imap4flags extension for 2.4 branch

[Deniss]

James,

I got another issue trying to migrate from current cyrus installation
(cyrus-2.4.17) to new one with imap4flags extension: bytecode is not
compatible with old compiled - "sieve runtime error: Invalid sieve
bytecode"

This happens when "fileinto" was used in compiled with cyrus-2.4.17
sieve script. Any other sieve target looks to work just fine.

Best,
  Deniss

On 2014.10.31. 6:16, James Cassell wrote:
> Deniss,
> 
> I believe I have found and fixed the issue.  There are two new commits
> on that branch.  Can you test it?
> 
> Thanks.
> 
> V/r,
> James Cassell
> 
> 
> On Wed, Oct 29, 2014, at 10:36 AM, Deniss wrote:
>> James,
>>
>> please look at backtrace:
>>
>> Program terminated with signal 11, Segmentation fault.
>> #0  append_fromstage (as=as at entry=0x3dfb16f2070, body=0x3dfb16f2a10,
>> stage=stage at entry=0x46586c0, internaldate=internaldate at entry=0,
>> flags=flags at entry=0x0, nolink=0) at append.c:468
>> 468         for (i = 0; i < flags->count ; i++) {
>>
>> looks like the problem is in append.c:468 where flags iterated but is
>> not initialized ?
>>
>> any sieve runtime error (Fileinto: Mailbox does not exist, Include can
>> not load script, etc) causes segfault of lmtpd
>>
>> Tested on on 2.4 master + patches from your imap4flags-2.4 branch
>>
>> Best,
>> Deniss
>>
>> On 2014.10.29. 13:43, James Cassell wrote:
>>> Deniss,
>>>
>>> I don't know.  My guess is that it is an error in the other patches I
>>> backported from master to 2.4.  What version are you using?  2.4.17
>>> stable?  Maybe someone on the cyrus-devel list might have an idea.
>>>
>>> Next week, I might try to backport my patches without requiring the
>>> other patches I (perhaps erroneously) backported.
>>>
>>> V/r,
>>> James Cassell
>>>
>>>
>>> On Tue, Oct 28, 2014, at 11:52 AM, Deniss wrote:
>>>> James,
>>>>
>>>> i testing your patch and have segfaults when sieve cant be include (the
>>>> file is missed):
>>>>
>>>> Oct 28 17:43:06 lv-test-2 lmtp[14400]: IOERROR: fstating sieve script
>>>> /var/imap/sieve/global/special-folders.bc: No such file or directory
>>>> Oct 28 17:43:06 lv-test-2 lmtp[14400]: sieve runtime error for testdev33
>>>> id <544FB87D.60803 at adsfasdfasd>: Include can not load script
>>>> Oct 28 17:43:06 lv-test-2 lmtp[14443]: executed
>>>> Oct 28 17:43:06 lv-test-2 kernel: lmtpd[14400]: segfault at 0 ip
>>>> 0000002923bc2a41 sp 000003b91dfeddb0 error 4 in lmtpd[2923ba9000+16f000]
>>>> Oct 28 17:43:06 lv-test-2 kernel: grsec: From 10.0.1.74: Segmentation
>>>> fault occurred at            (nil) in /C3/cyrus/bin/lmtpd[lmtpd:14400]
>>>> uid/euid:999/999 gid/egid:12/12, parent
>>>> /C3/cyrus/bin/master[master:14372] uid/euid:999/999 gid/egid:12/12
>>>> Oct 28 17:43:06 lv-test-2 kernel: grsec: From 10.0.1.74: denied resource
>>>> overstep by requesting 4096 for RLIMIT_CORE against limit 0 for
>>>> /C3/cyrus/bin/lmtpd[lmtpd:14400] uid/euid:999/999 gid/egid:12/12, parent
>>>> /C3/cyrus/bin/master[master:14372] uid/euid:999/999 gid/egid:12/12
>>>>
>>>>
>>>> without the patch sieve just ignore the file without segfaulting lmtpd.
>>>>
>>>>
>>>> Best,
>>>> Deniss
>>>>
>>>> On 2014.10.26. 10:41, James Cassell wrote:
>>>>> Hello Denniss,
>>>>>
>>>>> I have backported imap4flags to 2.4.17.  I backported 6 other patches to
>>>>> make backporting imap4flags easier.  There has been no code review of
>>>>> these patch backports.
>>>>>
>>>>> The code is available here:
>>>>> https://github.com/cyberpear/cyrus-imapd/commits/imap4flags-2.4
>>>>>
>>>>> I tried to make sure the bytecode generated on 2.4 would be readable by
>>>>> 2.5, though not the other way around.  I haven't tested this yet.
>>>>>
>>>>> I'd appreciate any reviews of the backported code.
>>>>>
>>>>> I'll probably send a pull request in a week or so for this code.
>>>>>
>>>>> V/r,
>>>>> James Cassell
>>>>>
>>>>>
>>>>> On Sat, Oct 25, 2014, at 10:40 AM, Deniss wrote:
>>>>>> hello James,
>>>>>>
>>>>>> thanks for implementing imap4flags sieve extension.
>>>>>> I'm looking to use the extension in production-ready 2.4 branch - 
>>>>>> "hasflag" test allows to add some logic while the "variables" extension 
>>>>>> is missed from cyrus.
>>>>>> Are there any caveats to backport imap4flags from master to 2.4 ?
>>>>>>
>>>>>> Best,
>>>>>> Deniss