Re: [igt-dev] [PATCH i-g-t] tests/i915/gem_lmem_swapping: Fix bug in test_smem_oom

["Melkaveri, Arjun" <arjun.melkaveri@intel.com>]

On Fri, Apr 01, 2022 at 01:59:28PM +0530, Ramalingam C wrote:
> On 2022-04-01 at 12:58:26 +0530, Arjun Melkaveri wrote:
> > ctx was not initialized with all physical engines,
> > after device reopening.
> > This change would resolve assert issue seen in __do_evict.
> > 
> > v2: Added missing intel_ctx_destroy.
> > 
> > Cc: Ashutosh Dixit <ashutosh.dixit@intel.com>
> > Cc: Ramalingam C <ramalingam.c@intel.com>
> > Signed-off-by: Arjun Melkaveri <arjun.melkaveri@intel.com>
> > ---
> >  tests/i915/gem_lmem_swapping.c | 4 ++++
> >  1 file changed, 4 insertions(+)
> > 
> > diff --git a/tests/i915/gem_lmem_swapping.c b/tests/i915/gem_lmem_swapping.c
> > index 43f0688f..432607b0 100644
> > --- a/tests/i915/gem_lmem_swapping.c
> > +++ b/tests/i915/gem_lmem_swapping.c
> > @@ -428,10 +428,14 @@ static void test_smem_oom(int i915,
> >  
> >  		fill_params(i915, &params, region, 0, 1, true);
> >  
> > +		ctx = intel_ctx_create_all_physical(fd);
> > +		__gem_context_set_persistence(fd, ctx->id, false);
> > +
> igt fixture already creates same kind of context and pass it here. Could
> we reuse it across all the child and parent processes?
> 
> Or we need independed context only?
> 
> Ram.
As we are repoening device , this has to be independed context
"int fd = gem_reopen_driver(i915);" 
> 
> >  		igt_install_exit_handler(smem_oom_exit_handler);
> >  		__do_evict(fd, ctx, &region->region, &params,
> >  			   params.seed + child + 1);
> >  
> > +		intel_ctx_destroy(fd, ctx);
> >  		close(fd);
> >  	}
> >  
> > -- 
> > 2.25.1
> >