Re: radeon panics kernels

On Fri, Oct 4, 2019 at 11:39 Steve Kargl <sgk_at_troutmask.apl.washington.edu>
wrote:

> On Thu, Oct 03, 2019 at 11:52:38PM +0200, Hans Petter Selasky wrote:
> > On 2019-10-03 22:26, Steve Kargl wrote:
> > > On Thu, Oct 03, 2019 at 03:05:27PM +0200, Hans Petter Selasky wrote:
> > >>
> > >> If you leave the port debug knob for drm-current-kmod AS-IS, I think
> you
> > >> can get away with:
> > >>
> > >> make DEBUG_FLAGS="-g"
> > >>
> > >> Then re-load the vmcore file in GDB/KGDB from ports (!) and add the
> > >> symbol files for the modules loaded. Then get the backtrace using bt
> > >> command.
> > >>
> > >> BTW: Did you try drm-devel-kmod for 13-current?
> > >>
> > >
> > > Took a bit of trial and error.  If I skip the panic
> > > and trap frames (#0 through #8). I find the backtrace
> > > that follows by sig.  If I move to frame #11, I see
> > >
> > > (kgdb) frame 11
> > > #11 r100_mm_rreg_slow (rdev=0xfffff80135766a70, reg=<optimized out>)
> > >      at
> /usr/ports/graphics/drm-current-kmod/work/kms-drm-2d2852e/drivers/gpu/drm/radeon/r100.c:4114
> > > 4114            writel(reg, ((void __iomem *)rdev->rmmio) +
> RADEON_MM_INDEX);
> > > (kgdb) p rdev->rmmio
> > > $3 = (void *) 0x0
> > >
> > > So, your guess of a NULL pointer seems correct.
> >
> > Can you do:
> >
> > set print pretty on
> > print *rdev
> >
>
> This produces close to 3400 lines of output.  Do you want me to
> send it to the list or directly to you?


Please put on gist, pastebin, etc and share the link.


>
> --
> Steve
>