47251Re: running valgrind with vim
- Aug 5, 2007Dominique Pelle wrote:
> > > Hi,Thanks for digging into this.
> > >
> > > Valgrind memory checker finds several errors in vim-7.1 (patches 1-50)
> > > when built with --enable-multibyte:
> > >
> > > $ ./configure --enable-multibyte
> > > $ make CFLAGS="-O0 -g"
> > > $ cd src/testdir
> > > $ valgrind ../vim -u unix.vim -U NONE --noplugin -s dotest.in
> > > test45.in 2> valgrind-test45.txt
> > It's a lot of work to check out the warnings, so please make sure there
> > are no false warnings.
> I believe I have a patch to fix errors reported by valgrind in test case 45
> but please review it.
> Error was:
> ==14428== Conditional jump or move depends on uninitialised value(s)
> ==14428== at 0x80FDD10: utfc_ptr2len (mbyte.c:1676)
> ==14428== by 0x813EB6B: win_line (screen.c:3586)
> ==14428== by 0x813BBF0: win_update (screen.c:1760)
> ==14428== by 0x8139FFE: update_screen (screen.c:522)
> ==14428== by 0x80CD1CF: main_loop (main.c:1088)
> ==14428== by 0x80CCF72: main (main.c:939)
> Variable p_extra at screen.c:3586 was initialized by fill_foldcolumn()
> with 'foldcolumn' spaces (3 spaces in test case 45). 'p_extra' pointer
> is actually in this case the local 'extra' array. The rest of 'p_extra'
> buffer is uninitialized. vim_line() then loops on those 3 spaces but
> accesses the next characters inside utfc_ptr2len() when multi_byte
> is enabled, hence causing valgrind to complain when displaying
> the last space.
> Bug can be simply triggered when doing ":set foldcolumn" and when
> vim was built 'multi_byte'. The following command is enough to trigger
> an error with valgrind:
> $ vim -u NONE -c 'set foldcolumn=4' 2> valgrind.log
> I attach the patch to fix the problem. It basically adds NUL character
> at the end of 'foldcolumn' spaces when. This patch actually fixes all errors
> reported by valgrind in test case 45.
> Without patch, bahavior of calling utfc_ptr2len() at line screen.c:3586
> may be undetermined. It's also slower since it may need to call
> utf_ptr2len() whereas with patch, function utfc_ptr2len() returns
> immediately at mbyte.c:1676.
I spotted a few other places where p_extra may point to text that is not
NUL terminated and where utfc_ptr2len() may not work correctly. That
won't give an uninitialized read error, but it may still be wrong.
From "know your smileys":
=):-) Uncle Sam
/// Bram Moolenaar -- Bram@... -- http://www.Moolenaar.net \\\
/// sponsor Vim, vote for features -- http://www.Vim.org/sponsor/ \\\
\\\ download, build and distribute -- http://www.A-A-P.org ///
\\\ help me help AIDS victims -- http://ICCF-Holland.org ///
You received this message from the "vim_dev" maillist.
For more information, visit http://www.vim.org/maillist.php
- << Previous post in topic