[LITMUS^RT] litmus-dev Digest, Vol 81, Issue 5

[Ricardo Teixeira]

Hi Björn.

I meant the IRQ field in the trace logs. The -x parameter did not bring the
records relative to UNLOCK as you said. Looking at ftdump records, I
noticed that the UNLOCK_START and UNLOCK_END records are interleaved by
SCHED and SCHED2 records (see below). Could this be why these records are
not being counted? The protocol I am testing is not suspension-based, but
it can call the litmus_reschedule() method at the end of unlock()
plugin method. I'll try with the standard tools to get the records of the
other processors, as you suggested. Thank you again.

UNLOCK_START seq: 372140 pid: 5086 timestamp: 7003579867914 cpu: 0 type: RT
irq: 1 irqc: 01
SCHED_START seq: 372141 pid: 5086 timestamp: 7003579871038 cpu: 0 type:
UNKNOWN irq: 0 irqc: 00
SCHED_END seq: 372142 pid: 5086 timestamp: 7003579875406 cpu: 0 type: RT
irq: 0 irqc: 00
SCHED2_START seq: 372143 pid: 5086 timestamp: 7003579875842 cpu: 0 type: RT
irq: 0 irqc: 00
SCHED2_END seq: 372144 pid: 5086 timestamp: 7003579876844 cpu: 0 type: RT
irq: 0 irqc: 00
UNLOCK_END seq: 372145 pid: 5086 timestamp: 7003579877156 cpu: 0 type: RT
irq: 0 irqc: 00

Em qua, 15 de mai de 2019 às 07:00, <litmus-dev-request at lists.litmus-rt.org>
escreveu:

> Send litmus-dev mailing list submissions to
>         litmus-dev at lists.litmus-rt.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
>         https://lists.litmus-rt.org/listinfo/litmus-dev
> or, via email, send a message with subject or body 'help' to
>         litmus-dev-request at lists.litmus-rt.org
>
> You can reach the person managing the list at
>         litmus-dev-owner at lists.litmus-rt.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of litmus-dev digest..."
>
>
> Today's Topics:
>
>    1. Re: litmus-dev Digest, Vol 81, Issue 3 (Björn Brandenburg)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Wed, 15 May 2019 09:30:12 +0200
> From: Björn Brandenburg <bbb at mpi-sws.org>
> To: litmus-dev at lists.litmus-rt.org
> Subject: Re: [LITMUS^RT] litmus-dev Digest, Vol 81, Issue 3
> Message-ID: <78F281B5-EC75-4510-B1C1-C11F0D63A5BE at mpi-sws.org>
> Content-Type: text/plain; charset="utf-8"
>
> Hi Ricardo,
>
> > On 15. May 2019, at 02:44, Ricardo Teixeira <ricardo.btxr at gmail.com>
> wrote:
> >
> >> Below are the kernel messages for the overhead and schedule tracers.
>
> these messages look reasonable; I don’t see anything suggesting that there
> might be a problem.
>
> >> When I'm looking for UNLOCK events through ftdump, I noticed that
> almost all UNLOCK_START records are generated with IRQ enabled for both
> protocols I'm running.
>
> What do you mean with “almost all [...] with IRQ enabled”? Do you have
> different lock entry paths, some with IRQs enabled and some without?
>
> Or do you mean the IRQ field in the trace records? That’s a snapshot of
> the core’s IRQ *counter*, which is reset when it is read. It’s perfectly
> normal for this number to be non-zero for a _START event — on average,
> you’d expect there to have been some interrupts since the last time that an
> event was recorded.
>
> >> Is this why these records are not being counted? Is there any parameter
> so I can collect these records?
>
> Yes, -x disables interrupt filtering, but, again, this should not be an
> issue for _START events.
>
>
> https://github.com/LITMUS-RT/feather-trace-tools/blob/master/src/ft2csv.c#L397
>
> You had mentioned in some email that you use “inherited” scripts that
> produce a single .ft file — I have no idea what these scripts are, where
> they come from, or what they do, so I can’t really help you debug them. I
> would recommend to use the standard tools as described in the tutorial.
>
>
> https://github.com/LITMUS-RT/feather-trace-tools/blob/master/doc/howto-trace-and-process-overheads.md
>
> At the very least, see if the problem persists if you follow that
> approach.
>
> Regards,
> Björn
>
>
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <
> http://lists.litmus-rt.org/pipermail/litmus-dev/attachments/20190515/824b17dc/attachment-0001.html
> >
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: smime.p7s
> Type: application/pkcs7-signature
> Size: 2334 bytes
> Desc: not available
> URL: <
> http://lists.litmus-rt.org/pipermail/litmus-dev/attachments/20190515/824b17dc/attachment-0001.bin
> >
>
> ------------------------------
>
> Subject: Digest Footer
>
> _______________________________________________
> litmus-dev mailing list
> litmus-dev at lists.litmus-rt.org
> https://lists.litmus-rt.org/listinfo/litmus-dev
>
>
> ------------------------------
>
> End of litmus-dev Digest, Vol 81, Issue 5
> *****************************************
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.litmus-rt.org/pipermail/litmus-dev/attachments/20190515/6acc2cc0/attachment.html>