Bug#264884: globfree() double-frees
At Tue, 14 Sep 2004 11:39:42 +0900,
GOTO Masanori wrote:
> My gdb log is:
>
> sh-3.00$ LD_LIBRARY_PATH=/usr/lib/debug gdb ./T.glob
> GNU gdb 6.1-debian
> ...
> This GDB was configured as "i386-linux"...(no debugging symbols found)...Using host libthread_db library "/usr/lib/debug/libthread_db.so.1".
>
> (gdb) b test29
> Breakpoint 1 at 0x804f1e5
> (gdb) run
> Starting program: /home/tet/test_sets/TESTROOT/tset/LSB.os/genuts/glob/T.glob
> (no debugging symbols found)...Detaching after fork from child process 7263.
> Detaching after fork from child process 7264.
>
> Breakpoint 1, 0x0804f1e5 in test29 ()
> (gdb) b globfree
> Breakpoint 2 at 0x400c2d36: file ../sysdeps/generic/glob.c, line 1100.
> (gdb) c
> Continuing.
>
> Breakpoint 2, 0x400c2d36 in *__GI_globfree (pglob=0x0) at ../sysdeps/generic/glob.c:1100
> 1100 {
> (gdb) c
> Continuing.
>
> Breakpoint 2, 0x400c2d36 in *__GI_globfree (pglob=0x0) at ../sysdeps/generic/glob.c:1100
> 1100 {
> (gdb)
> Continuing.
>
> There's no problem. pglob = NULL. Using MALLOC_CHECK_=3 or valgrind
> does not report such problems... (valgrind says there may be some
> leaks; it should be reported separately).
>
> If it's uninitialized value issue, it may be glob() problem rathar
> than globfree(). Could you confirm it?
I didn't reappear this problem. Moreover, if globfree() returns
GLOB_ABORTED, globfree() should not be called from looking at source
code.
But to make sure LSB conformance, I applied it for only sarge.
However after sarge, I drop it because it's unsure and I think the bug
is existed another places.
If you have any evidences to reappear this problem, please tell me.
If there's bug, we should fix the problem, even if your mission is to
make sarge LSB conformance and that job is finished when sarge is
released.
Regards,
-- gotom
Reply to: