Re: [PATCH i-g-t] tests/intel/xe_oa: Get stream status on read error

[Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>]

On Mon, Apr 14, 2025 at 04:19:11PM -0700, Dixit, Ashutosh wrote:
>On Mon, 14 Apr 2025 15:19:12 -0700, Umesh Nerlige Ramappa wrote:
>>
>> On Thu, Apr 10, 2025 at 01:33:44PM -0700, Dixit, Ashutosh wrote:
>> > On Wed, 09 Apr 2025 16:34:41 -0700, Umesh Nerlige Ramappa wrote:
>> >>
>> >> Whenever read fails, get stream status so we know the state of OA status
>> >> bits.
>> >>
>> >> Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
>> >> ---
>> >>  tests/intel/xe_oa.c | 6 ++++++
>> >>  1 file changed, 6 insertions(+)
>> >>
>> >> diff --git a/tests/intel/xe_oa.c b/tests/intel/xe_oa.c
>> >> index 67acc74b3279..f3521c3d602e 100644
>> >> --- a/tests/intel/xe_oa.c
>> >> +++ b/tests/intel/xe_oa.c
>> >> @@ -2294,6 +2294,7 @@ static void test_polling_small_buf(void)
>> >>	errno = 0;
>> >>	ret = read(stream_fd, buf, sizeof(buf));
>> >>	igt_assert_eq(ret, -1);
>> >> +	get_stream_status(stream_fd);
>> >>	igt_assert_eq(errno, ENOSPC);
>> >>
>> >>	/* Poll with 0 timeout and expect POLLIN flag to be set */
>> >> @@ -2568,6 +2569,7 @@ test_enable_disable(const struct drm_xe_engine_class_instance *hwe)
>> >>	errno = 0;
>> >>	ret = read(stream_fd, buf, sizeof(buf));
>> >>	igt_assert_eq(ret, -1);
>> >> +	get_stream_status(stream_fd);
>> >>	igt_assert_eq(errno, EINVAL);
>> >>
>> >>	do_ioctl(stream_fd, DRM_XE_OBSERVATION_IOCTL_ENABLE, 0);
>> >> @@ -2584,6 +2586,7 @@ test_enable_disable(const struct drm_xe_engine_class_instance *hwe)
>> >>	/* Ensure num_reports can be read */
>> >>	while ((ret = read(stream_fd, buf, sizeof(buf))) < 0 && errno == EINTR)
>> >>		;
>> >> +	get_stream_status(stream_fd);
>> >>	igt_assert_eq(ret, sizeof(buf));
>> >
>> > Here read() is not failing, so maybe we should drop this one? Otherwise:
>>
>> But here we could get an EIO for some strange reason and would be good to
>> know if a specific status bit was set. The idea was to call
>> get_stream_status in all places where read was called.
>
>It's fine, I went ahead and merged this along with some other patches.

oh, thanks for pushing these.

Umesh
>
>> Thanks,
>> Umesh
>>
>> >
>> > Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
>> >
>> >>
>> >>	__perf_close(stream_fd);
>> >> @@ -2727,6 +2730,7 @@ test_non_sampling_read_error(void)
>> >>
>> >>	ret = read(stream_fd, buf, sizeof(buf));
>> >>	igt_assert_eq(ret, -1);
>> >> +	get_stream_status(stream_fd);
>> >>	igt_assert_eq(errno, EINVAL);
>> >>
>> >>	__perf_close(stream_fd);
>> >> @@ -2768,6 +2772,7 @@ test_disabled_read_error(void)
>> >>
>> >>	ret = read(stream_fd, buf, sizeof(buf));
>> >>	igt_assert_eq(ret, -1);
>> >> +	get_stream_status(stream_fd);
>> >>	igt_assert_eq(errno, EINVAL);
>> >>
>> >>	__perf_close(stream_fd);
>> >> @@ -2786,6 +2791,7 @@ test_disabled_read_error(void)
>> >>
>> >>	ret = read(stream_fd, buf, sizeof(buf));
>> >>	igt_assert_eq(ret, -1);
>> >> +	get_stream_status(stream_fd);
>> >>	igt_assert_eq(errno, EINVAL);
>> >>
>> >>	do_ioctl(stream_fd, DRM_XE_OBSERVATION_IOCTL_ENABLE, 0);
>> >> --
>> >> 2.43.0
>> >>