re-ordering across FINISHED_ROUND observed

re-ordering across FINISHED_ROUND observed

[Milian Wolff]

[-- Attachment #1: Type: text/plain, Size: 1215 bytes --]

Hey all,

I have a perf.data file which contradicts the documentation:

> Ordering
> 
> The events are not necessarily in time stamp order, as they can be
> collected in parallel on different CPUs. If the events should be
> processed in time order they need to be sorted first. It is possible
> to only do a partial sort using the FINISHED_ROUND event header (see
> below). perf record guarantees that there is no reordering over a
> FINISHED_ROUND.

I see one round finishing, the maximum sample time for that round is:

52468794009888

Then the next round finishes, and the first sample in this round has time:

52468793922208

I have the perf.data file, it's large though (431MB uncompressed). The .xz 
compressed file is only ~6.2MB and can be found here:

http://milianw.de/files/perf.data.heaptrack.normal.xz

So, is the documentation wrong and there can actually be reordering over a 
FINISHED_ROUND? Or is that an indication of a bug somewhere? I'm running 
4.9.11-1-ARCH with perf from perf-core-for-mingo-4.11-20170220-1197-
g062f29461d1e.

Thanks

-- 
Milian Wolff | milian.wolff@kdab.com | Software Engineer
KDAB (Deutschland) GmbH&Co KG, a KDAB Group company
Tel: +49-30-521325470
KDAB - The Qt Experts

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5903 bytes --]