Quoting Antonio Argenziano (2017-12-19 21:16:40)
> Since commit: drm/i915/scheduler: Support user-defined priorities, the
> driver support an extra context param to set context's priority. Add
> tests for that interface and update invalid tests.
>
> v2:
> - Add arg size validation test. (Chris)
> - Add arg value overflow test. (Chris)
> - Add test for unsupported platforms. (Chris)
> - Feed interface with all priority values and in random order. (Chris)
>
> Signed-off-by: Antonio Argenziano <antonio.argenziano@xxxxxxxxx>
> Cc: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
> Cc: Michal Winiarski <michal.winiarski@xxxxxxxxx>
> ---
> + igt_subtest("user-set-priority-invalid-value") {
> + int prio_values[PRIO_RANGE - USER_PRIO_RANGE];
> + for (int i = 0; i < (PRIO_RANGE - USER_PRIO_RANGE); i++)
> + prio_values[i] = i + (MAX_USER_SET_PRIO + 1);
> + igt_permute_array(prio_values, ARRAY_SIZE(prio_values), igt_exchange_int);
> +
> + arg.ctx_id = gem_context_create(fd);
So arg.ctx_id points to different contexts depending on which subtests
are run, beside the context leak. But that does also make a good
suggestion that you do want to check both the default context (0) and a
user created context for handling (it may matter not today, but we do
already depend on the equivalence, i.e. that we can probe the default
context to determine support for user contexts). Do you feel like
parameterising the tests yet? :)
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
