perf/poll: another perf_fuzzer lockup
From: Vince Weaver
Date: Thu Nov 21 2013 - 13:51:36 EST
So I'm not sure what to make of this one. I was trying to reproduce the
sw_pagetable soft lockup and got the following instead. I think it's
perf_event related as the fuzzer does stomp on poll() with perf_event fds
but I am never sure exactly how to interpret the backtraces.
This is with 3.12 on a core2 machine.
Vince
[ 199.784001] ------------[ cut here ]------------
[ 199.784001] WARNING: CPU: 0 PID: 2801 at kernel/watchdog.c:245 watchdog_overflow_callback+0x9b/0xa6()
[ 199.784001] Watchdog detected hard LOCKUP on cpu 0
[ 199.784001] Modules linked in: cpufreq_userspace cpufreq_stats cpufreq_powersave cpufreq_conservative f71882fg evdev mcs7830 usbnet coretemp pcspkr psmouse serio_raw acpi_cpufreq wmi video ohci_pci processor button ohci_hcd i2c_nforce2 thermal_sys sg ehci_pci ehci_hcd sd_mod usbcore usb_common
[ 199.784001] CPU: 0 PID: 2801 Comm: perf_fuzzer Not tainted 3.12.0 #4
[ 199.784001] Hardware name: AOpen DE7000/nMCP7ALPx-DE R1.06 Oct.19.2012, BIOS 080015 10/19/2012
[ 199.784001] 00000000000000f5 ffff88011fc07bc8 ffffffff8151d8ec 00000000000000f5
[ 199.784001] ffff88011fc07c18 ffff88011fc07c08 ffffffff8103cda9 ffff88011fc07bf8
[ 199.784001] ffffffff810a137f ffff88011b313400 0000000000000000 ffff88011fc07d48
[ 199.784001] Call Trace:
[ 199.784001] <NMI> [<ffffffff8151d8ec>] dump_stack+0x49/0x5d
[ 199.784001] [<ffffffff8103cda9>] warn_slowpath_common+0x81/0x9b
[ 199.784001] [<ffffffff810a137f>] ? watchdog_overflow_callback+0x9b/0xa6
[ 199.784001] [<ffffffff8103ce66>] warn_slowpath_fmt+0x46/0x48
[ 199.784001] [<ffffffff810a137f>] watchdog_overflow_callback+0x9b/0xa6
[ 199.784001] [<ffffffff810cbb94>] __perf_event_overflow+0x137/0x1c1
[ 199.784001] [<ffffffff810cc23a>] perf_event_overflow+0x14/0x16
[ 199.784001] [<ffffffff81018fbd>] intel_pmu_handle_irq+0x2b8/0x34d
[ 199.784001] [<ffffffff81522253>] perf_event_nmi_handler+0x2d/0x4a
[ 199.784001] [<ffffffff81521b86>] nmi_handle+0x5e/0x13a
[ 199.784001] [<ffffffff81521d0a>] do_nmi+0xa8/0x2c0
[ 199.784001] [<ffffffff81521337>] end_repeat_nmi+0x1e/0x2e
[ 199.784001] [<ffffffff81520abf>] ? _raw_spin_lock+0x26/0x2a
[ 199.784001] [<ffffffff81520abf>] ? _raw_spin_lock+0x26/0x2a
[ 199.784001] [<ffffffff81520abf>] ? _raw_spin_lock+0x26/0x2a
[ 199.784001] <<EOE>> [<ffffffff81069ed5>] load_balance+0x2f2/0x609
[ 199.784001] [<ffffffff8106a3d8>] idle_balance+0x9a/0x10c
[ 199.784001] [<ffffffff8151f5a6>] __schedule+0x290/0x54b
[ 199.784001] [<ffffffff8105beae>] ? __hrtimer_start_range_ns+0x2ed/0x2ff
[ 199.784001] [<ffffffff8151fbbf>] schedule+0x64/0x66
[ 199.784001] [<ffffffff8151ed18>] schedule_hrtimeout_range_clock+0xed/0x134
[ 199.784001] [<ffffffff8105b659>] ? update_rmtp+0x65/0x65
[ 199.784001] [<ffffffff8105beee>] ? hrtimer_start_range_ns+0x14/0x16
[ 199.784001] [<ffffffff8151ed72>] schedule_hrtimeout_range+0x13/0x15
[ 199.784001] [<ffffffff8112274d>] poll_schedule_timeout+0x48/0x64
[ 199.784001] [<ffffffff81123453>] do_sys_poll+0x437/0x4d5
[ 199.784001] [<ffffffff811228cf>] ? __pollwait+0xcc/0xcc
[ 199.784001] [<ffffffff811228cf>] ? __pollwait+0xcc/0xcc
[ 199.784001] [<ffffffff811228cf>] ? __pollwait+0xcc/0xcc
[ 199.784001] [<ffffffff811228cf>] ? __pollwait+0xcc/0xcc
[ 199.784001] [<ffffffff811228cf>] ? __pollwait+0xcc/0xcc
[ 199.784001] [<ffffffff811228cf>] ? __pollwait+0xcc/0xcc
[ 199.784001] [<ffffffff811228cf>] ? __pollwait+0xcc/0xcc
[ 199.784001] [<ffffffff811228cf>] ? __pollwait+0xcc/0xcc
[ 199.784001] [<ffffffff811228cf>] ? __pollwait+0xcc/0xcc
[ 199.784001] [<ffffffff81123544>] SyS_poll+0x53/0xbc
[ 199.784001] [<ffffffff81527b56>] system_call_fastpath+0x1a/0x1f
[ 199.784001] ---[ end trace 1de1e48ddacc5178 ]---
[ 199.784001] perf samples too long (7070598 > 10000), lowering kernel.perf_event_max_sample_rate to 12500
[ 199.784001] INFO: NMI handler (perf_event_nmi_handler) took too long to run: 933.271 msecs
[ 199.784001] perf samples too long (7015378 > 20000), lowering kernel.perf_event_max_sample_rate to 6250
[ 199.784001] perf samples too long (6960587 > 40000), lowering kernel.perf_event_max_sample_rate to 3250
[ 199.784001] perf samples too long (6906224 > 76923), lowering kernel.perf_event_max_sample_rate to 1750
[ 199.784001] perf samples too long (6852286 > 142857), lowering kernel.perf_event_max_sample_rate to 1000
[ 199.784001] perf samples too long (6798770 > 250000), lowering kernel.perf_event_max_sample_rate to 500
[ 199.784001] perf samples too long (6745671 > 500000), lowering kernel.perf_event_max_sample_rate to 250
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/