Asus C101P Chromeboot fails to boot with Linux 5.2

Asus C101P Chromeboot fails to boot with Linux 5.2

[Alex Dewar]

Hi all,

Since upgrading to the 5.2 kernel, my Chromebook is failing to boot. The
Asus C101P is based on the RK3399 SoC and uses the rk3399-gru-bob device
tree. It used to boot with the 5.1 kernel and mostly worked, with the
exception of broken suspend and resume.

When I try to boot the screen just gets flooded with messages like this:
http://users.sussex.ac.uk/~ad374/boot_fail.jpg

I'm using Arch Linux ARM's linux-aarch64 package, source here:
https://archlinuxarm.org/packages/aarch64/linux-aarch64/files/PKGBUILD

Best,
Alex

Re: Asus C101P Chromeboot fails to boot with Linux 5.2

[Emil Renner Berthing]

Hi Alex,

On Fri, 12 Jul 2019 at 17:02, Alex Dewar <alex.dewar@gmx.co.uk> wrote:
> When I try to boot the screen just gets flooded with messages like this:
> http://users.sussex.ac.uk/~ad374/boot_fail.jpg

Those seem to be only audit messages. You can try booting with audit=0
on the kernel command line to get rid of those messages, so you have a
better chance of catching the real issue. (Or conclude that audit is
what is broken.)

/Emil

Re: Asus C101P Chromeboot fails to boot with Linux 5.2

[Alex Dewar]

Hi Emil,

I've tried booting with: audit=0 console=tty0 console=ttyS2,115200n8
earlyprintk=ttyS2,115200n8 init=/sbin/init root=PARTUUID=%U/PARTNROFF=1
rootwait rw noinitrd

Now I don't get the audit messages, but no other message is displayed
either. It just alternates between a flashing cursor and a multicoloured
square in the middle of the screen. The system isn't totally
unresponsive though. If I press the power button it still shuts down
properly.

I've just tested it again and confirmed that it does boot with the
v5.1.11 kernel.

Best,
Alex

Re: Asus C101P Chromeboot fails to boot with Linux 5.2

[Alex Dewar]

I just built v5.1 and v5.2 from source, without the Arch patches, and
get the same result: 5.1 boots and 5.2 doesn't. I tried adding
ignore_loglevel, but still just get a blinking cursor. So it's
definitely a regression.

I can have a go at bisecting this if it would help? Or is there
something else we could try first?

Alex

[REGRESSION] Xorg segfaults on Asus Chromebook CP101 with Linux v5.2 (was Asus C101P Chromeboot fails to boot with Linux 5.2)

[Alex Dewar]

Hi all,

I initially thought my machine was failing to boot entirely, but it
turns out it was just failing to start the display manager. I managed to
escape to a tty by hammering the keyboard a bit.

I suspect the culprit is the rockchip_vpu driver (in staging/media),
which has been renamed to hantro in this merge window. When I run startx
from a terminal, X fails to start and Xorg segfaults (log here:
http://users.sussex.ac.uk/~ad374/xorg.log). X seems to work without any
issues in v5.1.

I've also tried running trace on the Xorg process, but the output was
pretty verbose. I can share if that would be helpful though.

Best,
Alex

Re: [REGRESSION] Xorg segfaults on Asus Chromebook CP101 with Linux v5.2 (was Asus C101P Chromeboot fails to boot with Linux 5.2)

[Greg Kroah-Hartman]

On Sat, Jul 13, 2019 at 12:38:45PM +0100, Alex Dewar wrote:
> Hi all,
> 
> I initially thought my machine was failing to boot entirely, but it
> turns out it was just failing to start the display manager. I managed to
> escape to a tty by hammering the keyboard a bit.
> 
> I suspect the culprit is the rockchip_vpu driver (in staging/media),
> which has been renamed to hantro in this merge window. When I run startx
> from a terminal, X fails to start and Xorg segfaults (log here:
> http://users.sussex.ac.uk/~ad374/xorg.log). X seems to work without any
> issues in v5.1.
> 
> I've also tried running trace on the Xorg process, but the output was
> pretty verbose. I can share if that would be helpful though.

Can you run 'git bisect' to find the offending commit?

thanks,

greg k-h

Re: [REGRESSION] Xorg segfaults on Asus Chromebook CP101 with Linux v5.2 (was Asus C101P Chromeboot fails to boot with Linux 5.2)

[Heiko Stuebner]

Hi,

Am Samstag, 13. Juli 2019, 13:38:45 CEST schrieb Alex Dewar:
> I initially thought my machine was failing to boot entirely, but it
> turns out it was just failing to start the display manager. I managed to
> escape to a tty by hammering the keyboard a bit.
> 
> I suspect the culprit is the rockchip_vpu driver (in staging/media),
> which has been renamed to hantro in this merge window. When I run startx
> from a terminal, X fails to start and Xorg segfaults (log here:
> http://users.sussex.ac.uk/~ad374/xorg.log). X seems to work without any
> issues in v5.1.

5.2 also has support for Panfrost (Mali-Midgard GPUs) but I'm not
sure if it already can support X11 yet and your X11 log mentions
libglamoregl in the segfault stack trace.

Apart from it bisect that Greg suggested you could also just try
blacklisting either panfrost or vpu kernel modules
/etc/udev/somewhere . This would prevent them from loading

Hope that helps
Heiko

Re: [REGRESSION] Xorg segfaults on Asus Chromebook CP101 with Linux v5.2 (was Asus C101P Chromeboot fails to boot with Linux 5.2)

[Alex Dewar]

On 13/07/2019 16:17, Heiko Stuebner wrote:
> Hi,
>
> Am Samstag, 13. Juli 2019, 13:38:45 CEST schrieb Alex Dewar:
>> I initially thought my machine was failing to boot entirely, but it
>> turns out it was just failing to start the display manager. I managed to
>> escape to a tty by hammering the keyboard a bit.
>>
>> I suspect the culprit is the rockchip_vpu driver (in staging/media),
>> which has been renamed to hantro in this merge window. When I run startx
>> from a terminal, X fails to start and Xorg segfaults (log here:
>> http://users.sussex.ac.uk/~ad374/xorg.log). X seems to work without any
>> issues in v5.1.
>
> 5.2 also has support for Panfrost (Mali-Midgard GPUs) but I'm not
> sure if it already can support X11 yet and your X11 log mentions
> libglamoregl in the segfault stack trace.
>
> Apart from it bisect that Greg suggested you could also just try
> blacklisting either panfrost or vpu kernel modules
> /etc/udev/somewhere . This would prevent them from loading
>
> Hope that helps
> Heiko
>
>

Hi Heiko,

Thanks for this. I blacklisted the panfrost driver and X magically
started working again.

I'll try to do a bisect later to find the offending commit though.

In related news, it also seems that the sound and wifi drivers aren't
working either in 5.2 (although I need to do a bit more testing to
confirm the latter).

Best,
Alex

Re: [REGRESSION] Xorg segfaults on Asus Chromebook CP101 with Linux v5.2 (was Asus C101P Chromeboot fails to boot with Linux 5.2)

[Ezequiel Garcia]

On Sat, 13 Jul 2019 at 13:43, Alex Dewar <alex.dewar@gmx.co.uk> wrote:
>
> On 13/07/2019 16:17, Heiko Stuebner wrote:
> > Hi,
> >
> > Am Samstag, 13. Juli 2019, 13:38:45 CEST schrieb Alex Dewar:
> >> I initially thought my machine was failing to boot entirely, but it
> >> turns out it was just failing to start the display manager. I managed to
> >> escape to a tty by hammering the keyboard a bit.
> >>
> >> I suspect the culprit is the rockchip_vpu driver (in staging/media),
> >> which has been renamed to hantro in this merge window. When I run startx
> >> from a terminal, X fails to start and Xorg segfaults (log here:
> >> http://users.sussex.ac.uk/~ad374/xorg.log). X seems to work without any
> >> issues in v5.1.
> >
> > 5.2 also has support for Panfrost (Mali-Midgard GPUs) but I'm not
> > sure if it already can support X11 yet and your X11 log mentions
> > libglamoregl in the segfault stack trace.
> >
> > Apart from it bisect that Greg suggested you could also just try
> > blacklisting either panfrost or vpu kernel modules
> > /etc/udev/somewhere . This would prevent them from loading
> >
> > Hope that helps
> > Heiko
> >
> >
>
> Hi Heiko,
>
> Thanks for this. I blacklisted the panfrost driver and X magically
> started working again.
>
> I'll try to do a bisect later to find the offending commit though.
>
> In related news, it also seems that the sound and wifi drivers aren't
> working either in 5.2 (although I need to do a bit more testing to
> confirm the latter).
>

Adding myself and Tomeu.

Perhaps we need to disable Panfrost from defconfig from now?

Regards,
Eze