[Dovecot] UIDL bug

[Timo Sirainen]

On 6.10.2004, at 00:56, dovecot at spam.turbolink.net wrote:

> #5  0x0805c653 in mbox_lock (ibox=0x40151300, lock_type=1, 
> lock_id_r=0xbffffa0c) at mbox-lock.c:500
> #6  0x080604bf in mbox_sync (ibox=0x809d108, flags=MBOX_SYNC_REWRITE) 
> at mbox-sync.c:1249
> #7  0x0805ac2d in mbox_storage_close (box=0x809d108) at 
> mbox-storage.c:805
> #8  0x08074fcd in mailbox_close (box=0x0) at mail-storage.c:296
> #9  0x080528f1 in client_destroy (client=0x809cc88) at client.c:190

fr 9
p client->cmd

It looks like some command isn't being deinitialized properly. I found 
one possible reason, but I don't think it should happen at least 
often..

>>> And lots more of this one with IMAP users only:
>>>
>>> file mbox-sync-rewrite.c: line 460 (mbox_sync_rewrite): assertion
>>> failed: (mails[idx].space > 0)
>>
>> Yep, this happened with me too. CVS has much more understandable (and
>> better working and slightly faster) mbox rewriting code. But it has
>> some other new syncing problems that I haven't yet fixed.
>
> ok.  I'm starting to understand more of the dovecot source so I'll 
> check
> it out.

I think mbox code in CVS works again. Please try and see if it fixes 
even the first problem.

>>> #0  0x0807d63f in mail_cache_merge_bitmask (cache=0x6e6ca9a0,
>>> buffer=0x80b6b58, field=0, data=0x4027ebf8, data_size=4)
>>>     at mail-cache-compress.c:29
>>> 29                      buf_data_size =
>>> cache->fields[buf_field].field.field_size;
>>
> (gdb) p cache
> $5 = (struct mail_cache *) 0x6e6ca9a0
> (gdb) p cache->fields
> Cannot access memory at address 0x6e6ca9dc

Either corrupted backtrace or Dovecot somehow corrupted the stack while 
running. I added several asserts in CVS to catch all possible mistakes 
here.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 186 bytes
Desc: This is a digitally signed message part
URL: <http://dovecot.org/pipermail/dovecot/attachments/20041007/203a2fc5/attachment-0001.bin>