From: Mathieu Poirier <mathieu.poirier@linaro.org> To: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: "linux-arm-kernel@lists.infradead.org" <linux-arm-kernel@lists.infradead.org>, "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>, linux-doc@vger.kernel.org, Chunyan Zhang <zhang.chunyan@linaro.org>, Mike Leach <mike.leach@arm.com>, "Jeremiassen, Tor" <tor@ti.com>, Al Grant <al.grant@arm.com>, Rabin Vincent <rabin@rab.in> Subject: Re: [PATCH V8 18/23] coresight: etm-perf: new PMU driver for ETM tracers Date: Thu, 28 Jan 2016 14:12:23 -0700 [thread overview] Message-ID: <CANLsYkyOEOW=8Q1Xc3r9Cb9Oz-juo8S8Pbh5AybUVaQrCi4Esg@mail.gmail.com> (raw) In-Reply-To: <87powl678z.fsf@ashishki-desk.ger.corp.intel.com> On 28 January 2016 at 08:42, Alexander Shishkin <alexander.shishkin@linux.intel.com> wrote: > Mathieu Poirier <mathieu.poirier@linaro.org> writes: > >>> I'd like to understand all the potential failures here, because it's >>> really a good idea to keep those to a minimum for the sake of >>> consistency. That is, if the user succeeded in creating an event, about >>> the only good reason for the event not starting is a filled up buffer. >> >> Enabling a path should fail when one or many components of that path >> are already enabled by an ongoing trace session. This situation is >> quite likely to happen since in a lot of design tracers share the link >> and sinks. > > Yes, but provided that we don't get interference from sysfs users > (which, I guess, could be blocked out while etm perf events exist), this > part shouldn't fail, as nobody else should be using these links and > sinks but etm events and those are safe from overlapping because of > PERF_PMU_CAP_EXCLUSIVE. Or am I missing something? Interference from sysfs are being dealt with by drvdata->mode. As such a user can't muddle with perf sessions so things are good on that front. Reading the comment in [1], mutual exclusion rules aren't that clear to me. Perf is aware of the context and CPUs a session is slated for but has no insight on the path from source to sink. Tracing an "ls" command on CPU1 and "uname" on CPU3 might look very distinct to Perf but can easily share the same plumbing. If setting PERF_PMU_CAP_EXCLUSIVE in the PMU definition guarantees that a single session on that PMU can be active at any given time, then yes, the only cause of failutre is a buffer full condition. [1]. http://lxr.free-electrons.com/source/kernel/events/core.c#L3608 > >>> This is why it makes a lot of sense to keep all the >>> coresight_build_path()/coresight_enable_path() to the .event_init() >>> phase and let them fail early, if they should fail. >> >> If we do enable enable paths in .event_init() we can't support >> multiple concurrent trace session (see explanation above). The >> ultimate design is to have a source directly connected to a sink but >> so far none of the coresight topologies I've seen have been wired like >> that. > > So if we call dibs on those paths early (like event_init early), in such > a way that nobody but other etm events can use them, we should be ok, I > think. Elements get out of reach from sysfs as soon as a path is built, so we are already covered there. > > Regards, > -- > Alex
WARNING: multiple messages have this Message-ID (diff)
From: mathieu.poirier@linaro.org (Mathieu Poirier) To: linux-arm-kernel@lists.infradead.org Subject: [PATCH V8 18/23] coresight: etm-perf: new PMU driver for ETM tracers Date: Thu, 28 Jan 2016 14:12:23 -0700 [thread overview] Message-ID: <CANLsYkyOEOW=8Q1Xc3r9Cb9Oz-juo8S8Pbh5AybUVaQrCi4Esg@mail.gmail.com> (raw) In-Reply-To: <87powl678z.fsf@ashishki-desk.ger.corp.intel.com> On 28 January 2016 at 08:42, Alexander Shishkin <alexander.shishkin@linux.intel.com> wrote: > Mathieu Poirier <mathieu.poirier@linaro.org> writes: > >>> I'd like to understand all the potential failures here, because it's >>> really a good idea to keep those to a minimum for the sake of >>> consistency. That is, if the user succeeded in creating an event, about >>> the only good reason for the event not starting is a filled up buffer. >> >> Enabling a path should fail when one or many components of that path >> are already enabled by an ongoing trace session. This situation is >> quite likely to happen since in a lot of design tracers share the link >> and sinks. > > Yes, but provided that we don't get interference from sysfs users > (which, I guess, could be blocked out while etm perf events exist), this > part shouldn't fail, as nobody else should be using these links and > sinks but etm events and those are safe from overlapping because of > PERF_PMU_CAP_EXCLUSIVE. Or am I missing something? Interference from sysfs are being dealt with by drvdata->mode. As such a user can't muddle with perf sessions so things are good on that front. Reading the comment in [1], mutual exclusion rules aren't that clear to me. Perf is aware of the context and CPUs a session is slated for but has no insight on the path from source to sink. Tracing an "ls" command on CPU1 and "uname" on CPU3 might look very distinct to Perf but can easily share the same plumbing. If setting PERF_PMU_CAP_EXCLUSIVE in the PMU definition guarantees that a single session on that PMU can be active at any given time, then yes, the only cause of failutre is a buffer full condition. [1]. http://lxr.free-electrons.com/source/kernel/events/core.c#L3608 > >>> This is why it makes a lot of sense to keep all the >>> coresight_build_path()/coresight_enable_path() to the .event_init() >>> phase and let them fail early, if they should fail. >> >> If we do enable enable paths in .event_init() we can't support >> multiple concurrent trace session (see explanation above). The >> ultimate design is to have a source directly connected to a sink but >> so far none of the coresight topologies I've seen have been wired like >> that. > > So if we call dibs on those paths early (like event_init early), in such > a way that nobody but other etm events can use them, we should be ok, I > think. Elements get out of reach from sysfs as soon as a path is built, so we are already covered there. > > Regards, > -- > Alex
next prev parent reply other threads:[~2016-01-28 21:12 UTC|newest] Thread overview: 108+ messages / expand[flat|nested] mbox.gz Atom feed top 2016-01-14 21:45 [PATCH V8 00/23] Coresight integration with perf Mathieu Poirier 2016-01-14 21:45 ` Mathieu Poirier 2016-01-14 21:45 ` [PATCH V8 01/23] coresight: associating path with session rather than tracer Mathieu Poirier 2016-01-14 21:45 ` Mathieu Poirier 2016-01-14 21:45 ` [PATCH V8 02/23] coresight: add API to get sink from path Mathieu Poirier 2016-01-14 21:45 ` Mathieu Poirier 2016-01-14 21:45 ` [PATCH V8 03/23] coresight: moving PM runtime operations to core framework Mathieu Poirier 2016-01-14 21:45 ` Mathieu Poirier 2016-01-14 21:45 ` [PATCH V8 04/23] coresight: etm3x: moving etm_readl/writel to header file Mathieu Poirier 2016-01-14 21:45 ` Mathieu Poirier 2016-01-14 21:45 ` [PATCH V8 05/23] coresight: etm3x: moving sysFS entries to dedicated file Mathieu Poirier 2016-01-14 21:45 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 06/23] coresight: etm3x: unlocking tracers in default arch init Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 07/23] coresight: etm3x: splitting struct etm_drvdata Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 08/23] coresight: etm3x: adding operation mode for etm_enable() Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 09/23] coresight: etm3x: set progbit to stop trace collection Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 10/23] coresight: etm3x: changing default trace configuration Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 11/23] coresight: etm3x: consolidating initial config Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 12/23] coresight: etm3x: implementing user/kernel mode tracing Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 13/23] coresight: etm3x: implementing perf_enable/disable() API Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 14/23] coresight: etb10: moving to local atomic operations Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 15/23] coresight: etb10: adding operation mode for sink->enable() Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 16/23] coresight: etb10: implementing AUX API Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-26 15:53 ` Alexander Shishkin 2016-01-26 15:53 ` Alexander Shishkin 2016-01-27 20:55 ` Mathieu Poirier 2016-01-27 20:55 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 17/23] coresight: updating documentation to reflect integration with perf Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 18/23] coresight: etm-perf: new PMU driver for ETM tracers Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-26 15:27 ` Alexander Shishkin 2016-01-26 15:27 ` Alexander Shishkin 2016-01-27 18:33 ` Mathieu Poirier 2016-01-27 18:33 ` Mathieu Poirier 2016-01-28 15:42 ` Alexander Shishkin 2016-01-28 15:42 ` Alexander Shishkin 2016-01-28 21:12 ` Mathieu Poirier [this message] 2016-01-28 21:12 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 19/23] coresight: introducing a global trace ID function Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 20/23] perf tools: making function set_max_cpu_num() non static Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-25 20:46 ` Mathieu Poirier 2016-01-25 20:46 ` Mathieu Poirier 2016-01-25 21:12 ` Arnaldo Carvalho de Melo 2016-01-25 21:12 ` Arnaldo Carvalho de Melo 2016-01-25 21:29 ` Arnaldo Carvalho de Melo 2016-01-25 21:29 ` Arnaldo Carvalho de Melo 2016-01-26 17:08 ` Mathieu Poirier 2016-01-26 17:08 ` Mathieu Poirier 2016-01-26 18:51 ` Arnaldo Carvalho de Melo 2016-01-26 18:51 ` Arnaldo Carvalho de Melo 2016-01-27 16:24 ` Mathieu Poirier 2016-01-27 16:24 ` Mathieu Poirier 2016-02-03 10:15 ` [tip:perf/core] perf cpumap: Auto initialize cpu__max_{node,cpu} tip-bot for Arnaldo Carvalho de Melo 2016-01-14 21:46 ` [PATCH V8 21/23] perf tools: adding perf_evlist to *info_priv_size() Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-25 20:48 ` Mathieu Poirier 2016-01-25 20:48 ` Mathieu Poirier 2016-01-25 21:08 ` Arnaldo Carvalho de Melo 2016-01-25 21:08 ` Arnaldo Carvalho de Melo 2016-01-26 14:27 ` Adrian Hunter 2016-01-26 14:27 ` Adrian Hunter 2016-01-26 14:33 ` Arnaldo Carvalho de Melo 2016-01-26 14:33 ` Arnaldo Carvalho de Melo 2016-01-29 10:14 ` Adrian Hunter 2016-01-29 10:14 ` Adrian Hunter 2016-02-03 10:17 ` [tip:perf/core] perf auxtrace: Add perf_evlist pointer " tip-bot for Mathieu Poirier 2016-01-14 21:46 ` [PATCH V8 22/23] perf tools: making coresight PMU listable Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-25 20:49 ` Mathieu Poirier 2016-01-25 20:49 ` Mathieu Poirier 2016-01-25 21:10 ` Arnaldo Carvalho de Melo 2016-01-25 21:10 ` Arnaldo Carvalho de Melo 2016-01-29 10:24 ` Adrian Hunter 2016-01-29 10:24 ` Adrian Hunter 2016-01-14 21:46 ` [PATCH V8 23/23] perf tools: adding coresight etm PMU record capabilities Mathieu Poirier 2016-01-14 21:46 ` Mathieu Poirier 2016-01-25 20:51 ` Mathieu Poirier 2016-01-25 20:51 ` Mathieu Poirier 2016-01-25 21:10 ` Arnaldo Carvalho de Melo 2016-01-25 21:10 ` Arnaldo Carvalho de Melo 2016-01-29 10:34 ` Adrian Hunter 2016-01-29 10:34 ` Adrian Hunter 2016-01-29 17:37 ` Mathieu Poirier 2016-01-29 17:37 ` Mathieu Poirier 2016-01-29 21:12 ` Arnaldo Carvalho de Melo 2016-01-29 21:12 ` Arnaldo Carvalho de Melo 2016-01-29 22:24 ` Mathieu Poirier 2016-01-29 22:24 ` Mathieu Poirier 2016-02-02 16:20 ` Mathieu Poirier 2016-02-02 16:20 ` Mathieu Poirier 2016-02-02 16:41 ` Arnaldo Carvalho de Melo 2016-02-02 16:41 ` Arnaldo Carvalho de Melo 2016-02-03 16:11 ` Mathieu Poirier 2016-02-03 16:11 ` Mathieu Poirier
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to='CANLsYkyOEOW=8Q1Xc3r9Cb9Oz-juo8S8Pbh5AybUVaQrCi4Esg@mail.gmail.com' \ --to=mathieu.poirier@linaro.org \ --cc=al.grant@arm.com \ --cc=alexander.shishkin@linux.intel.com \ --cc=linux-arm-kernel@lists.infradead.org \ --cc=linux-doc@vger.kernel.org \ --cc=linux-kernel@vger.kernel.org \ --cc=mike.leach@arm.com \ --cc=rabin@rab.in \ --cc=tor@ti.com \ --cc=zhang.chunyan@linaro.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.