[Dovecot] Re: More test46 feedback

[dovecot at spam.turbolink.net]

On Tue, 28 Sep 2004, Timo Sirainen wrote:

> On 28.9.2004, at 00:29, dovecot at spam.turbolink.net wrote:
>
> > Here's another error message I just started getting with test46:
> >
> > Error: POP3(test2): file mbox-lock.c: line 493 (mbox_lock): assertion
> > failed: (lock_type == F_RDLCK ||ibox->mbox_lock_type != F_RDLCK)
>
> Could you get gdb backtrace from this? I know this happens when you try
> to copy a message into the same mbox, but why would anyone do that?

I haven't been able to reproduce it, I've only seen it happen in the logs
after the fact on a few uses mboxes.  I did get a core dump last time:

#0  0x4004bc31 in kill () from /lib/libc.so.6
(gdb) bt
#0  0x4004bc31 in kill () from /lib/libc.so.6
#1  0x4004b8b8 in raise () from /lib/libc.so.6
#2  0x4004d17c in abort () from /lib/libc.so.6
#3  0x0807f188 in i_set_failure_file ()
#4  0x0807ed63 in i_panic ()
#5  0x0805c653 in mbox_lock ()
#6  0x080604bf in mbox_sync ()
#7  0x0805ac2d in mbox_is_valid_mask ()
#8  0x08074fd1 in mailbox_close ()
#9  0x080528f1 in client_destroy ()
#10 0x080854ff in _o_stream_create ()
#11 0x08082dc4 in io_loop_handler_run ()
#12 0x08082509 in io_loop_run ()
#13 0x08053fba in main ()
#14 0x40037bb4 in __libc_start_main () from /lib/libc.so.6
(gdb)

I've only seen it happen using POP3, so I don't think it has to do with
copying messages.

I did leave the new lazy writing turned on so maybe it's something in that
new code?

- Mike