Re: [regression] hard system lockup

Re: [regression] hard system lockup

[Rokas Kupstys]

On 2019-03-31 19:42, Rokas Kupstys wrote:
> Since 4.19 booting a kvm VM results in eventual system lockup. dmesg 
> is full of warnings of watchdog detecting cpu lockups. I posted log 
> and more details at https://bugzilla.kernel.org/show_bug.cgi?id=201959 
> and i also found out issue started appearing at 
> 9cf57731b63e37ed995b46690adc604891a9a28f
>
> Could someone take a look please?
>

Since noone replied neither in mailing list nor in bug tracker - was i 
supposed to report issue somewhere else?

-- 
Rokas Kupstys

Re: [regression] hard system lockup

[Maxim Levitsky]

On Sun, 2019-04-07 at 09:51 +0300, Rokas Kupstys wrote:
> On 2019-03-31 19:42, Rokas Kupstys wrote:
> > Since 4.19 booting a kvm VM results in eventual system lockup. dmesg 
> > is full of warnings of watchdog detecting cpu lockups. I posted log 
> > and more details at https://bugzilla.kernel.org/show_bug.cgi?id=201959 
> > and i also found out issue started appearing at 
> > 9cf57731b63e37ed995b46690adc604891a9a28f
> > 
> > Could someone take a look please?
> > 
> 
> Since noone replied neither in mailing list nor in bug tracker - was i 
> supposed to report issue somewhere else?
> 

Did you change any settings related to watchdog in /proc/sys/ ?

I distinclty remember that I once had some trouble with watchdog, but I don't
remember what it was,  when I disabled the watchdog prior to VM startup (it
supposed to improve latency) on my gaming system, which I don't have anymore.

You should ping the authors of commit you bisected, and CC the LKML.

Best regards,
	Maxim Levitsky

Re: [regression] hard system lockup

[Rokas Kupstys]

On 2019-04-07 14:53, Maxim Levitsky wrote:
> On Sun, 2019-04-07 at 09:51 +0300, Rokas Kupstys wrote:
>> On 2019-03-31 19:42, Rokas Kupstys wrote:
>>> Since 4.19 booting a kvm VM results in eventual system lockup. dmesg
>>> is full of warnings of watchdog detecting cpu lockups. I posted log
>>> and more details at https://bugzilla.kernel.org/show_bug.cgi?id=201959
>>> and i also found out issue started appearing at
>>> 9cf57731b63e37ed995b46690adc604891a9a28f
>>>
>>> Could someone take a look please?
>>>
>> Since noone replied neither in mailing list nor in bug tracker - was i
>> supposed to report issue somewhere else?
>>
> Did you change any settings related to watchdog in /proc/sys/ ?
>
> I distinclty remember that I once had some trouble with watchdog, but I don't
> remember what it was,  when I disabled the watchdog prior to VM startup (it
> supposed to improve latency) on my gaming system, which I don't have anymore.
>
> You should ping the authors of commit you bisected, and CC the LKML.
>
> Best regards,
> 	Maxim Levitsky
>
Indeed this was the case. And it seems i finally set out to bisect the 
issue soon after fix was landed. Issue was fixed in 
53464ca9130be5466ada7b9dd653059c3a26cad9 and (to my surprise) 5.0.6 
kernel no longer has this issue.

Thank you for reply and thanks to Thomas Gleixner for the fix.

-- 
Rokas Kupstys