Igt-dev Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: "Anirban, Sk" <sk.anirban@intel.com>
To: Soham Purkait <soham.purkait@intel.com>,
	<igt-dev@lists.freedesktop.org>,  <riana.tauro@intel.com>,
	<badal.nilawar@intel.com>, <kamil.konieczny@intel.com>,
	<vinay.belgaumkar@intel.com>
Cc: <anshuman.gupta@intel.com>
Subject: Re: [PATCH i-g-t 2/3] tests/intel/xe_pmu: Add check for the other drm clients using the device
Date: Thu, 30 Apr 2026 18:23:55 +0530	[thread overview]
Message-ID: <1b3a275e-0276-4cd2-9f48-c42b24e3f480@intel.com> (raw)
In-Reply-To: <20260402130026.2687520-3-soham.purkait@intel.com>

Hi Soham,

On 02-04-2026 06:30 pm, Soham Purkait wrote:
> If other processes are using the DRM device while idle residency test
> is running, GPU engines may be in use, and idle residency might be lower
> than expected leading to false positives.
> To help identify this issue, a check for the DRM clients, except the very
> test, is used and a warning message is printed if any such processes
> are found and the test is skipped.
>
> Signed-off-by: Soham Purkait <soham.purkait@intel.com>
> ---
>   tests/intel/xe_pmu.c | 27 +++++++++++++++++++++++++--
>   tests/meson.build    |  1 +
>   2 files changed, 26 insertions(+), 2 deletions(-)
>
> diff --git a/tests/intel/xe_pmu.c b/tests/intel/xe_pmu.c
> index fb4b871e7..c907b6207 100644
> --- a/tests/intel/xe_pmu.c
> +++ b/tests/intel/xe_pmu.c
> @@ -100,6 +100,7 @@
>   #include "igt_sysfs.h"
>   
>   #include "xe/xe_gt.h"
> +#include "igt_drm_clients.h"
>   #include "xe/xe_ioctl.h"
>   #include "xe/xe_spin.h"
>   #include "xe/xe_sriov_admin.h"
> @@ -865,17 +866,39 @@ static void engine_activity_multi_client(int fd, struct drm_xe_engine_class_inst
>   
>   static void test_gt_c6_idle(int xe, unsigned int gt)
>   {
> -	int pmu_fd;
> +	int pmu_fd, num_clients = 0;
>   	uint64_t pmu_config;
>   	uint64_t ts[2];
> -	unsigned long slept, start, end;
> +	unsigned long slept, start, end, wait = 100000;
>   	uint64_t val;
> +	struct igt_drm_client_proc *procs = NULL;
>   
>   	/* Get the PMU config for the gt-c6 event */
>   	pmu_config = get_event_config(gt, NULL, "gt-c6-residency");
>   
>   	pmu_fd = open_pmu(xe, pmu_config);
>   
> +	/* Wait for other processes using the DRM device ends */
> +	for (int pass = 0; pass < 50; pass++) {
> +		if (!igt_drm_get_client_procs(xe, getpid(), NULL))
> +			break;
> +		usleep(wait);
> +	}
IMO there is no need of waiting here again as we will be checking the C6 
state multiple times later on.
> +
> +	num_clients = igt_drm_get_client_procs(xe, getpid(), &procs);
> +
> +	/* Check if any process is still using the DRM device */
> +	if (num_clients) {
> +		for (unsigned int i = 0; i < num_clients; i++) {
> +			igt_warn("%s (pid=%u client=%lu) is using DRM device.\n",
> +				 procs[i].print_name, procs[i].pid, procs[i].id);
> +		}
> +		free(procs);
> +		procs = NULL;
> +		igt_skip("As other processes are using the DRM device, "
> +			 "measured idle_residency may be lower than expected.\n");
> +	}
It will better if you accommodate the functionality in a single 
function, so that we can simply call it whenever

the below line of code fails for all the C6 related tests.


Thanks,

Anirban

> +
>   	igt_require_f(igt_wait(xe_gt_is_in_c6(xe, gt), 1000, 10), "GT %d should be in C6\n", gt);
>   
>   	/* While idle check full RC6. */
> diff --git a/tests/meson.build b/tests/meson.build
> index 26d9345ec..e682ed9dd 100644
> --- a/tests/meson.build
> +++ b/tests/meson.build
> @@ -425,6 +425,7 @@ extra_dependencies = {
>   	'sw_sync': [ libatomic ],
>   	'xe_fault_injection': [ lib_igt_xe_oa ],
>   	'xe_oa': [ lib_igt_xe_oa ],
> +	'xe_pmu': [ lib_igt_drm_clients ],
>           'xe_compute': [ igt_deps,lib_igt_perf,lib_igt_profiling,math ],
>   }
>   

  reply	other threads:[~2026-04-30 12:54 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-04-02 13:00 [PATCH i-g-t 0/3] Guard gt-c6-idle test against busy DRM clients Soham Purkait
2026-04-02 13:00 ` [PATCH i-g-t 1/3] lib/igt_drm_clients: Add helper to list and count the DRM client processes Soham Purkait
2026-04-02 13:00 ` [PATCH i-g-t 2/3] tests/intel/xe_pmu: Add check for the other drm clients using the device Soham Purkait
2026-04-30 12:53   ` Anirban, Sk [this message]
2026-04-02 13:00 ` [PATCH i-g-t 3/3] HAX: Add gt-c6-idle test to xe-fast-feedback.testlist Soham Purkait
2026-04-02 17:30 ` ✓ Xe.CI.BAT: success for Guard gt-c6-idle test against busy DRM clients Patchwork
2026-04-02 17:42 ` ✓ i915.CI.BAT: " Patchwork
2026-04-03  4:33 ` ✗ Xe.CI.FULL: failure " Patchwork
2026-04-03 18:31 ` ✗ i915.CI.Full: " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1b3a275e-0276-4cd2-9f48-c42b24e3f480@intel.com \
    --to=sk.anirban@intel.com \
    --cc=anshuman.gupta@intel.com \
    --cc=badal.nilawar@intel.com \
    --cc=igt-dev@lists.freedesktop.org \
    --cc=kamil.konieczny@intel.com \
    --cc=riana.tauro@intel.com \
    --cc=soham.purkait@intel.com \
    --cc=vinay.belgaumkar@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox