Re: [igt-dev] [PATCH i-g-t] tests/i915/gem_exec_schedule: Resolve missing context for independent subtest

["Dixit, Ashutosh" <ashutosh.dixit@intel.com>]

On Sun, 29 Aug 2021 13:15:59 -0700, Arjun Melkaveri wrote:
>
> @@ -380,22 +381,24 @@ static void independent(int fd, const intel_ctx_t *ctx, unsigned int engine,
>		if (!gem_class_can_store_dword(fd, e->class))
>			continue;
>
> +		tmp_ctx = intel_ctx_create(fd, &ctx->cfg);
>		if (spin == NULL) {
>			spin = __igt_spin_new(fd,
>					      .ahnd = ahnd,
> -					      .ctx = ctx,
> +					      .ctx = tmp_ctx,
>					      .engine = e->flags,
>					      .flags = flags);
>		} else {
>			struct drm_i915_gem_execbuffer2 eb = {
>				.buffer_count = 1,
>				.buffers_ptr = to_user_pointer(&spin->obj[IGT_SPIN_BATCH]),
> +				.rsvd1 = tmp_ctx->id,

.rsvd1 is definitely missing which should explain the assert for
gem_execbuf, but looks like we should not need a new "tmp_ctx" but should
be able to submit against the original "ctx", so all we need is:

				.rsvd1 = ctx->id,