This a CC of a mail to email@example.com (I could not find an archive; the given
link is dead):
the attached patch fixes the usage of an uninitialized 'user_groups'
This variable will be initialized by the get_groups() method, which
will not be called in all cases (only when '-G' flag is given). But the
grp_update() function (which uses this variable) can be called in other
cases also (e.g. on RHL systems which are adding user-groups).
Description of problem:
Created attachment 91187 [details]
fixes usage of uninitialized 'user_groups' variable
The combination of glibc-2.3.2 and libsafe causes this to segfault on RH9 and
Mandrake 9.1. We didn't discover this before because libsafe and older glibc
didn't trigger a segfault.
I have confirmed that the segfault no longer happens on RH9 with this patch to
*** Bug 106218 has been marked as a duplicate of this bug. ***
Can we please apply this to rawhide? This is still an issue in FC1.
Could you look at shadow cvs from time to time, please:
2003-06-30 Tomasz Kloczko <firstname.lastname@example.org>
* src/useradd.c, src/usermod.c:
Added initializing memory in variables when get_groups() function is not
and memory allocated for user_groups is not initialized (in both useradd
usermod when -G options is used and in usermod when -l option is used).
That causing segfaults sometimes.
This fix is importand but not critical because usermod and usermod aren't
Problem reported and fixed by Alexey Voinov <email@example.com>.
I just noticed that nalin checked in the fix into CVS on January 7th
for rawhide shadow-utils-4.0.3-17. Perhaps external contributors
should review the SRPM to be safe.
usermod.c hunk is missing in the shadow-4.0.3-uninitialized.patch from
Created attachment 97144 [details]
Arg... re-opening and bugging nalin...
btw, shadow-utils 188.8.131.52 has been released which contains this fix