The absence of the above files would indicate a failure in the kernel,On 11/09, Robert Bragg wrote:
> +
> +igt_main
> +{
> + igt_skip_on_simulation();
> +
> + igt_fixture {
> + struct stat sb;
> + int ret;
> +
> + drm_fd = drm_open_driver_render(DRIVER_INTEL);
> + devid = intel_get_drm_devid(drm_fd);
> + device = drm_get_card();
> +
> + igt_require(IS_HASWELL(devid));
> + igt_require(lookup_hsw_render_basic_id());
> +
> + ret = stat("/proc/sys/dev/i915/perf_stream_paranoid", &sb);
> + igt_require(ret == 0);
> + ret = stat("/proc/sys/dev/i915/oa_max_sample_rate", &sb);
> + igt_require(ret == 0);
so would it not be more apt to assert, rather than skip ?
> +
> + gt_frequency_range_save();
> +
> + write_u64_file("/proc/sys/dev/i915/perf_stream_paranoid", 1); Don't we also want to ensure that the oa_max_sample_rate is also in a
"good" starting state before we begin, especially since we ensure that
we leave in its default state when cleaning up ?
Anyway, I think it all looks pretty reasonable to me and it looks like
we have a good amount of coverage, so you can have my r-b with Chris'
comment addressed.