[igt-dev] [PATCH i-g-t v2] tests/xe: Add basic idle residency test
Gupta, Anshuman
anshuman.gupta at intel.com
Mon Jun 26 07:11:36 UTC 2023
> -----Original Message-----
> From: Tauro, Riana <riana.tauro at intel.com>
> Sent: Monday, June 26, 2023 11:49 AM
> To: Gupta, Anshuman <anshuman.gupta at intel.com>; igt-
> dev at lists.freedesktop.org; kamil.konieczny at linux.intel.com
> Cc: Dixit, Ashutosh <ashutosh.dixit at intel.com>; Nilawar, Badal
> <badal.nilawar at intel.com>; Belgaumkar, Vinay
> <vinay.belgaumkar at intel.com>; Sundaresan, Sujaritha
> <sujaritha.sundaresan at intel.com>
> Subject: Re: [PATCH i-g-t v2] tests/xe: Add basic idle residency test
>
> Hi Anshuman
>
> On 6/22/2023 11:04 AM, Gupta, Anshuman wrote:
> >
> >
> >> -----Original Message-----
> >> From: Tauro, Riana <riana.tauro at intel.com>
> >> Sent: Thursday, June 22, 2023 10:40 AM
> >> To: Gupta, Anshuman <anshuman.gupta at intel.com>; igt-
> >> dev at lists.freedesktop.org
> >> Cc: Dixit, Ashutosh <ashutosh.dixit at intel.com>; Nilawar, Badal
> >> <badal.nilawar at intel.com>; kamil.konieczny at linux.intel.com;
> >> Belgaumkar, Vinay <vinay.belgaumkar at intel.com>; Sundaresan, Sujaritha
> >> <sujaritha.sundaresan at intel.com>
> >> Subject: Re: [PATCH i-g-t v2] tests/xe: Add basic idle residency test
> >>
> >> Hi Anshuman
> >>
> >> Thanks for the review
> >>
> >> On 6/21/2023 5:31 PM, Gupta, Anshuman wrote:
> >>>
> >>>
> >>>> -----Original Message-----
> >>>> From: Tauro, Riana <riana.tauro at intel.com>
> >>>> Sent: Wednesday, June 21, 2023 3:21 PM
> >>>> To: igt-dev at lists.freedesktop.org
> >>>> Cc: Tauro, Riana <riana.tauro at intel.com>; Gupta, Anshuman
> >>>> <anshuman.gupta at intel.com>; Dixit, Ashutosh
> >>>> <ashutosh.dixit at intel.com>; Nilawar, Badal
> >>>> <badal.nilawar at intel.com>; kamil.konieczny at linux.intel.com
> >>>> Subject: [PATCH i-g-t v2] tests/xe: Add basic idle residency test
> >>>>
> >>>> The test reads idle residency within a time interval and checks if
> >>>> its within the tolerance
> >>>>
> >>>> Patch for gtidle properties
> >>>> https://patchwork.freedesktop.org/series/119262/
> >>>>
> >>>> v2:
> >>>> - rename file to xe_pm_residency
> >>>> - add kernel patch in commit message (Kamil)
> >>>>
> >>>> Signed-off-by: Riana Tauro <riana.tauro at intel.com>
> >>>> ---
> >>>> tests/meson.build | 1 +
> >>>> tests/xe/xe_pm_residency.c | 117
> >>>> +++++++++++++++++++++++++++++++++++++
> >>>> 2 files changed, 118 insertions(+)
> >>>> create mode 100644 tests/xe/xe_pm_residency.c
> >>>>
> >>>> diff --git a/tests/meson.build b/tests/meson.build index
> >>>> f908ae885..f877b7a11 100644
> >>>> --- a/tests/meson.build
> >>>> +++ b/tests/meson.build
> >>>> @@ -264,6 +264,7 @@ xe_progs = [
> >>>> 'xe_module_load',
> >>>> 'xe_noexec_ping_pong',
> >>>> 'xe_pm',
> >>>> + 'xe_pm_residency',
> >>>> 'xe_prime_self_import',
> >>>> 'xe_query',
> >>>> 'xe_vm',
> >>>> diff --git a/tests/xe/xe_pm_residency.c
> >>>> b/tests/xe/xe_pm_residency.c new file mode 100644 index
> >>>> 000000000..33f840325
> >>>> --- /dev/null
> >>>> +++ b/tests/xe/xe_pm_residency.c
> >>>> @@ -0,0 +1,117 @@
> >>>> +// SPDX-License-Identifier: MIT
> >>>> +/*
> >>>> + * Copyright © 2023 Intel Corporation */
> >>>> +
> >>>> +/**
> >>>> + * TEST: Test gtidle properties
> >>>> + * Category: Software building block
> >>>> + * Sub-category: Power Management
> >>>> + * Functionality: GT C States
> >>>> + * Test category: functionality test */
> >>>> +
> >>>> +#include <limits.h>
> >>>> +
> >>>> +#include "igt.h"
> >>>> +#include "igt_sysfs.h"
> >>>> +
> >>>> +#include "xe/xe_query.h"
> >>>> +
> >>>> +#define SLEEP_DURATION 3000 /* in milliseconds */
> >>>> +
> >>>> +const double tolerance = 0.1;
> >>>> +
> >>>> +#define assert_within_epsilon(x, ref, tol) \
> >>>> + igt_assert_f((double)(x) <= (1.0 + (tol)) * (double)(ref) && \
> >>>> + (double)(x) >= (1.0 - (tol)) * (double)(ref), \
> >>>> + "'%s' != '%s' (%f not within +%.1f%%/-%.1f%% tolerance of
> >>>> %f)\n",\
> >>>> + #x, #ref, (double)(x), \
> >>>> + (tol) * 100.0, (tol) * 100.0, \
> >>>> + (double)(ref))
> >>> This might get used at several places, how about moving it to a library ?
> >> Will move this to lib. lib/igt_pm should be okay?
> > I think igt_aux.c would be suitable, @kmail can provide more input on that.
>
>
> Will send this as a separate patch as moving to lib will affect some of the i915
> tests as well.
>
> Need to fix the other tests as they will cause build warnings
Agree, this can be taken up in separate patch series.
As you are adding new test, it needs to add.
IGT_TEST_DESCRIPTION() and igt_describe().
With that,
Reviewed-by: Anshuman Gupta <anshuman.gupta at intel.com>
>
> Thanks
> Riana
>
> >
> >>>> +
> >>>> +static unsigned int measured_usleep(unsigned int usec) {
> >>>> + struct timespec ts = { };
> >>>> + unsigned int slept;
> >>>> +
> >>>> + slept = igt_nsec_elapsed(&ts);
> >>>> + igt_assert(slept == 0);
> >>>> + do {
> >>>> + usleep(usec - slept);
> >>> slept was nsec, then why aren't we converting it to usec ?
> >> The first time slept is 0. From the next loop run, it is converted to usec.
> >>>> + slept = igt_nsec_elapsed(&ts) / 1000;
> >>>> + } while (slept < usec);
> >>>> +
> >>>> + return igt_nsec_elapsed(&ts) / 1000; }
> >>>> +
> >>>> +static bool in_gt_c6(int sysfs, int gt) {
> >>>> + char path[PATH_MAX];
> >>>> + char gt_c_state[8];
> >>>> +
> >>>> + sprintf(path, "device/gt%d/gtidle/idle_status", gt);
> >>>> + if (igt_sysfs_scanf(sysfs, path, "%s", gt_c_state) < 0)
> >>>> + return false;
> >>>> +
> >>>> + return strcmp(gt_c_state, "gt-c6") == 0; }
> >>>> +
> >>>> +static unsigned long read_idle_residency(int sysfs, int gt) {
> >>>> + unsigned long residency;
> >>>> + char path[PATH_MAX];
> >>>> +
> >>>> + residency = 0;
> >>>> + sprintf(path, "device/gt%d/gtidle/idle_residency_ms", gt);
> >>> IMO we need something like igt_debugfs_gt_dir() here for sysfs.
> >> sysfs lib also has this igt_sysfs_gt_open but the path is specific to
> >> i915. If this needs to be added, should it be a new lib ? or this can
> >> be done for the tile patches?
> > Hmm this can be pursue later once we will resolve the path of sysfs.
> > Thanks,
> > Anshuman Gupta.
> >>
> >> Thanks
> >> Riana
> >>> Thanks,
> >>> Anshuman Gupta.
> >>>> + igt_assert(igt_sysfs_scanf(sysfs, path, "%lu", &residency) == 1);
> >>>> + return residency;
> >>>> +}
> >>>> +
> >>>> +/**
> >>>> + * SUBTEST: idle-residency
> >>>> + * Description: basic residency test to check idle residency
> >>>> + * within a time interval
> >>>> + * Run type: FULL
> >>>> + */
> >>>> +static void test_idle_residency(int sysfs, int gt) {
> >>>> + unsigned long elapsed_ms, residency_start, residency_end;
> >>>> +
> >>>> + assert(igt_wait(in_gt_c6(sysfs, gt), 1000, 1));
> >>>> +
> >>>> + residency_start = read_idle_residency(sysfs, gt);
> >>>> + elapsed_ms = measured_usleep(SLEEP_DURATION * 1000) / 1000;
> >>>> + residency_end = read_idle_residency(sysfs, gt);
> >>>> +
> >>>> + igt_info("Measured %lums of idle residency in %lums\n",
> >>>> + residency_end - residency_start, elapsed_ms);
> >>>> +
> >>>> + assert_within_epsilon(residency_end - residency_start,
> >>>> +elapsed_ms, tolerance); }
> >>>> +
> >>>> +igt_main
> >>>> +{
> >>>> + int fd, gt;
> >>>> + static int sysfs = -1;
> >>>> +
> >>>> + igt_fixture {
> >>>> + fd = drm_open_driver(DRIVER_XE);
> >>>> +
> >>>> + xe_device_get(fd);
> >>>> + igt_require(!IS_PONTEVECCHIO(xe_dev_id(fd)));
> >>>> +
> >>>> + sysfs = igt_sysfs_open(fd);
> >>>> + igt_assert(sysfs != -1);
> >>>> + }
> >>>> +
> >>>> + igt_subtest("idle-residency")
> >>>> + xe_for_each_gt(fd, gt)
> >>>> + test_idle_residency(sysfs, gt);
> >>>> +
> >>>> + igt_fixture {
> >>>> + close(sysfs);
> >>>> + xe_device_put(fd);
> >>>> + close(fd);
> >>>> + }
> >>>> +}
> >>>> --
> >>>> 2.40.0
> >>>
More information about the igt-dev
mailing list