Bug 699754

Summary:	sys-kernel/gentoo-sources-4.19.82: libvirt-guests are stuck in reboot-loop
Product:	Gentoo Linux	Reporter:	tastytea <gentoo>
Component:	Current packages	Assignee:	Gentoo Kernel Bug Wranglers and Kernel Maintainers <kernel>
Status:	RESOLVED FIXED
Severity:	normal	CC:	fabian, jasmin+gentoo, randy
Priority:	Normal
Version:	unspecified
Hardware:	AMD64
OS:	Linux
URL:	https://lkml.org/lkml/2019/11/11/194
Whiteboard:
Package list:		Runtime testing required:	---
Bug Depends on:	700416
Bug Blocks:
Attachments:	libvirt log of guest dmesg excerpt

Description tastytea 2019-11-10 16:24:16 UTC

After upgrading from 4.19.72 to 4.19.82, my VMs are stuck in a reboot-loop. The boot up to Grub is fine, but when I select an item in the boot menu, the guest reboots after a short while. I can't see any output. I don't see anything unusual in the logs but I'll attach the log from libvirt and dmesg. After I switch back to .72, the guests boots fine. The same issue manifests when booting from "CD-ROM".

The .82-kernel is compiled with sys-devel/gcc-9.2.0-r2, the old was compiled with 8.3. The only change in the kernel config is CONFIG_GCC_VERSION.

I tried to reproduce the issue and found a possible clue: The affected machines both have an Intel Core 2 DUO CPU (T7300 and P8700). The machine that is not affected has an AMD CPU (A10-9700).

Reproducible: Always

Comment 1 tastytea 2019-11-10 16:25:31 UTC

Created attachment 595666 [details]
libvirt log of guest

Comment 2 tastytea 2019-11-10 16:26:16 UTC

Created attachment 595668 [details]
dmesg excerpt

Comment 3 tastytea 2019-11-10 22:19:44 UTC

I recompiled the kernel with sys-devel/gcc-8.3.0-r1 to be sure it's not that; the issue persists.

Comment 4 Stanislav Romanov 2019-11-11 18:52:32 UTC

Same issue after upgrading from linux-4.14.143-gentoo to linux-4.14.152-gentoo on the VM host. After reboot host works fine, but guests with linux-4.14.143-gentoo are stuck in a bootloop.
All kernels compiled with sys-devel/gcc-8.3.0-r1

Comment 5 Mike Pagano gentoo-dev

2019-11-14 16:19:08 UTC

Are you able to do a git bisect between the last working kernel and the first on working kernel?

http://www.reactivated.net/weblog/archives/2006/01/using-git-bisect-to-find-buggy-kernel-patches/

Comment 6 tastytea 2019-11-14 18:25:54 UTC

Yes, I'll get right to work. :-)

Comment 7 tastytea 2019-11-15 18:31:32 UTC

496cf984a60edb5534118a596613cc9971e406e8 is the first bad commit (commit 16cfacc8085782dab8e365979356ce1ca87fd6cc upstream).

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=496cf984a60edb5534118a596613cc9971e406e8

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=16cfacc8085782dab8e365979356ce1ca87fd6cc

Comment 8 Thomas Deutschmann (RETIRED) gentoo-dev

2019-11-17 23:10:12 UTC

Upstream patch: bf03d4f9334728bf7c8ffc7de787df48abd6340e 

Will be in next stable kernels.

Comment 9 tastytea 2019-11-18 03:34:00 UTC

I can confirm the fix works with 4.19.82.