From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Rafael J. Wysocki" Subject: Re: kernel oops and panic in acpi_atomic_read under 2.6.39.3. call trace included Date: Mon, 22 Aug 2011 20:47:10 +0200 Message-ID: <201108222047.11086.rjw@sisk.pl> References: <201108171751.51648.rick@microway.com> <201108182343.57312.rjw@sisk.pl> <8338f9871d1f52f9376a1eca319a7866.squirrel@www.microway.com> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: Received: from ogre.sisk.pl ([217.79.144.158]:57834 "EHLO ogre.sisk.pl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752204Ab1HVSp0 (ORCPT ); Mon, 22 Aug 2011 14:45:26 -0400 In-Reply-To: <8338f9871d1f52f9376a1eca319a7866.squirrel@www.microway.com> Sender: linux-acpi-owner@vger.kernel.org List-Id: linux-acpi@vger.kernel.org To: rick@microway.com Cc: linux-kernel@vger.kernel.org, Richard Houghton , ACPI Devel Mailing List , Len Brown , Matthew Garrett On Monday, August 22, 2011, rick@microway.com wrote: > > On Thursday, August 18, 2011, Rafael J. Wysocki wrote: > >> Hi, > >> > >> It's better to report ACPI failues to linux-acpi. > >> > >> On Wednesday, August 17, 2011, Rick Warner wrote: > >> > Hi all, > >> > > >> > I am getting a kernel oops/panic on a dual xeon system that is the > >> master of a > >> > 60 node HPC cluster. This is happening while stress testing the > >> system > >> > including significant network traffic. The OS is openSuse 11.4. > >> > > >> > We are running a custom compiled 2.6.39.3 kernel on the systems due to > >> a bug > >> > in the stock kernel 11.4 provided (igb driver related). After 1-3 > >> days of > >> > heavy testing, the master node locks up with the caps lock and scroll > >> lock > >> > keys on the keyboard blinking with the following output captured via a > >> serial > >> > console: > >> > > >> > [381920.681113] BUG: unable to handle kernel NULL pointer dereference > >> at > >> > (null) > >> > [381920.689067] IP: [] acpi_atomic_read+0xe3/0x120 > >> > [381920.695187] PGD 30c27a067 PUD 16efe6067 PMD 0 > >> > [381920.699782] Oops: 0000 [#1] PREEMPT SMP > >> > [381920.703866] last sysfs file: > >> > /sys/devices/system/cpu/cpu15/cache/index2/shared_cpu_map > >> > [381920.711868] CPU 6 > >> > [381920.713800] Modules linked in: md5 ipmi_devintf ipmi_si > >> ipmi_msghandler > >> > nfsd lockd nfs_acl auth_rpcgss sunrpc ipt_MASQUERADE iptable_mangle > >> > iptable_nat nf_nat nf_conntrack_ipv4 nf_conntrack nf_defrag_ipv4 > >> iptable_filter > >> > ip_tables x_tables af_packet edd cpufreq_conservative > >> cpufreq_userspace > >> > cpufreq_powersave acpi_cpufreq mperf xfs dm_mod ioatdma i7core_edac > >> edac_core > >> > sr_mod cdrom joydev igb i2c_i801 sg button ghes hed iTCO_wdt > >> > iTCO_vendor_support dca pcspkr ext4 jbd2 crc16 raid456 > >> async_raid6_recov > >> > async_pq raid6_pq async_xor xor async_memcpy async_tx raid10 raid1 > >> raid0 fan > >> > processor thermal thermal_sys ata_generic pata_atiixp arcmsr > >> > [381920.771623] > >> > [381920.773210] Pid: 12701, comm: cluster Not tainted > >> 2.6.39.3-microwaycustom > >> > #1 Supermicro X8DTH-i/6/iF/6F/X8DTH > >> > [381920.783292] RIP: 0010:[] [] > >> > acpi_atomic_read+0xe3/0x120 > >> > [381920.791853] RSP: 0000:ffff88063fc47d98 EFLAGS: 00010046 > >> > [381920.797260] RAX: 0000000000000000 RBX: 00000000bf7b5ff0 RCX: > >> ffffffff81a3cdd0 > >> > [381920.804486] RDX: 00000000bf7b6010 RSI: 00000000bf7b6000 RDI: > >> > ffff88062d4b95c0 > >> > [381920.811712] RBP: ffff88063fc47dc8 R08: ffff88063fc47d98 R09: > >> 0000000000000002 > >> > [381920.818940] R10: 0000000000000083 R11: 0000000000000010 R12: > >> > ffffc90003044c20 > >> > [381920.826168] R13: ffff88063fc47de0 R14: 0000000000000000 R15: > >> > 0000000000000000 > >> > [381920.833392] FS: 0000000000000000(0000) GS:ffff88063fc40000(0000) > >> > knlGS:0000000000000000 > >> > [381920.841571] CS: 0010 DS: 002b ES: 002b CR0: 0000000080050033 > >> > [381920.847409] CR2: 0000000000000000 CR3: 00000001d2d0f000 CR4: > >> > 00000000000006e0 > >> > [381920.854635] DR0: 0000000000000000 DR1: 0000000000000000 DR2: > >> > 0000000000000000 > >> > [381920.861861] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: > >> 0000000000000400 > >> > [381920.869088] Process cluster (pid: 12701, threadinfo > >> ffff880028284000, task > >> > ffff880550184380) > >> > [381920.877608] Stack: > >> > [381920.879717] ffffffff81a3cdd0 00000000bf7b5ff0 ffff88062b2cc140 > >> ffffc90003044ca8 > >> > [381920.887288] ffff88062b2cc140 0000000000000001 ffff88063fc47e08 > >> ffffffffa002b21f > >> > [381920.894862] 0000000000000000 0000000000000000 ffff88062b2cc140 > >> > 0000000000000000 > >> > [381920.902435] Call Trace: > >> > [381920.904969] > >> > [381920.907201] [] ghes_read_estatus+0x2f/0x170 > >> [ghes] > >> > [381920.913726] [] ghes_notify_nmi+0xd8/0x1b0 > >> [ghes] > >> > [381920.920076] [] notifier_call_chain+0x3f/0x80 > >> > [381920.926080] [] > >> __atomic_notifier_call_chain+0x48/0x70 > >> > [381920.932864] [] > >> atomic_notifier_call_chain+0x11/0x20 > >> > [381920.939473] [] notify_die+0x2e/0x30 > >> > [381920.944691] [] do_nmi+0xa2/0x270 > >> > [381920.949649] [] nmi+0x20/0x30 > >> > [381920.954269] [] ? native_write_msr_safe+0xa/0x10 > >> > [381920.960536] <> > >> > [381920.962725] > >> > [381920.965132] [] intel_pmu_disable_all+0x3e/0x120 > >> > [381920.971399] [] x86_pmu_disable+0x4a/0x50 > >> > [381920.977066] [] perf_pmu_disable+0x2b/0x40 > >> > [381920.982817] [] perf_event_task_tick+0x218/0x270 > >> > [381920.989082] [] scheduler_tick+0xdd/0x2c0 > >> > [381920.994740] [] update_process_times+0x67/0x80 > >> > [381921.000832] [] tick_sched_timer+0x5f/0xc0 > >> > [381921.006578] [] ? tick_nohz_handler+0x100/0x100 > >> > [381921.012758] [] __run_hrtimer+0x12d/0x280 > >> > [381921.018414] [] hrtimer_interrupt+0xb7/0x1e0 > >> > [381921.024343] [] > >> smp_apic_timer_interrupt+0x67/0xa0 > >> > [381921.030793] [] apic_timer_interrupt+0x13/0x20 > >> > [381921.036891] > >> > [381921.038913] Code: fc 10 74 1f 77 08 41 80 fc 08 75 48 eb 0e 41 80 > >> fc 20 74 > >> > 17 41 80 fc 40 75 3a eb 15 8a 00 0f b6 c0 eb 11 66 8b 00 0f b7 c0 eb > >> 09 <8b> > >> > 00 89 c0 eb 03 48 8b 00 49 89 45 00 e8 8e 2b e2 ff eb 1b 0f > >> > [381921.059462] RIP [] acpi_atomic_read+0xe3/0x120 > >> > [381921.065669] RSP > >> > [381921.069242] CR2: 0000000000000000 > >> > [381921.072645] ---[ end trace 52697bfc73a34a90 ]--- > >> > [381921.077343] Kernel panic - not syncing: Fatal exception in > >> interrupt > >> > [381921.083784] Pid: 12701, comm: cluster Tainted: G D > >> 2.6.39.3- > >> > microwaycustom #1 > >> > [381921.091788] Call Trace: > >> > [381921.094333] [] panic+0x9f/0x1da > >> > [381921.099881] [] oops_end+0xdc/0xf0 > >> > [381921.104952] [] no_context+0xf1/0x260 > >> > [381921.110277] [] > >> __bad_area_nosemaphore+0x155/0x200 > >> > [381921.116739] [] bad_area_nosemaphore+0xe/0x10 > >> > [381921.122766] [] do_page_fault+0x366/0x530 > >> > [381921.128440] [] ? > >> __perf_event_overflow+0xa9/0x220 > >> > [381921.134896] [] ? > >> perf_event_update_userpage+0x9b/0xe0 > >> > [381921.141698] [] ? > >> intel_pmu_enable_all+0xc9/0x1a0 > >> > [381921.148061] [] ? > >> x86_perf_event_set_period+0xdf/0x170 > >> > [381921.154852] [] page_fault+0x25/0x30 > >> > [381921.160081] [] ? acpi_atomic_read+0xe3/0x120 > >> > [381921.166094] [] ? acpi_atomic_read+0x59/0x120 > >> > [381921.172107] [] ghes_read_estatus+0x2f/0x170 > >> [ghes] > >> > [381921.178639] [] ghes_notify_nmi+0xd8/0x1b0 > >> [ghes] > >> > [381921.185004] [] notifier_call_chain+0x3f/0x80 > >> > [381921.191003] [] > >> __atomic_notifier_call_chain+0x48/0x70 > >> > [381921.197789] [] > >> atomic_notifier_call_chain+0x11/0x20 > >> > [381921.204399] [] notify_die+0x2e/0x30 > >> > [381921.209624] [] do_nmi+0xa2/0x270 > >> > [381921.214584] [] nmi+0x20/0x30 > >> > [381921.219208] [] ? native_write_msr_safe+0xa/0x10 > >> > [381921.225470] <> [] > >> > intel_pmu_disable_all+0x3e/0x120 > >> > [381921.233174] [] x86_pmu_disable+0x4a/0x50 > >> > [381921.238836] [] perf_pmu_disable+0x2b/0x40 > >> > [381921.244582] [] perf_event_task_tick+0x218/0x270 > >> > [381921.250855] [] scheduler_tick+0xdd/0x2c0 > >> > [381921.256520] [] update_process_times+0x67/0x80 > >> > [381921.262618] [] tick_sched_timer+0x5f/0xc0 > >> > [381921.268363] [] ? tick_nohz_handler+0x100/0x100 > >> > [381921.274540] [] __run_hrtimer+0x12d/0x280 > >> > [381921.280198] [] hrtimer_interrupt+0xb7/0x1e0 > >> > [381921.286125] [] > >> smp_apic_timer_interrupt+0x67/0xa0 > >> > [381921.292578] [] apic_timer_interrupt+0x13/0x20 > >> > [381921.298673] > >> > > >> > > >> > I recompiled the kernel again, disabling the tickless feature as I saw > >> "nohz" > >> > early in the call trace. After that, it reproduced again but the call > >> trace > >> > only changed slightly having tick_init_highres in there instead of > >> > tick_nohz_handler. I have that call trace if it is desired as well. > >> It is > >> > nearly identical though. > >> > > >> > I am currently trying a custom 2.6.36.4 on the system, but will need > >> up to 3 > >> > days before I know if the problem exists there as well. > >> > > >> > Any ideas on this? > >> > >> It looks like a wrong address in ghes_read_estatus(). I'll have a look > >> at it > >> later today. > > > > Hmm. Please apply the appended patch and see if the error is still > > reproducible (it's on top of the current mainline, so you may need to > > adjust it by hand). > > > > Thanks, > > Rafael > > > > --- > > drivers/acpi/apei/ghes.c | 3 +++ > > 1 file changed, 3 insertions(+) > > > > Index: linux/drivers/acpi/apei/ghes.c > > =================================================================== > > --- linux.orig/drivers/acpi/apei/ghes.c > > +++ linux/drivers/acpi/apei/ghes.c > > @@ -398,6 +398,9 @@ static int ghes_read_estatus(struct ghes > > u32 len; > > int rc; > > > > + if (!g) > > + return -EINVAL; > > + > > rc = acpi_atomic_read(&buf_paddr, &g->error_status_address); > > if (rc) { > > if (!silent && printk_ratelimit()) > > > > Thanks for the quick patch! Unfortunately, the system crashed again with > the same call trace. I'll list it below in case there are any differences > I missed. I have currently added a test for (!&g->error_status_address) > just like your test on (!g) to verify that isn't a null pointer either, > and will start tests back up with that change. Please let me know if you > have any other suggestions or patches for me to try. That was a long shot, but perhaps you can check what source code corresponds to ghes_read_estatus+0x38 in your kernel? Rafael