On Sunday, September 01, 2013 11:54:10 AM Viresh Kumar wrote: > On 31 August 2013 06:06, Stephen Boyd <sboyd@xxxxxxxxxxxxxx> wrote: > > Yes that patch may reduce the chance of the race condition but I > > don't believe it removes it entirely. I believe this bug still > > exists in linux-next. Consider the scenario where CPU1 is going > > down. > > > > __cpufreq_remove_dev() > > ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP); > > __cpufreq_governor() > > policy->governor->governor(policy, CPUFREQ_GOV_STOP); > > cpufreq_governor_dbs() > > case CPUFREQ_GOV_STOP: > > mutex_destroy(&cpu_cdbs->timer_mutex) > > cpu_cdbs->cur_policy = NULL; > > <PREEMPT> > > store() > > __cpufreq_set_policy() > > ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS); > > __cpufreq_governor() > > policy->governor->governor(policy, CPUFREQ_GOV_LIMITS); > > case CPUFREQ_GOV_LIMITS: > > mutex_lock(&cpu_cdbs->timer_mutex); <-- Warning (destroyed mutex) > > if (policy->max < cpu_cdbs->cur_policy->cur) <- cur_policy == NULL > > Some of the crashes you reported would be fixed by the patches I sent > today morning. > > Let me know if anything else is left for latest linux-next... > > Btw, I am facing another crash which I am not sure how to fix.. It > came with your script: This isn't a crash, but a WARN_ON_ONCE() triggering. The comment in kref_get() explains when that occurs, so we seem to have a race between store_scaling_min_freq() and CPU removal. > [ 190.850481] ------------[ cut here ]------------ > [ 190.850489] WARNING: CPU: 3 PID: 14140 at > /home/arm/work/kernel/mywork/linux.git/include/linux/kref.h:47 > kobject_get+0x42/0x50() > [ 190.850490] Modules linked in: nfsd nfs fscache lockd arc4 iwldvm > mac80211 i915 iwlwifi drm_kms_helper nfs_acl auth_rpcgss cfg80211 > sunrpc drm joyd > ev thinkpad_acpi snd_hda_codec_hdmi snd_seq_midi > snd_hda_codec_conexant oid_registry btusb snd_rawmidi snd_hda_intel > snd_hda_codec i2c_algo_bit rfcomm > snd_seq_midi_event bnep psmouse snd_seq snd_hwdep snd_pcm bluetooth > snd_timer snd_seq_device parport_pc snd_page_alloc ppdev tpm_tis snd > soundcore lp > c_ich lp parport video serio_raw mac_hid wmi nvram binfmt_misc btrfs > raid6_pq e1000e ptp pps_core xor sdhci_pci sdhci zlib_deflate > libcrc32c > [ 190.850563] CPU: 3 PID: 14140 Comm: sh Not tainted 3.11.0-rc7-custom #39 > [ 190.850567] Hardware name: LENOVO 4236G50/4236G50, BIOS 83ET70WW > (1.40 ) 06/12/2012 > [ 190.850571] 000000000000002f ffff8800c8bdfc38 ffffffff816746c3 > 0000000000000007 > [ 190.850580] 0000000000000000 ffff8800c8bdfc78 ffffffff8104cf8c > ffff88011e5f9b18 > [ 190.850587] ffff880118eaf000 0000000000000001 0000000000000202 > 0000000000000008 > [ 190.850593] Call Trace: > [ 190.850607] [<ffffffff816746c3>] dump_stack+0x46/0x58 > [ 190.850615] [<ffffffff8104cf8c>] warn_slowpath_common+0x8c/0xc0 > [ 190.850622] [<ffffffff8104cfda>] warn_slowpath_null+0x1a/0x20 > [ 190.850629] [<ffffffff81324e02>] kobject_get+0x42/0x50 > [ 190.850638] [<ffffffff81533ab0>] cpufreq_cpu_get+0x80/0xc0 > [ 190.850647] [<ffffffff81533c11>] cpufreq_get_policy+0x21/0x120 > [ 190.850655] [<ffffffff81533fdf>] store_scaling_min_freq+0x3f/0xa0 > [ 190.850666] [<ffffffff816785b6>] ? down_write+0x16/0x40 > [ 190.850674] [<ffffffff81533000>] store+0x70/0xb0 > [ 190.850683] [<ffffffff811f2582>] sysfs_write_file+0xe2/0x170 > [ 190.850693] [<ffffffff81181e8e>] vfs_write+0xce/0x200 > [ 190.850700] [<ffffffff81182392>] SyS_write+0x52/0xa0 > [ 190.850707] [<ffffffff81683882>] system_call_fastpath+0x16/0x1b -- I speak only for myself. Rafael J. Wysocki, Intel Open Source Technology Center. -- To unsubscribe from this list: send the line "unsubscribe cpufreq" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
