[vbox-dev] vbox kernel modules cause hardlock on kernel 4.2.0-rc2

[Larry Finger]

On 07/17/2015 05:29 AM, Jon Christopherson wrote:
> On 07/17/2015 04:13 AM, Ramshankar wrote:
>> On 07/16/2015 06:15 PM, Ramshankar wrote:
>>> Oops, misread your original report. It *is* the WARN panic. I'll try to
>>> provide a test build soon to fix this issue.
>>
>> Please let me know if this build works works for you:
>> https://www.virtualbox.org/download/testcase/VirtualBox-5.0.1-101671-Linux_amd64.run
>>
>>
>> Regards,
>> Ram.
>>
>>>> I'm not sure how that is happening.
>>>>
>>>> This changeset
>>>> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=2ecd9d29abb171d6e97a4f3eb29d7456a11401b7
>>>>
>>>>
>>>> reverts to the old assumptions of the caller disabling preemption while
>>>> calling preempt_notifier_register()/preempt_notifier_unregister().
>>>>
>>>> We would get into the "slower" path but i don't think that should really
>>>> cause the kernel panic like before.
>>>>
>>>> We would need to call the newly introduced functions
>>>> preempt_notifier_inc() and preempt_notifier_dec() but that should be a
>>>> performance issue and not a fatal panic afaict unless "WARN" ends up
>>>> panicking the system in which case it should be a different panic and
>>>> not the original oops you posted.
>>>>
>>>> Regards,
>>>> Ram,
>>>>
>>>>
>>>> _______________________________________________
>>>> vbox-dev mailing list
>>>> vbox-dev at virtualbox.org
>>>> https://www.virtualbox.org/mailman/listinfo/vbox-dev
>>>
>>> _______________________________________________
>>> vbox-dev mailing list
>>> vbox-dev at virtualbox.org
>>> https://www.virtualbox.org/mailman/listinfo/vbox-dev
>>
> Hello,
>
> This build works fine on my machine with 4.2.0-rc2 kernel, thanks!

What got changed in VirtualBox for this test build to work? Could you please 
provide a patch in advance of the release of 5.0.2?

Although I did not have a problem with my original tests of kernel 4.2-rcX, this 
failure happened to me as soon as commit 2ecd9d29abb1 landed in my source. For 
the moment, I can run by reverting that kernel commit. I could push a kernel 
patch requesting this reversion, but as that commit seems to help performance 
for KVM, that might not be a good idea.

Thanks,

Larry