public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Mike Travis <travis@sgi.com>
To: Rusty Russell <rusty@rustcorp.com.au>
Cc: Ingo Molnar <mingo@elte.hu>, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 3/3] Fix locking error caused by e76bd8d9850c2296a7e8e24c9dce9b5e6b55fe2f
Date: Tue, 25 Nov 2008 08:52:49 -0800	[thread overview]
Message-ID: <492C2D61.2050406@sgi.com> (raw)
In-Reply-To: <200811250959.20652.rusty@rustcorp.com.au>

Rusty Russell wrote:
> We can't call cpuset_cpus_allowed_locked() with the rq lock held.
> However, the rq lock merely protects us from (1) cpu_online_mask changing
> and (2) someone else changing p->cpus_allowed.
> 
> The first can't happen because we're being called from a cpu hotplug
> notifier.  The second doesn't really matter: we are forcing the task off
> a CPU it was affine to, so we're not doing very well anyway.
> 
> So we remove the rq lock from this path, and all is good.
> 
> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Tested-by: Mike Travis <travis@sgi.com>

 112# ./offline-test
possible cpus: 0-3, last cpu: 3
=== 0 > /sys/devices/system/cpu/cpu1/online
=== 0 > /sys/devices/system/cpu/cpu2/online
=== 0 > /sys/devices/system/cpu/cpu3/online
online cpus: 0, last cpu: 0
=== 1 > /sys/devices/system/cpu/cpu1/online
=== 1 > /sys/devices/system/cpu/cpu2/online
=== 1 > /sys/devices/system/cpu/cpu3/online

[ 1142.012421] numa_remove_cpu cpu 1 node 0: mask now 0
[ 1142.013646] CPU 1 is now offline
[ 1143.050422] numa_remove_cpu cpu 2 node 1: mask now 3
[ 1143.051661] CPU 2 is now offline
[ 1144.089438] numa_remove_cpu cpu 3 node 1: mask now
[ 1144.090661] CPU 3 is now offline
[ 1144.115065] SMP alternatives: switching to UP code


[ 1145.234266] SMP alternatives: switching to SMP code
[ 1145.370548] Booting processor 1 APIC 0x1 ip 0x6000
[ 1142.038605] Initializing CPU#1
[ 1142.038605] Calibrating delay using timer specific routine.. 4000.04 BogoMIPS (lpj=2000023)
[ 1142.038605] CPU: L1 I Cache: 64K (64 bytes/line), D cache 64K (64 bytes/line)
[ 1142.038605] CPU: L2 Cache: 1024K (64 bytes/line)
[ 1142.038605] CPU 1/0x1 -> Node 0
[ 1142.038605] CPU: Physical Processor ID: 0
[ 1142.038605] CPU: Processor Core ID: 1
[ 1142.038605] numa_add_cpu cpu 1 node 0: mask now 0-1
[ 1142.038605] x86 PAT enabled: cpu 1, old 0x7040600070406, new 0x7010600070106
[ 1145.455476] CPU1: <7>Switched to high resolution mode on CPU 1
[ 1145.610706] Dual-Core AMD Opteron(tm) Processor 2212 stepping 02
[ 1146.637468] Booting processor 2 APIC 0x2 ip 0x6000
[ 1143.076576] Initializing CPU#2
[ 1143.076576] Calibrating delay using timer specific routine.. 4000.09 BogoMIPS (lpj=2000049)
[ 1143.076576] CPU: L1 I Cache: 64K (64 bytes/line), D cache 64K (64 bytes/line)
[ 1143.076576] CPU: L2 Cache: 1024K (64 bytes/line)
[ 1143.076576] CPU 2/0x2 -> Node 1
[ 1143.076576] CPU: Physical Processor ID: 1
[ 1143.076576] CPU: Processor Core ID: 0
[ 1143.076576] numa_add_cpu cpu 2 node 1: mask now 2
[ 1143.076576] x86 PAT enabled: cpu 2, old 0x7040600070406, new 0x7010600070106
[ 1146.722688] CPU2: <7>Switched to high resolution mode on CPU 2
[ 1146.877389] Dual-Core AMD Opteron(tm) Processor 2212 stepping 02
[ 1147.903780] Booting processor 3 APIC 0x3 ip 0x6000
[ 1144.223262] Initializing CPU#3
[ 1144.223262] Calibrating delay using timer specific routine.. 4000.07 BogoMIPS (lpj=2000035)
[ 1144.223262] CPU: L1 I Cache: 64K (64 bytes/line), D cache 64K (64 bytes/line)
[ 1144.223262] CPU: L2 Cache: 1024K (64 bytes/line)
[ 1144.223262] CPU 3/0x3 -> Node 1
[ 1144.223262] CPU: Physical Processor ID: 1
[ 1144.223262] CPU: Processor Core ID: 1
[ 1144.223262] numa_add_cpu cpu 3 node 1: mask now 2-3
[ 1144.223262] x86 PAT enabled: cpu 3, old 0x7040600070406, new 0x7010600070106
[ 1147.988822] CPU3: <7>Switched to high resolution mode on CPU 3
[ 1148.144178] Dual-Core AMD Opteron(tm) Processor 2212 stepping 02


>  113# cat offline-test
#!/bin/bash

get_cpus()
{
        local which=$1; shift
        local list=$(cat /sys/devices/system/cpu/$which)
        local last=${list#0-}

        echo "$which cpus: $list, last cpu: $last" >/dev/stderr
        echo $last
}

set_cpu()
{
        local i state
        for ((i = 1; i <= $2; i++ ))
        do
                state=$(cat /sys/devices/system/cpu/cpu$i/online)
                [ $state != $1 ] && {
                        echo === "$1 > /sys/devices/system/cpu/cpu$i/online"
                        echo $1 > /sys/devices/system/cpu/cpu$i/online
                        sleep 1
                }
        done
}

typeset last=$(get_cpus possible)

set_cpu 0 $last

typeset ignore=$(get_cpus online)

set_cpu 1 $last

> diff --git a/kernel/sched.c b/kernel/sched.c
> index 1aa840a..7acf95f 100644
> --- a/kernel/sched.c
> +++ b/kernel/sched.c
> @@ -6126,7 +6126,6 @@ static int __migrate_task_irq(struct task_struct *p, int src_cpu, int 
> dest_cpu)
>   */
>  static void move_task_off_dead_cpu(int dead_cpu, struct task_struct *p)
>  {
> -	unsigned long flags;
>  	struct rq *rq;
>  	int dest_cpu;
>  	/* FIXME: Use cpumask_of_node here. */
> @@ -6146,10 +6145,8 @@ again:
>  
>  	/* No more Mr. Nice Guy. */
>  	if (dest_cpu >= nr_cpu_ids) {
> -		rq = task_rq_lock(p, &flags);
>  		cpuset_cpus_allowed_locked(p, &p->cpus_allowed);
>  		dest_cpu = cpumask_any_and(cpu_online_mask, &p->cpus_allowed);
> -		task_rq_unlock(rq, &flags);
>  
>  		/*
>  		 * Don't tell them about moving exiting tasks or


  reply	other threads:[~2008-11-25 16:53 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-11-24 16:08 [PULL] cpumask conversion patches for sched Rusty Russell
2008-11-24 17:17 ` Ingo Molnar
2008-11-24 17:22   ` Ingo Molnar
2008-11-24 17:23   ` Ingo Molnar
2008-11-24 19:19     ` [PATCH] sched: fix-local_cpu_mask Mike Travis
2008-11-25 15:10       ` Mike Travis
2008-11-24 23:27     ` [PULL] cpumask conversion patches for sched Rusty Russell
2008-11-26  6:59       ` Ingo Molnar
2008-11-24 23:27     ` [PATCH 1/3] Fix commits 7d1e6a9b and dcc30a35 for UP build Rusty Russell
2008-11-24 23:28       ` [PATCH 2/3] Fix commit 0e3900e6 " Rusty Russell
2008-11-24 23:29         ` [PATCH 3/3] Fix locking error caused by e76bd8d9850c2296a7e8e24c9dce9b5e6b55fe2f Rusty Russell
2008-11-25 16:52           ` Mike Travis [this message]
2008-11-25 16:55           ` [PATCH 1/1] Slight fix for locking-fix patch to remove compiler warning Mike Travis
2008-11-26  7:56             ` Rusty Russell

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=492C2D61.2050406@sgi.com \
    --to=travis@sgi.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=rusty@rustcorp.com.au \
    /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