Re: Problem Report: i915 driver most probably causing freezes on FreeBSD 13.1-RELEASE
- In reply to: Jan Beich : "Re: Problem Report: i915 driver most probably causing freezes on FreeBSD 13.1-RELEASE"
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Fri, 09 Sep 2022 14:14:44 UTC
On 2022-09-21 21:00, Jan Beich wrote:
> (Disclaimer: I'm on wlroots (Wayland), so can't help much with Xorg
> (X11) issues)
>
Hi Jan. Thank you for replying, any hint is much appreciated indeed.
> Beñat Dagorret <lists@nrd.fr> writes:
>
>> Hello,
>>
>> I'm having an issue with the i915kms driver on a fresh FreeBSD
>> 13.1-RELEASE install. The system boots and shows the Display Manager
>> (LightDM) without any problem, but after logging in, X crashes leaving
>> a black screen and a dead keyboard. The crash occurs always about 3
>> seconds after user login, no matter there was any user activity. I
>> need to shutdown the system remotely from my laptop through ssh.
>
> Can you trigger OpenGL crash via graphics/kmscube or glmark2-drm from
> benchmarks/glmark2 when running under console?
I don't know these tools but I will look further and give them a try to
see what happens.
>> vgapci0@pci0:0:2:0: class=0x030000 rev=0x06 hdr=0x00 vendor=0x8086
>> device=0x0402 subvendor=0x103c subdevice=0x21d0
>> vendor = 'Intel Corporation'
>> device = 'Xeon E3-1200 v3/4th Gen Core Processor Integrated
>> Graphics Controller'
>> class = display
>> subclass = VGA
>> bar [10] = type Memory, range 64, base 0xf7800000, size 4194304,
>> enabled
>> bar [18] = type Prefetchable Memory, range 64, base 0xe0000000,
>> size 268435456, enabled
>> bar [20] = type I/O Port, range 32, base 0xf000, size 64,
>> enabled
>> cap 05[90] = MSI supports 1 message enabled with 1 message
>> cap 01[d0] = powerspec 2 supports D0 D3 current D0
>> cap 13[a4] = PCI Advanced Features: FLR TP
>
> 0x0402 is Haswell, so OpenGL is driven by either i965 or crocus. In
> graphics/mesa-dri
> crocus is currently disabled by default unless
> MESA_LOADER_DRIVER_OVERRIDE=crocus, see
> https://gitlab.freedesktop.org/mesa/mesa/-/commit/988762679ccf
Yes it is Haswell. When the xf86-video-intel driver is loaded, instead
of modesetting, i965 gets loaded too.
> src/loader/meson.build:
> _prefer_crocus = get_option('prefer-crocus')
> if _prefer_crocus == 'auto'
> _prefer_crocus = 'false'
> endif
>
> Maybe try crocus from graphics/mesa-devel. i965 is gone upstream (Mesa
> >= 22.0.0).
>
Thank you for the tip. I think I remember seeing crocus in Xorg.0.log
once I booted the system from a live Linux distro. I didn't know what it
was and where it was coming from (I had radeon/glamor cards until now).
I will check the Xorg.log from the Linux distro again and definitely
give crocus a try.
>> [ 11.994]
>> X.Org X Server 1.20.14
>> X Protocol Version 11, Revision 0
>
> Try upgrading xorg-server to 21.1.4 by switching to /latest package
> repo.
> 1.20.14 on /quarterly will be unsupported after 2022Q3 branch EOL
> around
> 2022-10-01 (~1 week from now).
Will try this too, it's a fresh install so nothing to lose and so much
to learn :-)
>> [ 12.593] (II) event5 - Logitech USB Optical Mouse, class 0/0, rev
>> 2.00/72.00, addr 3: is tagged by udev as: Mouse
>> [ 12.594] (II) event5 - Logitech USB Optical Mouse, class 0/0, rev
>> 2.00/72.00, addr 3: device is a pointer
>
> Looking at the end of Xorg.log I don't see anything suspicious. Is it
> really the log from when Xorg crashed and not from a separate instance
> run by the Display Manager?
I didn't see anything suspicious either, and the xf86-video-intel
version of the Xorg.log is quite clean too. And yes, I think it is the
log from Xorg (located at /var/log/Xorg.0.log). But it's not modified by
the crash: no messages are appended when the server freezes. Unless I'm
missing something here.
Thank you again, I'm pretty confident about the crocus option.
Regards.