That Nasty Linux Kernel Lockup Bug Is Still Unresolved

[phoronix]

Phoronix: That Nasty Linux Kernel Lockup Bug Is Still Unresolved

Nearly one month ago back during the Linux 3.18 release candidates there was a worrisome regression uncovered by kernel developers, but now with the Linux 3.19 merge window nearly over, that issue still has yet to be firmly addressed...

http://www.phoronix.com/vr.php?view=MTg2NzA

[GreatEmerald]

You know, it would be pretty funny if this turned out to be a hardware issue.

[bitman]

You know, it would be pretty funny if this turned out to be a hardware issue.

Issue hitting many different people and not being present with older kernel does not sound like hardware issue at all.

[GreatEmerald]

Issue hitting many different people and not being present with older kernel does not sound like hardware issue at all.

Why not? Many different people seem to be having many different problems that result in the same symptoms. And it not being present in older kernels might just be a byproduct of the older kernels not making use of some hardware functionality. Which, of course, would still be nice, because it makes it possible to track down what said functionality could be and results in accidentally finding and fixing other bugs.

[bitman]

Why not? Many different people seem to be having many different problems that result in the same symptoms. And it not being present in older kernels might just be a byproduct of the older kernels not making use of some hardware functionality. Which, of course, would still be nice, because it makes it possible to track down what said functionality could be and results in accidentally finding and fixing other bugs.

Even if so kernel warrants a fix. Like workaround for hw quirk or disabling feature completely for said hardware.

[calc]

Is this the issue where khugepaged eats all cpu and hangs the system for a while then starts working again before repeating itself?

I've been seeing that issue for a while now and I run drm-intel-nightly kernels. My system has 32GB ram, usually with around 28GB free, so I was a bit surprised it would be having memory related issues.

[DeepDayze]

Is this the issue where khugepaged eats all cpu and hangs the system for a while then starts working again before repeating itself?

I've been seeing that issue for a while now and I run drm-intel-nightly kernels. My system has 32GB ram, usually with around 28GB free, so I was a bit surprised it would be having memory related issues.

If that's the case then the kernel devs should take a look at that daemon to see where the problem lies. Have you ever strace'd it to get an idea what's happening so the devs can take a look at it and pinpoint the problem?

[bitman]

Is this the issue where khugepaged eats all cpu and hangs the system for a while then starts working again before repeating itself?

I've been seeing that issue for a while now and I run drm-intel-nightly kernels. My system has 32GB ram, usually with around 28GB free, so I was a bit surprised it would be having memory related issues.

hmm you could be right. because i am experiencing lockups and i indeed use hugepages for my VMs.

[devsk]

Even if so kernel warrants a fix. Like workaround for hw quirk or disabling feature completely for said hardware.

Looks like something in kernel is stomping HPET. See the latest update here: http://lkml.iu.edu/hypermail/linux/k...2.3/00666.html

So, one workaround is to pass hpet=disable to the kernel cmdline.