Assert failure fs/jbd/transaction.c
Got an assert failure while I was in X, and was unable to do anything that'll hit disk (like run dmesg -- couldn't read dmesg in from disk).
Not sure what else was in the dmesg buffer.
handle_t *journal_start(journal_t *journal, int nblocks) {
...
if (handle) {
J_ASSERT(handle->h_transaction->t_journal == journal);
handle->h_ref++;
return handle;
}This is against a bk pull of 2.6.4-rc1, somewhere around... 2004/03/02.
No idea how to trigger it.
Only file systems that use jbd are:
/dev/hda5 on / type ext3 (rw,data=journal,errors=remount-ro) /dev/hda6 on /usr type ext3 (rw,nodev,data=journal) /dev/hda7 on /home type ext3 (rw,nosuid,nodev,data=journal)
Stuff that *might* have affected it:CONFIG_CC_OPTIMIZE_FOR_SIZE=yCONFIG_EXT3_FS=yCONFIG_EXT3_FS_POSIX_ACL=yCONFIG_EXT3_FS_XATTR=yCONFIG_FS_MBCACHE=yCONFIG_FS_POSIX_ACL=yCONFIG_PREEMPT=yCONFIG_REGPARM=y
ramune@hasenpfeffer:ramune: cat /proc/version
Linux version 2.6.4-rc1 (ramune@hasenpfeffer) (gcc version 3.3.1) #12 Mon Mar 1 08:54:17 PST 2004
Debian/Woody, P4 2.4GHz system.
ramune@hasenpfeffer:ramune: lsmod|tail +2|wc -l
193
(Yeah, that's a lot of stuff -- it gets to the login: prompt faster
that way if everything I can rip out is a module, bootup decompression is obscenely slow on this system).
It's the first time I ever saw this error, so no idea how to trigger it. All disk access went dead.
-- DN
Daniel