diff options
| author | Karl Berry <karl@freefriends.org> | 2017-11-17 23:38:04 +0000 |
|---|---|---|
| committer | Karl Berry <karl@freefriends.org> | 2017-11-17 23:38:04 +0000 |
| commit | 82d58e473ee47e5ad8f3d6989347fb95ab7d36ea (patch) | |
| tree | e5ce320ac5e7f3cd932731de3877c522b65fe5f6 /Build | |
| parent | c3ade43e7a77c7ba8035aeb2b592b2f6ab0795fc (diff) | |
(elt_in_db): protect against pathological arguments
git-svn-id: svn://tug.org/texlive/trunk@45841 c570f23f-e606-0410-a88d-b1316a301751
Diffstat (limited to 'Build')
| -rw-r--r-- | Build/source/texk/kpathsea/ChangeLog | 5 | ||||
| -rw-r--r-- | Build/source/texk/kpathsea/db.c | 10 |
2 files changed, 15 insertions, 0 deletions
diff --git a/Build/source/texk/kpathsea/ChangeLog b/Build/source/texk/kpathsea/ChangeLog index c05a72df8f2..1cb0690e38c 100644 --- a/Build/source/texk/kpathsea/ChangeLog +++ b/Build/source/texk/kpathsea/ChangeLog @@ -1,3 +1,8 @@ +2017-11-18 Karl Berry <karl@tug.org> + + * db.c (elt_in_db): protect against either argument being NULL or + empty. From Doug McKenna, personal mail, 17nov17. + 2017-11-11 Akira Kakuto <kakuto@fuk.kindai.ac.jp> * tex-file.c: Again use a custom kpathsea_getlongpath() diff --git a/Build/source/texk/kpathsea/db.c b/Build/source/texk/kpathsea/db.c index 16d4941d81e..fb4658c1968 100644 --- a/Build/source/texk/kpathsea/db.c +++ b/Build/source/texk/kpathsea/db.c @@ -296,6 +296,7 @@ match (const_string filename, const_string path_elt) That is, the question is whether to try the db for a file looked up in PATH_ELT. If PATH_ELT == ".", for example, the answer is no. If PATH_ELT == "/usr/local/lib/texmf/fonts//tfm", the answer is yes. + If either string is NULL or empty, return false. In practice, ls-R is only needed for lengthy subdirectory comparisons, but there's no gain to checking PATH_ELT to see if it is @@ -307,6 +308,15 @@ elt_in_db (const_string db_dir, const_string path_elt) { boolean found = false; + /* If both strings are empty or null return false on the grounds that + it's useless to do anything further with such a strange case (which + likely never happens). In theory one could argue that the empty + string is a prefix of any other string, but let's just declare the + result otherwise. */ + if (db_dir == NULL || *db_dir == 0 + || path_elt == NULL || *path_elt == 0) + return false; + while (!found && FILECHARCASEEQ (*db_dir++, *path_elt++)) { /* If we've matched the entire db directory, it's good. */ if (*db_dir == 0) |