Re: Regression with mainline kernel on rpi4

[Maxime Ripard]

On Mon, Sep 20, 2021 at 04:47:31PM +0200, Maxime Ripard wrote:
> On Sat, Sep 04, 2021 at 10:40:29AM +0100, Sudip Mukherjee wrote:
> > Hi Maxime,
> > 
> > On Sat, Sep 4, 2021 at 10:10 AM Maxime Ripard <maxime@xxxxxxxxxx> wrote:
> > >
> > > On Fri, Sep 03, 2021 at 09:09:50PM +0100, Sudip Mukherjee wrote:
> > > > Hi Maxime,
> > > >
> > > > On Fri, Sep 3, 2021 at 5:03 PM Maxime Ripard <maxime@xxxxxxxxxx> wrote:
> > > > >
> > > > > Hi Sudip,
> > > > >
> > > > > On Thu, Sep 02, 2021 at 10:08:19AM +0100, Sudip Mukherjee wrote:
> > > > > > Hi All,
> > > > > >
> > > > >
> > > >
> > > > <snip>
> > > >
> > > > >
> > > > > >
> > > > > > You can see the complete dmesg at
> > > > > > https://openqa.qa.codethink.co.uk/tests/76#step/dmesg/8
> > > > >
> > > > > What test were you running?
> > > >
> > > > Nothing particular, its just a boot test that we do every night after
> > > > pulling from torvalds/linux.git
> > >
> > > What are you booting to then?
> > 
> > I am not sure I understood the question.
> > Its an Ubuntu image. The desktop environment is gnome. And as
> > mentioned earlier, we use the HEAD of linus tree every night to boot
> > the rpi4 and test that we can login via desktop environment and that
> > there is no regression in dmesg.
> 
> Looking at the CI, this isn't from a RPi but from qemu?
> 
> What defconfig are you using? How did you generate the Ubuntu image?
> Through debootstrap? Any additional package?

So qemu (at least on Fedora 34) doesn't seem to have an RPI4 target, nor
upstream:
https://github.com/qemu/qemu/blob/079b1252e9de384385c9da910262312ec2e574c8/hw/arm/raspi.c#L367

I've tested an Ubuntu 21.04 arm64 build (since it seems like it's what
you've been using), built using debootstrap + ubuntu-desktop, both with
and without a monitor attached, and up to the desktop once logged in.

I don't see any crash.

Maxime

Attachment: signature.asc
Description: PGP signature