Re: balsa segfaults or doesn't die properly occasionally

[Kacper Wysocki <kacperw online no>]

> I'll get back to you on the malloc thing..

Today's lockup was neither a seg or a hang before closing, but since I  
have balsa in debug mode all the time now I caught it as well..
Here are the relevant strace lines (it's hard to tell where the  
'relevantness' starts, but the strace is 4MB so I have to crop it..

--
lseek(22, 0, SEEK_SET)                  = 0
read(22, "-----BEGIN PGP SIGNED MESSAGE---"..., 1073) = 1073
getuid32()                              = 1001
open("/etc/passwd", O_RDONLY)           = 23
fcntl64(23, F_GETFD)                    = 0
fcntl64(23, F_SETFD, FD_CLOEXEC)        = 0
_llseek(23, 0, [0], SEEK_CUR)           = 0
fstat64(23, {st_mode=S_IFREG|0644, st_size=1516, ...}) = 0
mmap2(NULL, 1516, PROT_READ, MAP_SHARED, 23, 0) = 0x4330f000
_llseek(23, 1516, [1516], SEEK_SET)     = 0
munmap(0x4330f000, 1516)                = 0
close(23)                               = 0
stat64("/home/comotion/.gnupg/trustdb.gpg.lock", 0xbffff61c) = -1  
ENOENT (No such file or di
rectory)
pipe([23, 24])                          = 0
fcntl64(23, F_SETFD, FD_CLOEXEC)        = 0
pipe([25, 26])                          = 0
fcntl64(26, F_SETFD, FD_CLOEXEC)        = 0
pipe([27, 28])                          = 0
fcntl64(27, F_SETFD, FD_CLOEXEC)        = 0
clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID| 
SIGCHLD, child_tidptr=0x4
1cd36a8) = 6824
waitpid(6824, [WIFEXITED(s) && WEXITSTATUS(s) == 0], 0) = 6824
--- SIGCHLD (Child exited) @ 0 (0) ---
sigreturn()                             = ? (mask now [])
close(24)                               = 0
close(25)                               = 0
close(28)                               = 0
fcntl64(26, F_GETFL)                    = 0x1 (flags O_WRONLY)
fcntl64(26, F_SETFL, O_WRONLY|O_NONBLOCK) = 0
select(28, [23 27], [26], NULL, {1, 0}) = 1 (out [26], left {1, 0})
write(26, "-----BEGIN PGP SIGNED MESSAGE---"..., 1073) = 1073
select(28, [23 27], [26], NULL, {1, 0}) = 1 (out [26], left {0,  
999000})
close(26)                               = 0
select(28, [23 27], [], NULL, {1, 0})   = 0 (Timeout)
select(28, [23 27], [], NULL, {1, 0})   = 0 (Timeout)
select(28, [23 27], [], NULL, {1, 0})   = 0 (Timeout)
select(28, [23 27], [], NULL, {1, 0})   = 0 (Timeout)
select(28, [23 27], [], NULL, {1, 0})   = 0 (Timeout)
[message repeated, oh, about 400 times]
select(28, [23 27], [], NULL, {1, 0})   = 0 (Timeout)
select(28, [23 27], [], NULL, {1, 0} <unfinished ...>

[this might have been where I sigterm'ed the process]

I was halfway through writing an email when this happened. I'll keep  
y'all posted as I gather more information. Pointers in the right  
direction(s) are appreciated. gdb, maybe?

-Kacper

PGP signature