Re: [uml-devel] [PATCH v3] um: Switch clocksource to hrtimers

[Anton Ivanov <anton.ivanov@kot-begemot.co.uk>]

On 02/11/15 15:25, Richard Weinberger wrote:
> Am 02.11.2015 um 15:30 schrieb Anton Ivanov:
>> On 02/11/15 10:01, Richard Weinberger wrote:
>>> Am 02.11.2015 um 10:53 schrieb Anton Ivanov:
>>>> [snip]
>>>>
>>>>>>> I'm pretty sure that you don't see the issue as your Jessy userspace uses nanosleep periodically.
>>>>>> There are quite a few things running so this may indeed be the case.
>>>>>>
>>>>>> What do you use for userspace (so I can try to reproduce this and debug it)?
>>>>> Debian Squeeze amd64 with almost nothing running.
>>>>>
>>>>>      PID TTY      STAT   TIME COMMAND
>>>>>        2 ?        S      0:00 [kthreadd]
>>>>>        3 ?        S      0:00  \_ [ksoftirqd/0]
>>>>>        4 ?        S      0:00  \_ [kworker/0:0]
>>>>>        5 ?        S<     0:00  \_ [kworker/0:0H]
>>>>>        6 ?        S      0:00  \_ [kworker/u2:0]
>>>>>        7 ?        S      0:00  \_ [kdevtmpfs]
>>>>>        8 ?        S<     0:00  \_ [netns]
>>>>>        9 ?        S<     0:00  \_ [writeback]
>>>>>       10 ?        S      0:00  \_ [kworker/u2:1]
>>>>>       11 ?        S<     0:00  \_ [crypto]
>>>>>       12 ?        S      0:00  \_ [kworker/0:1]
>>>>>       13 ?        S<     0:00  \_ [bioset]
>>>>>       14 ?        S<     0:00  \_ [kblockd]
>>>>>       15 ?        S      0:00  \_ [kswapd0]
>>>>>       68 ?        S      0:00  \_ [fsnotify_mark]
>>>>>      221 ?        S<     0:00  \_ [bioset]
>>>>>      229 ?        S<     0:00  \_ [deferwq]
>>>>>      231 ?        S      0:00  \_ [jbd2/ubda-8]
>>>>>      232 ?        S<     0:00  \_ [ext4-rsv-conver]
>>>>>      233 ?        S<     0:00  \_ [kworker/0:1H]
>>>>>        1 ?        Ss     0:00 init [2]
>>>>>      271 ?        S<s    0:00 udevd --daemon
>>>>>      297 ?        S<     0:00  \_ udevd --daemon
>>>>>      298 ?        S<     0:00  \_ udevd --daemon
>>>>>      549 ?        Sl     0:00 /usr/sbin/rsyslogd -c4
>>>>>      580 ?        Ss     0:00 /usr/sbin/cron
>>>>>      595 ?        Ss     0:00 /usr/bin/dbus-daemon --system
>>>>>      609 ?        Ss     0:00 /usr/sbin/sshd
>>>>>      628 tty0     Ss     0:00 /bin/login --
>>>>>      631 tty0     S      0:00  \_ -bash
>>>>>      636 tty0     R+     0:00      \_ ps fax
>>>>>      629 tty1     Ss+    0:00 /sbin/getty 38400 tty1 linux
>>>>>      630 ttyS0    Ss+    0:00 /sbin/getty 115200 ttyS0 linux
>>>> I cannot reproduce that by dropping jessie into single user mode so I am going to build a squeeze image and re-test with that.
>>> I can also reproduce using init=/bin/bash.
>> Reproduced, I think I located the culprit.
>>
>> We still need to relay the "interruption" signal which broke a nanosleep.
>>
>> I am testing a fix, which so far is working. If the fix proves viable, I will test it as above (init=/bin/bash) and the normal tests I usually do anyway.
>>
>> If it works OK, I will submit v4.
> Yay!

Apologies for that. A few years back I found that I can mask most of the 
races which this patch is trying to address by pinning to a single CPU. 
As a result most of my test scripts have a taskset -c 0 in them.

The revised version where we restore the relay of nanosleep 
interruptions checks out for me, I just re-submitted it.

I will rebase my ubd patches on top of that and once they work, submit 
them as well.

A.

>
> Thanks,
> //richard the QA engineer ;)
>


------------------------------------------------------------------------------
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel