[Intel-gfx] PROBLEM: i915 causes complete desktop freezes in 4.15-rc5

Alexandru Chirvasitu achirvasub at gmail.com
Fri Jan 5 19:58:42 UTC 2018


OK, I then plan to try with the previous config, plus these
modifications:

CONFIG_PAGE_POISONING not set
CONFIG_SLUB_STATS=y

I'm a little puzzled about this:

On Fri, Jan 05, 2018 at 07:51:01PM +0000, Chris Wilson wrote:
> Quoting Alexandru Chirvasitu (2018-01-05 19:37:24)
> > I'll try. I need a bit of clarification:
> > 
> > On Fri, Jan 05, 2018 at 05:52:25PM +0000, Chris Wilson wrote:
> > > Quoting Alexandru Chirvasitu (2018-01-03 21:53:15)
> > > > All right, here's the dmesg from the kernel compiled from drm-tip (in
> > > > sync with upstream at the time of the compilation earlier today), with
> > > > 
> > > > CONFIG_DRM_I915_DEBUG_GEM=y
> > > > 
> > > > I crashed it by opening 20+ xterm windows. Doesn't always do it though
> > > > (tried this before).
> > > 
> > > Sorry, still stumped. It's still the same use-after-free and no asserts
> > > hit. Can you keep KASAN enabled but disable slab/page poisoning? Hmm, I
> > > think it has to be page poisoning doing the 0x6b as SLAB_POISON is
> > > disabled by default.
> > 
> > Would this have to be through a new compilation, or would it do to
> > just remove the poisoning kernel parameter on boot? 
> 
> CONFIG_PAGE_POISONING
> 
> > 
> > > (You could check by enabling slabstats can looking in sysfs.)
> > 
> > Sorry, I was having some trouble parsing this; what do I check again
> > and how?
> > 
> > Does 'slabstats' refer to another kernel config option or something
> > else?
> 
> It would be CONFIG_SLUB_STATS and then you would a
> /sys/kernel/slab/i915_dependency/ directory with all the details of the
> slab, and in particular a poison file showing the status (and allowing
> it to modified) of slab poisoning
> 
> Ok, it looks like poisoning is only available if CONFIG_SLUB_DEBUG is
> set. Since it is not set, don't worry about it ;)

I actually do have SLUB_DEBUG set to 'y' in the kernel whose dmesg I
last sent you.

> -Chris


More information about the Intel-gfx mailing list