The 5.14 kernel in rawhide seems to be unstable, having lockups and failed boots, didn't occur with 5.13

[stan]

I have an install of rawhide.  It is updated to latest packages in
rawhide.  One of those packages is the new 5.14 kernel.  It mostly
works, but I have seen three freezes during boot.  Two of them just
stopped after a screen update from systemd, but one of them (the
latest) said something about a deadlock involving console and cpu.  I
had to reboot, and the journal was corrupted, so I lost the message.

I have also had several (less than 5) freezes while going about normal
business.  No consistency, they just happen.

It is early in the development cycle for 5.14, so these will probably go
away, but I wanted the issue to be raised and noted.

I had no issues whatsoever with the 5.13 kernels.

[stan]

On Fri, 9 Jul 2021 06:38:08 -0700
stan via kernel  wrote:
> I have an install of rawhide.  It is updated to latest packages in
> rawhide.  One of those packages is the new 5.14 kernel.  It mostly
> works, but I have seen three freezes during boot.  Two of them just
> stopped after a screen update from systemd, but one of them (the
> latest) said something about a deadlock involving console and cpu.  I
> had to reboot, and the journal was corrupted, so I lost the message.
>
> I have also had several (less than 5) freezes while going about normal
> business.  No consistency, they just happen.
>
> It is early in the development cycle for 5.14, so these will probably
> go away, but I wanted the issue to be raised and noted.

I installed the latest kernel today, 
Name        : kernel
Version     : 5.14.0
Release     : 0.rc0.20210708gite9f1cbc0c411.13.fc35
and it has issues.  It boots successfully to multiuser, but when I try
to start X, it freezes the machine.

I've attached oops from journalctl during boot.  I don't know if they
are the problem, but they probably shouldn't be happening.  Should I
open a bug for this?

[Chris Murphy]

On Sat, Jul 10, 2021 at 8:24 PM stan via kernel
 wrote:
> On Fri, 9 Jul 2021 06:38:08 -0700
> stan via kernel  wrote:
> > I have an install of rawhide.  It is updated to latest packages in
> > rawhide.  One of those packages is the new 5.14 kernel.  It mostly
> > works, but I have seen three freezes during boot.  Two of them just
> > stopped after a screen update from systemd, but one of them (the
> > latest) said something about a deadlock involving console and cpu.  I
> > had to reboot, and the journal was corrupted, so I lost the message.
> >
> > I have also had several (less than 5) freezes while going about normal
> > business.  No consistency, they just happen.
> >
> > It is early in the development cycle for 5.14, so these will probably
> > go away, but I wanted the issue to be raised and noted.
>
> I installed the latest kernel today,
> Name        : kernel
> Version     : 5.14.0
> Release     : 0.rc0.20210708gite9f1cbc0c411.13.fc35
> and it has issues.  It boots successfully to multiuser, but when I try
> to start X, it freezes the machine.
>
> I've attached oops from journalctl during boot.  I don't know if they
> are the problem, but they probably shouldn't be happening.  Should I
> open a bug for this?

You might ask in #fedora-kernel but my two cents is I generally don't
file bugs on rawhide kernels in the RHBZ. Instead I try to figure out
what part of the kernels is blowing up, and use a current version of
the MAINTAINERS file [1] to figure out where to send an email: I'm
expecting X, I'm seeing Y, here's a dmesg. In short.

There's a ton of changes happening in the rc0 kernels, they aren't
release candidates yet. The merge window just closed, and it's 7-8
weeks of stabilization. You can expect to be asked to bisect if it
sounds like a regression; and to test a patch. If you can't, just say
so.

Branch is in about 4 weeks [2]. We will see current rawhide become
Fedora 35, which looks like it'll be kernel 4.14-rc5 by then. Any sort
of show stopper type bugs I'll file in RHBZ with a URL pointing to the
email I started with the upstream maintainers of the problem driver or
kernel component. That's because these bugs are potentially release
blocking bugs that'll need to get fixed before either the beta or
final milestone. And for that I use the blockerbugs app [3] to propose
it.

[1]
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/MAIN...

[2]
https://fedorapeople.org/groups/schedule/f-35/f-35-key-tasks.html

[3]
https://qa.fedoraproject.org/blockerbugs/propose_bug

[stan]

On Mon, 12 Jul 2021 01:11:42 -0600
Chris Murphy  wrote:
> You might ask in #fedora-kernel but my two cents is I generally don't
> file bugs on rawhide kernels in the RHBZ. Instead I try to figure out
> what part of the kernels is blowing up, and use a current version of
> the MAINTAINERS file [1] to figure out where to send an email: I'm
> expecting X, I'm seeing Y, here's a dmesg. In short.
>
> There's a ton of changes happening in the rc0 kernels, they aren't
> release candidates yet. The merge window just closed, and it's 7-8
> weeks of stabilization. You can expect to be asked to bisect if it
> sounds like a regression; and to test a patch. If you can't, just say
> so.
>
> Branch is in about 4 weeks [2]. We will see current rawhide become
> Fedora 35, which looks like it'll be kernel 4.14-rc5 by then. Any sort
> of show stopper type bugs I'll file in RHBZ with a URL pointing to the
> email I started with the upstream maintainers of the problem driver or
> kernel component. That's because these bugs are potentially release
> blocking bugs that'll need to get fixed before either the beta or
> final milestone. And for that I use the blockerbugs app [3] to propose
> it.
>
> [1]
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/MAIN...
>
> [2]
> https://fedorapeople.org/groups/schedule/f-35/f-35-key-tasks.html
>
> [3]
> https://qa.fedoraproject.org/blockerbugs/propose_bug

Thanks!