2016-11-15 21:19 GMT+03:00 Fabian Vogt <[email protected]>: > Hi! > > Am Dienstag, 15. November 2016, 20:13:06 CET schrieb Matwey V. Kornilov: >> Hello, >> >> I am running 4.4.24-4-default from devel:ARM:Factory:Contrib:RaspberryPi3 >> >> and sometimes see the following when do readings from /sys like >> >> # cat /sys/devices/system/cpu/cpu?/cpufreq/cpuinfo_cur_freq > > Interesting, I haven't seen that error during tests. IceWM polls the current > frequency and I also ran cpupower in a loop, so I guess I would've hit it. > Is there a reliable way to reproduce it? > Just "while true; cat cpuinfo_cur_freq; done" ? >
No reliable way, sometimes while loop works well, sometimes it hangs at first cat after boot. >> [ 9138.093049] INFO: task cat:4444 blocked for more than 50 seconds. >> [ 9138.093058] Not tainted 4.4.24-4-default #1 >> [ 9138.093060] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" >> disables this message. >> [ 9138.093073] cat D ffff800000086c28 0 4444 4426 >> 0x00000001 >> [ 9138.093076] Call trace: >> [ 9138.093090] [<ffff800000086c28>] __switch_to+0x90/0xa8 >> [ 9138.093097] [<ffff800000749d14>] __schedule+0x234/0x758 >> [ 9138.093100] [<ffff80000074a278>] schedule+0x40/0xa0 >> [ 9138.093105] [<ffff80000074d25c>] schedule_timeout+0x214/0x280 >> [ 9138.093109] [<ffff80000074ae8c>] wait_for_common+0xb4/0x168 >> [ 9138.093113] [<ffff80000074af68>] wait_for_completion+0x28/0x38 >> [ 9138.093119] [<ffff8000005dd6f0>] rpi_firmware_property_list+0x120/0x280 >> [ 9138.093122] [<ffff8000005dd8d0>] rpi_firmware_property+0x80/0xc8 >> [ 9138.093133] [<ffff7ffffc115088>] bcm2835_cpufreq_get_clock+0x38/0x88 >> [bcm2835_cpufreq] >> [ 9138.093138] [<ffff7ffffc1150f4>] bcm2835_cpufreq_get+0x1c/0x28 >> [bcm2835_cpufreq] >> [ 9138.093144] [<ffff8000005d1264>] __cpufreq_get+0x34/0xc8 >> [ 9138.093148] [<ffff8000005d3304>] show_cpuinfo_cur_freq+0x24/0x78 >> [ 9138.093151] [<ffff8000005d1208>] show+0x48/0x70 >> [ 9138.093157] [<ffff8000002cb990>] sysfs_kf_seq_show+0xc0/0x158 >> [ 9138.093161] [<ffff8000002c9f94>] kernfs_seq_show+0x44/0x50 >> [ 9138.093166] [<ffff80000026cdac>] seq_read+0x17c/0x438 >> [ 9138.093169] [<ffff8000002cac34>] kernfs_fop_read+0x124/0x1b0 >> [ 9138.093175] [<ffff800000246430>] __vfs_read+0x48/0xf8 >> [ 9138.093178] [<ffff800000246e38>] vfs_read+0x88/0x160 >> [ 9138.093182] [<ffff800000247bc4>] SyS_read+0x54/0xb0 >> [ 9138.093187] [<ffff800000085e04>] el0_svc_naked+0x38/0x3c >> >> So, what do you think? Is it known? >> Probably I just need to try kernel 4.8.6 from openSUSE:Factory:ARM? > > I found https://github.com/raspberrypi/linux/issues/1364 , which fits > the backtracke by the kernel quite well. You could try different versions > of the firmware or even blacklist bcm2835_cpufreq as a workaround. > > Only the Leap 42.2 kernel has the cpufreq driver, so if you try the TW > kernel you will likely not hit this issue. This does mean that the CPU runs > at 800 MHz though. > > Good luck, > Fabian > > > -- > SUSE Linux GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany > GF: Felix Imendörffer, Jane Smithard, Graham Norton > HRB 21284 (AG Nürnberg) -- With best regards, Matwey V. Kornilov http://blog.matwey.name xmpp://[email protected] -- To unsubscribe, e-mail: [email protected] To contact the owner, e-mail: [email protected]
