i915/gem_mmap_offset: Ignore ENOSPC error for making residency execbuf

Platforms without ppgtt may be affected by ENOSPC when object is too big
to fit in gtt. Still we expect object will be cleared in pagefault path
so we can move forward even if execbuf is failing. To verify object
is cleared in both paths (execbuf/pagefault handler) we select them
randomly.

Fixes: https://gitlab.freedesktop.org/drm/intel/-/issues/6973



v2: Add explanation for random execution of make_resident() (Kamil)

Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
Cc: Kamil Konieczny <kamil.konieczny@linux.intel.com>
Reviewed-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>

tests/i915/gem_mmap_offset.c

@@ -102,7 +102,7 @@ static void make_resident(int i915, uint32_t batch, uint32_t handle)
 		.buffer_count = ARRAY_SIZE(obj),
 	};
 
-	gem_execbuf(i915, &eb);
+	__gem_execbuf(i915, &eb);
 	if (obj[1].handle != batch)
 		gem_close(i915, obj[1].handle);
 }
@@ -704,7 +704,9 @@ static void *thread_clear(void *data)
 		size = npages << 12;
 
 		igt_assert_eq(__gem_create_in_memory_region_list(i915, &handle, &size, 0, &arg->region, 1), 0);
-		make_resident(i915, batch, handle);
+		/* Zero-init bo in execbuf or pagefault handler path randomly */
+		if (random() & 1)
+			make_resident(i915, batch, handle);
 
 		ptr = __mmap_offset(i915, handle, 0, size,
 				    PROT_READ | PROT_WRITE,