mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "stephane eranian" <eranian@googlemail.com>
To: "Thomas Gleixner" <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org, akpm@linux-foundation.org,
	mingo@elte.hu, x86@kernel.org, andi@firstfloor.org,
	sfr@canb.auug.org.au
Subject: Re: [patch 02/24] perfmon: base code
Date: Thu, 27 Nov 2008 18:47:54 +0100	[thread overview]
Message-ID: <7c86c4470811270947q585cba5vf8bdb875962a1856@mail.gmail.com> (raw)
In-Reply-To: <alpine.LFD.2.00.0811271638350.3325@localhost.localdomain>

Thomas,



On Thu, Nov 27, 2008 at 6:24 PM, Thomas Gleixner <tglx@linutronix.de> wrote:
> On Wed, 26 Nov 2008, eranian@googlemail.com wrote:
>
>> Index: o3/perfmon/perfmon_res.c
>
>> +/*
>> + * global information about all sessions
>> + */
>> +struct pfm_resources {
>> +     cpumask_t sys_cpumask;     /* bitmask of used cpus */
>> +     u32 thread_sessions; /* #num loaded per-thread sessions */
>> +};
>
> What's the purpose of this being a structure if it's just a single
> instance ?
>
There is a single instance.
I was just trying to regroup related fields together instead of having them as
separate variables. I can make the change.


>> +static struct pfm_resources pfm_res;
>> +
>> +static __cacheline_aligned_in_smp DEFINE_SPINLOCK(pfm_res_lock);
>
>> +/**
>> + * pfm_session_acquire - reserve a per-thread session
>> + *
>> + * return:
>> + *    0    : success
>> + *   -EBUSY: if conflicting session exist
>
>  Where ?

Not in the patchset, conflict can arise when you add system-wide sessions.

>
>> + */
>> +int pfm_session_acquire(void)
>> +{
>> +     unsigned long flags;
>> +     int ret = 0;
>> +
>> +     /*
>> +      * validy checks on cpu_mask have been done upstream
>> +      */
>
>  How please ? pfm_res.sys_cpumask is local to this file and you want
>  to check it under the lock and _before_ you increment
>  thread_sessions blindly.
>
Stale comment.

>
> +EXPORT_SYMBOL(pfm_session_allcpus_release);
>
> All what that code should do (in fact it does not) is preventing the
> mix of thread and system wide sessions.
>
True. That is a current limitation.

> It does neither need a cpumask nor tons of useless loops and debug
> outputs with zero value.
>
Well, the the cpumask is indeed needed but you don't see the reason
why in the patchset!

Perfmon in system-wide does not operate like Oprofile. In system-wide
a perfmon session (context) monitors only ONE CPU at a time. Each
session is independent of each other. You can therefore measure different
things on different CPUs. Reservation is thus done independently for each
CPU, therefore we need a cpu bitmask  to track allocation.

The Oprofile reservation you see is built on top of the cpumask reservation.
It tries to allocate in one call and atomically ALL the CPUs as this is the way
Oprofile operates. Thus it fails if one perfmon system-wide session or one
perfmon per-thread exists. I believe there is enough bitmask functions to
avoid loops as you pointed out. I will fix that.


> static int global_session;
> static int thread_sessions;
> static DEFINE_SPINLOCK(session_lock);
>
> int pfm_session_request(int global)
> {
>        unsigned long flags;
>        int res = -EBUSY;
>
>        spin_lock_irqsave(&session_lock, flags);
>
>        if (!global && !global_session) {
>                thread_sessions++;
>                res = 0;
>        }
>
>        if (global && !thread_sessions && !global_session) {
>                global_session = 1;
>                res = 0;
>        }
>
>        spin_unlock_irqrestore(&session_lock, flags);
>        return res;
> }
>
> void pfm_session_release(int global)
> {
>        unsigned long flags;
>
>        spin_lock_irqsave(&session_lock, flags);
>
>        if (global) {
>                WARN_ON(!global_session);
>                global_session = 0;
>        } else {
>                if (!global_session && thread_sessions)
>                        thread_session--;
>                else
>                        WARN();
>        }
>
>        spin_unlock_irqrestore(&session_lock, flags);
> }
>
> Would do it nicely including useful sanity checks and 70% less code.
>
> Thanks,
>
>        tglx
>

  reply	other threads:[~2008-11-27 17:48 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-11-26  8:42 eranian
2008-11-26 11:17 ` Andi Kleen
2008-11-27 11:23 ` Thomas Gleixner
2008-11-27 17:24 ` Thomas Gleixner
2008-11-27 17:47   ` stephane eranian [this message]
2008-11-27 18:28     ` Thomas Gleixner
2008-11-27 18:41       ` Andi Kleen
2008-11-27 18:36         ` Thomas Gleixner
2008-11-27 18:54           ` stephane eranian
2008-11-27 18:51       ` stephane eranian
2008-11-27 19:35         ` Thomas Gleixner
2008-11-27 19:49           ` stephane eranian
2008-11-27 20:59             ` Thomas Gleixner
  -- strict thread matches above, loose matches on Subject: below --
2008-11-25 21:35 eranian

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=7c86c4470811270947q585cba5vf8bdb875962a1856@mail.gmail.com \
    --to=eranian@googlemail.com \
    --cc=akpm@linux-foundation.org \
    --cc=andi@firstfloor.org \
    --cc=eranian@gmail.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=sfr@canb.auug.org.au \
    --cc=tglx@linutronix.de \
    --cc=x86@kernel.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: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®