Bug 202419 - KVM 4.14.94 regression due to "x86,kvm: move qemu/guest FPU switching out to vcpu_run"
Summary: KVM 4.14.94 regression due to "x86,kvm: move qemu/guest FPU switching out to ...
Status: RESOLVED CODE_FIX
Alias: None
Product: Virtualization
Classification: Unclassified
Component: kvm (show other bugs)
Hardware: x86-64 Linux
: P1 normal
Assignee: virtualization_kvm
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-01-25 17:52 UTC by Roman Mamedov
Modified: 2019-01-30 05:53 UTC (History)
4 users (show)

See Also:
Kernel Version: 4.14.94
Subsystem:
Regression: Yes
Bisected commit-id:

Attachments
dmesg of the crash (5.68 KB, text/plain)
2019-01-25 17:52 UTC, Roman Mamedov 		Details
Add an attachment (proposed patch, testcase, etc.)

Description Roman Mamedov 2019-01-25 17:52:08 UTC 
Created attachment 280773 [details]
dmesg of the crash

Hello,

I have a VM which works fine on 4.14.93, but started crashing on boot in a very puzzling way on 4.14.94 (dmesg attached) and 4.14.95.

Given the latter two have pretty much one KVM-on-x86 related change [1] since 4.14.93, sure enough, rolling back that one makes it work again (on 4.14.95).

It is really weird, how would an FPU change affect loading libraries? But I double-checked that there's no mistake and no changes are happening within the guest initrd, libraries and such. Only switching the kernel on the host between with/without that patch, determines whether the VM will boot or not.

Any idea what might be going on here?

Thanks

[1] https://patchwork.kernel.org/patch/10058409/
Note You need to log in before you can comment on or make changes to this bug.