[systemd-devel] [PATCH] journal: Fix sd_journal_enumerate_unique skipping values

[Jan Janssen]

On 2014-10-08 14:59, Zbigniew Jędrzejewski-Szmek wrote:
> On Wed, Oct 08, 2014 at 08:24:49AM +0200, Jan Janssen wrote:
>>
>>
>>> Gesendet: Mittwoch, 08. Oktober 2014 um 01:40 Uhr
>>> Von: "Zbigniew Jędrzejewski-Szmek" <zbyszek at in.waw.pl>
>>> An: "Jan Janssen" <medhefgo at web.de>
>>> Cc: systemd-devel at lists.freedesktop.org
>>> Betreff: Re: [systemd-devel] [PATCH] journal: Fix sd_journal_enumerate_unique skipping values
>>>
>>> On Mon, Oct 06, 2014 at 06:57:38PM +0200, Zbigniew Jędrzejewski-Szmek wrote:
>>>> On Mon, Oct 06, 2014 at 06:36:34PM +0200, Jan Janssen wrote:
>>>>> *bump*
>>>> Sorry, I'll look into this.
>>>
>>> Doesn't work. Both without or with your other patch
>>> sd_journal_enumerate_unique I get bogus results on my test case. It
>>> seems the issue is more complicated.
>>>
>>
>> That's odd. Care to elaborate what bogus results means? Are you even affected by the
>> bug in question without the patch?
>
> Yes, I have a VM where I get a smaller number from -F _BOOT_ID than from --list-boots
> (w/o your patches), and then the same smaller number with one or two of your patches.
> So results become consistent, but equally bad.
>
> Of course I can't know if this is exactly the same bug, but it certainly looks
> like it.
>

Sounds like maybe one of those calls end up interleaving journals from 
different machines?

Also, does removing the call to journal_file_object_release() in 
sd_journal_enumerate_unique() improve things or not? How about moving it 
after the if(found) where it was before the patch?

I'd love to investigate this, but I sadly don't have any journals that 
triggers this :(

Jan