diff options
| author |   jasper <jasper@openbsd.org> | 2011-11-01 13:52:21 +0000 |
|---|---|---|
| committer | jasper <jasper@openbsd.org> | 2011-11-01 13:52:21 +0000 |
| commit | 945a9a587895b1b0229296859cdf38ee3657a42c (patch) | |
| tree | 09792524a15958ed7d97aba0fcd3bfc5992138f3 /gnu/usr.bin/perl/cpan/Encode/Unicode/Unicode.xs | |
| parent | Fixup previous: don't forget to actually get the current rtable; (diff) | |
| download | wireguard-openbsd-945a9a587895b1b0229296859cdf38ee3657a42c.tar.xz wireguard-openbsd-945a9a587895b1b0229296859cdf38ee3657a42c.zip | |
Security fix for CVE-2011-2939, Perl decode_xs heap-based buffer overflow.
ok espie@ millert@
Diffstat (limited to 'gnu/usr.bin/perl/cpan/Encode/Unicode/Unicode.xs')
| -rw-r--r-- | gnu/usr.bin/perl/cpan/Encode/Unicode/Unicode.xs | 5 |
1 files changed, 4 insertions, 1 deletions
diff --git a/gnu/usr.bin/perl/cpan/Encode/Unicode/Unicode.xs b/gnu/usr.bin/perl/cpan/Encode/Unicode/Unicode.xs index 9741626bd31..d4f2e714ca2 100644 --- a/gnu/usr.bin/perl/cpan/Encode/Unicode/Unicode.xs +++ b/gnu/usr.bin/perl/cpan/Encode/Unicode/Unicode.xs @@ -246,7 +246,10 @@ CODE: This prevents allocating too much in the rogue case of a large input consisting initially of long sequence uft8-byte unicode chars followed by single utf8-byte chars. */ - STRLEN remaining = (e - s)/usize; + /* +1 + fixes Unicode.xs!decode_xs n-byte heap-overflow + */ + STRLEN remaining = (e - s)/usize + 1; /* +1 to avoid the leak */ STRLEN max_alloc = remaining + (8*1024*1024); STRLEN est_alloc = remaining * UTF8_MAXLEN; STRLEN newlen = SvLEN(result) + /* min(max_alloc, est_alloc) */ |