, 2, FS, TAB, CCI_L) \
_(ANY, lj_tab_clear, 1, FS, NIL, 0) \
- _(ANY, lj_tab_newkey, 3, S, P32, CCI_L) \
+ _(ANY, lj_tab_newkey, 3, S, PGC, CCI_L) \
_(ANY, lj_tab_len, 1, FL, INT, 0) \
_(ANY, lj_gc_step_jit, 2, FS, NIL, CCI_L) \
_(ANY, lj_gc_barrieruv, 2, FS, NIL, 0) \
- _(ANY, lj_mem_newgco, 2, FS, P32, CCI_L) \
- _(ANY, lj_math_random_step, 1, FS, NUM, CCI_CASTU64|CCI_RANDFPR)\
+ _(ANY, lj_mem_newgco, 2, FS, PGC, CCI_L) \
+ _(ANY, lj_math_random_step, 1, FS, NUM, CCI_CASTU64) \
_(ANY, lj_vm_modi, 2, FN, INT, 0) \
_(ANY, sinh, 1, N, NUM, XA_FP) \
_(ANY, cosh, 1, N, NUM, XA_FP) \
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_iropt.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_iropt.h
index 953740443cd..73aef0ef38e 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_iropt.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_iropt.h
@@ -1,6 +1,6 @@
/*
** Common header for IR emitter and optimizations.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_IROPT_H
@@ -36,12 +36,11 @@ static LJ_AINLINE IRRef lj_ir_nextins(jit_State *J)
return ref;
}
+LJ_FUNC TRef lj_ir_ggfload(jit_State *J, IRType t, uintptr_t ofs);
+
/* Interning of constants. */
LJ_FUNC TRef LJ_FASTCALL lj_ir_kint(jit_State *J, int32_t k);
-LJ_FUNC void lj_ir_k64_freeall(jit_State *J);
-LJ_FUNC TRef lj_ir_k64(jit_State *J, IROp op, cTValue *tv);
-LJ_FUNC TValue *lj_ir_k64_reserve(jit_State *J);
-LJ_FUNC cTValue *lj_ir_k64_find(jit_State *J, uint64_t u64);
+LJ_FUNC TRef lj_ir_k64(jit_State *J, IROp op, uint64_t u64);
LJ_FUNC TRef lj_ir_knum_u64(jit_State *J, uint64_t u64);
LJ_FUNC TRef lj_ir_knumint(jit_State *J, lua_Number n);
LJ_FUNC TRef lj_ir_kint64(jit_State *J, uint64_t u64);
@@ -49,6 +48,7 @@ LJ_FUNC TRef lj_ir_kgc(jit_State *J, GCobj *o, IRType t);
LJ_FUNC TRef lj_ir_kptr_(jit_State *J, IROp op, void *ptr);
LJ_FUNC TRef lj_ir_knull(jit_State *J, IRType t);
LJ_FUNC TRef lj_ir_kslot(jit_State *J, TRef key, IRRef slot);
+LJ_FUNC TRef lj_ir_ktrace(jit_State *J);
#if LJ_64
#define lj_ir_kintp(J, k) lj_ir_kint64(J, (uint64_t)(k))
@@ -75,8 +75,8 @@ static LJ_AINLINE TRef lj_ir_knum(jit_State *J, lua_Number n)
#define lj_ir_knum_tobit(J) lj_ir_knum_u64(J, U64x(43380000,00000000))
/* Special 128 bit SIMD constants. */
-#define lj_ir_knum_abs(J) lj_ir_k64(J, IR_KNUM, LJ_KSIMD(J, LJ_KSIMD_ABS))
-#define lj_ir_knum_neg(J) lj_ir_k64(J, IR_KNUM, LJ_KSIMD(J, LJ_KSIMD_NEG))
+#define lj_ir_ksimd(J, idx) \
+ lj_ir_ggfload(J, IRT_NUM, (uintptr_t)LJ_KSIMD(J, idx) - (uintptr_t)J2GG(J))
/* Access to constants. */
LJ_FUNC void lj_ir_kvalue(lua_State *L, TValue *tv, const IRIns *ir);
@@ -143,8 +143,8 @@ LJ_FUNC TRef LJ_FASTCALL lj_opt_narrow_cindex(jit_State *J, TRef key);
LJ_FUNC TRef lj_opt_narrow_arith(jit_State *J, TRef rb, TRef rc,
TValue *vb, TValue *vc, IROp op);
LJ_FUNC TRef lj_opt_narrow_unm(jit_State *J, TRef rc, TValue *vc);
-LJ_FUNC TRef lj_opt_narrow_mod(jit_State *J, TRef rb, TRef rc, TValue *vc);
-LJ_FUNC TRef lj_opt_narrow_pow(jit_State *J, TRef rb, TRef rc, TValue *vc);
+LJ_FUNC TRef lj_opt_narrow_mod(jit_State *J, TRef rb, TRef rc, TValue *vb, TValue *vc);
+LJ_FUNC TRef lj_opt_narrow_pow(jit_State *J, TRef rb, TRef rc, TValue *vb, TValue *vc);
LJ_FUNC IRType lj_opt_narrow_forl(jit_State *J, cTValue *forbase);
/* Optimization passes. */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_jit.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_jit.h
index 5a51357e077..92054e3df60 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_jit.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_jit.h
@@ -1,6 +1,6 @@
/*
** Common definitions for the JIT compiler.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_JIT_H
@@ -19,10 +19,11 @@
#define JIT_F_SSE4_1 0x00000040
#define JIT_F_PREFER_IMUL 0x00000080
#define JIT_F_LEA_AGU 0x00000100
+#define JIT_F_BMI2 0x00000200
/* Names for the CPU-specific flags. Must match the order above. */
#define JIT_F_CPU_FIRST JIT_F_SSE2
-#define JIT_F_CPUSTRING "\4SSE2\4SSE3\6SSE4.1\3AMD\4ATOM"
+#define JIT_F_CPUSTRING "\4SSE2\4SSE3\6SSE4.1\3AMD\4ATOM\4BMI2"
#elif LJ_TARGET_ARM
#define JIT_F_ARMV6_ 0x00000010
#define JIT_F_ARMV6T2_ 0x00000020
@@ -45,12 +46,16 @@
#define JIT_F_CPU_FIRST JIT_F_SQRT
#define JIT_F_CPUSTRING "\4SQRT\5ROUND"
#elif LJ_TARGET_MIPS
-#define JIT_F_MIPS32R2 0x00000010
+#define JIT_F_MIPSXXR2 0x00000010
/* Names for the CPU-specific flags. Must match the order above. */
-#define JIT_F_CPU_FIRST JIT_F_MIPS32R2
+#define JIT_F_CPU_FIRST JIT_F_MIPSXXR2
+#if LJ_TARGET_MIPS32
#define JIT_F_CPUSTRING "\010MIPS32R2"
#else
+#define JIT_F_CPUSTRING "\010MIPS64R2"
+#endif
+#else
#define JIT_F_CPU_FIRST 0
#define JIT_F_CPUSTRING ""
#endif
@@ -178,14 +183,26 @@ LJ_STATIC_ASSERT(SNAP_CONT == TREF_CONT);
#define SNAP(slot, flags, ref) (((SnapEntry)(slot) << 24) + (flags) + (ref))
#define SNAP_TR(slot, tr) \
(((SnapEntry)(slot) << 24) + ((tr) & (TREF_CONT|TREF_FRAME|TREF_REFMASK)))
+#if !LJ_FR2
#define SNAP_MKPC(pc) ((SnapEntry)u32ptr(pc))
+#endif
#define SNAP_MKFTSZ(ftsz) ((SnapEntry)(ftsz))
#define snap_ref(sn) ((sn) & 0xffff)
#define snap_slot(sn) ((BCReg)((sn) >> 24))
#define snap_isframe(sn) ((sn) & SNAP_FRAME)
-#define snap_pc(sn) ((const BCIns *)(uintptr_t)(sn))
#define snap_setref(sn, ref) (((sn) & (0xffff0000&~SNAP_NORESTORE)) | (ref))
+static LJ_AINLINE const BCIns *snap_pc(SnapEntry *sn)
+{
+#if LJ_FR2
+ uint64_t pcbase;
+ memcpy(&pcbase, sn, sizeof(uint64_t));
+ return (const BCIns *)(pcbase >> 8);
+#else
+ return (const BCIns *)(uintptr_t)*sn;
+#endif
+}
+
/* Snapshot and exit numbers. */
typedef uint32_t SnapNo;
typedef uint32_t ExitNo;
@@ -307,6 +324,45 @@ enum {
LJ_KSIMD__MAX
};
+enum {
+#if LJ_TARGET_X86ORX64
+ LJ_K64_TOBIT, /* 2^52 + 2^51 */
+ LJ_K64_2P64, /* 2^64 */
+ LJ_K64_M2P64, /* -2^64 */
+#if LJ_32
+ LJ_K64_M2P64_31, /* -2^64 or -2^31 */
+#else
+ LJ_K64_M2P64_31 = LJ_K64_M2P64,
+#endif
+#endif
+#if LJ_TARGET_MIPS
+ LJ_K64_2P31, /* 2^31 */
+#if LJ_64
+ LJ_K64_2P63, /* 2^63 */
+ LJ_K64_M2P64, /* -2^64 */
+#endif
+#endif
+ LJ_K64__MAX,
+};
+
+enum {
+#if LJ_TARGET_X86ORX64
+ LJ_K32_M2P64_31, /* -2^64 or -2^31 */
+#endif
+#if LJ_TARGET_PPC
+ LJ_K32_2P52_2P31, /* 2^52 + 2^31 */
+ LJ_K32_2P52, /* 2^52 */
+#endif
+#if LJ_TARGET_PPC || LJ_TARGET_MIPS
+ LJ_K32_2P31, /* 2^31 */
+#endif
+#if LJ_TARGET_MIPS64
+ LJ_K32_2P63, /* 2^63 */
+ LJ_K32_M2P64, /* -2^64 */
+#endif
+ LJ_K32__MAX
+};
+
/* Get 16 byte aligned pointer to SIMD constant. */
#define LJ_KSIMD(J, n) \
((TValue *)(((intptr_t)&J->ksimd[2*(n)] + 15) & ~(intptr_t)15))
@@ -323,13 +379,14 @@ enum {
/* Fold state is used to fold instructions on-the-fly. */
typedef struct FoldState {
IRIns ins; /* Currently emitted instruction. */
- IRIns left; /* Instruction referenced by left operand. */
- IRIns right; /* Instruction referenced by right operand. */
+ IRIns left[2]; /* Instruction referenced by left operand. */
+ IRIns right[2]; /* Instruction referenced by right operand. */
} FoldState;
/* JIT compiler state. */
typedef struct jit_State {
GCtrace cur; /* Current trace. */
+ GCtrace *curfinal; /* Final address of current trace (set during asm). */
lua_State *L; /* Current Lua state. */
const BCIns *pc; /* Current PC. */
@@ -359,8 +416,9 @@ typedef struct jit_State {
int32_t framedepth; /* Current frame depth. */
int32_t retdepth; /* Return frame depth (count of RETF). */
- MRef k64; /* Pointer to chained array of 64 bit constants. */
TValue ksimd[LJ_KSIMD__MAX*2+1]; /* 16 byte aligned SIMD constants. */
+ TValue k64[LJ_K64__MAX]; /* Common 8 byte constants used by backends. */
+ uint32_t k32[LJ_K32__MAX]; /* Ditto for 4 byte constants. */
IRIns *irbuf; /* Temp. IR instruction buffer. Biased with REF_BIAS. */
IRRef irtoplim; /* Upper limit of instuction buffer (biased). */
@@ -381,7 +439,7 @@ typedef struct jit_State {
GCRef *trace; /* Array of traces. */
TraceNo freetrace; /* Start of scan for next free trace. */
MSize sizetrace; /* Size of trace array. */
- TValue *ktracep; /* Pointer to K64Array slot with GCtrace pointer. */
+ IRRef1 ktrace; /* Reference to KGC with GCtrace. */
IRRef1 chain[IR__MAX]; /* IR instruction skip-list chain anchors. */
TRef slot[LJ_MAX_JSLOTS+LJ_STACK_EXTRA]; /* Stack slot map. */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_lex.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_lex.c
index 0187c4267a2..2d2f8194cfd 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_lex.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_lex.c
@@ -1,6 +1,6 @@
/*
** Lexical analyzer.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_lex.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_lex.h
index 7fed1fd6137..33fa8657262 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_lex.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_lex.h
@@ -1,6 +1,6 @@
/*
** Lexical analyzer.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_LEX_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_lib.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_lib.c
index 8bdf6912a8d..b8638de6a08 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_lib.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_lib.c
@@ -1,6 +1,6 @@
/*
** Library function support.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_lib_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_lib.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_lib.h
index dbc04055b44..37ec9d78007 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_lib.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_lib.h
@@ -1,6 +1,6 @@
/*
** Library function support.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_LIB_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_load.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_load.c
index 7783eee0504..c046402dbed 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_load.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_load.c
@@ -1,6 +1,6 @@
/*
** Load and dump code.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include <errno.h>
@@ -166,6 +166,8 @@ LUA_API int lua_dump(lua_State *L, lua_Writer writer, void *data)
return 1;
}
+
+
/* -- Luajittex needs this one because it overloads loadfile -- */
LUALIB_API int RESERVED_load_aux_JIT(lua_State *L, int status, int envarg)
{
@@ -182,3 +184,4 @@ LUALIB_API int RESERVED_load_aux_JIT(lua_State *L, int status, int envarg)
return 2;
}
}
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.c
index 3eaee0546e7..77035bf72a4 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.c
@@ -1,6 +1,6 @@
/*
** Machine code management.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_mcode_c
@@ -204,8 +204,8 @@ static void mcode_protect(jit_State *J, int prot)
/* -- MCode area allocation ----------------------------------------------- */
-#if LJ_TARGET_X64
-#define mcode_validptr(p) ((p) && (uintptr_t)(p) < (uintptr_t)1<<47)
+#if LJ_64
+#define mcode_validptr(p) (p)
#else
#define mcode_validptr(p) ((p) && (uintptr_t)(p) < 0xffff0000)
#endif
@@ -221,8 +221,8 @@ static void *mcode_alloc(jit_State *J, size_t sz)
*/
#if LJ_TARGET_MIPS
/* Use the middle of the 256MB-aligned region. */
- uintptr_t target = ((uintptr_t)(void *)lj_vm_exit_handler & 0xf0000000u) +
- 0x08000000u;
+ uintptr_t target = ((uintptr_t)(void *)lj_vm_exit_handler &
+ ~(uintptr_t)0x0fffffffu) + 0x08000000u;
#else
uintptr_t target = (uintptr_t)(void *)lj_vm_exit_handler & ~(uintptr_t)0xffff;
#endif
@@ -230,7 +230,8 @@ static void *mcode_alloc(jit_State *J, size_t sz)
/* First try a contiguous area below the last one. */
uintptr_t hint = J->mcarea ? (uintptr_t)J->mcarea - sz : 0;
int i;
- for (i = 0; i < 32; i++) { /* 32 attempts ought to be enough ... */
+ /* Limit probing iterations, depending on the available pool size. */
+ for (i = 0; i < LJ_TARGET_JUMPRANGE; i++) {
if (mcode_validptr(hint)) {
void *p = mcode_alloc_at(J, hint, sz, MCPROT_GEN);
@@ -239,11 +240,11 @@ static void *mcode_alloc(jit_State *J, size_t sz)
return p;
if (p) mcode_free(J, p, sz); /* Free badly placed area. */
}
- /* Next try probing pseudo-random addresses. */
+ /* Next try probing 64K-aligned pseudo-random addresses. */
do {
- hint = (0x78fb ^ LJ_PRNG_BITS(J, 15)) << 16; /* 64K aligned. */
- } while (!(hint + sz < range));
- hint = target + hint - (range>>1);
+ hint = LJ_PRNG_BITS(J, LJ_TARGET_JUMPRANGE-16) << 16;
+ } while (!(hint + sz < range+range));
+ hint = target + hint - range;
}
lj_trace_err(J, LJ_TRERR_MCODEAL); /* Give up. OS probably ignores hints? */
return NULL;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.h
index 503e001974f..f0847e931e5 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_mcode.h
@@ -1,6 +1,6 @@
/*
** Machine code management.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_MCODE_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_meta.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_meta.c
index c7993ad6264..0bd4d8429b6 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_meta.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_meta.c
@@ -1,6 +1,6 @@
/*
** Metamethod handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_meta.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_meta.h
index 52e85cba2d1..73b45724735 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_meta.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_meta.h
@@ -1,6 +1,6 @@
/*
** Metamethod handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_META_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_obj.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_obj.c
index 1daea817ace..ee33aeb3a8e 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_obj.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_obj.c
@@ -1,6 +1,6 @@
/*
** Miscellaneous object handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_obj_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_obj.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_obj.h
index 059eb13250a..52372c3e7e5 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_obj.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_obj.h
@@ -1,6 +1,6 @@
/*
** LuaJIT VM tags, values and objects.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -843,12 +843,16 @@ static LJ_AINLINE void setlightudV(TValue *o, void *p)
#endif
#if LJ_FR2
-#define setcont(o, f) ((o)->u64 = (uint64_t)(uintptr_t)(void *)(f))
+#define contptr(f) ((void *)(f))
+#define setcont(o, f) ((o)->u64 = (uint64_t)(uintptr_t)contptr(f))
#elif LJ_64
+#define contptr(f) \
+ ((void *)(uintptr_t)(uint32_t)((intptr_t)(f) - (intptr_t)lj_vm_asm_begin))
#define setcont(o, f) \
((o)->u64 = (uint64_t)(void *)(f) - (uint64_t)lj_vm_asm_begin)
#else
-#define setcont(o, f) setlightudV((o), (void *)(f))
+#define contptr(f) ((void *)(f))
+#define setcont(o, f) setlightudV((o), contptr(f))
#endif
#define tvchecklive(L, o) \
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_dce.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_dce.c
index e5a096af561..2417f3242aa 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_dce.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_dce.c
@@ -1,6 +1,6 @@
/*
** DCE: Dead Code Elimination. Pre-LOOP only -- ASM already performs DCE.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_opt_dce_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_fold.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_fold.c
index e1d136915cb..acbf36a5c7c 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_fold.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_fold.c
@@ -2,7 +2,7 @@
** FOLD: Constant Folding, Algebraic Simplifications and Reassociation.
** ABCelim: Array Bounds Check Elimination.
** CSE: Common-Subexpression Elimination.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_opt_fold_c
@@ -136,8 +136,8 @@
/* Some local macros to save typing. Undef'd at the end. */
#define IR(ref) (&J->cur.ir[(ref)])
#define fins (&J->fold.ins)
-#define fleft (&J->fold.left)
-#define fright (&J->fold.right)
+#define fleft (J->fold.left)
+#define fright (J->fold.right)
#define knumleft (ir_knum(fleft)->n)
#define knumright (ir_knum(fright)->n)
@@ -173,8 +173,6 @@ LJFOLD(ADD KNUM KNUM)
LJFOLD(SUB KNUM KNUM)
LJFOLD(MUL KNUM KNUM)
LJFOLD(DIV KNUM KNUM)
-LJFOLD(NEG KNUM KNUM)
-LJFOLD(ABS KNUM KNUM)
LJFOLD(ATAN2 KNUM KNUM)
LJFOLD(LDEXP KNUM KNUM)
LJFOLD(MIN KNUM KNUM)
@@ -187,6 +185,15 @@ LJFOLDF(kfold_numarith)
return lj_ir_knum(J, y);
}
+LJFOLD(NEG KNUM FLOAD)
+LJFOLD(ABS KNUM FLOAD)
+LJFOLDF(kfold_numabsneg)
+{
+ lua_Number a = knumleft;
+ lua_Number y = lj_vm_foldarith(a, a, fins->o - IR_ADD);
+ return lj_ir_knum(J, y);
+}
+
LJFOLD(LDEXP KNUM KINT)
LJFOLDF(kfold_ldexp)
{
@@ -347,6 +354,11 @@ static uint64_t kfold_int64arith(uint64_t k1, uint64_t k2, IROp op)
case IR_BAND: k1 &= k2; break;
case IR_BOR: k1 |= k2; break;
case IR_BXOR: k1 ^= k2; break;
+ case IR_BSHL: k1 <<= (k2 & 63); break;
+ case IR_BSHR: k1 = (int32_t)((uint32_t)k1 >> (k2 & 63)); break;
+ case IR_BSAR: k1 >>= (k2 & 63); break;
+ case IR_BROL: k1 = (int32_t)lj_rol((uint32_t)k1, (k2 & 63)); break;
+ case IR_BROR: k1 = (int32_t)lj_ror((uint32_t)k1, (k2 & 63)); break;
#endif
default: UNUSED(k2); lua_assert(0); break;
}
@@ -436,14 +448,14 @@ LJFOLDF(kfold_int64comp)
#if LJ_HASFFI
uint64_t a = ir_k64(fleft)->u64, b = ir_k64(fright)->u64;
switch ((IROp)fins->o) {
- case IR_LT: return CONDFOLD(a < b);
- case IR_GE: return CONDFOLD(a >= b);
- case IR_LE: return CONDFOLD(a <= b);
- case IR_GT: return CONDFOLD(a > b);
- case IR_ULT: return CONDFOLD((uint64_t)a < (uint64_t)b);
- case IR_UGE: return CONDFOLD((uint64_t)a >= (uint64_t)b);
- case IR_ULE: return CONDFOLD((uint64_t)a <= (uint64_t)b);
- case IR_UGT: return CONDFOLD((uint64_t)a > (uint64_t)b);
+ case IR_LT: return CONDFOLD((int64_t)a < (int64_t)b);
+ case IR_GE: return CONDFOLD((int64_t)a >= (int64_t)b);
+ case IR_LE: return CONDFOLD((int64_t)a <= (int64_t)b);
+ case IR_GT: return CONDFOLD((int64_t)a > (int64_t)b);
+ case IR_ULT: return CONDFOLD(a < b);
+ case IR_UGE: return CONDFOLD(a >= b);
+ case IR_ULE: return CONDFOLD(a <= b);
+ case IR_UGT: return CONDFOLD(a > b);
default: lua_assert(0); return FAILFOLD;
}
#else
@@ -502,7 +514,7 @@ LJFOLDF(kfold_strref_snew)
PHIBARRIER(ir);
fins->op2 = emitir(IRTI(IR_ADD), ir->op2, fins->op2); /* Clobbers fins! */
fins->op1 = str;
- fins->ot = IRT(IR_STRREF, IRT_P32);
+ fins->ot = IRT(IR_STRREF, IRT_PGC);
return RETRYFOLD;
}
}
@@ -911,13 +923,13 @@ LJFOLDF(shortcut_round)
return NEXTFOLD;
}
-LJFOLD(ABS ABS KNUM)
+LJFOLD(ABS ABS FLOAD)
LJFOLDF(shortcut_left)
{
return LEFTFOLD; /* f(g(x)) ==> g(x) */
}
-LJFOLD(ABS NEG KNUM)
+LJFOLD(ABS NEG FLOAD)
LJFOLDF(shortcut_dropleft)
{
PHIBARRIER(fleft);
@@ -998,8 +1010,10 @@ LJFOLDF(simplify_nummuldiv_k)
if (n == 1.0) { /* x o 1 ==> x */
return LEFTFOLD;
} else if (n == -1.0) { /* x o -1 ==> -x */
+ IRRef op1 = fins->op1;
+ fins->op2 = (IRRef1)lj_ir_ksimd(J, LJ_KSIMD_NEG); /* Modifies fins. */
+ fins->op1 = op1;
fins->o = IR_NEG;
- fins->op2 = (IRRef1)lj_ir_knum_neg(J);
return RETRYFOLD;
} else if (fins->o == IR_MUL && n == 2.0) { /* x * 2 ==> x + x */
fins->o = IR_ADD;
@@ -1651,6 +1665,14 @@ LJFOLDF(simplify_shiftk_andk)
fins->op2 = (IRRef1)lj_ir_kint(J, k);
fins->ot = IRTI(IR_BAND);
return RETRYFOLD;
+ } else if (irk->o == IR_KINT64) {
+ uint64_t k = kfold_int64arith(ir_k64(irk)->u64, fright->i, (IROp)fins->o);
+ IROpT ot = fleft->ot;
+ fins->op1 = fleft->op1;
+ fins->op1 = (IRRef1)lj_opt_fold(J);
+ fins->op2 = (IRRef1)lj_ir_kint64(J, k);
+ fins->ot = ot;
+ return RETRYFOLD;
}
return NEXTFOLD;
}
@@ -1666,6 +1688,47 @@ LJFOLDF(simplify_andk_shiftk)
return NEXTFOLD;
}
+LJFOLD(BAND BOR KINT)
+LJFOLD(BOR BAND KINT)
+LJFOLDF(simplify_andor_k)
+{
+ IRIns *irk = IR(fleft->op2);
+ PHIBARRIER(fleft);
+ if (irk->o == IR_KINT) {
+ int32_t k = kfold_intop(irk->i, fright->i, (IROp)fins->o);
+ /* (i | k1) & k2 ==> i & k2, if (k1 & k2) == 0. */
+ /* (i & k1) | k2 ==> i | k2, if (k1 | k2) == -1. */
+ if (k == (fins->o == IR_BAND ? 0 : -1)) {
+ fins->op1 = fleft->op1;
+ return RETRYFOLD;
+ }
+ }
+ return NEXTFOLD;
+}
+
+LJFOLD(BAND BOR KINT64)
+LJFOLD(BOR BAND KINT64)
+LJFOLDF(simplify_andor_k64)
+{
+#if LJ_HASFFI
+ IRIns *irk = IR(fleft->op2);
+ PHIBARRIER(fleft);
+ if (irk->o == IR_KINT64) {
+ uint64_t k = kfold_int64arith(ir_k64(irk)->u64,
+ ir_k64(fright)->u64, (IROp)fins->o);
+ /* (i | k1) & k2 ==> i & k2, if (k1 & k2) == 0. */
+ /* (i & k1) | k2 ==> i | k2, if (k1 | k2) == -1. */
+ if (k == (fins->o == IR_BAND ? (uint64_t)0 : ~(uint64_t)0)) {
+ fins->op1 = fleft->op1;
+ return RETRYFOLD;
+ }
+ }
+ return NEXTFOLD;
+#else
+ UNUSED(J); lua_assert(0); return FAILFOLD;
+#endif
+}
+
/* -- Reassociation ------------------------------------------------------- */
LJFOLD(ADD ADD KINT)
@@ -2393,10 +2456,14 @@ retry:
if (fins->op1 >= J->cur.nk) {
key += (uint32_t)IR(fins->op1)->o << 10;
*fleft = *IR(fins->op1);
+ if (fins->op1 < REF_TRUE)
+ fleft[1] = IR(fins->op1)[1];
}
if (fins->op2 >= J->cur.nk) {
key += (uint32_t)IR(fins->op2)->o;
*fright = *IR(fins->op2);
+ if (fins->op2 < REF_TRUE)
+ fright[1] = IR(fins->op2)[1];
} else {
key += (fins->op2 & 0x3ffu); /* Literal mask. Must include IRCONV_*MASK. */
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_loop.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_loop.c
index 7978f3eeafc..04c6d06ddff 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_loop.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_loop.c
@@ -1,6 +1,6 @@
/*
** LOOP: Loop Optimizations.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_opt_loop_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_mem.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_mem.c
index 5549b0d02dd..cc177d39e8b 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_mem.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_mem.c
@@ -3,7 +3,7 @@
** AA: Alias Analysis using high-level semantic disambiguation.
** FWD: Load Forwarding (L2L) + Store Forwarding (S2L).
** DSE: Dead-Store Elimination.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_opt_mem_c
@@ -22,8 +22,8 @@
/* Some local macros to save typing. Undef'd at the end. */
#define IR(ref) (&J->cur.ir[(ref)])
#define fins (&J->fold.ins)
-#define fleft (&J->fold.left)
-#define fright (&J->fold.right)
+#define fleft (J->fold.left)
+#define fright (J->fold.right)
/*
** Caveat #1: return value is not always a TRef -- only use with tref_ref().
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_narrow.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_narrow.c
index b1ab5ba8eeb..cd96ca4b4f9 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_narrow.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_narrow.c
@@ -1,7 +1,7 @@
/*
** NARROW: Narrowing of numbers to integers (double to int32_t).
** STRIPOV: Stripping of overflow checks.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_opt_narrow_c
@@ -517,18 +517,24 @@ static int numisint(lua_Number n)
return (n == (lua_Number)lj_num2int(n));
}
+/* Convert string to number. Error out for non-numeric string values. */
+static TRef conv_str_tonum(jit_State *J, TRef tr, TValue *o)
+{
+ if (tref_isstr(tr)) {
+ tr = emitir(IRTG(IR_STRTO, IRT_NUM), tr, 0);
+ /* Would need an inverted STRTO for this rare and useless case. */
+ if (!lj_strscan_num(strV(o), o)) /* Convert in-place. Value used below. */
+ lj_trace_err(J, LJ_TRERR_BADTYPE); /* Punt if non-numeric. */
+ }
+ return tr;
+}
+
/* Narrowing of arithmetic operations. */
TRef lj_opt_narrow_arith(jit_State *J, TRef rb, TRef rc,
TValue *vb, TValue *vc, IROp op)
{
- if (tref_isstr(rb)) {
- rb = emitir(IRTG(IR_STRTO, IRT_NUM), rb, 0);
- lj_strscan_num(strV(vb), vb);
- }
- if (tref_isstr(rc)) {
- rc = emitir(IRTG(IR_STRTO, IRT_NUM), rc, 0);
- lj_strscan_num(strV(vc), vc);
- }
+ rb = conv_str_tonum(J, rb, vb);
+ rc = conv_str_tonum(J, rc, vc);
/* Must not narrow MUL in non-DUALNUM variant, because it loses -0. */
if ((op >= IR_ADD && op <= (LJ_DUALNUM ? IR_MUL : IR_SUB)) &&
tref_isinteger(rb) && tref_isinteger(rc) &&
@@ -543,24 +549,21 @@ TRef lj_opt_narrow_arith(jit_State *J, TRef rb, TRef rc,
/* Narrowing of unary minus operator. */
TRef lj_opt_narrow_unm(jit_State *J, TRef rc, TValue *vc)
{
- if (tref_isstr(rc)) {
- rc = emitir(IRTG(IR_STRTO, IRT_NUM), rc, 0);
- lj_strscan_num(strV(vc), vc);
- }
+ rc = conv_str_tonum(J, rc, vc);
if (tref_isinteger(rc)) {
if ((uint32_t)numberVint(vc) != 0x80000000u)
return emitir(IRTGI(IR_SUBOV), lj_ir_kint(J, 0), rc);
rc = emitir(IRTN(IR_CONV), rc, IRCONV_NUM_INT);
}
- return emitir(IRTN(IR_NEG), rc, lj_ir_knum_neg(J));
+ return emitir(IRTN(IR_NEG), rc, lj_ir_ksimd(J, LJ_KSIMD_NEG));
}
/* Narrowing of modulo operator. */
-TRef lj_opt_narrow_mod(jit_State *J, TRef rb, TRef rc, TValue *vc)
+TRef lj_opt_narrow_mod(jit_State *J, TRef rb, TRef rc, TValue *vb, TValue *vc)
{
TRef tmp;
- if (tvisstr(vc) && !lj_strscan_num(strV(vc), vc))
- lj_trace_err(J, LJ_TRERR_BADTYPE);
+ rb = conv_str_tonum(J, rb, vb);
+ rc = conv_str_tonum(J, rc, vc);
if ((LJ_DUALNUM || (J->flags & JIT_F_OPT_NARROW)) &&
tref_isinteger(rb) && tref_isinteger(rc) &&
(tvisint(vc) ? intV(vc) != 0 : !tviszero(vc))) {
@@ -577,10 +580,11 @@ TRef lj_opt_narrow_mod(jit_State *J, TRef rb, TRef rc, TValue *vc)
}
/* Narrowing of power operator or math.pow. */
-TRef lj_opt_narrow_pow(jit_State *J, TRef rb, TRef rc, TValue *vc)
+TRef lj_opt_narrow_pow(jit_State *J, TRef rb, TRef rc, TValue *vb, TValue *vc)
{
- if (tvisstr(vc) && !lj_strscan_num(strV(vc), vc))
- lj_trace_err(J, LJ_TRERR_BADTYPE);
+ rb = conv_str_tonum(J, rb, vb);
+ rb = lj_ir_tonum(J, rb); /* Left arg is always treated as an FP number. */
+ rc = conv_str_tonum(J, rc, vc);
/* Narrowing must be unconditional to preserve (-x)^i semantics. */
if (tvisint(vc) || numisint(numV(vc))) {
int checkrange = 0;
@@ -591,8 +595,6 @@ TRef lj_opt_narrow_pow(jit_State *J, TRef rb, TRef rc, TValue *vc)
checkrange = 1;
}
if (!tref_isinteger(rc)) {
- if (tref_isstr(rc))
- rc = emitir(IRTG(IR_STRTO, IRT_NUM), rc, 0);
/* Guarded conversion to integer! */
rc = emitir(IRTGI(IR_CONV), rc, IRCONV_INT_NUM|IRCONV_CHECK);
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_sink.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_sink.c
index 975ee831d91..929ccb61877 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_sink.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_sink.c
@@ -1,6 +1,6 @@
/*
** SINK: Allocation Sinking and Store Sinking.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_opt_sink_c
@@ -153,10 +153,9 @@ static void sink_remark_phi(jit_State *J)
remark = 0;
for (ir = IR(J->cur.nins-1); ir->o == IR_PHI; ir--) {
IRIns *irl = IR(ir->op1), *irr = IR(ir->op2);
- if (((irl->t.irt ^ irr->t.irt) & IRT_MARK))
- remark = 1;
- else if (irl->prev == irr->prev)
+ if (!((irl->t.irt ^ irr->t.irt) & IRT_MARK) && irl->prev == irr->prev)
continue;
+ remark |= (~(irl->t.irt & irr->t.irt) & IRT_MARK);
irt_setmark(IR(ir->op1)->t);
irt_setmark(IR(ir->op2)->t);
}
@@ -166,8 +165,8 @@ static void sink_remark_phi(jit_State *J)
/* Sweep instructions and tag sunken allocations and stores. */
static void sink_sweep_ins(jit_State *J)
{
- IRIns *ir, *irfirst = IR(J->cur.nk);
- for (ir = IR(J->cur.nins-1) ; ir >= irfirst; ir--) {
+ IRIns *ir, *irbase = IR(REF_BASE);
+ for (ir = IR(J->cur.nins-1) ; ir >= irbase; ir--) {
switch (ir->o) {
case IR_ASTORE: case IR_HSTORE: case IR_FSTORE: case IR_XSTORE: {
IRIns *ira = sink_checkalloc(J, ir);
@@ -217,6 +216,12 @@ static void sink_sweep_ins(jit_State *J)
break;
}
}
+ for (ir = IR(J->cur.nk); ir < irbase; ir++) {
+ irt_clearmark(ir->t);
+ ir->prev = REGSP_INIT;
+ if (irt_is64(ir->t) && ir->o != IR_KNULL)
+ ir++;
+ }
}
/* Allocation sinking and store sinking.
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_split.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_split.c
index 6def4161aa3..fc9352042ed 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_split.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_opt_split.c
@@ -1,6 +1,6 @@
/*
** SPLIT: Split 64 bit IR instructions into 32 bit IR instructions.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_opt_split_c
@@ -16,6 +16,7 @@
#include "lj_jit.h"
#include "lj_ircall.h"
#include "lj_iropt.h"
+#include "lj_dispatch.h"
#include "lj_vm.h"
/* SPLIT pass:
@@ -353,6 +354,8 @@ static void split_ir(jit_State *J)
ir->prev = ref; /* Identity substitution for loword. */
hisubst[ref] = 0;
}
+ if (irt_is64(ir->t) && ir->o != IR_KNULL)
+ ref++;
}
/* Process old IR instructions. */
@@ -433,7 +436,8 @@ static void split_ir(jit_State *J)
nir->o = IR_CONV; /* Pass through loword. */
nir->op2 = (IRT_INT << 5) | IRT_INT;
hi = split_emit(J, IRT(ir->o == IR_NEG ? IR_BXOR : IR_BAND, IRT_SOFTFP),
- hisubst[ir->op1], hisubst[ir->op2]);
+ hisubst[ir->op1],
+ lj_ir_kint(J, (int32_t)(0x7fffffffu + (ir->o == IR_NEG))));
break;
case IR_SLOAD:
if ((nir->op2 & IRSLOAD_CONVERT)) { /* Convert from int to number. */
@@ -448,6 +452,11 @@ static void split_ir(jit_State *J)
case IR_STRTO:
hi = split_emit(J, IRT(IR_HIOP, IRT_SOFTFP), nref, nref);
break;
+ case IR_FLOAD:
+ lua_assert(ir->op1 == REF_NIL);
+ hi = lj_ir_kint(J, *(int32_t*)((char*)J2GG(J) + ir->op2 + LJ_LE*4));
+ nir->op2 += LJ_BE*4;
+ break;
case IR_XLOAD: {
IRIns inslo = *nir; /* Save/undo the emit of the lo XLOAD. */
J->cur.nins--;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_parse.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_parse.c
index 610c861480e..08f7cfa6ac5 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_parse.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_parse.c
@@ -1,6 +1,6 @@
/*
** Lua parser (source code -> bytecode).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -1282,12 +1282,14 @@ static void fscope_end(FuncState *fs)
MSize idx = gola_new(ls, NAME_BREAK, VSTACK_LABEL, fs->pc);
ls->vtop = idx; /* Drop break label immediately. */
gola_resolve(ls, bl, idx);
+ } else { /* Need the fixup step to propagate the breaks. */
+ gola_fixup(ls, bl);
return;
- } /* else: need the fixup step to propagate the breaks. */
- } else if (!(bl->flags & FSCOPE_GOLA)) {
- return;
+ }
+ }
+ if ((bl->flags & FSCOPE_GOLA)) {
+ gola_fixup(ls, bl);
}
- gola_fixup(ls, bl);
}
/* Mark scope as having an upvalue. */
@@ -2177,6 +2179,8 @@ static void assign_adjust(LexState *ls, BCReg nvars, BCReg nexps, ExpDesc *e)
bcemit_nil(fs, reg, (BCReg)extra);
}
}
+ if (nexps > nvars)
+ ls->fs->freereg -= nexps - nvars; /* Drop leftover regs. */
}
/* Recursively parse assignment statement. */
@@ -2210,8 +2214,6 @@ static void parse_assignment(LexState *ls, LHSVarList *lh, BCReg nvars)
return;
}
assign_adjust(ls, nvars, nexps, &e);
- if (nexps > nvars)
- ls->fs->freereg -= nexps - nvars; /* Drop leftover regs. */
}
/* Assign RHS to LHS and recurse downwards. */
expr_init(&e, VNONRELOC, ls->fs->freereg-1);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_parse.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_parse.h
index b383c8971b1..ceeab6994ff 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_parse.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_parse.h
@@ -1,6 +1,6 @@
/*
** Lua parser (source code -> bytecode).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_PARSE_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_profile.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_profile.c
index 95b0a621e84..116998e1e8f 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_profile.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_profile.c
@@ -1,6 +1,6 @@
/*
** Low-overhead profiling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_profile_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_profile.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_profile.h
index 14c3fc0881f..0cccfd78cd7 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_profile.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_profile.h
@@ -1,6 +1,6 @@
/*
** Low-overhead profiling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_PROFILE_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_record.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_record.c
index 90f53af3e6b..9d0469c4251 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_record.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_record.c
@@ -1,6 +1,6 @@
/*
** Trace recorder (bytecode -> SSA IR).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_record_c
@@ -51,7 +51,7 @@ static void rec_check_ir(jit_State *J)
{
IRRef i, nins = J->cur.nins, nk = J->cur.nk;
lua_assert(nk <= REF_BIAS && nins >= REF_BIAS && nins < 65536);
- for (i = nins-1; i >= nk; i--) {
+ for (i = nk; i < nins; i++) {
IRIns *ir = IR(i);
uint32_t mode = lj_ir_mode[ir->o];
IRRef op1 = ir->op1;
@@ -61,7 +61,10 @@ static void rec_check_ir(jit_State *J)
case IRMref: lua_assert(op1 >= nk);
lua_assert(i >= REF_BIAS ? op1 < i : op1 > i); break;
case IRMlit: break;
- case IRMcst: lua_assert(i < REF_BIAS); continue;
+ case IRMcst: lua_assert(i < REF_BIAS);
+ if (irt_is64(ir->t) && ir->o != IR_KNULL)
+ i++;
+ continue;
}
switch (irm_op2(mode)) {
case IRMnone: lua_assert(op2 == 0); break;
@@ -84,30 +87,48 @@ static void rec_check_slots(jit_State *J)
BCReg s, nslots = J->baseslot + J->maxslot;
int32_t depth = 0;
cTValue *base = J->L->base - J->baseslot;
- lua_assert(J->baseslot >= 1 && J->baseslot < LJ_MAX_JSLOTS);
- lua_assert(J->baseslot == 1 || (J->slot[J->baseslot-1] & TREF_FRAME));
+ lua_assert(J->baseslot >= 1+LJ_FR2 && J->baseslot < LJ_MAX_JSLOTS);
+ lua_assert(J->baseslot == 1+LJ_FR2 || (J->slot[J->baseslot-1] & TREF_FRAME));
lua_assert(nslots < LJ_MAX_JSLOTS);
for (s = 0; s < nslots; s++) {
TRef tr = J->slot[s];
if (tr) {
cTValue *tv = &base[s];
IRRef ref = tref_ref(tr);
- IRIns *ir;
- lua_assert(ref >= J->cur.nk && ref < J->cur.nins);
- ir = IR(ref);
- lua_assert(irt_t(ir->t) == tref_t(tr));
+ IRIns *ir = NULL; /* Silence compiler. */
+ if (!LJ_FR2 || ref || !(tr & (TREF_FRAME | TREF_CONT))) {
+ lua_assert(ref >= J->cur.nk && ref < J->cur.nins);
+ ir = IR(ref);
+ lua_assert(irt_t(ir->t) == tref_t(tr));
+ }
if (s == 0) {
lua_assert(tref_isfunc(tr));
+#if LJ_FR2
+ } else if (s == 1) {
+ lua_assert((tr & ~TREF_FRAME) == 0);
+#endif
} else if ((tr & TREF_FRAME)) {
GCfunc *fn = gco2func(frame_gc(tv));
BCReg delta = (BCReg)(tv - frame_prev(tv));
+#if LJ_FR2
+ if (ref)
+ lua_assert(ir_knum(ir)->u64 == tv->u64);
+ tr = J->slot[s-1];
+ ir = IR(tref_ref(tr));
+#endif
lua_assert(tref_isfunc(tr));
if (tref_isk(tr)) lua_assert(fn == ir_kfunc(ir));
- lua_assert(s > delta ? (J->slot[s-delta] & TREF_FRAME) : (s == delta));
+ lua_assert(s > delta + LJ_FR2 ? (J->slot[s-delta] & TREF_FRAME)
+ : (s == delta + LJ_FR2));
depth++;
} else if ((tr & TREF_CONT)) {
+#if LJ_FR2
+ if (ref)
+ lua_assert(ir_knum(ir)->u64 == tv->u64);
+#else
lua_assert(ir_kptr(ir) == gcrefp(tv->gcr, void));
- lua_assert((J->slot[s+1] & TREF_FRAME));
+#endif
+ lua_assert((J->slot[s+1+LJ_FR2] & TREF_FRAME));
depth++;
} else {
if (tvisnumber(tv))
@@ -159,10 +180,10 @@ static TRef sload(jit_State *J, int32_t slot)
/* Get TRef for current function. */
static TRef getcurrf(jit_State *J)
{
- if (J->base[-1])
- return J->base[-1];
- lua_assert(J->baseslot == 1);
- return sloadt(J, -1, IRT_FUNC, IRSLOAD_READONLY);
+ if (J->base[-1-LJ_FR2])
+ return J->base[-1-LJ_FR2];
+ lua_assert(J->baseslot == 1+LJ_FR2);
+ return sloadt(J, -1-LJ_FR2, IRT_FUNC, IRSLOAD_READONLY);
}
/* Compare for raw object equality.
@@ -506,7 +527,6 @@ static LoopEvent rec_for(jit_State *J, const BCIns *fori, int isforl)
static LoopEvent rec_iterl(jit_State *J, const BCIns iterins)
{
BCReg ra = bc_a(iterins);
- lua_assert(!LJ_FR2); /* TODO_FR2: handle different frame setup. */
if (!tref_isnil(getslot(J, ra))) { /* Looping back? */
J->base[ra-1] = J->base[ra]; /* Copy result of ITERC to control var. */
J->maxslot = ra-1+bc_b(J->pc[-1]);
@@ -643,8 +663,8 @@ static TRef rec_call_specialize(jit_State *J, GCfunc *fn, TRef tr)
GCproto *pt = funcproto(fn);
/* Too many closures created? Probably not a monomorphic function. */
if (pt->flags >= PROTO_CLC_POLY) { /* Specialize to prototype instead. */
- TRef trpt = emitir(IRT(IR_FLOAD, IRT_P32), tr, IRFL_FUNC_PC);
- emitir(IRTG(IR_EQ, IRT_P32), trpt, lj_ir_kptr(J, proto_bc(pt)));
+ TRef trpt = emitir(IRT(IR_FLOAD, IRT_PGC), tr, IRFL_FUNC_PC);
+ emitir(IRTG(IR_EQ, IRT_PGC), trpt, lj_ir_kptr(J, proto_bc(pt)));
(void)lj_ir_kgc(J, obj2gco(pt), IRT_PROTO); /* Prevent GC of proto. */
return tr;
}
@@ -675,22 +695,31 @@ static void rec_call_setup(jit_State *J, BCReg func, ptrdiff_t nargs)
{
RecordIndex ix;
TValue *functv = &J->L->base[func];
- TRef *fbase = &J->base[func];
+ TRef kfunc, *fbase = &J->base[func];
ptrdiff_t i;
- lua_assert(!LJ_FR2); /* TODO_FR2: handle different frame setup. */
- for (i = 0; i <= nargs; i++)
- (void)getslot(J, func+i); /* Ensure func and all args have a reference. */
+ (void)getslot(J, func); /* Ensure func has a reference. */
+ for (i = 1; i <= nargs; i++)
+ (void)getslot(J, func+LJ_FR2+i); /* Ensure all args have a reference. */
if (!tref_isfunc(fbase[0])) { /* Resolve __call metamethod. */
ix.tab = fbase[0];
copyTV(J->L, &ix.tabv, functv);
if (!lj_record_mm_lookup(J, &ix, MM_call) || !tref_isfunc(ix.mobj))
lj_trace_err(J, LJ_TRERR_NOMM);
- for (i = ++nargs; i > 0; i--) /* Shift arguments up. */
- fbase[i] = fbase[i-1];
+ for (i = ++nargs; i > LJ_FR2; i--) /* Shift arguments up. */
+ fbase[i+LJ_FR2] = fbase[i+LJ_FR2-1];
+#if LJ_FR2
+ fbase[2] = fbase[0];
+#endif
fbase[0] = ix.mobj; /* Replace function. */
functv = &ix.mobjv;
}
- fbase[0] = TREF_FRAME | rec_call_specialize(J, funcV(functv), fbase[0]);
+ kfunc = rec_call_specialize(J, funcV(functv), fbase[0]);
+#if LJ_FR2
+ fbase[0] = kfunc;
+ fbase[1] = TREF_FRAME;
+#else
+ fbase[0] = kfunc | TREF_FRAME;
+#endif
J->maxslot = (BCReg)nargs;
}
@@ -700,8 +729,8 @@ void lj_record_call(jit_State *J, BCReg func, ptrdiff_t nargs)
rec_call_setup(J, func, nargs);
/* Bump frame. */
J->framedepth++;
- J->base += func+1;
- J->baseslot += func+1;
+ J->base += func+1+LJ_FR2;
+ J->baseslot += func+1+LJ_FR2;
}
/* Record tail call. */
@@ -717,7 +746,9 @@ void lj_record_tailcall(jit_State *J, BCReg func, ptrdiff_t nargs)
func += cbase;
}
/* Move func + args down. */
- memmove(&J->base[-1], &J->base[func], sizeof(TRef)*(J->maxslot+1));
+ if (LJ_FR2 && J->baseslot == 2)
+ J->base[func+1] = TREF_FRAME;
+ memmove(&J->base[-1-LJ_FR2], &J->base[func], sizeof(TRef)*(J->maxslot+1+LJ_FR2));
/* Note: the new TREF_FRAME is now at J->base[-1] (even for slot #0). */
/* Tailcalls can form a loop, so count towards the loop unroll limit. */
if (++J->tailcalled > J->loopunroll)
@@ -758,9 +789,9 @@ void lj_record_ret(jit_State *J, BCReg rbase, ptrdiff_t gotresults)
(void)getslot(J, rbase+i); /* Ensure all results have a reference. */
while (frame_ispcall(frame)) { /* Immediately resolve pcall() returns. */
BCReg cbase = (BCReg)frame_delta(frame);
- if (--J->framedepth < 0)
+ if (--J->framedepth <= 0)
lj_trace_err(J, LJ_TRERR_NYIRETL);
- lua_assert(J->baseslot > 1);
+ lua_assert(J->baseslot > 1+LJ_FR2);
gotresults++;
rbase += cbase;
J->baseslot -= (BCReg)cbase;
@@ -784,7 +815,7 @@ void lj_record_ret(jit_State *J, BCReg rbase, ptrdiff_t gotresults)
BCReg cbase = (BCReg)frame_delta(frame);
if (--J->framedepth < 0) /* NYI: return of vararg func to lower frame. */
lj_trace_err(J, LJ_TRERR_NYIRETL);
- lua_assert(J->baseslot > 1);
+ lua_assert(J->baseslot > 1+LJ_FR2);
rbase += cbase;
J->baseslot -= (BCReg)cbase;
J->base -= cbase;
@@ -794,8 +825,7 @@ void lj_record_ret(jit_State *J, BCReg rbase, ptrdiff_t gotresults)
BCIns callins = *(frame_pc(frame)-1);
ptrdiff_t nresults = bc_b(callins) ? (ptrdiff_t)bc_b(callins)-1 :gotresults;
BCReg cbase = bc_a(callins);
- GCproto *pt = funcproto(frame_func(frame - (cbase+1-LJ_FR2)));
- lua_assert(!LJ_FR2); /* TODO_FR2: handle different frame teardown. */
+ GCproto *pt = funcproto(frame_func(frame - (cbase+1+LJ_FR2)));
if ((pt->flags & PROTO_NOJIT))
lj_trace_err(J, LJ_TRERR_CJITOFF);
if (J->framedepth == 0 && J->pt && frame == J->L->base - 1) {
@@ -808,13 +838,13 @@ void lj_record_ret(jit_State *J, BCReg rbase, ptrdiff_t gotresults)
lj_snap_add(J);
}
for (i = 0; i < nresults; i++) /* Adjust results. */
- J->base[i-1] = i < gotresults ? J->base[rbase+i] : TREF_NIL;
+ J->base[i-1-LJ_FR2] = i < gotresults ? J->base[rbase+i] : TREF_NIL;
J->maxslot = cbase+(BCReg)nresults;
if (J->framedepth > 0) { /* Return to a frame that is part of the trace. */
J->framedepth--;
- lua_assert(J->baseslot > cbase+1);
- J->baseslot -= cbase+1;
- J->base -= cbase+1;
+ lua_assert(J->baseslot > cbase+1+LJ_FR2);
+ J->baseslot -= cbase+1+LJ_FR2;
+ J->base -= cbase+1+LJ_FR2;
} else if (J->parent == 0 && J->exitno == 0 &&
!bc_isret(bc_op(J->cur.startins))) {
/* Return to lower frame would leave the loop in a root trace. */
@@ -824,13 +854,13 @@ void lj_record_ret(jit_State *J, BCReg rbase, ptrdiff_t gotresults)
} else { /* Return to lower frame. Guard for the target we return to. */
TRef trpt = lj_ir_kgc(J, obj2gco(pt), IRT_PROTO);
TRef trpc = lj_ir_kptr(J, (void *)frame_pc(frame));
- emitir(IRTG(IR_RETF, IRT_P32), trpt, trpc);
+ emitir(IRTG(IR_RETF, IRT_PGC), trpt, trpc);
J->retdepth++;
J->needsnap = 1;
- lua_assert(J->baseslot == 1);
+ lua_assert(J->baseslot == 1+LJ_FR2);
/* Shift result slots up and clear the slots of the new frame below. */
- memmove(J->base + cbase, J->base-1, sizeof(TRef)*nresults);
- memset(J->base-1, 0, sizeof(TRef)*(cbase+1));
+ memmove(J->base + cbase, J->base-1-LJ_FR2, sizeof(TRef)*nresults);
+ memset(J->base-1-LJ_FR2, 0, sizeof(TRef)*(cbase+1+LJ_FR2));
}
} else if (frame_iscont(frame)) { /* Return to continuation frame. */
ASMFunction cont = frame_contf(frame);
@@ -839,32 +869,39 @@ void lj_record_ret(jit_State *J, BCReg rbase, ptrdiff_t gotresults)
lj_trace_err(J, LJ_TRERR_NYIRETL);
J->baseslot -= (BCReg)cbase;
J->base -= cbase;
- J->maxslot = cbase-2;
+ J->maxslot = cbase-(2<<LJ_FR2);
if (cont == lj_cont_ra) {
/* Copy result to destination slot. */
BCReg dst = bc_a(*(frame_contpc(frame)-1));
J->base[dst] = gotresults ? J->base[cbase+rbase] : TREF_NIL;
- if (dst >= J->maxslot) J->maxslot = dst+1;
+ if (dst >= J->maxslot) {
+ J->maxslot = dst+1;
+ }
} else if (cont == lj_cont_nop) {
/* Nothing to do here. */
} else if (cont == lj_cont_cat) {
BCReg bslot = bc_b(*(frame_contpc(frame)-1));
TRef tr = gotresults ? J->base[cbase+rbase] : TREF_NIL;
- if (bslot != cbase-2) { /* Concatenate the remainder. */
+ if (bslot != J->maxslot) { /* Concatenate the remainder. */
TValue *b = J->L->base, save; /* Simulate lower frame and result. */
- J->base[cbase-2] = tr;
- copyTV(J->L, &save, b-2);
- if (gotresults) copyTV(J->L, b-2, b+rbase); else setnilV(b-2);
+ J->base[J->maxslot] = tr;
+ copyTV(J->L, &save, b-(2<<LJ_FR2));
+ if (gotresults)
+ copyTV(J->L, b-(2<<LJ_FR2), b+rbase);
+ else
+ setnilV(b-(2<<LJ_FR2));
J->L->base = b - cbase;
- tr = rec_cat(J, bslot, cbase-2);
+ tr = rec_cat(J, bslot, cbase-(2<<LJ_FR2));
b = J->L->base + cbase; /* Undo. */
J->L->base = b;
- copyTV(J->L, b-2, &save);
+ copyTV(J->L, b-(2<<LJ_FR2), &save);
}
if (tr) { /* Store final result. */
BCReg dst = bc_a(*(frame_contpc(frame)-1));
J->base[dst] = tr;
- if (dst >= J->maxslot) J->maxslot = dst+1;
+ if (dst >= J->maxslot) {
+ J->maxslot = dst+1;
+ }
} /* Otherwise continue with another __concat call. */
} else {
/* Result type already specialized. */
@@ -873,7 +910,7 @@ void lj_record_ret(jit_State *J, BCReg rbase, ptrdiff_t gotresults)
} else {
lj_trace_err(J, LJ_TRERR_NYIRETL); /* NYI: handle return to C frame. */
}
- lua_assert(J->baseslot >= 1);
+ lua_assert(J->baseslot >= 1+LJ_FR2);
}
/* -- Metamethod handling ------------------------------------------------- */
@@ -882,16 +919,16 @@ void lj_record_ret(jit_State *J, BCReg rbase, ptrdiff_t gotresults)
static BCReg rec_mm_prep(jit_State *J, ASMFunction cont)
{
BCReg s, top = cont == lj_cont_cat ? J->maxslot : curr_proto(J->L)->framesize;
-#if LJ_64
- TRef trcont = lj_ir_kptr(J, (void *)((int64_t)cont-(int64_t)lj_vm_asm_begin));
+#if LJ_FR2
+ J->base[top] = lj_ir_k64(J, IR_KNUM, u64ptr(contptr(cont)));
+ J->base[top+1] = TREF_CONT;
#else
- TRef trcont = lj_ir_kptr(J, (void *)cont);
+ J->base[top] = lj_ir_kptr(J, contptr(cont)) | TREF_CONT;
#endif
- J->base[top] = trcont | TREF_CONT;
J->framedepth++;
for (s = J->maxslot; s < top; s++)
J->base[s] = 0; /* Clear frame gap to avoid resurrecting previous refs. */
- return top+1;
+ return top+1+LJ_FR2;
}
/* Record metamethod lookup. */
@@ -910,7 +947,7 @@ int lj_record_mm_lookup(jit_State *J, RecordIndex *ix, MMS mm)
cTValue *mo;
if (LJ_HASFFI && udtype == UDTYPE_FFI_CLIB) {
/* Specialize to the C library namespace object. */
- emitir(IRTG(IR_EQ, IRT_P32), ix->tab, lj_ir_kptr(J, udataV(&ix->tabv)));
+ emitir(IRTG(IR_EQ, IRT_PGC), ix->tab, lj_ir_kptr(J, udataV(&ix->tabv)));
} else {
/* Specialize to the type of userdata. */
TRef tr = emitir(IRT(IR_FLOAD, IRT_U8), ix->tab, IRFL_UDATA_UDTYPE);
@@ -939,7 +976,13 @@ int lj_record_mm_lookup(jit_State *J, RecordIndex *ix, MMS mm)
}
/* The cdata metatable is treated as immutable. */
if (LJ_HASFFI && tref_iscdata(ix->tab)) goto immutable_mt;
+#if LJ_GC64
+ /* TODO: fix ARM32 asm_fload(), so we can use this for all archs. */
+ ix->mt = mix.tab = lj_ir_ggfload(J, IRT_TAB,
+ GG_OFS(g.gcroot[GCROOT_BASEMT+itypemap(&ix->tabv)]));
+#else
ix->mt = mix.tab = lj_ir_ktab(J, mt);
+#endif
goto nocheck;
}
ix->mt = mt ? mix.tab : TREF_NIL;
@@ -969,9 +1012,9 @@ static TRef rec_mm_arith(jit_State *J, RecordIndex *ix, MMS mm)
BCReg func = rec_mm_prep(J, mm == MM_concat ? lj_cont_cat : lj_cont_ra);
TRef *base = J->base + func;
TValue *basev = J->L->base + func;
- base[1] = ix->tab; base[2] = ix->key;
- copyTV(J->L, basev+1, &ix->tabv);
- copyTV(J->L, basev+2, &ix->keyv);
+ base[1+LJ_FR2] = ix->tab; base[2+LJ_FR2] = ix->key;
+ copyTV(J->L, basev+1+LJ_FR2, &ix->tabv);
+ copyTV(J->L, basev+2+LJ_FR2, &ix->keyv);
if (!lj_record_mm_lookup(J, ix, mm)) { /* Lookup mm on 1st operand. */
if (mm != MM_unm) {
ix->tab = ix->key;
@@ -982,8 +1025,10 @@ static TRef rec_mm_arith(jit_State *J, RecordIndex *ix, MMS mm)
lj_trace_err(J, LJ_TRERR_NOMM);
}
ok:
- lua_assert(!LJ_FR2); /* TODO_FR2: handle different frame setup. */
base[0] = ix->mobj;
+#if LJ_FR2
+ base[1] = 0;
+#endif
copyTV(J->L, basev+0, &ix->mobjv);
lj_record_call(J, func, 2);
return 0; /* No result yet. */
@@ -999,8 +1044,9 @@ static TRef rec_mm_len(jit_State *J, TRef tr, TValue *tv)
BCReg func = rec_mm_prep(J, lj_cont_ra);
TRef *base = J->base + func;
TValue *basev = J->L->base + func;
- lua_assert(!LJ_FR2); /* TODO_FR2: handle different frame setup. */
base[0] = ix.mobj; copyTV(J->L, basev+0, &ix.mobjv);
+ base += LJ_FR2;
+ basev += LJ_FR2;
base[1] = tr; copyTV(J->L, basev+1, tv);
#if LJ_52
base[2] = tr; copyTV(J->L, basev+2, tv);
@@ -1020,11 +1066,10 @@ static TRef rec_mm_len(jit_State *J, TRef tr, TValue *tv)
static void rec_mm_callcomp(jit_State *J, RecordIndex *ix, int op)
{
BCReg func = rec_mm_prep(J, (op&1) ? lj_cont_condf : lj_cont_condt);
- TRef *base = J->base + func;
- TValue *tv = J->L->base + func;
- lua_assert(!LJ_FR2); /* TODO_FR2: handle different frame setup. */
- base[0] = ix->mobj; base[1] = ix->val; base[2] = ix->key;
- copyTV(J->L, tv+0, &ix->mobjv);
+ TRef *base = J->base + func + LJ_FR2;
+ TValue *tv = J->L->base + func + LJ_FR2;
+ base[-LJ_FR2] = ix->mobj; base[1] = ix->val; base[2] = ix->key;
+ copyTV(J->L, tv-LJ_FR2, &ix->mobjv);
copyTV(J->L, tv+1, &ix->valv);
copyTV(J->L, tv+2, &ix->keyv);
lj_record_call(J, func, 2);
@@ -1257,8 +1302,8 @@ static TRef rec_idx_key(jit_State *J, RecordIndex *ix, IRRef *rbref,
if ((MSize)k < t->asize) { /* Currently an array key? */
TRef arrayref;
rec_idx_abc(J, asizeref, ikey, t->asize);
- arrayref = emitir(IRT(IR_FLOAD, IRT_P32), ix->tab, IRFL_TAB_ARRAY);
- return emitir(IRT(IR_AREF, IRT_P32), arrayref, ikey);
+ arrayref = emitir(IRT(IR_FLOAD, IRT_PGC), ix->tab, IRFL_TAB_ARRAY);
+ return emitir(IRT(IR_AREF, IRT_PGC), arrayref, ikey);
} else { /* Currently not in array (may be an array extension)? */
emitir(IRTGI(IR_ULE), asizeref, ikey); /* Inv. bounds check. */
if (k == 0 && tref_isk(key))
@@ -1298,13 +1343,13 @@ static TRef rec_idx_key(jit_State *J, RecordIndex *ix, IRRef *rbref,
*rbguard = J->guardemit;
hm = emitir(IRTI(IR_FLOAD), ix->tab, IRFL_TAB_HMASK);
emitir(IRTGI(IR_EQ), hm, lj_ir_kint(J, (int32_t)t->hmask));
- node = emitir(IRT(IR_FLOAD, IRT_P32), ix->tab, IRFL_TAB_NODE);
+ node = emitir(IRT(IR_FLOAD, IRT_PGC), ix->tab, IRFL_TAB_NODE);
kslot = lj_ir_kslot(J, key, hslot / sizeof(Node));
- return emitir(IRTG(IR_HREFK, IRT_P32), node, kslot);
+ return emitir(IRTG(IR_HREFK, IRT_PGC), node, kslot);
}
}
/* Fall back to a regular hash lookup. */
- return emitir(IRT(IR_HREF, IRT_P32), ix->tab, key);
+ return emitir(IRT(IR_HREF, IRT_PGC), ix->tab, key);
}
/* Determine whether a key is NOT one of the fast metamethod names. */
@@ -1341,11 +1386,10 @@ TRef lj_record_idx(jit_State *J, RecordIndex *ix)
handlemm:
if (tref_isfunc(ix->mobj)) { /* Handle metamethod call. */
BCReg func = rec_mm_prep(J, ix->val ? lj_cont_nop : lj_cont_ra);
- TRef *base = J->base + func;
- TValue *tv = J->L->base + func;
- lua_assert(!LJ_FR2); /* TODO_FR2: handle different frame setup. */
- base[0] = ix->mobj; base[1] = ix->tab; base[2] = ix->key;
- setfuncV(J->L, tv+0, funcV(&ix->mobjv));
+ TRef *base = J->base + func + LJ_FR2;
+ TValue *tv = J->L->base + func + LJ_FR2;
+ base[-LJ_FR2] = ix->mobj; base[1] = ix->tab; base[2] = ix->key;
+ setfuncV(J->L, tv-LJ_FR2, funcV(&ix->mobjv));
copyTV(J->L, tv+1, &ix->tabv);
copyTV(J->L, tv+2, &ix->keyv);
if (ix->val) {
@@ -1387,7 +1431,7 @@ TRef lj_record_idx(jit_State *J, RecordIndex *ix)
IRType t = itype2irt(oldv);
TRef res;
if (oldv == niltvg(J2G(J))) {
- emitir(IRTG(IR_EQ, IRT_P32), xref, lj_ir_kkptr(J, niltvg(J2G(J))));
+ emitir(IRTG(IR_EQ, IRT_PGC), xref, lj_ir_kkptr(J, niltvg(J2G(J))));
res = TREF_NIL;
} else {
res = emitir(IRTG(loadop, t), xref, 0);
@@ -1417,7 +1461,7 @@ TRef lj_record_idx(jit_State *J, RecordIndex *ix)
if (hasmm)
emitir(IRTG(loadop, IRT_NIL), xref, 0); /* Guard for nil value. */
else if (xrefop == IR_HREF)
- emitir(IRTG(oldv == niltvg(J2G(J)) ? IR_EQ : IR_NE, IRT_P32),
+ emitir(IRTG(oldv == niltvg(J2G(J)) ? IR_EQ : IR_NE, IRT_PGC),
xref, lj_ir_kkptr(J, niltvg(J2G(J))));
if (ix->idxchain && lj_record_mm_lookup(J, ix, MM_newindex)) {
lua_assert(hasmm);
@@ -1428,7 +1472,7 @@ TRef lj_record_idx(jit_State *J, RecordIndex *ix)
TRef key = ix->key;
if (tref_isinteger(key)) /* NEWREF needs a TValue as a key. */
key = emitir(IRTN(IR_CONV), key, IRCONV_NUM_INT);
- xref = emitir(IRT(IR_NEWREF, IRT_P32), ix->tab, key);
+ xref = emitir(IRT(IR_NEWREF, IRT_PGC), ix->tab, key);
keybarrier = 0; /* NEWREF already takes care of the key barrier. */
#ifdef LUAJIT_ENABLE_TABLE_BUMP
if ((J->flags & JIT_F_OPT_SINK)) /* Avoid a separate flag. */
@@ -1438,7 +1482,7 @@ TRef lj_record_idx(jit_State *J, RecordIndex *ix)
} else if (!lj_opt_fwd_wasnonnil(J, loadop, tref_ref(xref))) {
/* Cannot derive that the previous value was non-nil, must do checks. */
if (xrefop == IR_HREF) /* Guard against store to niltv. */
- emitir(IRTG(IR_NE, IRT_P32), xref, lj_ir_kkptr(J, niltvg(J2G(J))));
+ emitir(IRTG(IR_NE, IRT_PGC), xref, lj_ir_kkptr(J, niltvg(J2G(J))));
if (ix->idxchain) { /* Metamethod lookup required? */
/* A check for NULL metatable is cheaper (hoistable) than a load. */
if (!mt) {
@@ -1460,7 +1504,7 @@ TRef lj_record_idx(jit_State *J, RecordIndex *ix)
emitir(IRT(IR_TBAR, IRT_NIL), ix->tab, 0);
/* Invalidate neg. metamethod cache for stores with certain string keys. */
if (!nommstr(J, ix->key)) {
- TRef fref = emitir(IRT(IR_FREF, IRT_P32), ix->tab, IRFL_TAB_NOMM);
+ TRef fref = emitir(IRT(IR_FREF, IRT_PGC), ix->tab, IRFL_TAB_NOMM);
emitir(IRT(IR_FSTORE, IRT_U8), fref, lj_ir_kint(J, 0));
}
J->needsnap = 1;
@@ -1535,7 +1579,11 @@ static TRef rec_upvalue(jit_State *J, uint32_t uv, TRef val)
goto noconstify;
kfunc = lj_ir_kfunc(J, J->fn);
emitir(IRTG(IR_EQ, IRT_FUNC), fn, kfunc);
- J->base[-1] = TREF_FRAME | kfunc;
+#if LJ_FR2
+ J->base[-2] = kfunc;
+#else
+ J->base[-1] = kfunc | TREF_FRAME;
+#endif
fn = kfunc;
}
tr = lj_record_constify(J, uvval(uvp));
@@ -1546,13 +1594,17 @@ noconstify:
/* Note: this effectively limits LJ_MAX_UPVAL to 127. */
uv = (uv << 8) | (hashrot(uvp->dhash, uvp->dhash + HASH_BIAS) & 0xff);
if (!uvp->closed) {
+ uref = tref_ref(emitir(IRTG(IR_UREFO, IRT_PGC), fn, uv));
/* In current stack? */
if (uvval(uvp) >= tvref(J->L->stack) &&
uvval(uvp) < tvref(J->L->maxstack)) {
int32_t slot = (int32_t)(uvval(uvp) - (J->L->base - J->baseslot));
if (slot >= 0) { /* Aliases an SSA slot? */
+ emitir(IRTG(IR_EQ, IRT_PGC),
+ REF_BASE,
+ emitir(IRT(IR_ADD, IRT_PGC), uref,
+ lj_ir_kint(J, (slot - 1 - LJ_FR2) * -8)));
slot -= (int32_t)J->baseslot; /* Note: slot number may be negative! */
- /* NYI: add IR to guard that it's still aliasing the same slot. */
if (val == 0) {
return getslot(J, slot);
} else {
@@ -1562,10 +1614,12 @@ noconstify:
}
}
}
- uref = tref_ref(emitir(IRTG(IR_UREFO, IRT_P32), fn, uv));
+ emitir(IRTG(IR_UGT, IRT_PGC),
+ emitir(IRT(IR_SUB, IRT_PGC), uref, REF_BASE),
+ lj_ir_kint(J, (J->baseslot + J->maxslot) * 8));
} else {
needbarrier = 1;
- uref = tref_ref(emitir(IRTG(IR_UREFC, IRT_P32), fn, uv));
+ uref = tref_ref(emitir(IRTG(IR_UREFC, IRT_PGC), fn, uv));
}
if (val == 0) { /* Upvalue load */
IRType t = itype2irt(uvval(uvp));
@@ -1640,11 +1694,14 @@ static void rec_func_setup(jit_State *J)
static void rec_func_vararg(jit_State *J)
{
GCproto *pt = J->pt;
- BCReg s, fixargs, vframe = J->maxslot+1;
+ BCReg s, fixargs, vframe = J->maxslot+1+LJ_FR2;
lua_assert((pt->flags & PROTO_VARARG));
if (J->baseslot + vframe + pt->framesize >= LJ_MAX_JSLOTS)
lj_trace_err(J, LJ_TRERR_STACKOV);
- J->base[vframe-1] = J->base[-1]; /* Copy function up. */
+ J->base[vframe-1-LJ_FR2] = J->base[-1-LJ_FR2]; /* Copy function up. */
+#if LJ_FR2
+ J->base[vframe-1] = TREF_FRAME;
+#endif
/* Copy fixarg slots up and set their original slots to nil. */
fixargs = pt->numparams < J->maxslot ? pt->numparams : J->maxslot;
for (s = 0; s < fixargs; s++) {
@@ -1693,8 +1750,11 @@ static int select_detect(jit_State *J)
BCIns ins = J->pc[1];
if (bc_op(ins) == BC_CALLM && bc_b(ins) == 2 && bc_c(ins) == 1) {
cTValue *func = &J->L->base[bc_a(ins)];
- if (tvisfunc(func) && funcV(func)->c.ffid == FF_select)
+ if (tvisfunc(func) && funcV(func)->c.ffid == FF_select) {
+ TRef kfunc = lj_ir_kfunc(J, funcV(func));
+ emitir(IRTG(IR_EQ, IRT_FUNC), getslot(J, bc_a(ins)), kfunc);
return 1;
+ }
}
return 0;
}
@@ -1703,8 +1763,10 @@ static int select_detect(jit_State *J)
static void rec_varg(jit_State *J, BCReg dst, ptrdiff_t nresults)
{
int32_t numparams = J->pt->numparams;
- ptrdiff_t nvararg = frame_delta(J->L->base-1) - numparams - 1;
+ ptrdiff_t nvararg = frame_delta(J->L->base-1) - numparams - 1 - LJ_FR2;
lua_assert(frame_isvarg(J->L->base-1));
+ if (LJ_FR2 && dst > J->maxslot)
+ J->base[dst-1] = 0; /* Prevent resurrection of unrelated slot. */
if (J->framedepth > 0) { /* Simple case: varargs defined on-trace. */
ptrdiff_t i;
if (nvararg < 0) nvararg = 0;
@@ -1715,10 +1777,10 @@ static void rec_varg(jit_State *J, BCReg dst, ptrdiff_t nresults)
J->maxslot = dst + (BCReg)nresults;
}
for (i = 0; i < nresults; i++)
- J->base[dst+i] = i < nvararg ? getslot(J, i - nvararg - 1) : TREF_NIL;
+ J->base[dst+i] = i < nvararg ? getslot(J, i - nvararg - 1 - LJ_FR2) : TREF_NIL;
} else { /* Unknown number of varargs passed to trace. */
- TRef fr = emitir(IRTI(IR_SLOAD), 0, IRSLOAD_READONLY|IRSLOAD_FRAME);
- int32_t frofs = 8*(1+numparams)+FRAME_VARG;
+ TRef fr = emitir(IRTI(IR_SLOAD), LJ_FR2, IRSLOAD_READONLY|IRSLOAD_FRAME);
+ int32_t frofs = 8*(1+LJ_FR2+numparams)+FRAME_VARG;
if (nresults >= 0) { /* Known fixed number of results. */
ptrdiff_t i;
if (nvararg > 0) {
@@ -1729,11 +1791,11 @@ static void rec_varg(jit_State *J, BCReg dst, ptrdiff_t nresults)
else
emitir(IRTGI(IR_EQ), fr,
lj_ir_kint(J, (int32_t)frame_ftsz(J->L->base-1)));
- vbase = emitir(IRTI(IR_SUB), REF_BASE, fr);
- vbase = emitir(IRT(IR_ADD, IRT_P32), vbase, lj_ir_kint(J, frofs-8));
+ vbase = emitir(IRT(IR_SUB, IRT_IGC), REF_BASE, fr);
+ vbase = emitir(IRT(IR_ADD, IRT_PGC), vbase, lj_ir_kint(J, frofs-8));
for (i = 0; i < nload; i++) {
- IRType t = itype2irt(&J->L->base[i-1-nvararg]);
- TRef aref = emitir(IRT(IR_AREF, IRT_P32),
+ IRType t = itype2irt(&J->L->base[i-1-LJ_FR2-nvararg]);
+ TRef aref = emitir(IRT(IR_AREF, IRT_PGC),
vbase, lj_ir_kint(J, (int32_t)i));
TRef tr = emitir(IRTG(IR_VLOAD, t), aref, 0);
if (irtype_ispri(t)) tr = TREF_PRI(t); /* Canonicalize primitives. */
@@ -1779,15 +1841,16 @@ static void rec_varg(jit_State *J, BCReg dst, ptrdiff_t nresults)
}
if (idx != 0 && idx <= nvararg) {
IRType t;
- TRef aref, vbase = emitir(IRTI(IR_SUB), REF_BASE, fr);
- vbase = emitir(IRT(IR_ADD, IRT_P32), vbase, lj_ir_kint(J, frofs-8));
- t = itype2irt(&J->L->base[idx-2-nvararg]);
- aref = emitir(IRT(IR_AREF, IRT_P32), vbase, tridx);
+ TRef aref, vbase = emitir(IRT(IR_SUB, IRT_IGC), REF_BASE, fr);
+ vbase = emitir(IRT(IR_ADD, IRT_PGC), vbase,
+ lj_ir_kint(J, frofs-(8<<LJ_FR2)));
+ t = itype2irt(&J->L->base[idx-2-LJ_FR2-nvararg]);
+ aref = emitir(IRT(IR_AREF, IRT_PGC), vbase, tridx);
tr = emitir(IRTG(IR_VLOAD, t), aref, 0);
if (irtype_ispri(t)) tr = TREF_PRI(t); /* Canonicalize primitives. */
}
- J->base[dst-2] = tr;
- J->maxslot = dst-1;
+ J->base[dst-2-LJ_FR2] = tr;
+ J->maxslot = dst-1-LJ_FR2;
J->bcskip = 2; /* Skip CALLM + select. */
} else {
nyivarg:
@@ -1836,10 +1899,10 @@ static TRef rec_cat(jit_State *J, BCReg baseslot, BCReg topslot)
break;
}
xbase = ++trp;
- tr = hdr = emitir(IRT(IR_BUFHDR, IRT_P32),
+ tr = hdr = emitir(IRT(IR_BUFHDR, IRT_PGC),
lj_ir_kptr(J, &J2G(J)->tmpbuf), IRBUFHDR_RESET);
do {
- tr = emitir(IRT(IR_BUFPUT, IRT_P32), tr, *trp++);
+ tr = emitir(IRT(IR_BUFPUT, IRT_PGC), tr, *trp++);
} while (trp <= top);
tr = emitir(IRT(IR_BUFSTR, IRT_STR), tr, hdr);
J->maxslot = (BCReg)(xbase - J->base);
@@ -1880,7 +1943,15 @@ static void rec_comp_fixup(jit_State *J, const BCIns *pc, int cond)
const BCIns *npc = pc + 2 + (cond ? bc_j(jmpins) : 0);
SnapShot *snap = &J->cur.snap[J->cur.nsnap-1];
/* Set PC to opposite target to avoid re-recording the comp. in side trace. */
+#if LJ_FR2
+ SnapEntry *flink = &J->cur.snapmap[snap->mapofs + snap->nent];
+ uint64_t pcbase;
+ memcpy(&pcbase, flink, sizeof(uint64_t));
+ pcbase = (pcbase & 0xff) | (u64ptr(npc) << 8);
+ memcpy(flink, &pcbase, sizeof(uint64_t));
+#else
J->cur.snapmap[snap->mapofs + snap->nent] = SNAP_MKPC(npc);
+#endif
J->needsnap = 1;
if (bc_a(jmpins) < J->maxslot) J->maxslot = bc_a(jmpins);
lj_snap_shrink(J); /* Shrink last snapshot if possible. */
@@ -2156,14 +2227,14 @@ void lj_record_ins(jit_State *J)
case BC_MODVN: case BC_MODVV:
recmod:
if (tref_isnumber_str(rb) && tref_isnumber_str(rc))
- rc = lj_opt_narrow_mod(J, rb, rc, rcv);
+ rc = lj_opt_narrow_mod(J, rb, rc, rbv, rcv);
else
rc = rec_mm_arith(J, &ix, MM_mod);
break;
case BC_POW:
if (tref_isnumber_str(rb) && tref_isnumber_str(rc))
- rc = lj_opt_narrow_pow(J, lj_ir_tonum(J, rb), rc, rcv);
+ rc = lj_opt_narrow_pow(J, rb, rc, rbv, rcv);
else
rc = rec_mm_arith(J, &ix, MM_pow);
break;
@@ -2178,7 +2249,13 @@ void lj_record_ins(jit_State *J)
case BC_MOV:
/* Clear gap of method call to avoid resurrecting previous refs. */
- if (ra > J->maxslot) J->base[ra-1] = 0;
+ if (ra > J->maxslot) {
+#if LJ_FR2
+ memset(J->base + J->maxslot, 0, (ra - J->maxslot) * sizeof(TRef));
+#else
+ J->base[ra-1] = 0;
+#endif
+ }
break;
case BC_KSTR: case BC_KNUM: case BC_KPRI:
break;
@@ -2186,6 +2263,8 @@ void lj_record_ins(jit_State *J)
rc = lj_ir_kint(J, (int32_t)(int16_t)rc);
break;
case BC_KNIL:
+ if (LJ_FR2 && ra > J->maxslot)
+ J->base[ra-1] = 0;
while (ra <= rc)
J->base[ra++] = TREF_NIL;
if (rc >= J->maxslot) J->maxslot = rc+1;
@@ -2247,14 +2326,14 @@ void lj_record_ins(jit_State *J)
/* -- Calls and vararg handling ----------------------------------------- */
case BC_ITERC:
- J->base[ra] = getslot(J, ra-3-LJ_FR2);
- J->base[ra+1] = getslot(J, ra-2-LJ_FR2);
- J->base[ra+2] = getslot(J, ra-1-LJ_FR2);
+ J->base[ra] = getslot(J, ra-3);
+ J->base[ra+1+LJ_FR2] = getslot(J, ra-2);
+ J->base[ra+2+LJ_FR2] = getslot(J, ra-1);
{ /* Do the actual copy now because lj_record_call needs the values. */
TValue *b = &J->L->base[ra];
- copyTV(J->L, b, b-3-LJ_FR2);
- copyTV(J->L, b+1, b-2-LJ_FR2);
- copyTV(J->L, b+2, b-1-LJ_FR2);
+ copyTV(J->L, b, b-3);
+ copyTV(J->L, b+1+LJ_FR2, b-2);
+ copyTV(J->L, b+2+LJ_FR2, b-1);
}
lj_record_call(J, ra, (ptrdiff_t)rc-1);
break;
@@ -2377,7 +2456,12 @@ void lj_record_ins(jit_State *J)
/* rc == 0 if we have no result yet, e.g. pending __index metamethod call. */
if (bcmode_a(op) == BCMdst && rc) {
J->base[ra] = rc;
- if (ra >= J->maxslot) J->maxslot = ra+1;
+ if (ra >= J->maxslot) {
+#if LJ_FR2
+ if (ra > J->maxslot) J->base[ra-1] = 0;
+#endif
+ J->maxslot = ra+1;
+ }
}
#undef rav
@@ -2462,7 +2546,7 @@ void lj_record_setup(jit_State *J)
J->scev.idx = REF_NIL;
setmref(J->scev.pc, NULL);
- J->baseslot = 1; /* Invoking function is at base[-1]. */
+ J->baseslot = 1+LJ_FR2; /* Invoking function is at base[-1-LJ_FR2]. */
J->base = J->slot + J->baseslot;
J->maxslot = 0;
J->framedepth = 0;
@@ -2477,7 +2561,7 @@ void lj_record_setup(jit_State *J)
J->bc_extent = ~(MSize)0;
/* Emit instructions for fixed references. Also triggers initial IR alloc. */
- emitir_raw(IRT(IR_BASE, IRT_P32), J->parent, J->exitno);
+ emitir_raw(IRT(IR_BASE, IRT_PGC), J->parent, J->exitno);
for (i = 0; i <= 2; i++) {
IRIns *ir = IR(REF_NIL-i);
ir->i = 0;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_record.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_record.h
index cb184e8f6b9..93d374d2494 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_record.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_record.h
@@ -1,6 +1,6 @@
/*
** Trace recorder (bytecode -> SSA IR).
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_RECORD_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_snap.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_snap.c
index 8638d9edbfd..bb063c2b652 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_snap.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_snap.c
@@ -1,6 +1,6 @@
/*
** Snapshot handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_snap_c
@@ -68,10 +68,22 @@ static MSize snapshot_slots(jit_State *J, SnapEntry *map, BCReg nslots)
for (s = 0; s < nslots; s++) {
TRef tr = J->slot[s];
IRRef ref = tref_ref(tr);
+#if LJ_FR2
+ if (s == 1) { /* Ignore slot 1 in LJ_FR2 mode, except if tailcalled. */
+ if ((tr & TREF_FRAME))
+ map[n++] = SNAP(1, SNAP_FRAME | SNAP_NORESTORE, REF_NIL);
+ continue;
+ }
+ if ((tr & (TREF_FRAME | TREF_CONT)) && !ref) {
+ cTValue *base = J->L->base - J->baseslot;
+ tr = J->slot[s] = (tr & 0xff0000) | lj_ir_k64(J, IR_KNUM, base[s].u64);
+ ref = tref_ref(tr);
+ }
+#endif
if (ref) {
SnapEntry sn = SNAP_TR(s, tr);
IRIns *ir = &J->cur.ir[ref];
- if (!(sn & (SNAP_CONT|SNAP_FRAME)) &&
+ if ((LJ_FR2 || !(sn & (SNAP_CONT|SNAP_FRAME))) &&
ir->o == IR_SLOAD && ir->op1 == s && ref > retf) {
/* No need to snapshot unmodified non-inherited slots. */
if (!(ir->op2 & IRSLOAD_INHERIT))
@@ -90,34 +102,51 @@ static MSize snapshot_slots(jit_State *J, SnapEntry *map, BCReg nslots)
}
/* Add frame links at the end of the snapshot. */
-static BCReg snapshot_framelinks(jit_State *J, SnapEntry *map)
+static MSize snapshot_framelinks(jit_State *J, SnapEntry *map, uint8_t *topslot)
{
cTValue *frame = J->L->base - 1;
- cTValue *lim = J->L->base - J->baseslot;
+ cTValue *lim = J->L->base - J->baseslot + LJ_FR2;
GCfunc *fn = frame_func(frame);
cTValue *ftop = isluafunc(fn) ? (frame+funcproto(fn)->framesize) : J->L->top;
+#if LJ_FR2
+ uint64_t pcbase = (u64ptr(J->pc) << 8) | (J->baseslot - 2);
+ lua_assert(2 <= J->baseslot && J->baseslot <= 257);
+ memcpy(map, &pcbase, sizeof(uint64_t));
+#else
MSize f = 0;
- lua_assert(!LJ_FR2); /* TODO_FR2: store 64 bit PCs. */
map[f++] = SNAP_MKPC(J->pc); /* The current PC is always the first entry. */
+#endif
while (frame > lim) { /* Backwards traversal of all frames above base. */
if (frame_islua(frame)) {
+#if !LJ_FR2
map[f++] = SNAP_MKPC(frame_pc(frame));
+#endif
frame = frame_prevl(frame);
} else if (frame_iscont(frame)) {
+#if !LJ_FR2
map[f++] = SNAP_MKFTSZ(frame_ftsz(frame));
map[f++] = SNAP_MKPC(frame_contpc(frame));
+#endif
frame = frame_prevd(frame);
} else {
lua_assert(!frame_isc(frame));
+#if !LJ_FR2
map[f++] = SNAP_MKFTSZ(frame_ftsz(frame));
+#endif
frame = frame_prevd(frame);
continue;
}
if (frame + funcproto(frame_func(frame))->framesize > ftop)
ftop = frame + funcproto(frame_func(frame))->framesize;
}
+ *topslot = (uint8_t)(ftop - lim);
+#if LJ_FR2
+ lua_assert(sizeof(SnapEntry) * 2 == sizeof(uint64_t));
+ return 2;
+#else
lua_assert(f == (MSize)(1 + J->framedepth));
- return (BCReg)(ftop - lim);
+ return f;
+#endif
}
/* Take a snapshot of the current stack. */
@@ -127,16 +156,16 @@ static void snapshot_stack(jit_State *J, SnapShot *snap, MSize nsnapmap)
MSize nent;
SnapEntry *p;
/* Conservative estimate. */
- lj_snap_grow_map(J, nsnapmap + nslots + (MSize)J->framedepth+1);
+ lj_snap_grow_map(J, nsnapmap + nslots + (MSize)(LJ_FR2?2:J->framedepth+1));
p = &J->cur.snapmap[nsnapmap];
nent = snapshot_slots(J, p, nslots);
- snap->topslot = (uint8_t)snapshot_framelinks(J, p + nent);
+ snap->nent = (uint8_t)nent;
+ nent += snapshot_framelinks(J, p + nent, &snap->topslot);
snap->mapofs = (uint16_t)nsnapmap;
snap->ref = (IRRef1)J->cur.nins;
- snap->nent = (uint8_t)nent;
snap->nslots = (uint8_t)nslots;
snap->count = 0;
- J->cur.nsnapmap = (uint16_t)(nsnapmap + nent + 1 + J->framedepth);
+ J->cur.nsnapmap = (uint16_t)(nsnapmap + nent);
}
/* Add or merge a snapshot. */
@@ -284,8 +313,8 @@ void lj_snap_shrink(jit_State *J)
MSize n, m, nlim, nent = snap->nent;
uint8_t udf[SNAP_USEDEF_SLOTS];
BCReg maxslot = J->maxslot;
- BCReg minslot = snap_usedef(J, udf, snap_pc(map[nent]), maxslot);
BCReg baseslot = J->baseslot;
+ BCReg minslot = snap_usedef(J, udf, snap_pc(&map[nent]), maxslot);
maxslot += baseslot;
minslot += baseslot;
snap->nslots = (uint8_t)maxslot;
@@ -371,8 +400,8 @@ static TRef snap_replay_const(jit_State *J, IRIns *ir)
case IR_KPRI: return TREF_PRI(irt_type(ir->t));
case IR_KINT: return lj_ir_kint(J, ir->i);
case IR_KGC: return lj_ir_kgc(J, ir_kgc(ir), irt_t(ir->t));
- case IR_KNUM: return lj_ir_k64(J, IR_KNUM, ir_knum(ir));
- case IR_KINT64: return lj_ir_k64(J, IR_KINT64, ir_kint64(ir));
+ case IR_KNUM: case IR_KINT64:
+ return lj_ir_k64(J, (IROp)ir->o, ir_k64(ir)->u64);
case IR_KPTR: return lj_ir_kptr(J, ir_kptr(ir)); /* Continuation. */
default: lua_assert(0); return TREF_NIL; break;
}
@@ -445,7 +474,11 @@ void lj_snap_replay(jit_State *J, GCtrace *T)
goto setslot;
bloomset(seen, ref);
if (irref_isk(ref)) {
- tr = snap_replay_const(J, ir);
+ /* See special treatment of LJ_FR2 slot 1 in snapshot_slots() above. */
+ if (LJ_FR2 && (sn == SNAP(1, SNAP_FRAME | SNAP_NORESTORE, REF_NIL)))
+ tr = 0;
+ else
+ tr = snap_replay_const(J, ir);
} else if (!regsp_used(ir->prev)) {
pass23 = 1;
lua_assert(s != 0);
@@ -459,7 +492,7 @@ void lj_snap_replay(jit_State *J, GCtrace *T)
}
setslot:
J->slot[s] = tr | (sn&(SNAP_CONT|SNAP_FRAME)); /* Same as TREF_* flags. */
- J->framedepth += ((sn & (SNAP_CONT|SNAP_FRAME)) && s);
+ J->framedepth += ((sn & (SNAP_CONT|SNAP_FRAME)) && (s != LJ_FR2));
if ((sn & SNAP_FRAME))
J->baseslot = s+1;
}
@@ -555,8 +588,7 @@ void lj_snap_replay(jit_State *J, GCtrace *T)
if (irref_isk(irs->op2) && irref_isk((irs+1)->op2)) {
uint64_t k = (uint32_t)T->ir[irs->op2].i +
((uint64_t)T->ir[(irs+1)->op2].i << 32);
- val = lj_ir_k64(J, t == IRT_I64 ? IR_KINT64 : IR_KNUM,
- lj_ir_k64_find(J, k));
+ val = lj_ir_k64(J, t == IRT_I64 ? IR_KINT64 : IR_KNUM, k);
} else {
val = emitir_raw(IRT(IR_HIOP, t), val,
snap_pref(J, T, map, nent, seen, (irs+1)->op2));
@@ -599,7 +631,6 @@ static void snap_restoreval(jit_State *J, GCtrace *T, ExitState *ex,
}
if (LJ_UNLIKELY(bloomtest(rfilt, ref)))
rs = snap_renameref(T, snapno, ref, rs);
- lua_assert(!LJ_GC64); /* TODO_GC64: handle 64 bit references. */
if (ra_hasspill(regsp_spill(rs))) { /* Restore from spill slot. */
int32_t *sps = &ex->spill[regsp_spill(rs)];
if (irt_isinteger(t)) {
@@ -608,9 +639,11 @@ static void snap_restoreval(jit_State *J, GCtrace *T, ExitState *ex,
} else if (irt_isnum(t)) {
o->u64 = *(uint64_t *)sps;
#endif
- } else if (LJ_64 && irt_islightud(t)) {
+#if LJ_64 && !LJ_GC64
+ } else if (irt_islightud(t)) {
/* 64 bit lightuserdata which may escape already has the tag bits. */
o->u64 = *(uint64_t *)sps;
+#endif
} else {
lua_assert(!irt_ispri(t)); /* PRI refs never have a spill slot. */
setgcV(J->L, o, (GCobj *)(uintptr_t)*(GCSize *)sps, irt_toitype(t));
@@ -628,9 +661,11 @@ static void snap_restoreval(jit_State *J, GCtrace *T, ExitState *ex,
} else if (irt_isnum(t)) {
setnumV(o, ex->fpr[r-RID_MIN_FPR]);
#endif
- } else if (LJ_64 && irt_is64(t)) {
+#if LJ_64 && !LJ_GC64
+ } else if (irt_is64(t)) {
/* 64 bit values that already have the tag bits. */
o->u64 = ex->gpr[r-RID_MIN_GPR];
+#endif
} else if (irt_ispri(t)) {
setpriV(o, irt_toitype(t));
} else {
@@ -651,7 +686,7 @@ static void snap_restoredata(GCtrace *T, ExitState *ex,
uint64_t tmp;
if (irref_isk(ref)) {
if (ir->o == IR_KNUM || ir->o == IR_KINT64) {
- src = mref(ir->ptr, int32_t);
+ src = (int32_t *)&ir[1];
} else if (sz == 8) {
tmp = (uint64_t)(uint32_t)ir->i;
src = (int32_t *)&tmp;
@@ -688,8 +723,9 @@ static void snap_restoredata(GCtrace *T, ExitState *ex,
#else
if (LJ_BE && sz == 4) src++;
#endif
- }
+ } else
#endif
+ if (LJ_64 && LJ_BE && sz == 4) src++;
}
}
lua_assert(sz == 1 || sz == 2 || sz == 4 || sz == 8);
@@ -795,11 +831,15 @@ const BCIns *lj_snap_restore(jit_State *J, void *exptr)
SnapShot *snap = &T->snap[snapno];
MSize n, nent = snap->nent;
SnapEntry *map = &T->snapmap[snap->mapofs];
- SnapEntry *flinks = &T->snapmap[snap_nextofs(T, snap)-1];
+#if !LJ_FR2 || defined(LUA_USE_ASSERT)
+ SnapEntry *flinks = &T->snapmap[snap_nextofs(T, snap)-1-LJ_FR2];
+#endif
+#if !LJ_FR2
ptrdiff_t ftsz0;
+#endif
TValue *frame;
BloomFilter rfilt = snap_renamefilter(T, snapno);
- const BCIns *pc = snap_pc(map[nent]);
+ const BCIns *pc = snap_pc(&map[nent]);
lua_State *L = J->L;
/* Set interpreter PC to the next PC to get correct error messages. */
@@ -812,8 +852,10 @@ const BCIns *lj_snap_restore(jit_State *J, void *exptr)
}
/* Fill stack slots with data from the registers and spill slots. */
- frame = L->base-1;
+ frame = L->base-1-LJ_FR2;
+#if !LJ_FR2
ftsz0 = frame_ftsz(frame); /* Preserve link to previous frame in slot #0. */
+#endif
for (n = 0; n < nent; n++) {
SnapEntry sn = map[n];
if (!(sn & SNAP_NORESTORE)) {
@@ -836,14 +878,18 @@ const BCIns *lj_snap_restore(jit_State *J, void *exptr)
TValue tmp;
snap_restoreval(J, T, ex, snapno, rfilt, ref+1, &tmp);
o->u32.hi = tmp.u32.lo;
+#if !LJ_FR2
} else if ((sn & (SNAP_CONT|SNAP_FRAME))) {
- lua_assert(!LJ_FR2); /* TODO_FR2: store 64 bit PCs. */
/* Overwrite tag with frame link. */
setframe_ftsz(o, snap_slot(sn) != 0 ? (int32_t)*flinks-- : ftsz0);
L->base = o+1;
+#endif
}
}
}
+#if LJ_FR2
+ L->base += (map[nent+LJ_BE] & 0xff);
+#endif
lua_assert(map + nent == flinks);
/* Compute current stack top. */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_snap.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_snap.h
index 1d3379bc6f4..2c9ae3d643a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_snap.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_snap.h
@@ -1,6 +1,6 @@
/*
** Snapshot handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_SNAP_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_state.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_state.c
index 66bf439f822..632dd07e5f9 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_state.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_state.c
@@ -1,6 +1,6 @@
/*
** State and stack handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -180,7 +180,7 @@ static void close_state(lua_State *L)
g->allocf(g->allocd, G2GG(g), sizeof(GG_State), 0);
}
-#if LJ_64 && !(defined(LUAJIT_USE_VALGRIND) && defined(LUAJIT_USE_SYSMALLOC))
+#if LJ_64 && !LJ_GC64 && !(defined(LUAJIT_USE_VALGRIND) && defined(LUAJIT_USE_SYSMALLOC))
lua_State *lj_state_newstate(lua_Alloc f, void *ud)
#else
LUA_API lua_State *lua_newstate(lua_Alloc f, void *ud)
@@ -224,7 +224,7 @@ LUA_API lua_State *lua_newstate(lua_Alloc f, void *ud)
close_state(L);
return NULL;
}
- L->status = 0;
+ L->status = LUA_OK;
return L;
}
@@ -256,10 +256,10 @@ LUA_API void lua_close(lua_State *L)
#endif
for (i = 0;;) {
hook_enter(g);
- L->status = 0;
+ L->status = LUA_OK;
L->base = L->top = tvref(L->stack) + 1 + LJ_FR2;
L->cframe = NULL;
- if (lj_vm_cpcall(L, NULL, NULL, cpfinalize) == 0) {
+ if (lj_vm_cpcall(L, NULL, NULL, cpfinalize) == LUA_OK) {
if (++i >= 10) break;
lj_gc_separateudata(g, 1); /* Separate udata again. */
if (gcref(g->gc.mmudata) == NULL) /* Until nothing is left to do. */
@@ -274,7 +274,7 @@ lua_State *lj_state_new(lua_State *L)
lua_State *L1 = lj_mem_newobj(L, lua_State);
L1->gct = ~LJ_TTHREAD;
L1->dummy_ffid = FF_C;
- L1->status = 0;
+ L1->status = LUA_OK;
L1->stacksize = 0;
setmref(L1->stack, NULL);
L1->cframe = NULL;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_state.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_state.h
index e128d3211b3..02a0eafa332 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_state.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_state.h
@@ -1,6 +1,6 @@
/*
** State and stack handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_STATE_H
@@ -28,7 +28,7 @@ static LJ_AINLINE void lj_state_checkstack(lua_State *L, MSize need)
LJ_FUNC lua_State *lj_state_new(lua_State *L);
LJ_FUNC void LJ_FASTCALL lj_state_free(global_State *g, lua_State *L);
-#if LJ_64
+#if LJ_64 && !LJ_GC64 && !(defined(LUAJIT_USE_VALGRIND) && defined(LUAJIT_USE_SYSMALLOC))
LJ_FUNC lua_State *lj_state_newstate(lua_Alloc f, void *ud);
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_str.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_str.c
index 086072c3807..ec11ab00c9e 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_str.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_str.c
@@ -1,6 +1,6 @@
/*
** String handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_str_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_str.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_str.h
index bb306c142ba..85c1e405bd7 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_str.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_str.h
@@ -1,6 +1,6 @@
/*
** String handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_STR_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.c
index 7c7d81d360f..d7893ce981b 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.c
@@ -1,6 +1,6 @@
/*
** String formatting.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include <stdio.h>
@@ -98,11 +98,15 @@ char * LJ_FASTCALL lj_strfmt_wint(char *p, int32_t k)
uint32_t u = (uint32_t)k;
if (k < 0) { u = (uint32_t)-k; *p++ = '-'; }
if (u < 10000) {
- if (u < 10) goto dig1; if (u < 100) goto dig2; if (u < 1000) goto dig3;
+ if (u < 10) goto dig1;
+ if (u < 100) goto dig2;
+ if (u < 1000) goto dig3;
} else {
uint32_t v = u / 10000; u -= v * 10000;
if (v < 10000) {
- if (v < 10) goto dig5; if (v < 100) goto dig6; if (v < 1000) goto dig7;
+ if (v < 10) goto dig5;
+ if (v < 100) goto dig6;
+ if (v < 1000) goto dig7;
} else {
uint32_t w = v / 10000; v -= w * 10000;
if (w >= 10) WINT_R(w, 10, 10)
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.h
index a38bc7c38e1..6e1d9017e62 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt.h
@@ -1,6 +1,6 @@
/*
** String formatting.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_STRFMT_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt_num.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt_num.c
index ab84fda76d2..9271f68a117 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt_num.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_strfmt_num.c
@@ -1,6 +1,6 @@
/*
** String formatting for floating-point numbers.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
** Contributed by Peter Cawley.
*/
@@ -138,7 +138,7 @@ static uint32_t nd_mul2k(uint32_t* nd, uint32_t ndhi, uint32_t k,
}
if (carry_in) {
nd[++ndhi] = carry_in; carry_in = 0;
- if(start++ == ndlo) ++ndlo;
+ if (start++ == ndlo) ++ndlo;
}
k -= ND_MUL2K_MAX_SHIFT;
}
@@ -364,6 +364,7 @@ static char *lj_strfmt_wfnum(SBuf *sb, SFormat sf, lua_Number n, char *p)
/* Precision is sufficiently low that rescaling will probably work. */
if ((ndebias = rescale_e[e >> 6])) {
t.n = n * rescale_n[e >> 6];
+ if (LJ_UNLIKELY(!e)) t.n *= 1e10, ndebias -= 10;
t.u64 -= 2; /* Convert 2ulp below (later we convert 2ulp above). */
nd[0] = 0x100000 | (t.u32.hi & 0xfffff);
e = ((t.u32.hi >> 20) & 0x7ff) - 1075 - (ND_MUL2K_MAX_SHIFT < 29);
@@ -479,7 +480,7 @@ static char *lj_strfmt_wfnum(SBuf *sb, SFormat sf, lua_Number n, char *p)
if ((prec | (sf & STRFMT_F_ALT))) {
/* Emit fractional part. */
p[1] = '.'; p += 2;
- prec -= (q - p); p = q; /* Account for the digits already emitted. */
+ prec -= (MSize)(q - p); p = q; /* Account for digits already emitted. */
/* Then emit chunks of 9 digits (this may emit 8 digits too many). */
for (i = ndhi; (int32_t)prec > 0 && i != ndlo; prec -= 9) {
i = (i - 1) & 0x3f;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.c
index 7c5171328e8..f5f35c96028 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.c
@@ -1,6 +1,6 @@
/*
** String scanning.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include <math.h>
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.h
index b3ee9e528df..6fb0dda08cd 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_strscan.h
@@ -1,6 +1,6 @@
/*
** String scanning.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_STRSCAN_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_tab.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_tab.c
index 8011212f1b1..47c0cfd34af 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_tab.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_tab.c
@@ -1,6 +1,6 @@
/*
** Table handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -28,7 +28,6 @@ static LJ_AINLINE Node *hashmask(const GCtab *t, uint32_t hash)
#define hashlohi(t, lo, hi) hashmask((t), hashrot((lo), (hi)))
#define hashnum(t, o) hashlohi((t), (o)->u32.lo, ((o)->u32.hi << 1))
-#define hashptr(t, p) hashlohi((t), u32ptr(p), u32ptr(p) + HASH_BIAS)
#if LJ_GC64
#define hashgcref(t, r) \
hashlohi((t), (uint32_t)gcrefu(r), (uint32_t)(gcrefu(r) >> 32))
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_tab.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_tab.h
index f06fcf611ad..71e34945e8a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_tab.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_tab.h
@@ -1,6 +1,6 @@
/*
** Table handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_TAB_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_target.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_target.h
index abea8d5b216..8dcae957f0e 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_target.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_target.h
@@ -1,6 +1,6 @@
/*
** Definitions for target CPU.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_TARGET_H
@@ -55,7 +55,7 @@ typedef uint32_t RegSP;
/* Bitset for registers. 32 registers suffice for most architectures.
** Note that one set holds bits for both GPRs and FPRs.
*/
-#if LJ_TARGET_PPC || LJ_TARGET_MIPS
+#if LJ_TARGET_PPC || LJ_TARGET_MIPS || LJ_TARGET_ARM64
typedef uint64_t RegSet;
#else
typedef uint32_t RegSet;
@@ -69,7 +69,7 @@ typedef uint32_t RegSet;
#define rset_set(rs, r) (rs |= RID2RSET(r))
#define rset_clear(rs, r) (rs &= ~RID2RSET(r))
#define rset_exclude(rs, r) (rs & ~RID2RSET(r))
-#if LJ_TARGET_PPC || LJ_TARGET_MIPS
+#if LJ_TARGET_PPC || LJ_TARGET_MIPS || LJ_TARGET_ARM64
#define rset_picktop(rs) ((Reg)(__builtin_clzll(rs)^63))
#define rset_pickbot(rs) ((Reg)__builtin_ctzll(rs))
#else
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm.h
index 36959dbcc4e..5551b1f1cef 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm.h
@@ -1,6 +1,6 @@
/*
** Definitions for ARM CPUs.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_TARGET_ARM_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm64.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm64.h
index 57ab134f98e..520023ae218 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm64.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_arm64.h
@@ -1,6 +1,6 @@
/*
** Definitions for ARM64 CPUs.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_TARGET_ARM64_H
@@ -55,7 +55,8 @@ enum {
/* Make use of all registers, except for x18, fp, lr and sp. */
#define RSET_FIXED \
- (RID2RSET(RID_X18)|RID2RSET(RID_FP)|RID2RSET(RID_LR)|RID2RSET(RID_SP))
+ (RID2RSET(RID_X18)|RID2RSET(RID_FP)|RID2RSET(RID_LR)|RID2RSET(RID_SP)|\
+ RID2RSET(RID_GL))
#define RSET_GPR (RSET_RANGE(RID_MIN_GPR, RID_MAX_GPR) - RSET_FIXED)
#define RSET_FPR RSET_RANGE(RID_MIN_FPR, RID_MAX_FPR)
#define RSET_ALL (RSET_GPR|RSET_FPR)
@@ -73,25 +74,256 @@ enum {
#define REGARG_LASTFPR RID_D7
#define REGARG_NUMFPR 8
+/* -- Spill slots --------------------------------------------------------- */
+
+/* Spill slots are 32 bit wide. An even/odd pair is used for FPRs.
+**
+** SPS_FIXED: Available fixed spill slots in interpreter frame.
+** This definition must match with the vm_arm64.dasc file.
+** Pre-allocate some slots to avoid sp adjust in every root trace.
+**
+** SPS_FIRST: First spill slot for general use. Reserve min. two 32 bit slots.
+*/
+#define SPS_FIXED 4
+#define SPS_FIRST 2
+
+#define SPOFS_TMP 0
+
+#define sps_scale(slot) (4 * (int32_t)(slot))
+#define sps_align(slot) (((slot) - SPS_FIXED + 3) & ~3)
+
+/* -- Exit state ---------------------------------------------------------- */
+
+/* This definition must match with the *.dasc file(s). */
+typedef struct {
+ lua_Number fpr[RID_NUM_FPR]; /* Floating-point registers. */
+ intptr_t gpr[RID_NUM_GPR]; /* General-purpose registers. */
+ int32_t spill[256]; /* Spill slots. */
+} ExitState;
+
+/* Highest exit + 1 indicates stack check. */
+#define EXITSTATE_CHECKEXIT 1
+
+/* Return the address of a per-trace exit stub. */
+static LJ_AINLINE uint32_t *exitstub_trace_addr_(uint32_t *p, uint32_t exitno)
+{
+ while (*p == (LJ_LE ? 0xd503201f : 0x1f2003d5)) p++; /* Skip A64I_NOP. */
+ return p + 3 + exitno;
+}
+/* Avoid dependence on lj_jit.h if only including lj_target.h. */
+#define exitstub_trace_addr(T, exitno) \
+ exitstub_trace_addr_((MCode *)((char *)(T)->mcode + (T)->szmcode), (exitno))
+
/* -- Instructions -------------------------------------------------------- */
+/* ARM64 instructions are always little-endian. Swap for ARM64BE. */
+#if LJ_BE
+#define A64I_LE(x) (lj_bswap(x))
+#else
+#define A64I_LE(x) (x)
+#endif
+
/* Instruction fields. */
#define A64F_D(r) (r)
-#define A64F_N(r) ((r) << 5)
-#define A64F_A(r) ((r) << 10)
-#define A64F_M(r) ((r) << 16)
+#define A64F_N(r) ((r) << 5)
+#define A64F_A(r) ((r) << 10)
+#define A64F_M(r) ((r) << 16)
+#define A64F_IMMS(x) ((x) << 10)
+#define A64F_IMMR(x) ((x) << 16)
#define A64F_U16(x) ((x) << 5)
+#define A64F_U12(x) ((x) << 10)
#define A64F_S26(x) (x)
-#define A64F_S19(x) ((x) << 5)
+#define A64F_S19(x) (((uint32_t)(x) & 0x7ffffu) << 5)
+#define A64F_S14(x) ((x) << 5)
+#define A64F_S9(x) ((x) << 12)
+#define A64F_BIT(x) ((x) << 19)
+#define A64F_SH(sh, x) (((sh) << 22) | ((x) << 10))
+#define A64F_EX(ex) (A64I_EX | ((ex) << 13))
+#define A64F_EXSH(ex,x) (A64I_EX | ((ex) << 13) | ((x) << 10))
+#define A64F_FP8(x) ((x) << 13)
+#define A64F_CC(cc) ((cc) << 12)
+#define A64F_LSL16(x) (((x) / 16) << 21)
+#define A64F_BSH(sh) ((sh) << 10)
typedef enum A64Ins {
+ A64I_S = 0x20000000,
+ A64I_X = 0x80000000,
+ A64I_EX = 0x00200000,
+ A64I_ON = 0x00200000,
+ A64I_K12 = 0x1a000000,
+ A64I_K13 = 0x18000000,
+ A64I_LS_U = 0x01000000,
+ A64I_LS_S = 0x00800000,
+ A64I_LS_R = 0x01200800,
+ A64I_LS_SH = 0x00001000,
+ A64I_LS_UXTWx = 0x00004000,
+ A64I_LS_SXTWx = 0x0000c000,
+ A64I_LS_SXTXx = 0x0000e000,
+ A64I_LS_LSLx = 0x00006000,
+
+ A64I_ADDw = 0x0b000000,
+ A64I_ADDx = 0x8b000000,
+ A64I_ADDSw = 0x2b000000,
+ A64I_ADDSx = 0xab000000,
+ A64I_NEGw = 0x4b0003e0,
+ A64I_NEGx = 0xcb0003e0,
+ A64I_SUBw = 0x4b000000,
+ A64I_SUBx = 0xcb000000,
+ A64I_SUBSw = 0x6b000000,
+ A64I_SUBSx = 0xeb000000,
+
+ A64I_MULw = 0x1b007c00,
+ A64I_MULx = 0x9b007c00,
+ A64I_SMULL = 0x9b207c00,
+
+ A64I_ANDw = 0x0a000000,
+ A64I_ANDx = 0x8a000000,
+ A64I_ANDSw = 0x6a000000,
+ A64I_ANDSx = 0xea000000,
+ A64I_EORw = 0x4a000000,
+ A64I_EORx = 0xca000000,
+ A64I_ORRw = 0x2a000000,
+ A64I_ORRx = 0xaa000000,
+ A64I_TSTw = 0x6a00001f,
+ A64I_TSTx = 0xea00001f,
+
+ A64I_CMPw = 0x6b00001f,
+ A64I_CMPx = 0xeb00001f,
+ A64I_CMNw = 0x2b00001f,
+ A64I_CMNx = 0xab00001f,
+ A64I_CCMPw = 0x7a400000,
+ A64I_CCMPx = 0xfa400000,
+ A64I_CSELw = 0x1a800000,
+ A64I_CSELx = 0x9a800000,
+
+ A64I_ASRw = 0x13007c00,
+ A64I_ASRx = 0x9340fc00,
+ A64I_LSLx = 0xd3400000,
+ A64I_LSRx = 0xd340fc00,
+ A64I_SHRw = 0x1ac02000,
+ A64I_SHRx = 0x9ac02000, /* lsl/lsr/asr/ror x0, x0, x0 */
+ A64I_REVw = 0x5ac00800,
+ A64I_REVx = 0xdac00c00,
+
+ A64I_EXTRw = 0x13800000,
+ A64I_EXTRx = 0x93c00000,
+ A64I_SBFMw = 0x13000000,
+ A64I_SBFMx = 0x93400000,
+ A64I_SXTBw = 0x13001c00,
+ A64I_SXTHw = 0x13003c00,
+ A64I_SXTW = 0x93407c00,
+ A64I_UBFMw = 0x53000000,
+ A64I_UBFMx = 0xd3400000,
+ A64I_UXTBw = 0x53001c00,
+ A64I_UXTHw = 0x53003c00,
+
+ A64I_MOVw = 0x2a0003e0,
+ A64I_MOVx = 0xaa0003e0,
+ A64I_MVNw = 0x2a2003e0,
+ A64I_MVNx = 0xaa2003e0,
+ A64I_MOVKw = 0x72800000,
+ A64I_MOVKx = 0xf2800000,
A64I_MOVZw = 0x52800000,
A64I_MOVZx = 0xd2800000,
+ A64I_MOVNw = 0x12800000,
+ A64I_MOVNx = 0x92800000,
+
+ A64I_LDRB = 0x39400000,
+ A64I_LDRH = 0x79400000,
+ A64I_LDRw = 0xb9400000,
+ A64I_LDRx = 0xf9400000,
A64I_LDRLw = 0x18000000,
A64I_LDRLx = 0x58000000,
- A64I_NOP = 0xd503201f,
+ A64I_STRB = 0x39000000,
+ A64I_STRH = 0x79000000,
+ A64I_STRw = 0xb9000000,
+ A64I_STRx = 0xf9000000,
+ A64I_STPw = 0x29000000,
+ A64I_STPx = 0xa9000000,
+ A64I_LDPw = 0x29400000,
+ A64I_LDPx = 0xa9400000,
+
A64I_B = 0x14000000,
+ A64I_BCC = 0x54000000,
+ A64I_BL = 0x94000000,
A64I_BR = 0xd61f0000,
+ A64I_BLR = 0xd63f0000,
+ A64I_TBZ = 0x36000000,
+ A64I_TBNZ = 0x37000000,
+ A64I_CBZ = 0x34000000,
+ A64I_CBNZ = 0x35000000,
+
+ A64I_NOP = 0xd503201f,
+
+ /* FP */
+ A64I_FADDd = 0x1e602800,
+ A64I_FSUBd = 0x1e603800,
+ A64I_FMADDd = 0x1f400000,
+ A64I_FMSUBd = 0x1f408000,
+ A64I_FNMADDd = 0x1f600000,
+ A64I_FNMSUBd = 0x1f608000,
+ A64I_FMULd = 0x1e600800,
+ A64I_FDIVd = 0x1e601800,
+ A64I_FNEGd = 0x1e614000,
+ A64I_FABS = 0x1e60c000,
+ A64I_FSQRTd = 0x1e61c000,
+ A64I_LDRs = 0xbd400000,
+ A64I_LDRd = 0xfd400000,
+ A64I_STRs = 0xbd000000,
+ A64I_STRd = 0xfd000000,
+ A64I_LDPs = 0x2d400000,
+ A64I_LDPd = 0x6d400000,
+ A64I_STPs = 0x2d000000,
+ A64I_STPd = 0x6d000000,
+ A64I_FCMPd = 0x1e602000,
+ A64I_FCMPZd = 0x1e602008,
+ A64I_FCSELd = 0x1e600c00,
+ A64I_FRINTMd = 0x1e654000,
+ A64I_FRINTPd = 0x1e64c000,
+ A64I_FRINTZd = 0x1e65c000,
+
+ A64I_FCVT_F32_F64 = 0x1e624000,
+ A64I_FCVT_F64_F32 = 0x1e22c000,
+ A64I_FCVT_F32_S32 = 0x1e220000,
+ A64I_FCVT_F64_S32 = 0x1e620000,
+ A64I_FCVT_F32_U32 = 0x1e230000,
+ A64I_FCVT_F64_U32 = 0x1e630000,
+ A64I_FCVT_F32_S64 = 0x9e220000,
+ A64I_FCVT_F64_S64 = 0x9e620000,
+ A64I_FCVT_F32_U64 = 0x9e230000,
+ A64I_FCVT_F64_U64 = 0x9e630000,
+ A64I_FCVT_S32_F64 = 0x1e780000,
+ A64I_FCVT_S32_F32 = 0x1e380000,
+ A64I_FCVT_U32_F64 = 0x1e790000,
+ A64I_FCVT_U32_F32 = 0x1e390000,
+ A64I_FCVT_S64_F64 = 0x9e780000,
+ A64I_FCVT_S64_F32 = 0x9e380000,
+ A64I_FCVT_U64_F64 = 0x9e790000,
+ A64I_FCVT_U64_F32 = 0x9e390000,
+
+ A64I_FMOV_S = 0x1e204000,
+ A64I_FMOV_D = 0x1e604000,
+ A64I_FMOV_R_S = 0x1e260000,
+ A64I_FMOV_S_R = 0x1e270000,
+ A64I_FMOV_R_D = 0x9e660000,
+ A64I_FMOV_D_R = 0x9e670000,
+ A64I_FMOV_DI = 0x1e601000,
} A64Ins;
+typedef enum A64Shift {
+ A64SH_LSL, A64SH_LSR, A64SH_ASR, A64SH_ROR
+} A64Shift;
+
+typedef enum A64Extend {
+ A64EX_UXTB, A64EX_UXTH, A64EX_UXTW, A64EX_UXTX,
+ A64EX_SXTB, A64EX_SXTH, A64EX_SXTW, A64EX_SXTX,
+} A64Extend;
+
+/* ARM condition codes. */
+typedef enum A64CC {
+ CC_EQ, CC_NE, CC_CS, CC_CC, CC_MI, CC_PL, CC_VS, CC_VC,
+ CC_HI, CC_LS, CC_GE, CC_LT, CC_GT, CC_LE, CC_AL,
+ CC_HS = CC_CS, CC_LO = CC_CC
+} A64CC;
+
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_mips.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_mips.h
index bafa817a38a..740687b355d 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_mips.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_mips.h
@@ -1,6 +1,6 @@
/*
** Definitions for MIPS CPUs.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_TARGET_MIPS_H
@@ -32,6 +32,7 @@ enum {
RID_MAX,
RID_ZERO = RID_R0,
RID_TMP = RID_RA,
+ RID_GP = RID_R28,
/* Calling conventions. */
RID_RET = RID_R2,
@@ -74,46 +75,64 @@ enum {
/* -- Register sets ------------------------------------------------------- */
-/* Make use of all registers, except ZERO, TMP, SP, SYS1, SYS2 and JGL. */
+/* Make use of all registers, except ZERO, TMP, SP, SYS1, SYS2, JGL and GP. */
#define RSET_FIXED \
(RID2RSET(RID_ZERO)|RID2RSET(RID_TMP)|RID2RSET(RID_SP)|\
- RID2RSET(RID_SYS1)|RID2RSET(RID_SYS2)|RID2RSET(RID_JGL))
+ RID2RSET(RID_SYS1)|RID2RSET(RID_SYS2)|RID2RSET(RID_JGL)|RID2RSET(RID_GP))
#define RSET_GPR (RSET_RANGE(RID_MIN_GPR, RID_MAX_GPR) - RSET_FIXED)
#if LJ_SOFTFP
-#define RSET_FPR 0
+#define RSET_FPR 0
#else
+#if LJ_32
#define RSET_FPR \
(RID2RSET(RID_F0)|RID2RSET(RID_F2)|RID2RSET(RID_F4)|RID2RSET(RID_F6)|\
RID2RSET(RID_F8)|RID2RSET(RID_F10)|RID2RSET(RID_F12)|RID2RSET(RID_F14)|\
RID2RSET(RID_F16)|RID2RSET(RID_F18)|RID2RSET(RID_F20)|RID2RSET(RID_F22)|\
RID2RSET(RID_F24)|RID2RSET(RID_F26)|RID2RSET(RID_F28)|RID2RSET(RID_F30))
+#else
+#define RSET_FPR RSET_RANGE(RID_MIN_FPR, RID_MAX_FPR)
+#endif
#endif
-#define RSET_ALL (RSET_GPR|RSET_FPR)
-#define RSET_INIT RSET_ALL
+#define RSET_ALL (RSET_GPR|RSET_FPR)
+#define RSET_INIT RSET_ALL
#define RSET_SCRATCH_GPR \
(RSET_RANGE(RID_R1, RID_R15+1)|\
- RID2RSET(RID_R24)|RID2RSET(RID_R25)|RID2RSET(RID_R28))
+ RID2RSET(RID_R24)|RID2RSET(RID_R25))
#if LJ_SOFTFP
#define RSET_SCRATCH_FPR 0
#else
+#if LJ_32
#define RSET_SCRATCH_FPR \
(RID2RSET(RID_F0)|RID2RSET(RID_F2)|RID2RSET(RID_F4)|RID2RSET(RID_F6)|\
RID2RSET(RID_F8)|RID2RSET(RID_F10)|RID2RSET(RID_F12)|RID2RSET(RID_F14)|\
RID2RSET(RID_F16)|RID2RSET(RID_F18))
+#else
+#define RSET_SCRATCH_FPR RSET_RANGE(RID_F0, RID_F24)
+#endif
#endif
#define RSET_SCRATCH (RSET_SCRATCH_GPR|RSET_SCRATCH_FPR)
#define REGARG_FIRSTGPR RID_R4
+#if LJ_32
#define REGARG_LASTGPR RID_R7
#define REGARG_NUMGPR 4
+#else
+#define REGARG_LASTGPR RID_R11
+#define REGARG_NUMGPR 8
+#endif
#if LJ_ABI_SOFTFP
#define REGARG_FIRSTFPR 0
#define REGARG_LASTFPR 0
#define REGARG_NUMFPR 0
#else
#define REGARG_FIRSTFPR RID_F12
+#if LJ_32
#define REGARG_LASTFPR RID_F14
#define REGARG_NUMFPR 2
+#else
+#define REGARG_LASTFPR RID_F19
+#define REGARG_NUMFPR 8
+#endif
#endif
/* -- Spill slots --------------------------------------------------------- */
@@ -125,7 +144,11 @@ enum {
**
** SPS_FIRST: First spill slot for general use.
*/
+#if LJ_32
#define SPS_FIXED 5
+#else
+#define SPS_FIXED 4
+#endif
#define SPS_FIRST 4
#define SPOFS_TMP 0
@@ -140,7 +163,7 @@ typedef struct {
#if !LJ_SOFTFP
lua_Number fpr[RID_NUM_FPR]; /* Floating-point registers. */
#endif
- int32_t gpr[RID_NUM_GPR]; /* General-purpose registers. */
+ intptr_t gpr[RID_NUM_GPR]; /* General-purpose registers. */
int32_t spill[256]; /* Spill slots. */
} ExitState;
@@ -169,32 +192,41 @@ static LJ_AINLINE uint32_t *exitstub_trace_addr_(uint32_t *p)
#define MIPSF_F(r) ((r) << 6)
#define MIPSF_A(n) ((n) << 6)
#define MIPSF_M(n) ((n) << 11)
+#define MIPSF_L(n) ((n) << 6)
typedef enum MIPSIns {
+ MIPSI_D = 0x38,
+ MIPSI_DV = 0x10,
+ MIPSI_D32 = 0x3c,
/* Integer instructions. */
- MIPSI_MOVE = 0x00000021,
+ MIPSI_MOVE = 0x00000025,
MIPSI_NOP = 0x00000000,
MIPSI_LI = 0x24000000,
MIPSI_LU = 0x34000000,
MIPSI_LUI = 0x3c000000,
- MIPSI_ADDIU = 0x24000000,
+ MIPSI_AND = 0x00000024,
MIPSI_ANDI = 0x30000000,
+ MIPSI_OR = 0x00000025,
MIPSI_ORI = 0x34000000,
+ MIPSI_XOR = 0x00000026,
MIPSI_XORI = 0x38000000,
+ MIPSI_NOR = 0x00000027,
+
+ MIPSI_SLT = 0x0000002a,
+ MIPSI_SLTU = 0x0000002b,
MIPSI_SLTI = 0x28000000,
MIPSI_SLTIU = 0x2c000000,
MIPSI_ADDU = 0x00000021,
+ MIPSI_ADDIU = 0x24000000,
+ MIPSI_SUB = 0x00000022,
MIPSI_SUBU = 0x00000023,
MIPSI_MUL = 0x70000002,
- MIPSI_AND = 0x00000024,
- MIPSI_OR = 0x00000025,
- MIPSI_XOR = 0x00000026,
- MIPSI_NOR = 0x00000027,
- MIPSI_SLT = 0x0000002a,
- MIPSI_SLTU = 0x0000002b,
+ MIPSI_DIV = 0x0000001a,
+ MIPSI_DIVU = 0x0000001b,
+
MIPSI_MOVZ = 0x0000000a,
MIPSI_MOVN = 0x0000000b,
MIPSI_MFHI = 0x00000010,
@@ -204,19 +236,24 @@ typedef enum MIPSIns {
MIPSI_SLL = 0x00000000,
MIPSI_SRL = 0x00000002,
MIPSI_SRA = 0x00000003,
- MIPSI_ROTR = 0x00200002, /* MIPS32R2 */
+ MIPSI_ROTR = 0x00200002, /* MIPSXXR2 */
+ MIPSI_DROTR = 0x0020003a,
+ MIPSI_DROTR32 = 0x0020003e,
MIPSI_SLLV = 0x00000004,
MIPSI_SRLV = 0x00000006,
MIPSI_SRAV = 0x00000007,
- MIPSI_ROTRV = 0x00000046, /* MIPS32R2 */
+ MIPSI_ROTRV = 0x00000046, /* MIPSXXR2 */
+ MIPSI_DROTRV = 0x00000056,
- MIPSI_SEB = 0x7c000420, /* MIPS32R2 */
- MIPSI_SEH = 0x7c000620, /* MIPS32R2 */
- MIPSI_WSBH = 0x7c0000a0, /* MIPS32R2 */
+ MIPSI_SEB = 0x7c000420, /* MIPSXXR2 */
+ MIPSI_SEH = 0x7c000620, /* MIPSXXR2 */
+ MIPSI_WSBH = 0x7c0000a0, /* MIPSXXR2 */
+ MIPSI_DSBH = 0x7c0000a4,
MIPSI_B = 0x10000000,
MIPSI_J = 0x08000000,
MIPSI_JAL = 0x0c000000,
+ MIPSI_JALX = 0x74000000,
MIPSI_JR = 0x00000008,
MIPSI_JALR = 0x0000f809,
@@ -229,7 +266,9 @@ typedef enum MIPSIns {
/* Load/store instructions. */
MIPSI_LW = 0x8c000000,
+ MIPSI_LD = 0xdc000000,
MIPSI_SW = 0xac000000,
+ MIPSI_SD = 0xfc000000,
MIPSI_LB = 0x80000000,
MIPSI_SB = 0xa0000000,
MIPSI_LH = 0x84000000,
@@ -241,6 +280,50 @@ typedef enum MIPSIns {
MIPSI_LDC1 = 0xd4000000,
MIPSI_SDC1 = 0xf4000000,
+ /* MIPS64 instructions. */
+ MIPSI_DADD = 0x0000002c,
+ MIPSI_DADDI = 0x60000000,
+ MIPSI_DADDU = 0x0000002d,
+ MIPSI_DADDIU = 0x64000000,
+ MIPSI_DSUB = 0x0000002e,
+ MIPSI_DSUBU = 0x0000002f,
+ MIPSI_DDIV = 0x0000001e,
+ MIPSI_DDIVU = 0x0000001f,
+ MIPSI_DMULT = 0x0000001c,
+ MIPSI_DMULTU = 0x0000001d,
+
+ MIPSI_DSLL = 0x00000038,
+ MIPSI_DSRL = 0x0000003a,
+ MIPSI_DSLLV = 0x00000014,
+ MIPSI_DSRLV = 0x00000016,
+ MIPSI_DSRA = 0x0000003b,
+ MIPSI_DSRAV = 0x00000017,
+ MIPSI_DSRA32 = 0x0000003f,
+ MIPSI_DSLL32 = 0x0000003c,
+ MIPSI_DSRL32 = 0x0000003e,
+ MIPSI_DSHD = 0x7c000164,
+
+ MIPSI_AADDU = LJ_32 ? MIPSI_ADDU : MIPSI_DADDU,
+ MIPSI_AADDIU = LJ_32 ? MIPSI_ADDIU : MIPSI_DADDIU,
+ MIPSI_ASUBU = LJ_32 ? MIPSI_SUBU : MIPSI_DSUBU,
+ MIPSI_AL = LJ_32 ? MIPSI_LW : MIPSI_LD,
+ MIPSI_AS = LJ_32 ? MIPSI_SW : MIPSI_SD,
+
+ /* Extract/insert instructions. */
+ MIPSI_DEXTM = 0x7c000001,
+ MIPSI_DEXTU = 0x7c000002,
+ MIPSI_DEXT = 0x7c000003,
+ MIPSI_DINSM = 0x7c000005,
+ MIPSI_DINSU = 0x7c000006,
+ MIPSI_DINS = 0x7c000007,
+
+ MIPSI_RINT_D = 0x4620001a,
+ MIPSI_RINT_S = 0x4600001a,
+ MIPSI_RINT = 0x4400001a,
+ MIPSI_FLOOR_D = 0x4620000b,
+ MIPSI_CEIL_D = 0x4620000a,
+ MIPSI_ROUND_D = 0x46200008,
+
/* FP instructions. */
MIPSI_MOV_S = 0x46000006,
MIPSI_MOV_D = 0x46200006,
@@ -265,24 +348,30 @@ typedef enum MIPSIns {
MIPSI_CVT_W_D = 0x46200024,
MIPSI_CVT_S_W = 0x46800020,
MIPSI_CVT_D_W = 0x46800021,
+ MIPSI_CVT_S_L = 0x46a00020,
+ MIPSI_CVT_D_L = 0x46a00021,
MIPSI_TRUNC_W_S = 0x4600000d,
MIPSI_TRUNC_W_D = 0x4620000d,
+ MIPSI_TRUNC_L_S = 0x46000009,
+ MIPSI_TRUNC_L_D = 0x46200009,
MIPSI_FLOOR_W_S = 0x4600000f,
MIPSI_FLOOR_W_D = 0x4620000f,
MIPSI_MFC1 = 0x44000000,
MIPSI_MTC1 = 0x44800000,
+ MIPSI_DMTC1 = 0x44a00000,
+ MIPSI_DMFC1 = 0x44200000,
MIPSI_BC1F = 0x45000000,
MIPSI_BC1T = 0x45010000,
MIPSI_C_EQ_D = 0x46200032,
+ MIPSI_C_OLT_S = 0x46000034,
MIPSI_C_OLT_D = 0x46200034,
MIPSI_C_ULT_D = 0x46200035,
MIPSI_C_OLE_D = 0x46200036,
MIPSI_C_ULE_D = 0x46200037,
-
} MIPSIns;
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_ppc.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_ppc.h
index bbf22390c46..c5c991a377a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_ppc.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_ppc.h
@@ -1,6 +1,6 @@
/*
** Definitions for PPC CPUs.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_TARGET_PPC_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_x86.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_x86.h
index 289f83e1c8b..356f792459b 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_target_x86.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_target_x86.h
@@ -1,6 +1,6 @@
/*
** Definitions for x86 and x64 CPUs.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_TARGET_X86_H
@@ -22,7 +22,7 @@
_(XMM0) _(XMM1) _(XMM2) _(XMM3) _(XMM4) _(XMM5) _(XMM6) _(XMM7)
#endif
#define VRIDDEF(_) \
- _(MRM)
+ _(MRM) _(RIP)
#define RIDENUM(name) RID_##name,
@@ -31,6 +31,7 @@ enum {
FPRDEF(RIDENUM) /* Floating-point registers (FPRs). */
RID_MAX,
RID_MRM = RID_MAX, /* Pseudo-id for ModRM operand. */
+ RID_RIP = RID_MAX+5, /* Pseudo-id for RIP (x64 only), rm bits = 5. */
/* Calling conventions. */
RID_SP = RID_ESP,
@@ -63,8 +64,10 @@ enum {
/* -- Register sets ------------------------------------------------------- */
-/* Make use of all registers, except the stack pointer. */
-#define RSET_GPR (RSET_RANGE(RID_MIN_GPR, RID_MAX_GPR)-RID2RSET(RID_ESP))
+/* Make use of all registers, except the stack pointer (and maybe DISPATCH). */
+#define RSET_GPR (RSET_RANGE(RID_MIN_GPR, RID_MAX_GPR) \
+ - RID2RSET(RID_ESP) \
+ - LJ_GC64*RID2RSET(RID_DISPATCH))
#define RSET_FPR (RSET_RANGE(RID_MIN_FPR, RID_MAX_FPR))
#define RSET_ALL (RSET_GPR|RSET_FPR)
#define RSET_INIT RSET_ALL
@@ -189,12 +192,18 @@ typedef struct {
#define XO_f20f(o) ((uint32_t)(0x0ff2fc + (0x##o<<24)))
#define XO_f30f(o) ((uint32_t)(0x0ff3fc + (0x##o<<24)))
+#define XV_660f38(o) ((uint32_t)(0x79e2c4 + (0x##o<<24)))
+#define XV_f20f38(o) ((uint32_t)(0x7be2c4 + (0x##o<<24)))
+#define XV_f20f3a(o) ((uint32_t)(0x7be3c4 + (0x##o<<24)))
+#define XV_f30f38(o) ((uint32_t)(0x7ae2c4 + (0x##o<<24)))
+
/* This list of x86 opcodes is not intended to be complete. Opcodes are only
** included when needed. Take a look at DynASM or jit.dis_x86 to see the
** whole mess.
*/
typedef enum {
/* Fixed length opcodes. XI_* prefix. */
+ XI_O16 = 0x66,
XI_NOP = 0x90,
XI_XCHGa = 0x90,
XI_CALL = 0xe8,
@@ -212,6 +221,7 @@ typedef enum {
XI_PUSHi8 = 0x6a,
XI_TESTb = 0x84,
XI_TEST = 0x85,
+ XI_INT3 = 0xcc,
XI_MOVmi = 0xc7,
XI_GROUP5 = 0xff,
@@ -231,7 +241,14 @@ typedef enum {
XI_FSCALE = 0xfdd9,
XI_FYL2X = 0xf1d9,
+ /* VEX-encoded instructions. XV_* prefix. */
+ XV_RORX = XV_f20f3a(f0),
+ XV_SARX = XV_f30f38(f7),
+ XV_SHLX = XV_660f38(f7),
+ XV_SHRX = XV_f20f38(f7),
+
/* Variable-length opcodes. XO_* prefix. */
+ XO_OR = XO_(0b),
XO_MOV = XO_(8b),
XO_MOVto = XO_(89),
XO_MOVtow = XO_66(89),
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_trace.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_trace.c
index 7970aba6ae0..d85b47f8013 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_trace.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_trace.c
@@ -1,6 +1,6 @@
/*
** Trace management.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_trace_c
@@ -117,15 +117,26 @@ static void perftools_addtrace(GCtrace *T)
}
#endif
-/* Allocate space for copy of trace. */
-static GCtrace *trace_save_alloc(jit_State *J)
+/* Allocate space for copy of T. */
+GCtrace * LJ_FASTCALL lj_trace_alloc(lua_State *L, GCtrace *T)
{
size_t sztr = ((sizeof(GCtrace)+7)&~7);
- size_t szins = (J->cur.nins-J->cur.nk)*sizeof(IRIns);
+ size_t szins = (T->nins-T->nk)*sizeof(IRIns);
size_t sz = sztr + szins +
- J->cur.nsnap*sizeof(SnapShot) +
- J->cur.nsnapmap*sizeof(SnapEntry);
- return lj_mem_newt(J->L, (MSize)sz, GCtrace);
+ T->nsnap*sizeof(SnapShot) +
+ T->nsnapmap*sizeof(SnapEntry);
+ GCtrace *T2 = lj_mem_newt(L, (MSize)sz, GCtrace);
+ char *p = (char *)T2 + sztr;
+ T2->gct = ~LJ_TTRACE;
+ T2->marked = 0;
+ T2->traceno = 0;
+ T2->ir = (IRIns *)p - T->nk;
+ T2->nins = T->nins;
+ T2->nk = T->nk;
+ T2->nsnap = T->nsnap;
+ T2->nsnapmap = T->nsnapmap;
+ memcpy(p, T->ir + T->nk, szins);
+ return T2;
}
/* Save current trace by copying and compacting it. */
@@ -139,12 +150,12 @@ static void trace_save(jit_State *J, GCtrace *T)
setgcrefp(J2G(J)->gc.root, T);
newwhite(J2G(J), T);
T->gct = ~LJ_TTRACE;
- T->ir = (IRIns *)p - J->cur.nk;
- memcpy(p, J->cur.ir+J->cur.nk, szins);
+ T->ir = (IRIns *)p - J->cur.nk; /* The IR has already been copied above. */
p += szins;
TRACE_APPENDVEC(snap, nsnap, SnapShot)
TRACE_APPENDVEC(snapmap, nsnapmap, SnapEntry)
J->cur.traceno = 0;
+ J->curfinal = NULL;
setgcrefp(J->trace[T->traceno], T);
lj_gc_barriertrace(J2G(J), T->traceno);
lj_gdbjit_addtrace(J, T);
@@ -284,7 +295,6 @@ int lj_trace_flushall(lua_State *L)
memset(J->penalty, 0, sizeof(J->penalty));
/* Free the whole machine code and invalidate all exit stub groups. */
lj_mcode_free(J);
- lj_ir_k64_freeall(J);
memset(J->exitstubgroup, 0, sizeof(J->exitstubgroup));
lj_vmevent_send(L, TRACE,
setstrV(L, L->top++, lj_str_newlit(L, "flush"));
@@ -297,13 +307,42 @@ void lj_trace_initstate(global_State *g)
{
jit_State *J = G2J(g);
TValue *tv;
- /* Initialize SIMD constants. */
+
+ /* Initialize aligned SIMD constants. */
tv = LJ_KSIMD(J, LJ_KSIMD_ABS);
tv[0].u64 = U64x(7fffffff,ffffffff);
tv[1].u64 = U64x(7fffffff,ffffffff);
tv = LJ_KSIMD(J, LJ_KSIMD_NEG);
tv[0].u64 = U64x(80000000,00000000);
tv[1].u64 = U64x(80000000,00000000);
+
+ /* Initialize 32/64 bit constants. */
+#if LJ_TARGET_X86ORX64
+ J->k64[LJ_K64_TOBIT].u64 = U64x(43380000,00000000);
+#if LJ_32
+ J->k64[LJ_K64_M2P64_31].u64 = U64x(c1e00000,00000000);
+#endif
+ J->k64[LJ_K64_2P64].u64 = U64x(43f00000,00000000);
+ J->k32[LJ_K32_M2P64_31] = LJ_64 ? 0xdf800000 : 0xcf000000;
+#endif
+#if LJ_TARGET_X86ORX64 || LJ_TARGET_MIPS64
+ J->k64[LJ_K64_M2P64].u64 = U64x(c3f00000,00000000);
+#endif
+#if LJ_TARGET_PPC
+ J->k32[LJ_K32_2P52_2P31] = 0x59800004;
+ J->k32[LJ_K32_2P52] = 0x59800000;
+#endif
+#if LJ_TARGET_PPC || LJ_TARGET_MIPS
+ J->k32[LJ_K32_2P31] = 0x4f000000;
+#endif
+#if LJ_TARGET_MIPS
+ J->k64[LJ_K64_2P31].u64 = U64x(41e00000,00000000);
+#if LJ_64
+ J->k64[LJ_K64_2P63].u64 = U64x(43e00000,00000000);
+ J->k32[LJ_K32_2P63] = 0x5f000000;
+ J->k32[LJ_K32_M2P64] = 0xdf800000;
+#endif
+#endif
}
/* Free everything associated with the JIT compiler state. */
@@ -318,7 +357,6 @@ void lj_trace_freestate(global_State *g)
}
#endif
lj_mcode_free(J);
- lj_ir_k64_freeall(J);
lj_mem_freevec(g, J->snapmapbuf, J->sizesnapmap, SnapEntry);
lj_mem_freevec(g, J->snapbuf, J->sizesnap, SnapShot);
lj_mem_freevec(g, J->irbuf + J->irbotlim, J->irtoplim - J->irbotlim, IRIns);
@@ -403,7 +441,7 @@ static void trace_start(jit_State *J)
J->postproc = LJ_POST_NONE;
lj_resetsplit(J);
J->retryrec = 0;
- J->ktracep = NULL;
+ J->ktrace = 0;
setgcref(J->cur.startpt, obj2gco(J->pt));
L = J->L;
@@ -415,6 +453,12 @@ static void trace_start(jit_State *J)
if (J->parent) {
setintV(L->top++, J->parent);
setintV(L->top++, J->exitno);
+ } else {
+ BCOp op = bc_op(*J->pc);
+ if (op == BC_CALLM || op == BC_CALL || op == BC_ITERC) {
+ setintV(L->top++, J->exitno); /* Parent of stitched trace. */
+ setintV(L->top++, -1);
+ }
}
);
lj_record_setup(J);
@@ -427,7 +471,7 @@ static void trace_stop(jit_State *J)
BCOp op = bc_op(J->cur.startins);
GCproto *pt = &gcref(J->cur.startpt)->pt;
TraceNo traceno = J->cur.traceno;
- GCtrace *T = trace_save_alloc(J); /* Do this first. May throw OOM. */
+ GCtrace *T = J->curfinal;
lua_State *L;
switch (op) {
@@ -479,9 +523,6 @@ static void trace_stop(jit_State *J)
lj_mcode_commit(J, J->cur.mcode);
J->postproc = LJ_POST_NONE;
trace_save(J, T);
- if (J->ktracep) { /* Patch K64Array slot with the final GCtrace pointer. */
- setgcV(J->L, J->ktracep, obj2gco(T), LJ_TTRACE);
- }
L = J->L;
lj_vmevent_send(L, TRACE,
@@ -515,6 +556,10 @@ static int trace_abort(jit_State *J)
J->postproc = LJ_POST_NONE;
lj_mcode_abort(J);
+ if (J->curfinal) {
+ lj_trace_free(J2G(J), J->curfinal);
+ J->curfinal = NULL;
+ }
if (tvisnumber(L->top-1))
e = (TraceError)numberVint(L->top-1);
if (e == LJ_TRERR_MCODELM) {
@@ -849,7 +894,7 @@ int LJ_FASTCALL lj_trace_exit(jit_State *J, void *exptr)
ERRNO_RESTORE
switch (bc_op(*pc)) {
case BC_CALLM: case BC_CALLMT:
- return (int)((BCReg)(L->top - L->base) - bc_a(*pc) - bc_c(*pc) + LJ_FR2);
+ return (int)((BCReg)(L->top - L->base) - bc_a(*pc) - bc_c(*pc) - LJ_FR2);
case BC_RETM:
return (int)((BCReg)(L->top - L->base) + 1 - bc_a(*pc) - bc_d(*pc));
case BC_TSETM:
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_trace.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_trace.h
index 6faa1aa3cde..22cae741f3f 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_trace.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_trace.h
@@ -1,6 +1,6 @@
/*
** Trace management.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_TRACE_H
@@ -23,6 +23,7 @@ LJ_FUNC_NORET void lj_trace_err(jit_State *J, TraceError e);
LJ_FUNC_NORET void lj_trace_err_info(jit_State *J, TraceError e);
/* Trace management. */
+LJ_FUNC GCtrace * LJ_FASTCALL lj_trace_alloc(lua_State *L, GCtrace *T);
LJ_FUNC void LJ_FASTCALL lj_trace_free(global_State *g, GCtrace *T);
LJ_FUNC void lj_trace_reenableproto(GCproto *pt);
LJ_FUNC void lj_trace_flushproto(global_State *g, GCproto *pt);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_traceerr.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_traceerr.h
index 9c01ffb7ac5..1363c4f3272 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_traceerr.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_traceerr.h
@@ -1,6 +1,6 @@
/*
** Trace compiler error messages.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/* This file may be included multiple times with different TREDEF macros. */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_udata.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_udata.c
index 71697a42fb9..bd0321b834b 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_udata.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_udata.c
@@ -1,6 +1,6 @@
/*
** Userdata handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_udata_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_udata.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_udata.h
index 81c15816da5..f271a42d327 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_udata.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_udata.h
@@ -1,6 +1,6 @@
/*
** Userdata handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_UDATA_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_vm.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_vm.h
index be35295d261..1cc7eed782f 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_vm.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_vm.h
@@ -1,6 +1,6 @@
/*
** Assembler VM interface definitions.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_VM_H
@@ -17,6 +17,10 @@ LJ_ASMF int lj_vm_cpcall(lua_State *L, lua_CFunction func, void *ud,
LJ_ASMF int lj_vm_resume(lua_State *L, TValue *base, int nres1, ptrdiff_t ef);
LJ_ASMF_NORET void LJ_FASTCALL lj_vm_unwind_c(void *cframe, int errcode);
LJ_ASMF_NORET void LJ_FASTCALL lj_vm_unwind_ff(void *cframe);
+#if LJ_ABI_WIN && LJ_TARGET_X86
+LJ_ASMF_NORET void LJ_FASTCALL lj_vm_rtlunwind(void *cframe, void *excptrec,
+ void *unwinder, int errcode);
+#endif
LJ_ASMF void lj_vm_unwind_c_eh(void);
LJ_ASMF void lj_vm_unwind_ff_eh(void);
#if LJ_TARGET_X86ORX64
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.c
index 3acdd2534e6..86640804bed 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.c
@@ -1,6 +1,6 @@
/*
** VM event handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#include <stdio.h>
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.h
index b9495b4f9ec..050fb4dd24a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_vmevent.h
@@ -1,6 +1,6 @@
/*
** VM event handling.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LJ_VMEVENT_H
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_vmmath.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_vmmath.c
index e6622650503..b231d3e811e 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_vmmath.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_vmmath.c
@@ -1,6 +1,6 @@
/*
** Math helper functions for assembler VM.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#define lj_vmmath_c
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/ljamalg.c b/Build/source/libs/luajit/LuaJIT-src/src/ljamalg.c
index 4c44de27ad8..f1f28623fa9 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/ljamalg.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/ljamalg.c
@@ -1,6 +1,6 @@
/*
** LuaJIT core and libraries amalgamation.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
/*
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lua.h b/Build/source/libs/luajit/LuaJIT-src/src/lua.h
index fb4e69f4076..b918366cbdf 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lua.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lua.h
@@ -39,7 +39,8 @@
#define lua_upvalueindex(i) (LUA_GLOBALSINDEX-(i))
-/* thread status; 0 is OK */
+/* thread status */
+#define LUA_OK 0
#define LUA_YIELD 1
#define LUA_ERRRUN 2
#define LUA_ERRSYNTAX 3
@@ -102,11 +103,9 @@ typedef LUA_NUMBER lua_Number;
/* type for integer functions */
typedef LUA_INTEGER lua_Integer;
-
/* communication with LuaJiTTeX */
LUA_API int luajittex_choose_hash_function;
-
/*
** state manipulation
*/
@@ -350,16 +349,21 @@ LUA_API void *lua_upvalueid (lua_State *L, int idx, int n);
LUA_API void lua_upvaluejoin (lua_State *L, int idx1, int n1, int idx2, int n2);
LUA_API int lua_loadx (lua_State *L, lua_Reader reader, void *dt,
const char *chunkname, const char *mode);
+LUA_API const lua_Number *lua_version (lua_State *L);
+LUA_API void lua_copy (lua_State *L, int fromidx, int toidx);
+LUA_API lua_Number lua_tonumberx (lua_State *L, int idx, int *isnum);
+LUA_API lua_Integer lua_tointegerx (lua_State *L, int idx, int *isnum);
#define LUA_OPEQ 0
#define LUA_OPLT 1
#define LUA_OPLE 2
-#define LUA_OK 0
-
+
/* see http://comments.gmane.org/gmane.comp.programming.swig/18673 */
# define lua_rawlen lua_objlen
+/* From Lua 5.3. */
+LUA_API int lua_isyieldable (lua_State *L);
struct lua_Debug {
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/luaconf.h b/Build/source/libs/luajit/LuaJIT-src/src/luaconf.h
index 4c24bb1a0e2..c2d29d94922 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/luaconf.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/luaconf.h
@@ -1,6 +1,6 @@
/*
** Configuration header.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef luaconf_h
@@ -37,7 +37,7 @@
#endif
#define LUA_LROOT "/usr/local"
#define LUA_LUADIR "/lua/5.1/"
-#define LUA_LJDIR "/luajit-2.1.0-beta2/"
+#define LUA_LJDIR "/luajit-2.1.0-beta3/"
#ifdef LUA_ROOT
#define LUA_JROOT LUA_ROOT
@@ -79,7 +79,7 @@
#define LUA_IGMARK "-"
#define LUA_PATH_CONFIG \
LUA_DIRSEP "\n" LUA_PATHSEP "\n" LUA_PATH_MARK "\n" \
- LUA_EXECDIR "\n" LUA_IGMARK
+ LUA_EXECDIR "\n" LUA_IGMARK "\n"
/* Quoting in error messages. */
#define LUA_QL(x) "'" x "'"
@@ -92,10 +92,6 @@
#define LUAI_GCMUL 200 /* Run GC at 200% of allocation speed. */
#define LUA_MAXCAPTURES 32 /* Max. pattern captures. */
-/* Compatibility with older library function names. */
-#define LUA_COMPAT_MOD /* OLD: math.mod, NEW: math.fmod */
-#define LUA_COMPAT_GFIND /* OLD: string.gfind, NEW: string.gmatch */
-
/* Configuration for the frontend (the luajit executable). */
#if defined(luajit_c)
#define LUA_PROGNAME "luajit" /* Fallback frontend name. */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/luajit.c b/Build/source/libs/luajit/LuaJIT-src/src/luajit.c
index 00e12bd32e3..1ca243011fb 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/luajit.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/luajit.c
@@ -1,6 +1,6 @@
/*
** LuaJIT frontend. Runs commands, scripts, read-eval-print (REPL) etc.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
**
** Major portions taken verbatim or adapted from the Lua interpreter.
** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
@@ -124,7 +124,7 @@ static int docall(lua_State *L, int narg, int clear)
#endif
lua_remove(L, base); /* remove traceback function */
/* force a complete garbage collection in case of errors */
- if (status != 0) lua_gc(L, LUA_GCCOLLECT, 0);
+ if (status != LUA_OK) lua_gc(L, LUA_GCCOLLECT, 0);
return status;
}
@@ -152,22 +152,15 @@ static void print_jit_status(lua_State *L)
putc('\n', stdout);
}
-static int getargs(lua_State *L, char **argv, int n)
+static void createargtable(lua_State *L, char **argv, int argc, int argf)
{
- int narg;
int i;
- int argc = 0;
- while (argv[argc]) argc++; /* count total number of arguments */
- narg = argc - (n + 1); /* number of arguments to the script */
- luaL_checkstack(L, narg + 3, "too many arguments to script");
- for (i = n+1; i < argc; i++)
- lua_pushstring(L, argv[i]);
- lua_createtable(L, narg, n + 1);
+ lua_createtable(L, argc - argf, argf);
for (i = 0; i < argc; i++) {
lua_pushstring(L, argv[i]);
- lua_rawseti(L, -2, i - n);
+ lua_rawseti(L, -2, i - argf);
}
- return narg;
+ lua_setglobal(L, "arg");
}
static int dofile(lua_State *L, const char *name)
@@ -256,9 +249,9 @@ static void dotty(lua_State *L)
const char *oldprogname = progname;
progname = NULL;
while ((status = loadline(L)) != -1) {
- if (status == 0) status = docall(L, 0, 0);
+ if (status == LUA_OK) status = docall(L, 0, 0);
report(L, status);
- if (status == 0 && lua_gettop(L) > 0) { /* any result to print? */
+ if (status == LUA_OK && lua_gettop(L) > 0) { /* any result to print? */
lua_getglobal(L, "print");
lua_insert(L, 1);
if (lua_pcall(L, lua_gettop(L)-1, 0, 0) != 0)
@@ -273,21 +266,30 @@ static void dotty(lua_State *L)
progname = oldprogname;
}
-static int handle_script(lua_State *L, char **argv, int n)
+static int handle_script(lua_State *L, char **argx)
{
int status;
- const char *fname;
- int narg = getargs(L, argv, n); /* collect arguments */
- lua_setglobal(L, "arg");
- fname = argv[n];
- if (strcmp(fname, "-") == 0 && strcmp(argv[n-1], "--") != 0)
+ const char *fname = argx[0];
+ if (strcmp(fname, "-") == 0 && strcmp(argx[-1], "--") != 0)
fname = NULL; /* stdin */
status = luaL_loadfile(L, fname);
- lua_insert(L, -(narg+1));
- if (status == 0)
+ if (status == LUA_OK) {
+ /* Fetch args from arg table. LUA_INIT or -e might have changed them. */
+ int narg = 0;
+ lua_getglobal(L, "arg");
+ if (lua_istable(L, -1)) {
+ do {
+ narg++;
+ lua_rawgeti(L, -narg, narg);
+ } while (!lua_isnil(L, -1));
+ lua_pop(L, 1);
+ lua_remove(L, -narg);
+ narg--;
+ } else {
+ lua_pop(L, 1);
+ }
status = docall(L, narg, 0);
- else
- lua_pop(L, narg);
+ }
return report(L, status);
}
@@ -384,7 +386,8 @@ static int dobytecode(lua_State *L, char **argv)
}
for (argv++; *argv != NULL; narg++, argv++)
lua_pushstring(L, *argv);
- return report(L, lua_pcall(L, narg, 0, 0));
+ report(L, lua_pcall(L, narg, 0, 0));
+ return -1;
}
/* check that argument has no extra characters at the end */
@@ -405,7 +408,7 @@ static int collectargs(char **argv, int *flags)
switch (argv[i][1]) { /* Check option. */
case '-':
notail(argv[i]);
- return (argv[i+1] != NULL ? i+1 : 0);
+ return i+1;
case '\0':
return i;
case 'i':
@@ -430,23 +433,23 @@ static int collectargs(char **argv, int *flags)
case 'b': /* LuaJIT extension */
if (*flags) return -1;
*flags |= FLAGS_EXEC;
- return 0;
+ return i+1;
case 'E':
*flags |= FLAGS_NOENV;
break;
default: return -1; /* invalid option */
}
}
- return 0;
+ return i;
}
-static int runargs(lua_State *L, char **argv, int n)
+static int runargs(lua_State *L, char **argv, int argn)
{
int i;
- for (i = 1; i < n; i++) {
+ for (i = 1; i < argn; i++) {
if (argv[i] == NULL) continue;
lua_assert(argv[i][0] == '-');
- switch (argv[i][1]) { /* option */
+ switch (argv[i][1]) {
case 'e': {
const char *chunk = argv[i] + 2;
if (*chunk == '\0') chunk = argv[++i];
@@ -460,10 +463,10 @@ static int runargs(lua_State *L, char **argv, int n)
if (*filename == '\0') filename = argv[++i];
lua_assert(filename != NULL);
if (dolibrary(L, filename))
- return 1; /* stop if file fails */
+ return 1;
break;
}
- case 'j': { /* LuaJIT extension */
+ case 'j': { /* LuaJIT extension. */
const char *cmd = argv[i] + 2;
if (*cmd == '\0') cmd = argv[++i];
lua_assert(cmd != NULL);
@@ -471,16 +474,16 @@ static int runargs(lua_State *L, char **argv, int n)
return 1;
break;
}
- case 'O': /* LuaJIT extension */
+ case 'O': /* LuaJIT extension. */
if (dojitopt(L, argv[i] + 2))
return 1;
break;
- case 'b': /* LuaJIT extension */
+ case 'b': /* LuaJIT extension. */
return dobytecode(L, argv+i);
default: break;
}
}
- return 0;
+ return LUA_OK;
}
static int handle_luainit(lua_State *L)
@@ -491,7 +494,7 @@ static int handle_luainit(lua_State *L)
const char *init = getenv(LUA_INIT);
#endif
if (init == NULL)
- return 0; /* status OK */
+ return LUA_OK;
else if (init[0] == '@')
return dofile(L, init+1);
else
@@ -508,45 +511,57 @@ static int pmain(lua_State *L)
{
struct Smain *s = &smain;
char **argv = s->argv;
- int script;
+ int argn;
int flags = 0;
globalL = L;
if (argv[0] && argv[0][0]) progname = argv[0];
- LUAJIT_VERSION_SYM(); /* linker-enforced version check */
- script = collectargs(argv, &flags);
- if (script < 0) { /* invalid args? */
+
+ LUAJIT_VERSION_SYM(); /* Linker-enforced version check. */
+
+ argn = collectargs(argv, &flags);
+ if (argn < 0) { /* Invalid args? */
print_usage();
s->status = 1;
return 0;
}
+
if ((flags & FLAGS_NOENV)) {
lua_pushboolean(L, 1);
lua_setfield(L, LUA_REGISTRYINDEX, "LUA_NOENV");
}
- lua_gc(L, LUA_GCSTOP, 0); /* stop collector during initialization */
- luaL_openlibs(L); /* open libraries */
+
+ /* Stop collector during library initialization. */
+ lua_gc(L, LUA_GCSTOP, 0);
+ luaL_openlibs(L);
lua_gc(L, LUA_GCRESTART, -1);
+
+ createargtable(L, argv, s->argc, argn);
+
if (!(flags & FLAGS_NOENV)) {
s->status = handle_luainit(L);
- if (s->status != 0) return 0;
+ if (s->status != LUA_OK) return 0;
}
+
if ((flags & FLAGS_VERSION)) print_version();
- s->status = runargs(L, argv, (script > 0) ? script : s->argc);
- if (s->status != 0) return 0;
- if (script) {
- s->status = handle_script(L, argv, script);
- if (s->status != 0) return 0;
+
+ s->status = runargs(L, argv, argn);
+ if (s->status != LUA_OK) return 0;
+
+ if (s->argc > argn) {
+ s->status = handle_script(L, argv + argn);
+ if (s->status != LUA_OK) return 0;
}
+
if ((flags & FLAGS_INTERACTIVE)) {
print_jit_status(L);
dotty(L);
- } else if (script == 0 && !(flags & (FLAGS_EXEC|FLAGS_VERSION))) {
+ } else if (s->argc == argn && !(flags & (FLAGS_EXEC|FLAGS_VERSION))) {
if (lua_stdin_is_tty()) {
print_version();
print_jit_status(L);
dotty(L);
} else {
- dofile(L, NULL); /* executes stdin as a file */
+ dofile(L, NULL); /* Executes stdin as a file. */
}
}
return 0;
@@ -555,7 +570,7 @@ static int pmain(lua_State *L)
int main(int argc, char **argv)
{
int status;
- lua_State *L = lua_open(); /* create state */
+ lua_State *L = lua_open();
if (L == NULL) {
l_message(argv[0], "cannot create state: not enough memory");
return EXIT_FAILURE;
@@ -565,6 +580,6 @@ int main(int argc, char **argv)
status = lua_cpcall(L, pmain, NULL);
report(L, status);
lua_close(L);
- return (status || smain.status) ? EXIT_FAILURE : EXIT_SUCCESS;
+ return (status || smain.status > 0) ? EXIT_FAILURE : EXIT_SUCCESS;
}
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/luajit.h b/Build/source/libs/luajit/LuaJIT-src/src/luajit.h
index 1d0a558cabc..708a5a11fdb 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/luajit.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/luajit.h
@@ -1,7 +1,7 @@
/*
** LuaJIT -- a Just-In-Time Compiler for Lua. http://luajit.org/
**
-** Copyright (C) 2005-2016 Mike Pall. All rights reserved.
+** Copyright (C) 2005-2017 Mike Pall. All rights reserved.
**
** Permission is hereby granted, free of charge, to any person obtaining
** a copy of this software and associated documentation files (the
@@ -30,10 +30,10 @@
#include "lua.h"
-#define LUAJIT_VERSION "LuaJIT 2.1.0-beta2"
+#define LUAJIT_VERSION "LuaJIT 2.1.0-beta3"
#define LUAJIT_VERSION_NUM 20100 /* Version 2.1.0 = 02.01.00. */
-#define LUAJIT_VERSION_SYM luaJIT_version_2_1_0_beta2
-#define LUAJIT_COPYRIGHT "Copyright (C) 2005-2016 Mike Pall"
+#define LUAJIT_VERSION_SYM luaJIT_version_2_1_0_beta3
+#define LUAJIT_COPYRIGHT "Copyright (C) 2005-2017 Mike Pall"
#define LUAJIT_URL "http://luajit.org/"
/* Modes for luaJIT_setmode. */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lualib.h b/Build/source/libs/luajit/LuaJIT-src/src/lualib.h
index f2d9b486e6f..aed323428eb 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lualib.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lualib.h
@@ -1,6 +1,6 @@
/*
** Standard library header.
-** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
*/
#ifndef _LUALIB_H
@@ -23,7 +23,7 @@
#define LUA_FFILIBNAME "ffi"
#define LUA_BITLIBNAME_32 "bit32"
-
+
LUALIB_API int luaopen_base(lua_State *L);
LUALIB_API int luaopen_math(lua_State *L);
LUALIB_API int luaopen_string(lua_State *L);
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/msvcbuild.bat b/Build/source/libs/luajit/LuaJIT-src/src/msvcbuild.bat
index 9c6b5320ae7..71bde7598dc 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/msvcbuild.bat
+++ b/Build/source/libs/luajit/LuaJIT-src/src/msvcbuild.bat
@@ -1,5 +1,5 @@
@rem Script to build LuaJIT with MSVC.
-@rem Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+@rem Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
@rem
@rem Either open a "Visual Studio .NET Command Prompt"
@rem (Note that the Express Edition does not contain an x64 compiler)
@@ -14,12 +14,13 @@
@if not defined INCLUDE goto :FAIL
@setlocal
-@set LJCOMPILE=cl /nologo /c /O2 /W3 /D_CRT_SECURE_NO_DEPRECATE
+@set LJCOMPILE=cl /nologo /c /O2 /W3 /D_CRT_SECURE_NO_DEPRECATE /D_CRT_STDIO_INLINE=__declspec(dllexport)__inline
@set LJLINK=link /nologo
@set LJMT=mt /nologo
@set LJLIB=lib /nologo /nodefaultlib
@set DASMDIR=..\dynasm
@set DASM=%DASMDIR%\dynasm.lua
+@set DASC=vm_x86.dasc
@set LJDLLNAME=lua51.dll
@set LJLIBNAME=lua51.lib
@set ALL_LIB=lib_base.c lib_math.c lib_bit.c lib_string.c lib_table.c lib_io.c lib_os.c lib_package.c lib_debug.c lib_jit.c lib_ffi.c
@@ -39,7 +40,12 @@ if exist minilua.exe.manifest^
@set LJARCH=x86
@set LJCOMPILE=%LJCOMPILE% /arch:SSE2
:X64
-minilua %DASM% -LN %DASMFLAGS% -o host\buildvm_arch.h vm_x86.dasc
+@if "%1" neq "gc64" goto :NOGC64
+@shift
+@set DASC=vm_x64.dasc
+@set LJCOMPILE=%LJCOMPILE% /DLUAJIT_ENABLE_GC64
+:NOGC64
+minilua %DASM% -LN %DASMFLAGS% -o host\buildvm_arch.h %DASC%
@if errorlevel 1 goto :BAD
%LJCOMPILE% /I "." /I %DASMDIR% host\buildvm*.c
@@ -67,7 +73,7 @@ buildvm -m folddef -o lj_folddef.h lj_opt_fold.c
@if "%1" neq "debug" goto :NODEBUG
@shift
@set LJCOMPILE=%LJCOMPILE% /Zi
-@set LJLINK=%LJLINK% /debug
+@set LJLINK=%LJLINK% /debug /opt:ref /opt:icf /incremental:no
:NODEBUG
@if "%1"=="amalg" goto :AMALGDLL
@if "%1"=="static" goto :STATIC
@@ -99,6 +105,8 @@ if exist luajit.exe.manifest^
%LJMT% -manifest luajit.exe.manifest -outputresource:luajit.exe
@del *.obj *.manifest minilua.exe buildvm.exe
+@del host\buildvm_arch.h
+@del lj_bcdef.h lj_ffdef.h lj_libdef.h lj_recdef.h lj_folddef.h
@echo.
@echo === Successfully built LuaJIT for Windows/%LJARCH% ===
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/vm_arm.dasc b/Build/source/libs/luajit/LuaJIT-src/src/vm_arm.dasc
index 882884cdf76..780cc16e6de 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/vm_arm.dasc
+++ b/Build/source/libs/luajit/LuaJIT-src/src/vm_arm.dasc
@@ -1,6 +1,6 @@
|// Low-level VM code for ARM CPUs.
|// Bytecode interpreter, fast functions and helper functions.
-|// Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+|// Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
|
|.arch arm
|.section code_op, code_sub
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/vm_arm64.dasc b/Build/source/libs/luajit/LuaJIT-src/src/vm_arm64.dasc
index 7a881bdd34c..3eaf37638e6 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/vm_arm64.dasc
+++ b/Build/source/libs/luajit/LuaJIT-src/src/vm_arm64.dasc
@@ -1,6 +1,6 @@
|// Low-level VM code for ARM64 CPUs.
|// Bytecode interpreter, fast functions and helper functions.
-|// Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+|// Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
|
|.arch arm64
|.section code_op, code_sub
@@ -151,6 +151,21 @@
|.define FRAME_FUNC, #-16
|.define FRAME_PC, #-8
|
+|// Endian-specific defines.
+|.if ENDIAN_LE
+|.define LO, 0
+|.define OFS_RD, 2
+|.define OFS_RB, 3
+|.define OFS_RA, 1
+|.define OFS_OP, 0
+|.else
+|.define LO, 4
+|.define OFS_RD, 0
+|.define OFS_RB, 0
+|.define OFS_RA, 2
+|.define OFS_OP, 3
+|.endif
+|
|.macro decode_RA, dst, ins; ubfx dst, ins, #8, #8; .endmacro
|.macro decode_RB, dst, ins; ubfx dst, ins, #24, #8; .endmacro
|.macro decode_RC, dst, ins; ubfx dst, ins, #16, #8; .endmacro
@@ -236,12 +251,17 @@
|.macro mov_false, reg; movn reg, #0x8000, lsl #32; .endmacro
|.macro mov_true, reg; movn reg, #0x0001, lsl #48; .endmacro
|
-#define GL_J(field) (GG_OFS(J) + (int)offsetof(jit_State, field))
+#define GL_J(field) (GG_G2J + (int)offsetof(jit_State, field))
|
#define PC2PROTO(field) ((int)offsetof(GCproto, field)-(int)sizeof(GCproto))
|
|.macro hotcheck, delta
-| NYI
+| lsr CARG1, PC, #1
+| and CARG1, CARG1, #126
+| add CARG1, CARG1, #GG_G2DISP+GG_DISP2HOT
+| ldrh CARG2w, [GL, CARG1]
+| subs CARG2, CARG2, #delta
+| strh CARG2w, [GL, CARG1]
|.endmacro
|
|.macro hotloop
@@ -712,7 +732,7 @@ static void build_subroutines(BuildCtx *ctx)
| cmp CRET1, #1
| bhi ->vmeta_binop
|4:
- | ldrh RBw, [PC, #2]
+ | ldrh RBw, [PC, # OFS_RD]
| add PC, PC, #4
| add RB, PC, RB, lsl #2
| sub RB, RB, #0x20000
@@ -869,7 +889,7 @@ static void build_subroutines(BuildCtx *ctx)
| bl extern lj_meta_for // (lua_State *L, TValue *base)
| ldr INSw, [PC, #-4]
|.if JIT
- | uxtb TMP0, INS
+ | uxtb TMP0w, INSw
|.endif
| decode_RA RA, INS
| decode_RD RC, INS
@@ -1495,7 +1515,12 @@ static void build_subroutines(BuildCtx *ctx)
| bne ->fff_fallback
| checkint CARG1, ->fff_fallback
| mov CARG3, #1
- | mov CARG2, BASE // Points to stack. Little-endian.
+ | // Point to the char inside the integer in the stack slot.
+ |.if ENDIAN_LE
+ | mov CARG2, BASE
+ |.else
+ | add CARG2, BASE, #7
+ |.endif
|->fff_newstr:
| // CARG2 = str, CARG3 = len.
| str BASE, L->base
@@ -1698,7 +1723,7 @@ static void build_subroutines(BuildCtx *ctx)
| ands TMP0, PC, #FRAME_TYPE
| and TMP1, PC, #~FRAME_TYPEP
| bne >3
- | ldrb RAw, [PC, #-3]
+ | ldrb RAw, [PC, #-4+OFS_RA]
| lsl RA, RA, #3
| add TMP1, RA, #16
|3:
@@ -1732,7 +1757,20 @@ static void build_subroutines(BuildCtx *ctx)
|//-----------------------------------------------------------------------
|
|->vm_record: // Dispatch target for recording phase.
- | NYI
+ |.if JIT
+ | ldrb CARG1w, GL->hookmask
+ | tst CARG1, #HOOK_VMEVENT // No recording while in vmevent.
+ | bne >5
+ | // Decrement the hookcount for consistency, but always do the call.
+ | ldr CARG2w, GL->hookcount
+ | tst CARG1, #HOOK_ACTIVE
+ | bne >1
+ | sub CARG2w, CARG2w, #1
+ | tst CARG1, #LUA_MASKLINE|LUA_MASKCOUNT
+ | beq >1
+ | str CARG2w, GL->hookcount
+ | b >1
+ |.endif
|
|->vm_rethook: // Dispatch target for return hooks.
| ldrb TMP2w, GL->hookmask
@@ -1774,7 +1812,21 @@ static void build_subroutines(BuildCtx *ctx)
| b <4
|
|->vm_hotloop: // Hot loop counter underflow.
- | NYI
+ |.if JIT
+ | ldr LFUNC:CARG3, [BASE, FRAME_FUNC] // Same as curr_topL(L).
+ | add CARG1, GL, #GG_G2DISP+GG_DISP2J
+ | and LFUNC:CARG3, CARG3, #LJ_GCVMASK
+ | str PC, SAVE_PC
+ | ldr CARG3, LFUNC:CARG3->pc
+ | mov CARG2, PC
+ | str L, [GL, #GL_J(L)]
+ | ldrb CARG3w, [CARG3, #PC2PROTO(framesize)]
+ | str BASE, L->base
+ | add CARG3, BASE, CARG3, lsl #3
+ | str CARG3, L->top
+ | bl extern lj_trace_hot // (jit_State *J, const BCIns *pc)
+ | b <3
+ |.endif
|
|->vm_callhook: // Dispatch target for call hooks.
| mov CARG2, PC
@@ -1804,7 +1856,54 @@ static void build_subroutines(BuildCtx *ctx)
| br CRET1
|
|->cont_stitch: // Trace stitching.
- | NYI
+ |.if JIT
+ | // RA = resultptr, CARG4 = meta base
+ | ldr RBw, SAVE_MULTRES
+ | ldr INSw, [PC, #-4]
+ | ldr TRACE:CARG3, [CARG4, #-40] // Save previous trace.
+ | subs RB, RB, #8
+ | decode_RA RC, INS // Call base.
+ | and CARG3, CARG3, #LJ_GCVMASK
+ | beq >2
+ |1: // Move results down.
+ | ldr CARG1, [RA]
+ | add RA, RA, #8
+ | subs RB, RB, #8
+ | str CARG1, [BASE, RC, lsl #3]
+ | add RC, RC, #1
+ | bne <1
+ |2:
+ | decode_RA RA, INS
+ | decode_RB RB, INS
+ | add RA, RA, RB
+ |3:
+ | cmp RA, RC
+ | bhi >9 // More results wanted?
+ |
+ | ldrh RAw, TRACE:CARG3->traceno
+ | ldrh RCw, TRACE:CARG3->link
+ | cmp RCw, RAw
+ | beq ->cont_nop // Blacklisted.
+ | cmp RCw, #0
+ | bne =>BC_JLOOP // Jump to stitched trace.
+ |
+ | // Stitch a new trace to the previous trace.
+ | mov CARG1, #GL_J(exitno)
+ | str RAw, [GL, CARG1]
+ | mov CARG1, #GL_J(L)
+ | str L, [GL, CARG1]
+ | str BASE, L->base
+ | add CARG1, GL, #GG_G2J
+ | mov CARG2, PC
+ | bl extern lj_dispatch_stitch // (jit_State *J, const BCIns *pc)
+ | ldr BASE, L->base
+ | b ->cont_nop
+ |
+ |9: // Fill up results with nil.
+ | str TISNIL, [BASE, RC, lsl #3]
+ | add RC, RC, #1
+ | b <3
+ |.endif
|
|->vm_profhook: // Dispatch target for profiler hook.
#if LJ_HASPROFILE
@@ -1822,10 +1921,122 @@ static void build_subroutines(BuildCtx *ctx)
|//-- Trace exit handler -------------------------------------------------
|//-----------------------------------------------------------------------
|
+ |.macro savex_, a, b
+ | stp d..a, d..b, [sp, #a*8]
+ | stp x..a, x..b, [sp, #32*8+a*8]
+ |.endmacro
+ |
|->vm_exit_handler:
- | NYI
+ |.if JIT
+ | sub sp, sp, #(64*8)
+ | savex_, 0, 1
+ | savex_, 2, 3
+ | savex_, 4, 5
+ | savex_, 6, 7
+ | savex_, 8, 9
+ | savex_, 10, 11
+ | savex_, 12, 13
+ | savex_, 14, 15
+ | savex_, 16, 17
+ | savex_, 18, 19
+ | savex_, 20, 21
+ | savex_, 22, 23
+ | savex_, 24, 25
+ | savex_, 26, 27
+ | savex_, 28, 29
+ | stp d30, d31, [sp, #30*8]
+ | ldr CARG1, [sp, #64*8] // Load original value of lr.
+ | add CARG3, sp, #64*8 // Recompute original value of sp.
+ | mv_vmstate CARG4w, EXIT
+ | stp xzr, CARG3, [sp, #62*8] // Store 0/sp in RID_LR/RID_SP.
+ | sub CARG1, CARG1, lr
+ | ldr L, GL->cur_L
+ | lsr CARG1, CARG1, #2
+ | ldr BASE, GL->jit_base
+ | sub CARG1, CARG1, #2
+ | ldr CARG2w, [lr] // Load trace number.
+ | st_vmstate CARG4w
+ |.if ENDIAN_BE
+ | rev32 CARG2, CARG2
+ |.endif
+ | str BASE, L->base
+ | ubfx CARG2w, CARG2w, #5, #16
+ | str CARG1w, [GL, #GL_J(exitno)]
+ | str CARG2w, [GL, #GL_J(parent)]
+ | str L, [GL, #GL_J(L)]
+ | str xzr, GL->jit_base
+ | add CARG1, GL, #GG_G2J
+ | mov CARG2, sp
+ | bl extern lj_trace_exit // (jit_State *J, ExitState *ex)
+ | // Returns MULTRES (unscaled) or negated error code.
+ | ldr CARG2, L->cframe
+ | ldr BASE, L->base
+ | and sp, CARG2, #CFRAME_RAWMASK
+ | ldr PC, SAVE_PC // Get SAVE_PC.
+ | str L, SAVE_L // Set SAVE_L (on-trace resume/yield).
+ | b >1
+ |.endif
+ |
|->vm_exit_interp:
- | NYI
+ | // CARG1 = MULTRES or negated error code, BASE, PC and GL set.
+ |.if JIT
+ | ldr L, SAVE_L
+ |1:
+ | cmp CARG1w, #0
+ | blt >9 // Check for error from exit.
+ | lsl RC, CARG1, #3
+ | ldr LFUNC:CARG2, [BASE, FRAME_FUNC]
+ | movz TISNUM, #(LJ_TISNUM>>1)&0xffff, lsl #48
+ | movz TISNUMhi, #(LJ_TISNUM>>1)&0xffff, lsl #16
+ | movn TISNIL, #0
+ | and LFUNC:CARG2, CARG2, #LJ_GCVMASK
+ | str RCw, SAVE_MULTRES
+ | str BASE, L->base
+ | ldr CARG2, LFUNC:CARG2->pc
+ | str xzr, GL->jit_base
+ | mv_vmstate CARG4w, INTERP
+ | ldr KBASE, [CARG2, #PC2PROTO(k)]
+ | // Modified copy of ins_next which handles function header dispatch, too.
+ | ldrb RBw, [PC, # OFS_OP]
+ | ldr INSw, [PC], #4
+ | st_vmstate CARG4w
+ | cmp RBw, #BC_FUNCC+2 // Fast function?
+ | add TMP1, GL, INS, uxtb #3
+ | bhs >4
+ |2:
+ | cmp RBw, #BC_FUNCF // Function header?
+ | add TMP0, GL, RB, uxtb #3
+ | ldr RB, [TMP0, #GG_G2DISP]
+ | decode_RA RA, INS
+ | lsr TMP0, INS, #16
+ | csel RC, TMP0, RC, lo
+ | blo >5
+ | ldr CARG3, [BASE, FRAME_FUNC]
+ | sub RC, RC, #8
+ | add RA, BASE, RA, lsl #3 // Yes: RA = BASE+framesize*8, RC = nargs*8
+ | and LFUNC:CARG3, CARG3, #LJ_GCVMASK
+ |5:
+ | br RB
+ |
+ |4: // Check frame below fast function.
+ | ldr CARG1, [BASE, FRAME_PC]
+ | ands CARG2, CARG1, #FRAME_TYPE
+ | bne <2 // Trace stitching continuation?
+ | // Otherwise set KBASE for Lua function below fast function.
+ | ldr CARG3w, [CARG1, #-4]
+ | decode_RA CARG1, CARG3
+ | sub CARG2, BASE, CARG1, lsl #3
+ | ldr LFUNC:CARG3, [CARG2, #-32]
+ | and LFUNC:CARG3, CARG3, #LJ_GCVMASK
+ | ldr CARG3, LFUNC:CARG3->pc
+ | ldr KBASE, [CARG3, #PC2PROTO(k)]
+ | b <2
+ |
+ |9: // Rethrow error from the right C frame.
+ | neg CARG2, CARG1
+ | mov CARG1, L
+ | bl extern lj_err_throw // (lua_State *L, int errcode)
+ |.endif
|
|//-----------------------------------------------------------------------
|//-- Math helper functions ----------------------------------------------
@@ -1965,7 +2176,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
case BC_ISLT: case BC_ISGE: case BC_ISLE: case BC_ISGT:
| // RA = src1, RC = src2, JMP with RC = target
| ldr CARG1, [BASE, RA, lsl #3]
- | ldrh RBw, [PC, #2]
+ | ldrh RBw, [PC, # OFS_RD]
| ldr CARG2, [BASE, RC, lsl #3]
| add PC, PC, #4
| add RB, PC, RB, lsl #2
@@ -2022,7 +2233,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| // RA = src1, RC = src2, JMP with RC = target
| ldr CARG1, [BASE, RA, lsl #3]
| add RC, BASE, RC, lsl #3
- | ldrh RBw, [PC, #2]
+ | ldrh RBw, [PC, # OFS_RD]
| ldr CARG3, [RC]
| add PC, PC, #4
| add RB, PC, RB, lsl #2
@@ -2083,7 +2294,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| // RA = src, RC = str_const (~), JMP with RC = target
| ldr CARG1, [BASE, RA, lsl #3]
| mvn RC, RC
- | ldrh RBw, [PC, #2]
+ | ldrh RBw, [PC, # OFS_RD]
| ldr CARG2, [KBASE, RC, lsl #3]
| add PC, PC, #4
| movn TMP0, #~LJ_TSTR
@@ -2111,7 +2322,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| // RA = src, RC = num_const (~), JMP with RC = target
| ldr CARG1, [BASE, RA, lsl #3]
| add RC, KBASE, RC, lsl #3
- | ldrh RBw, [PC, #2]
+ | ldrh RBw, [PC, # OFS_RD]
| ldr CARG3, [RC]
| add PC, PC, #4
| add RB, PC, RB, lsl #2
@@ -2171,7 +2382,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
vk = op == BC_ISEQP;
| // RA = src, RC = primitive_type (~), JMP with RC = target
| ldr TMP0, [BASE, RA, lsl #3]
- | ldrh RBw, [PC, #2]
+ | ldrh RBw, [PC, # OFS_RD]
| add PC, PC, #4
| add RC, RC, #1
| add RB, PC, RB, lsl #2
@@ -2196,7 +2407,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
case BC_ISTC: case BC_ISFC: case BC_IST: case BC_ISF:
| // RA = dst or unused, RC = src, JMP with RC = target
- | ldrh RBw, [PC, #2]
+ | ldrh RBw, [PC, # OFS_RD]
| ldr TMP0, [BASE, RC, lsl #3]
| add PC, PC, #4
| mov_false TMP1
@@ -2443,7 +2654,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| str PC, SAVE_PC
| bl extern lj_meta_cat // (lua_State *L, TValue *top, int left)
| // Returns NULL (finished) or TValue * (metamethod).
- | ldrb RBw, [PC, #-1]
+ | ldrb RBw, [PC, #-4+OFS_RB]
| ldr BASE, L->base
| cbnz CRET1, ->vmeta_binop
| ldr TMP0, [BASE, RB, lsl #3]
@@ -3074,7 +3285,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| ins_callt
|
|5: // Tailcall to a fast function with a Lua frame below.
- | ldrb RAw, [PC, #-3]
+ | ldrb RAw, [PC, #-4+OFS_RA]
| sub CARG1, BASE, RA, lsl #3
| ldr LFUNC:CARG1, [CARG1, #-32]
| and LFUNC:CARG1, CARG1, #LJ_GCVMASK
@@ -3115,8 +3326,8 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
|.endif
| add RA, BASE, RA, lsl #3
| ldr TAB:RB, [RA, #-16]
- | ldrh TMP3w, [PC, #2]
- | ldr CARG1w, [RA, #-8] // Get index from control var.
+ | ldrh TMP3w, [PC, # OFS_RD]
+ | ldr CARG1w, [RA, #-8+LO] // Get index from control var.
| add PC, PC, #4
| add TMP3, PC, TMP3, lsl #2
| and TAB:RB, RB, #LJ_GCVMASK
@@ -3135,7 +3346,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| stp CARG1, TMP0, [RA]
| add CARG1, CARG1, #1
|3:
- | str CARG1w, [RA, #-8] // Update control var.
+ | str CARG1w, [RA, #-8+LO] // Update control var.
| mov PC, TMP3
|4:
| ins_next
@@ -3181,8 +3392,8 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
|5: // Despecialize bytecode if any of the checks fail.
| mov TMP0, #BC_JMP
| mov TMP1, #BC_ITERC
- | strb TMP0w, [PC, #-4]
- | strb TMP1w, [RC]
+ | strb TMP0w, [PC, #-4+OFS_OP]
+ | strb TMP1w, [RC, # OFS_OP]
| b <1
break;
@@ -3387,7 +3598,8 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
if (op == BC_FORI) {
| csel PC, RC, PC, gt
} else if (op == BC_JFORI) {
- | ldrh RCw, [RC, #-2]
+ | mov PC, RC
+ | ldrh RCw, [RC, #-4+OFS_RD]
} else if (op == BC_IFORL) {
| csel PC, RC, PC, le
}
@@ -3428,7 +3640,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
if (op == BC_FORI) {
| csel PC, RC, PC, hi
} else if (op == BC_JFORI) {
- | ldrh RCw, [RC, #-2]
+ | ldrh RCw, [RC, #-4+OFS_RD]
| bls =>BC_JLOOP
} else if (op == BC_IFORL) {
| csel PC, RC, PC, ls
@@ -3488,7 +3700,16 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
case BC_JLOOP:
|.if JIT
- | NYI
+ | // RA = base (ignored), RC = traceno
+ | ldr CARG1, [GL, #GL_J(trace)]
+ | mov CARG2w, #0 // Traces on ARM64 don't store the trace #, so use 0.
+ | ldr TRACE:RC, [CARG1, RC, lsl #3]
+ | st_vmstate CARG2w
+ | ldr RA, TRACE:RC->mcode
+ | str BASE, GL->jit_base
+ | str L, GL->tmpbuf.L
+ | sub sp, sp, #16 // See SPS_FIXED. Avoids sp adjust in every root trace.
+ | br RA
|.endif
break;
@@ -3546,10 +3767,12 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
case BC_IFUNCV:
| // BASE = new base, RA = BASE+framesize*8, CARG3 = LFUNC, RC = nargs*8
| ldr CARG1, L->maxstack
+ | movn TMP0, #~LJ_TFUNC
| add TMP2, BASE, RC
+ | add LFUNC:CARG3, CARG3, TMP0, lsl #47
| add RA, RA, RC
| add TMP0, RC, #16+FRAME_VARG
- | str LFUNC:CARG3, [TMP2], #8 // Store (untagged) copy of LFUNC.
+ | str LFUNC:CARG3, [TMP2], #8 // Store (tagged) copy of LFUNC.
| ldr KBASE, [PC, #-4+PC2PROTO(k)]
| cmp RA, CARG1
| str TMP0, [TMP2], #8 // Store delta + FRAME_VARG.
@@ -3736,8 +3959,8 @@ static void emit_asm_debug(BuildCtx *ctx)
"\t.uleb128 0x1\n"
"\t.sleb128 -8\n"
"\t.byte 30\n" /* Return address is in lr. */
- "\t.uleb128 1\n" /* augmentation length */
- "\t.byte 0x1b\n" /* pcrel|sdata4 */
+ "\t.uleb128 1\n" /* augmentation length */
+ "\t.byte 0x1b\n" /* pcrel|sdata4 */
"\t.byte 0xc\n\t.uleb128 31\n\t.uleb128 0\n" /* def_cfa sp */
"\t.align 3\n"
".LECIE2:\n\n");
@@ -3748,7 +3971,7 @@ static void emit_asm_debug(BuildCtx *ctx)
"\t.long .LASFDE3-.Lframe2\n"
"\t.long lj_vm_ffi_call-.\n"
"\t.long %d\n"
- "\t.uleb128 0\n" /* augmentation length */
+ "\t.uleb128 0\n" /* augmentation length */
"\t.byte 0xe\n\t.uleb128 32\n" /* def_cfa_offset */
"\t.byte 0x9d\n\t.uleb128 4\n" /* offset fp */
"\t.byte 0x9e\n\t.uleb128 3\n" /* offset lr */
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/vm_mips.dasc b/Build/source/libs/luajit/LuaJIT-src/src/vm_mips.dasc
index 03b3084e88a..1afd61187a7 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/vm_mips.dasc
+++ b/Build/source/libs/luajit/LuaJIT-src/src/vm_mips.dasc
@@ -1,6 +1,6 @@
|// Low-level VM code for MIPS CPUs.
|// Bytecode interpreter, fast functions and helper functions.
-|// Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+|// Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
|//
|// MIPS soft-float support contributed by Djordje Kovacevic and
|// Stefan Pejic from RT-RK.com, sponsored by Cisco Systems, Inc.
@@ -57,7 +57,7 @@
|.define TMP2, r14
|.define TMP3, r15
|
-|// Calling conventions.
+|// MIPS o32 calling convention.
|.define CFUNCADDR, r25
|.define CARG1, r4
|.define CARG2, r5
@@ -1875,7 +1875,8 @@ static void build_subroutines(BuildCtx *ctx)
|
|.ffunc string_char // Only handle the 1-arg case here.
| ffgccheck
- |. lw CARG3, HI(BASE)
+ |. nop
+ | lw CARG3, HI(BASE)
| lw CARG1, LO(BASE)
| li TMP1, 255
| xori AT, NARGS8:RC, 8 // Exactly 1 argument.
@@ -1902,7 +1903,8 @@ static void build_subroutines(BuildCtx *ctx)
|
|.ffunc string_sub
| ffgccheck
- |. addiu AT, NARGS8:RC, -16
+ |. nop
+ | addiu AT, NARGS8:RC, -16
| lw CARG3, 16+HI(BASE)
| lw TMP0, HI(BASE)
| lw STR:CARG1, LO(BASE)
@@ -1946,7 +1948,8 @@ static void build_subroutines(BuildCtx *ctx)
|.macro ffstring_op, name
| .ffunc string_ .. name
| ffgccheck
- |. lw CARG3, HI(BASE)
+ |. nop
+ | lw CARG3, HI(BASE)
| lw STR:CARG2, LO(BASE)
| beqz NARGS8:RC, ->fff_fallback
|. li AT, LJ_TSTR
@@ -4314,7 +4317,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| ins_next2
|
|7: // Possible table write barrier for the value. Skip valiswhite check.
- | barrierback TAB:RB, TMP3, TMP0, <2
+ | barrierback TAB:CARG2, TMP3, TMP0, <2
break;
case BC_TSETM:
@@ -4543,24 +4546,24 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
case BC_ISNEXT:
| // RA = base*8, RD = target (points to ITERN)
| addu RA, BASE, RA
- | lw TMP0, -24+HI(RA)
- | lw CFUNC:TMP1, -24+LO(RA)
- | lw TMP2, -16+HI(RA)
- | lw TMP3, -8+HI(RA)
+ | srl TMP0, RD, 1
+ | lw CARG1, -24+HI(RA)
+ | lw CFUNC:CARG2, -24+LO(RA)
+ | addu TMP0, PC, TMP0
+ | lw CARG3, -16+HI(RA)
+ | lw CARG4, -8+HI(RA)
| li AT, LJ_TFUNC
- | bne TMP0, AT, >5
- |. addiu TMP2, TMP2, -LJ_TTAB
- | lbu TMP1, CFUNC:TMP1->ffid
- | addiu TMP3, TMP3, -LJ_TNIL
- | srl TMP0, RD, 1
- | or TMP2, TMP2, TMP3
- | addiu TMP1, TMP1, -FF_next_N
- | addu TMP0, PC, TMP0
- | or TMP1, TMP1, TMP2
- | bnez TMP1, >5
- |. lui TMP2, (-(BCBIAS_J*4 >> 16) & 65535)
+ | bne CARG1, AT, >5
+ |. lui TMP2, (-(BCBIAS_J*4 >> 16) & 65535)
+ | lbu CARG2, CFUNC:CARG2->ffid
+ | addiu CARG3, CARG3, -LJ_TTAB
+ | addiu CARG4, CARG4, -LJ_TNIL
+ | or CARG3, CARG3, CARG4
+ | addiu CARG2, CARG2, -FF_next_N
+ | or CARG2, CARG2, CARG3
+ | bnez CARG2, >5
+ |. lui TMP1, 0xfffe
| addu PC, TMP0, TMP2
- | lui TMP1, 0xfffe
| ori TMP1, TMP1, 0x7fff
| sw r0, -8+LO(RA) // Initialize control var.
| sw TMP1, -8+HI(RA)
@@ -4570,7 +4573,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| li TMP3, BC_JMP
| li TMP1, BC_ITERC
| sb TMP3, -4+OFS_OP(PC)
- | addu PC, TMP0, TMP2
+ | addu PC, TMP0, TMP2
| b <1
|. sb TMP1, OFS_OP(PC)
break;
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/vm_ppc.dasc b/Build/source/libs/luajit/LuaJIT-src/src/vm_ppc.dasc
index be4356e7480..b4260ebc107 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/vm_ppc.dasc
+++ b/Build/source/libs/luajit/LuaJIT-src/src/vm_ppc.dasc
@@ -1,6 +1,6 @@
|// Low-level VM code for PowerPC 32 bit or 32on64 bit mode.
|// Bytecode interpreter, fast functions and helper functions.
-|// Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+|// Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
|
|.arch ppc
|.section code_op, code_sub
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/vm_x64.dasc b/Build/source/libs/luajit/LuaJIT-src/src/vm_x64.dasc
index a6b6853e1e9..a003fb4f6bf 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/vm_x64.dasc
+++ b/Build/source/libs/luajit/LuaJIT-src/src/vm_x64.dasc
@@ -1,6 +1,6 @@
|// Low-level VM code for x64 CPUs in LJ_GC64 mode.
|// Bytecode interpreter, fast functions and helper functions.
-|// Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+|// Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
|
|.arch x64
|.section code_op, code_sub
@@ -250,11 +250,11 @@
|// Macros to clear or set tags.
|.macro cleartp, reg; shl reg, 17; shr reg, 17; .endmacro
|.macro settp, reg, tp
-| mov64 ITYPE, ((int64_t)tp<<47)
+| mov64 ITYPE, ((uint64_t)tp<<47)
| or reg, ITYPE
|.endmacro
|.macro settp, dst, reg, tp
-| mov64 dst, ((int64_t)tp<<47)
+| mov64 dst, ((uint64_t)tp<<47)
| or dst, reg
|.endmacro
|.macro setint, reg
@@ -1105,11 +1105,11 @@ static void build_subroutines(BuildCtx *ctx)
| mov BASE, L:RB->base
| mov NARGS:RDd, TMP1d
| mov LFUNC:RB, [RA-16]
- | cleartp LFUNC:RB
| add NARGS:RDd, 1
| // This is fragile. L->base must not move, KBASE must always be defined.
| cmp KBASE, BASE // Continue with CALLT if flag set.
| je ->BC_CALLT_Z
+ | cleartp LFUNC:RB
| mov BASE, RA
| ins_call // Otherwise call resolved metamethod.
|
@@ -1469,7 +1469,7 @@ static void build_subroutines(BuildCtx *ctx)
| mov [BASE-16], CFUNC:RD
| mov [BASE-8], TMPR
|.if DUALNUM
- | mov64 RD, ((int64_t)LJ_TISNUM<<47)
+ | mov64 RD, ((uint64_t)LJ_TISNUM<<47)
| mov [BASE], RD
|.else
| mov qword [BASE], 0
@@ -1804,12 +1804,12 @@ static void build_subroutines(BuildCtx *ctx)
| jmp ->fff_res1
|
|.ffunc_n math_frexp
+ | mov RB, BASE
|.if X64WIN
- | lea CARG2, TMP1
+ | lea CARG2, TMP1 // Caveat: CARG2 == BASE
|.else
| lea CARG1, TMP1
|.endif
- | mov RB, BASE
| call extern frexp
| mov BASE, RB
| mov RBd, TMP1d
@@ -1826,13 +1826,12 @@ static void build_subroutines(BuildCtx *ctx)
| jmp ->fff_res
|
|.ffunc_n math_modf
+ | mov RB, BASE
|.if X64WIN
- | lea CARG2, [BASE-16]
+ | lea CARG2, [BASE-16] // Caveat: CARG2 == BASE
|.else
| lea CARG1, [BASE-16]
|.endif
- | mov PC, [BASE-8]
- | mov RB, BASE
| call extern modf
| mov BASE, RB
| mov PC, [BASE-8]
@@ -2325,7 +2324,8 @@ static void build_subroutines(BuildCtx *ctx)
|->cont_stitch: // Trace stitching.
|.if JIT
| // BASE = base, RC = result, RB = mbase
- | mov ITYPEd, [RB-24] // Save previous trace number.
+ | mov TRACE:ITYPE, [RB-40] // Save previous trace.
+ | cleartp TRACE:ITYPE
| mov TMPRd, MULTRES
| movzx RAd, PC_RA
| lea RA, [BASE+RA*8] // Call base.
@@ -2347,11 +2347,10 @@ static void build_subroutines(BuildCtx *ctx)
| cmp RC, RA
| ja >9 // More results wanted?
|
- | mov RA, [DISPATCH+DISPATCH_J(trace)]
- | mov TRACE:RD, [RA+ITYPE*8]
- | test TRACE:RD, TRACE:RD
+ | test TRACE:ITYPE, TRACE:ITYPE
| jz ->cont_nop
- | movzx RDd, word TRACE:RD->link
+ | movzx RBd, word TRACE:ITYPE->traceno
+ | movzx RDd, word TRACE:ITYPE->link
| cmp RDd, RBd
| je ->cont_nop // Blacklisted.
| test RDd, RDd
@@ -2402,12 +2401,11 @@ static void build_subroutines(BuildCtx *ctx)
| movzx RCd, byte [rbp-8] // Reconstruct exit number.
| mov RCH, byte [rbp-16]
| mov [rbp-8], r15; mov [rbp-16], r14
- | // Caveat: DISPATCH is rbx.
- | mov DISPATCH, [ebp]
- | mov RA, [DISPATCH+DISPATCH_GL(vmstate)] // Get trace number.
+ | // DISPATCH is preserved on-trace in LJ_GC64 mode.
+ | mov RAd, [DISPATCH+DISPATCH_GL(vmstate)] // Get trace number.
| set_vmstate EXIT
- | mov [DISPATCH+DISPATCH_J(exitno)], RC
- | mov [DISPATCH+DISPATCH_J(parent)], RA
+ | mov [DISPATCH+DISPATCH_J(exitno)], RCd
+ | mov [DISPATCH+DISPATCH_J(parent)], RAd
|.if X64WIN
| sub rsp, 16*8+4*8 // Room for SSE regs + save area.
|.else
@@ -2433,7 +2431,7 @@ static void build_subroutines(BuildCtx *ctx)
| mov CARG2, rsp
|.endif
| lea CARG1, [DISPATCH+GG_DISP2J]
- | mov dword [DISPATCH+DISPATCH_GL(jit_base)], 0
+ | mov qword [DISPATCH+DISPATCH_GL(jit_base)], 0
| call extern lj_trace_exit // (jit_State *J, ExitState *ex)
| // MULTRES or negated error code returned in eax (RD).
| mov RA, L:RB->cframe
@@ -2480,7 +2478,7 @@ static void build_subroutines(BuildCtx *ctx)
| mov KBASE, LFUNC:KBASE->pc
| mov KBASE, [KBASE+PC2PROTO(k)]
| mov L:RB->base, BASE
- | mov dword [DISPATCH+DISPATCH_GL(jit_base)], 0
+ | mov qword [DISPATCH+DISPATCH_GL(jit_base)], 0
| set_vmstate INTERP
| // Modified copy of ins_next which handles function header dispatch, too.
| mov RCd, [PC]
@@ -2504,7 +2502,7 @@ static void build_subroutines(BuildCtx *ctx)
| // Otherwise set KBASE for Lua function below fast function.
| movzx RCd, byte [RC-3]
| neg RC
- | mov LFUNC:KBASE, [BASE+RC*8-24]
+ | mov LFUNC:KBASE, [BASE+RC*8-32]
| cleartp LFUNC:KBASE
| mov KBASE, LFUNC:KBASE->pc
| mov KBASE, [KBASE+PC2PROTO(k)]
@@ -2639,6 +2637,7 @@ static void build_subroutines(BuildCtx *ctx)
| mov eax, CARG1d
| .if X64WIN; push rsi; mov rsi, CARG2; .endif
| push rbx
+ | xor ecx, ecx
| cpuid
| mov [rsi], eax
| mov [rsi+4], ebx
@@ -3516,7 +3515,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| ins_AD // RA = level, RD = target
| branchPC RD // Do this first to free RD.
| mov L:RB, SAVE_L
- | cmp dword L:RB->openupval, 0
+ | cmp aword L:RB->openupval, 0
| je >1
| mov L:RB->base, BASE
| lea CARG2, [BASE+RA*8] // Caveat: CARG2 == BASE
@@ -4050,7 +4049,7 @@ static void build_ins(BuildCtx *ctx, BCOp op, int defop)
| mov RC, [RA-24] // Copy control var. fb[1] = fb[-3].
| mov [RA], RB
| mov [RA+8], RC
- | mov LFUNC:RB, [RA-40] // Copy callable. fb[-1] = fb[-5]
+ | mov LFUNC:RB, [RA-40] // Copy callable. fb[-2] = fb[-5]
| mov [RA-16], LFUNC:RB
| mov NARGS:RDd, 2+1 // Handle like a regular 2-arg call.
| checkfunc LFUNC:RB, ->vmeta_call
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/vm_x86.dasc b/Build/source/libs/luajit/LuaJIT-src/src/vm_x86.dasc
index fcef660c420..211ae7b922a 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/vm_x86.dasc
+++ b/Build/source/libs/luajit/LuaJIT-src/src/vm_x86.dasc
@@ -1,6 +1,6 @@
|// Low-level VM code for x86 CPUs.
|// Bytecode interpreter, fast functions and helper functions.
-|// Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
+|// Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
|
|.if P64
|.arch x64
@@ -121,19 +121,68 @@
|//-----------------------------------------------------------------------
|.if not X64 // x86 stack layout.
|
-|.define CFRAME_SPACE, aword*7 // Delta for esp (see <--).
+|.if WIN
+|
+|.define CFRAME_SPACE, aword*9 // Delta for esp (see <--).
|.macro saveregs_
| push edi; push esi; push ebx
+| push extern lj_err_unwind_win
+| fs; push dword [0]
+| fs; mov [0], esp
| sub esp, CFRAME_SPACE
|.endmacro
-|.macro saveregs
-| push ebp; saveregs_
+|.macro restoreregs
+| add esp, CFRAME_SPACE
+| fs; pop dword [0]
+| pop edi // Short for esp += 4.
+| pop ebx; pop esi; pop edi; pop ebp
+|.endmacro
+|
+|.else
+|
+|.define CFRAME_SPACE, aword*7 // Delta for esp (see <--).
+|.macro saveregs_
+| push edi; push esi; push ebx
+| sub esp, CFRAME_SPACE
|.endmacro
|.macro restoreregs
| add esp, CFRAME_SPACE
| pop ebx; pop esi; pop edi; pop ebp
|.endmacro
|
+|.endif
+|
+|.macro saveregs
+| push ebp; saveregs_
+|.endmacro
+|
+|.if WIN
+|.define SAVE_ERRF, aword [esp+aword*19] // vm_pcall/vm_cpcall only.
+|.define SAVE_NRES, aword [esp+aword*18]
+|.define SAVE_CFRAME, aword [esp+aword*17]
+|.define SAVE_L, aword [esp+aword*16]
+|//----- 16 byte aligned, ^^^ arguments from C caller
+|.define SAVE_RET, aword [esp+aword*15] //<-- esp entering interpreter.
+|.define SAVE_R4, aword [esp+aword*14]
+|.define SAVE_R3, aword [esp+aword*13]
+|.define SAVE_R2, aword [esp+aword*12]
+|//----- 16 byte aligned
+|.define SAVE_R1, aword [esp+aword*11]
+|.define SEH_FUNC, aword [esp+aword*10]
+|.define SEH_NEXT, aword [esp+aword*9] //<-- esp after register saves.
+|.define UNUSED2, aword [esp+aword*8]
+|//----- 16 byte aligned
+|.define UNUSED1, aword [esp+aword*7]
+|.define SAVE_PC, aword [esp+aword*6]
+|.define TMP2, aword [esp+aword*5]
+|.define TMP1, aword [esp+aword*4]
+|//----- 16 byte aligned
+|.define ARG4, aword [esp+aword*3]
+|.define ARG3, aword [esp+aword*2]
+|.define ARG2, aword [esp+aword*1]
+|.define ARG1, aword [esp] //<-- esp while in interpreter.
+|//----- 16 byte aligned, ^^^ arguments for C callee
+|.else
|.define SAVE_ERRF, aword [esp+aword*15] // vm_pcall/vm_cpcall only.
|.define SAVE_NRES, aword [esp+aword*14]
|.define SAVE_CFRAME, aword [esp+aword*13]
@@ -154,6 +203,7 @@
|.define ARG2, aword [esp+aword*1]
|.define ARG1, aword [esp] //<-- esp while in interpreter.
|//----- 16 byte aligned, ^^^ arguments for C callee
+|.endif
|
|// FPARGx overlaps ARGx and ARG(x+1) on x86.
|.define FPARG3, qword [esp+qword*1]
@@ -554,6 +604,10 @@ static void build_subroutines(BuildCtx *ctx)
|.else
| mov eax, FCARG2 // Error return status for vm_pcall.
| mov esp, FCARG1
+ |.if WIN
+ | lea FCARG1, SEH_NEXT
+ | fs; mov [0], FCARG1
+ |.endif
|.endif
|->vm_unwind_c_eh: // Landing pad for external unwinder.
| mov L:RB, SAVE_L
@@ -577,6 +631,10 @@ static void build_subroutines(BuildCtx *ctx)
|.else
| and FCARG1, CFRAME_RAWMASK
| mov esp, FCARG1
+ |.if WIN
+ | lea FCARG1, SEH_NEXT
+ | fs; mov [0], FCARG1
+ |.endif
|.endif
|->vm_unwind_ff_eh: // Landing pad for external unwinder.
| mov L:RB, SAVE_L
@@ -590,6 +648,19 @@ static void build_subroutines(BuildCtx *ctx)
| set_vmstate INTERP
| jmp ->vm_returnc // Increments RD/MULTRES and returns.
|
+ |.if WIN and not X64
+ |->vm_rtlunwind@16: // Thin layer around RtlUnwind.
+ | // (void *cframe, void *excptrec, void *unwinder, int errcode)
+ | mov [esp], FCARG1 // Return value for RtlUnwind.
+ | push FCARG2 // Exception record for RtlUnwind.
+ | push 0 // Ignored by RtlUnwind.
+ | push dword [FCARG1+CFRAME_OFS_SEH]
+ | call extern RtlUnwind@16 // Violates ABI (clobbers too much).
+ | mov FCARG1, eax
+ | mov FCARG2, [esp+4] // errcode (for vm_unwind_c).
+ | ret // Jump to unwinder.
+ |.endif
+ |
|//-----------------------------------------------------------------------
|//-- Grow stack for calls -----------------------------------------------
|//-----------------------------------------------------------------------
@@ -3026,6 +3097,7 @@ static void build_subroutines(BuildCtx *ctx)
| mov eax, CARG1d
| .if X64WIN; push rsi; mov rsi, CARG2; .endif
| push rbx
+ | xor ecx, ecx
| cpuid
| mov [rsi], eax
| mov [rsi+4], ebx
@@ -3049,6 +3121,7 @@ static void build_subroutines(BuildCtx *ctx)
| mov eax, [esp+4] // Argument 1 is function number.
| push edi
| push ebx
+ | xor ecx, ecx
| cpuid
| mov edi, [esp+16] // Argument 2 is result area.
| mov [edi], eax