View previous topic :: View next topic |
Author |
Message |
lekto Apprentice
Joined: 20 Sep 2014 Posts: 170 Location: Ancient Rome
|
Posted: Fri Jun 15, 2018 6:49 pm Post subject: Stability issuse with >=gentoo-sources:4.16.11 |
|
|
Hi, I have stability issue with gentoo-sources-4.16.11 and newer. System get froze at random moments, but higher load cause it faster. Yesterday system (gentoo-sources-4.6.11) get crashed first time while watching youtube, second time after few hours of playing factorio. Today I was testing gentoo-sources-4.17.1, rebuilding libreoffice doesn't crash it, but valley benchmark crashed it in less than minute. Gentoo-sources-4.16.10 works great, last weekend I spent playing factorio and I didn't get any crash, now I running valley benchmark for about 3 hours and still it's ok.
emerge --info
.config |
|
Back to top |
|
|
Hu Moderator
Joined: 06 Mar 2007 Posts: 21490
|
Posted: Sat Jun 16, 2018 1:36 am Post subject: |
|
|
Is the problem reproducible with a vanilla kernel? If yes, which patch in the range (v4.16.10, v4.16.11] causes the problem? Git bisect can save you considerable time if you need to test to find the answer. |
|
Back to top |
|
|
lekto Apprentice
Joined: 20 Sep 2014 Posts: 170 Location: Ancient Rome
|
Posted: Sun Jun 17, 2018 3:48 pm Post subject: |
|
|
I used git bisect and I think I found it. It was bit tricky because when system is running I don't know if it's because kernel is ok or it didn't crash yet.
Code: | 95271aeb93d4681c65e2f94969b23ef6070367a6 is the first bad commit
commit 95271aeb93d4681c65e2f94969b23ef6070367a6
Author: Thomas Gleixner <tglx@linutronix.de>
Date: Thu May 10 20:42:48 2018 +0200
x86/speculation, KVM: Implement support for VIRT_SPEC_CTRL/LS_CFG
commit 47c61b3955cf712cadfc25635bf9bc174af030ea upstream
Add the necessary logic for supporting the emulated VIRT_SPEC_CTRL MSR to
x86_virt_spec_ctrl(). If either X86_FEATURE_LS_CFG_SSBD or
X86_FEATURE_VIRT_SPEC_CTRL is set then use the new guest_virt_spec_ctrl
argument to check whether the state must be modified on the host. The
update reuses speculative_store_bypass_update() so the ZEN-specific sibling
coordination can be reused.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
:040000 040000 1a076b17c0c8d0bf255c9aba856f1b75791deeb1 82bd77c23c8e1ade512f4afa5ce021cbad6b2ae8 M arch
|
|
|
Back to top |
|
|
Hu Moderator
Joined: 06 Mar 2007 Posts: 21490
|
Posted: Sun Jun 17, 2018 4:45 pm Post subject: |
|
|
Yes, tracing bugs is difficult without a reliable reproducer. In this case, I am a bit doubtful that this is the correct result. Your original problem report makes no mention of any of the kernel functionality that I expect the cited patch would modify. |
|
Back to top |
|
|
lekto Apprentice
Joined: 20 Sep 2014 Posts: 170 Location: Ancient Rome
|
Posted: Sun Jun 17, 2018 5:08 pm Post subject: |
|
|
Oops, when I was writing report I mentioned that factorio and valley benchmark was run in virtual machine (qemu + kvm + windows 10) but later I remove too much and I accidently remove this sentence too. |
|
Back to top |
|
|
Hu Moderator
Joined: 06 Mar 2007 Posts: 21490
|
Posted: Sun Jun 17, 2018 5:49 pm Post subject: |
|
|
In that case, your result seems reasonable. Please report the problem upstream, with as much detail as you can provide. |
|
Back to top |
|
|
tholin Apprentice
Joined: 04 Oct 2008 Posts: 200
|
Posted: Sat Jul 07, 2018 8:05 pm Post subject: |
|
|
Did you get any response from upstream on this problem? My own gaming VM has started to hang and I think it coincided with the microcode update for SSBD mitigation.
How do you reliably cause the hang on your system? |
|
Back to top |
|
|
|