Re: i915kms and chip resets on rsc0?

On Thu, Jan 28, 2021 at 08:48:06PM +0100, Emmanuel Vadot wrote:
> On Thu, 28 Jan 2021 11:15:10 -0800
> Steve Kargl <sgk_at_troutmask.apl.washington.edu> wrote:
> 
> > All,
> > 
> > I have finally gotten my old laptop somewhat back to it
> > 2 Dec 2020 working state.  The kernel/world/drm-current-kmod
> > from 2 Dec 2020 worked wonderfully.  Sadly, the two week
> > recover from a failed update has one final hurdle.
> > 
> > With nearly top-of-tree src/ and ports/
> > 
> > % cd /usr/src
> > % git log | more
> > commit 0e01ea872ee475d7aef11d21588504e2ef4eb32c
> > Author: Cy Schubert <cy_at_FreeBSD.org>
> > Date:   Wed Jan 27 21:52:08 2021 -0800
> > 
> > % pkg info | grep kmod
> > drm-current-kmod-5.4.62.g20210128 DRM modules for the linuxkpi-based KMS components
> > gpu-firmware-kmod-g20201213    Firmware modules for the linuxkpi-based KMS components
> > 

(snip)

> > Asynchronous wait on fence i915:Xorg[100156]:4 timed out (hint:0x24ad37f0S)
> > drmn0: GPU recovery timed out, cancelling all in-flight rendering.
> > drmn0: Resetting chip for hang on rcs0
> > 
> 
>  Does 5.4.62.g20210118 works ?

It works to the extent that the X server loads and
the fvwm2 window manager comes up with my normal 
desktop.  In fact, I'm typing this in a xterm at
that moment.

What I don't know is if the X server is ignoring the GPU.
It is unclear to me if the last "drmn0: Resetting chip
for hang on rcs0" is disabling some functionality.  xwininfo
reports

% xwininfo

xwininfo: Window id: 0xef (the root window) (has no name)

  Absolute upper-left X:  0
  Absolute upper-left Y:  0
  Relative upper-left X:  0
  Relative upper-left Y:  0
  Width: 1400
  Height: 1050
  Depth: 24
  Visual: 0x21
  Visual Class: TrueColor
  Border width: 0
  Class: InputOutput
  Colormap: 0x20 (installed)
  Bit Gravity State: ForgetGravity
  Window Gravity State: NorthWestGravity
  Backing Store State: NotUseful
  Save Under State: no
  Map State: IsViewable
  Override Redirect State: no
  Corners:  +0+0  -0+0  -0-0  +0-0
  -geometry 1400x1050+0+0

which looks like what I expect for 'startx -- -depth 24'.

-- 
Steve