[Intel-gfx] [igt-dev] [PATCH i-g-t] igt/amdgpu_amd_prime: Bail if we fail to create more contexts

Chris Wilson chris at chris-wilson.co.uk
Thu Dec 13 15:45:35 UTC 2018


Quoting Chris Wilson (2018-12-13 15:36:43)
> Quoting Antonio Argenziano (2018-12-13 15:28:00)
> > 
> > 
> > On 13/12/18 03:57, Chris Wilson wrote:
> > > amdgpu has started to report out of space after creating a few contexts.
> > > This is not the scope of this test as here we just verifying that fences
> > > created in amd can be imported and used for synchronisation by i915 and
> > > for that we just need at least one context created!
> > > 
> > > References: https://bugs.freedesktop.org/show_bug.cgi?id=109049
> > > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > 
> > LGTM.
> > 
> > Reviwed-by: Antonio Argenziano <antonio.argenziano at intel.com>
> > 
> > > ---
> > >   tests/amdgpu/amd_prime.c | 5 +++--
> > >   1 file changed, 3 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/tests/amdgpu/amd_prime.c b/tests/amdgpu/amd_prime.c
> > > index bda0ce83d..518c88963 100644
> > > --- a/tests/amdgpu/amd_prime.c
> > > +++ b/tests/amdgpu/amd_prime.c
> > > @@ -354,8 +354,8 @@ static void amd_to_i915(int i915, int amd, amdgpu_device_handle device)
> > 
> > doesn't i915_to_amd have the same issue?
> 
> Only if you manage to run out of swap in 2s or used gen11. We don't like
> to mention the feature improvements.

Actually, in i915 we use asynchronous destruction of contexts and so
immediately release the resources and can reallocate as required. But
amdgpu uses synchronous context destruction and so we have to hold on to
the context even though we only want to import the fence into i915.

If we run out of contexts here with i915 (even on icl), it's a kernel
bug.
-Chris


More information about the Intel-gfx mailing list