linux-kernel - Re: [BUG] soft lockup while booting machine with more than 700 cores

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20110210205648.GA10341@sgi.com>
Date:	Thu, 10 Feb 2011 14:56:48 -0600
From:	Jack Steiner <steiner@....com>
To:	Ingo Molnar <mingo@...e.hu>
Cc:	raz ben yehuda <raz@...lemp.com>, linux-kernel@...r.kernel.org,
	mingo@...hat.com, Peter Zijlstra <a.p.zijlstra@...llo.nl>,
	Mike Galbraith <efault@....de>, Cliff Wickman <cpw@....com>,
	Mike Travis <travis@....com>,
	Thomas Gleixner <tglx@...utronix.de>,
	"H. Peter Anvin" <hpa@...or.com>
Subject: Re: [BUG] soft lockup while booting machine with more than 700
	cores

On Thu, Feb 10, 2011 at 01:39:37PM +0100, Ingo Molnar wrote:
> 
> * raz ben yehuda <raz@...lemp.com> wrote:
> 
> > Mingo Hello
> > 
> > Bellow is a boot of a 2.6.32.19 kernel over a machine with more than 700 cores. I 
> > am failing to boot it due to a soft lockup in rebalance_domains area. I did not 
> > find anything related in mainline git and kernel's bugzilla.

Not sure what boot options you are using. We saw similar problems on large SGI UV
systems. See if booting with "nohz=off" helps (you might already have this).

We also noticed that the rebalance_domains() code references many per-cpu
run queue structures. All of the structures have identical offsets relative
to the size of a cache leaf. The result is that all index into the same lines in the
L3 caches. That causes many evictions. We tried an experimental to
stride the run queues at 128 byte offsets. That helped in some cases but the
results were mixed.  We are still experimenting with the patch.

--- jack



> > 
> > thank you
> > Raz
> > 
> > 
> >  [  929.614315] TCP cubic registered 
> >  [  929.614577] NET: Registered protocol family 17 
> >  [  930.785915] Bridge firewalling registered 
> >  [  930.928396] Freeing unused kernel memory: 1380k freed 
> >  =============================================================================== 
> >  Running /disklessrc 
> >  Mounting /proc 
> >  Creating /dev 
> >  Creating initial device nodes 
> > [  931.327841] usb 5-1: configuration #1 chosen from 1 choice 
> > [  931.657469] input: HP Virtual Keyboard as /class/input/input0 
> > [  931.671560] generic-usb 0003:03F0:1027.0001: input: USB HID v1.01 Keyboard [H
> > P Virtual Keyboard] on usb-0000:01:04.0-1/input0 
> >  [  931.911480] input: HP Virtual Keyboard as /class/input/input1 
> >  [  931.926135] generic-usb 0003:03F0:1027.0002: input: USB HID v1.01 Mouse [HP V
> >  irtual Keyboard] on usb-0000:01:04.0-1/input1 
> >  [  932.247432] scsi 0:0:0:0: Direct-Access     Generic  USB Flash Disk   0.00 PQ
> >  : 0 ANSI: 2 
> >  [  932.301626] sd 0:0:0:0: Attached scsi generic sg0 type 0 
> >  [  932.416279] sd 0:0:0:0: [sda] 7892992 512-byte logical blocks: (4.04 GB/3.76 
> >  GiB) 
> >  [  932.559424] sd 0:0:0:0: [sda] Write Protect is off 
> >  [  932.563238] sd 0:0:0:0: [sda] Assuming drive cache: write through 
> >  [  932.802006] sd 0:0:0:0: [sda] Assuming drive cache: write through 
> >  [  932.805070]  sda: sda1 
> >  [  934.315071] sd 0:0:0:0: [sda] Assuming drive cache: write through 
> >  [  934.318055] sd 0:0:0:0: [sda] Attached SCSI removable disk 
> >  Loading nfs module... [ 1011.681028] BUG: soft lockup - CPU#240 stuck for 62s! [
> >  swapper:0] 
> >  [ 1011.744482] Modules linked in: sunrpc(+) 
> >  [ 1011.789117] CPU 240: 
> >  [ 1011.828757] Modules linked in: sunrpc(+) 
> >  [ 1011.874003] Pid: 0, comm: swapper Not tainted 2.6.32.19-3.vSMP #2 vSMP 3.5 
> >  [ 1011.935843] RIP: 0010:[<ffffffff8105ac32>]  [<ffffffff8105ac32>] weighted_cpu
> >  load+0x12/0x20 
> >  [ 1012.051597] RSP: 0018:ffff89468e803c88  EFLAGS: 00010286 
> >  [ 1012.115020] RAX: 00000000000115c0 RBX: 0000000000000002 RCX: 000000000000001d
> >  [ 1012.162897] RDX: ffff8acd2e840000 RSI: 0000000000000002 RDI: 000000000000021d
> >  [ 1012.243858] RBP: ffffffff81033133 R08: 0000000000000200 R09: ffff894f0ca3d450
> >  [ 1012.309760] R10: 0000000000000000 R11: ffff89468e803dc0 R12: ffff89468e803c00
> >  [ 1012.358023] R13: 00000000000115c0 R14: ffffffff8104b6dc R15: ffffffff81046ea6
> >  [ 1012.417072] FS:  0000000000000000(0000) GS:ffff89468e800000(0000) knlGS:00000
> >  00000000000 
> >  [ 1012.494488] CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b 
> >  [ 1012.559412] CR2: 00000000008d3988 CR3: 0000000001001000 CR4: 00000000000026e0
> >  [ 1012.619828] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> >  [ 1012.675491] DR3: 0000000000000000 DR6: 0000000000000000 DR7: 0000000000000000
> >  [ 1012.739386] Call Trace: 
> >  [ 1012.790082]  <IRQ>  [<ffffffff81039705>] ? sched_clock+0x5/0x10 
> >  [ 1012.868687]  [<ffffffff8105ac6b>] ? source_load+0x2b/0x70 
> >  [ 1012.923473]  [<ffffffff810602d5>] ? find_busiest_group+0x1b5/0xa30 
> >  [ 1012.973482]  [<ffffffff81063487>] ? rebalance_domains+0x117/0x470 
> >  [ 1013.031838]  [<ffffffff81065a4e>] ? run_rebalance_domains+0x3e/0xe0 
> >  [ 1013.081837]  [<ffffffff8106fbbe>] ? __do_softirq+0xae/0x140 
> >  [ 1013.134496]  [<ffffffff81085da0>] ? ktime_get+0x50/0xd0 
> > [ 1013.182834]  [<ffffffff8103374c>] ? call_softirq+0x1c/0x30 
> >  [ 1013.246263]  [<ffffffff81035745>] ? do_softirq+0x65/0xa0 
> >  [ 1013.314801]  [<ffffffff8106fb0c>] ? irq_exit+0x7c/0x80 
> >  [ 1013.355605]  [<ffffffff81046eab>] ? smp_apic_timer_interrupt+0x6b/0xa0 
> >  [ 1013.391166]  [<ffffffff8104b6dc>] ? native_apic_msr_write+0x2c/0x40 
> >  [ 1013.391166]  [<ffffffff81033133>] ? apic_timer_interrupt+0x13/0x20 
> >  [ 1013.478307]  <EOI>  [<ffffffff8104dc92>] ? native_safe_halt+0x2/0x10 
> >  [ 1013.515916]  [<ffffffff8103a481>] ? default_idle+0x21/0x40 
> >  [ 1013.572168]  [<ffffffff81031537>] ? cpu_idle+0x57/0x90 
> >  [ 1112.445978] BUG: soft lockup - CPU#240 stuck for 62s! [swapper:0] 
> >  [ 1112.445978] Modules linked in: sunrpc(+) 
> 
> Interesting.
> 
> Could you boot up with just enough cores for it to not lock up, and run perf top and 
> see where the overhead is?
> 
> Thanks,
> 
> 	Ingo
--
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/