[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1308009910.23688.34.camel@atheurer-ubuntu10>
Date: Mon, 13 Jun 2011 19:05:10 -0500
From: Andrew Theurer <habanero@...ux.vnet.ibm.com>
To: Steven Rostedt <rostedt@...dmis.org>
Cc: Peter Zijlstra <peterz@...radead.org>,
linux-kernel@...r.kernel.org, paulmck <paulmck@...ux.vnet.ibm.com>
Subject: Re: BUG on 3.0-rc on commit
d72bce0e67e8afc6eb959f656013cbb577426f1e
On Mon, 2011-06-13 at 11:33 -0400, Steven Rostedt wrote:
> On Mon, 2011-06-13 at 10:06 -0500, Andrew Theurer wrote:
> > On Fri, 2011-06-10 at 18:27 +0200, Peter Zijlstra wrote:
> > > On Fri, 2011-06-10 at 10:34 -0500, Andrew Theurer wrote:
> > > > RIP: 0010:[<ffffffff8104e8c1>] [<ffffffff8104e8c1>] find_lowest_rq+0xa1/0x150
> > >
> > > If you've still got the vmlinux around, could you find out where in
> > > find_lowest_rq that RIP is?
> >
> > Does this help?
>
> I take it back, it looks like this does help.
>
> >
> >
> > from objdump -d -S
> >
> > ffffffff8104e820 <find_lowest_rq>:
> > ffffffff8104e820: 55 push %rbp
> > ffffffff8104e821: 48 89 e5 mov %rsp,%rbp
> > ffffffff8104e824: 48 83 ec 30 sub $0x30,%rsp
> > ffffffff8104e828: 48 89 5d d8 mov %rbx,-0x28(%rbp)
> > ffffffff8104e82c: 4c 89 65 e0 mov %r12,-0x20(%rbp)
> > ffffffff8104e830: 4c 89 6d e8 mov %r13,-0x18(%rbp)
> > ffffffff8104e834: 4c 89 75 f0 mov %r14,-0x10(%rbp)
> > ffffffff8104e838: 4c 89 7d f8 mov %r15,-0x8(%rbp)
> > ffffffff8104e83c: e8 3f bf 48 00 callq ffffffff814da780 <mcount>
> > ffffffff8104e841: 48 c7 c0 88 e8 00 00 mov $0xe888,%rax
> > ffffffff8104e848: 65 48 03 04 25 50 dc add %gs:0xdc50,%rax
> > ffffffff8104e84f: 00 00
> > ffffffff8104e851: 65 44 8b 2c 25 58 dc mov %gs:0xdc58,%r13d
> > ffffffff8104e858: 00 00
> > ffffffff8104e85a: 83 bf bc 01 00 00 01 cmpl $0x1,0x1bc(%rdi)
> > ffffffff8104e861: 4c 8b 20 mov (%rax),%r12
> > ffffffff8104e864: 48 8b 47 08 mov 0x8(%rdi),%rax
> > ffffffff8104e868: 8b 58 18 mov 0x18(%rax),%ebx
> > ffffffff8104e86b: 75 23 jne ffffffff8104e890 <find_lowest_rq+0x70>
> > ffffffff8104e86d: b8 ff ff ff ff mov $0xffffffff,%eax
> > ffffffff8104e872: 48 8b 5d d8 mov -0x28(%rbp),%rbx
> > ffffffff8104e876: 4c 8b 65 e0 mov -0x20(%rbp),%r12
> > ffffffff8104e87a: 4c 8b 6d e8 mov -0x18(%rbp),%r13
> > ffffffff8104e87e: 4c 8b 75 f0 mov -0x10(%rbp),%r14
> > ffffffff8104e882: 4c 8b 7d f8 mov -0x8(%rbp),%r15
> > ffffffff8104e886: c9 leaveq
> > ffffffff8104e887: c3 retq
> > ffffffff8104e888: 0f 1f 84 00 00 00 00 nopl 0x0(%rax,%rax,1)
> > ffffffff8104e88f: 00
> > ffffffff8104e890: 89 d8 mov %ebx,%eax
> > ffffffff8104e892: 49 c7 c7 c0 2a 01 00 mov $0x12ac0,%r15
> > ffffffff8104e899: 48 89 fe mov %rdi,%rsi
> > ffffffff8104e89c: 48 8b 04 c5 40 a3 bf mov -0x7e405cc0(,%rax,8),%rax
> > ffffffff8104e8a3: 81
> > ffffffff8104e8a4: 4c 89 e2 mov %r12,%rdx
> > ffffffff8104e8a7: 49 8b 84 07 88 08 00 mov 0x888(%r15,%rax,1),%rax
> > ffffffff8104e8ae: 00
> > ffffffff8104e8af: 48 83 c0 38 add $0x38,%rax
> > ffffffff8104e8b3: 48 89 c7 mov %rax,%rdi
> > ffffffff8104e8b6: e8 85 75 0a 00 callq ffffffff810f5e40 <cpupri_find>
> > ffffffff8104e8bb: 85 c0 test %eax,%eax
> > ffffffff8104e8bd: 74 ae je ffffffff8104e86d <find_lowest_rq+0x4d>
> > ffffffff8104e8bf: 89 d8 mov %ebx,%eax
> > ffffffff8104e8c1: 41 0f a3 1c 24 bt %ebx,(%r12)
>
> Following the asm here, I figured we are here:
>
> static int find_lowest_rq(struct task_struct *task)
> {
> struct sched_domain *sd;
> struct cpumask *lowest_mask = __get_cpu_var(local_cpu_mask);
> int this_cpu = smp_processor_id();
> int cpu = task_cpu(task);
>
> if (task->rt.nr_cpus_allowed == 1)
> return -1; /* No other targets possible */
>
> if (!cpupri_find(&task_rq(task)->rd->cpupri, task, lowest_mask))
> return -1; /* No targets found */
>
> /*
> * At this point we have built a mask of cpus representing the
> * lowest priority tasks in the system. Now we want to elect
> * the best one based on our affinity and topology.
> *
> * We prioritize the last cpu that the task executed on since
> * it is most likely cache-hot in that location.
> */
> if (cpumask_test_cpu(cpu, lowest_mask)) <<<----------- HERE
> return cpu;
>
>
> Looks to me that lowest_mask is NULL???
>
> Does this happen on boot up? Hmm, could be, by looking at Peter's patch,
> he could be waking them up and we are getting into this code before
> lowest_mask is defined.
>
> Could you try this patch?
Yes, and it works! Thanks very much!
>
> -- Steve
-Andrew
>
> diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c
> index 88725c9..46e3e09 100644
> --- a/kernel/sched_rt.c
> +++ b/kernel/sched_rt.c
> @@ -1239,6 +1239,10 @@ static int find_lowest_rq(struct task_struct *task)
> int this_cpu = smp_processor_id();
> int cpu = task_cpu(task);
>
> + /* Make sure everything is initialized first */
> + if (system_state != SYSTEM_RUNNING)
> + return -1;
> +
> if (task->rt.nr_cpus_allowed == 1)
> return -1; /* No other targets possible */
>
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists