[Intel-gfx] [PATCH 6/9] drm/i915/pmu: Export counters from all tiles
Umesh Nerlige Ramappa
umesh.nerlige.ramappa at intel.com
Thu Mar 30 17:33:33 UTC 2023
On Thu, Mar 30, 2023 at 02:01:42PM +0100, Tvrtko Ursulin wrote:
>
>On 30/03/2023 01:41, Umesh Nerlige Ramappa wrote:
>>From: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
>>
>>Start exporting frequency and RC6 counters from all tiles.
>>
>>Existing counters keep their names and config values and new one use the
>>namespace added in the previous patch, with the "-gtN" added to their
>>names.
>
>The part about keeping the names is not in the code any more. So something will have to give, either the commit text or the code.
>
>Even without that detail, I suspect someone might want to add them Co-developed-by since I *think* someone did some changes.
>>Interrupts counter is an odd one off. Because it is the global device
>>counters (not only GT) we choose not to add per tile versions for now.
>>
>>Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
>>Signed-off-by: Aravind Iddamsetty <aravind.iddamsetty at intel.com>
>>---
>> drivers/gpu/drm/i915/i915_pmu.c | 96 ++++++++++++++++++++++++++-------
>> 1 file changed, 77 insertions(+), 19 deletions(-)
>>
>>diff --git a/drivers/gpu/drm/i915/i915_pmu.c b/drivers/gpu/drm/i915/i915_pmu.c
>>index 5d1de98d86b4..2a5deabff088 100644
>>--- a/drivers/gpu/drm/i915/i915_pmu.c
>>+++ b/drivers/gpu/drm/i915/i915_pmu.c
>>@@ -548,8 +548,9 @@ config_status(struct drm_i915_private *i915, u64 config)
>> struct intel_gt *gt = to_gt(i915);
>> unsigned int gt_id = config_gt_id(config);
>>+ unsigned int max_gt_id = HAS_EXTRA_GT_LIST(i915) ? 1 : 0;
>>- if (gt_id)
>>+ if (gt_id > max_gt_id)
>> return -ENOENT;
>> switch (config_counter(config)) {
>>@@ -563,6 +564,8 @@ config_status(struct drm_i915_private *i915, u64 config)
>> return -ENODEV;
>> break;
>> case I915_PMU_INTERRUPTS:
>>+ if (gt_id)
>>+ return -ENOENT;
>> break;
>> case I915_PMU_RC6_RESIDENCY:
>> if (!gt->rc6.supported)
>>@@ -932,9 +935,9 @@ static const struct attribute_group i915_pmu_cpumask_attr_group = {
>> .attrs = i915_cpumask_attrs,
>> };
>>-#define __event(__config, __name, __unit) \
>>+#define __event(__counter, __name, __unit) \
>> { \
>>- .config = (__config), \
>>+ .counter = (__counter), \
>> .name = (__name), \
>> .unit = (__unit), \
>> }
>>@@ -975,15 +978,21 @@ create_event_attributes(struct i915_pmu *pmu)
>> {
>> struct drm_i915_private *i915 = container_of(pmu, typeof(*i915), pmu);
>> static const struct {
>>- u64 config;
>>+ unsigned int counter;
>> const char *name;
>> const char *unit;
>> } events[] = {
>>- __event(I915_PMU_ACTUAL_FREQUENCY, "actual-frequency", "M"),
>>- __event(I915_PMU_REQUESTED_FREQUENCY, "requested-frequency", "M"),
>>- __event(I915_PMU_INTERRUPTS, "interrupts", NULL),
>>- __event(I915_PMU_RC6_RESIDENCY, "rc6-residency", "ns"),
>>- __event(I915_PMU_SOFTWARE_GT_AWAKE_TIME, "software-gt-awake-time", "ns"),
>>+ __event(0, "actual-frequency", "M"),
>>+ __event(1, "requested-frequency", "M"),
>>+ __event(3, "rc6-residency", "ns"),
>>+ __event(4, "software-gt-awake-time", "ns"),
>>+ };
>>+ static const struct {
>>+ unsigned int counter;
>>+ const char *name;
>>+ const char *unit;
>>+ } global_events[] = {
>>+ __event(2, "interrupts", NULL),
>> };
>> static const struct {
>> enum drm_i915_pmu_engine_sample sample;
>>@@ -998,14 +1007,29 @@ create_event_attributes(struct i915_pmu *pmu)
>> struct i915_ext_attribute *i915_attr = NULL, *i915_iter;
>> struct attribute **attr = NULL, **attr_iter;
>> struct intel_engine_cs *engine;
>>- unsigned int i;
>>+ struct intel_gt *gt;
>>+ unsigned int i, j;
>> /* Count how many counters we will be exposing. */
>>- for (i = 0; i < ARRAY_SIZE(events); i++) {
>>- if (!config_status(i915, events[i].config))
>>+ /* per gt counters */
>
>Two comments one by another, two styles - the inconsistency hurts.
>
>Not sure why global events needed to be split out into a separate array? Like this below two loops are needed for each stage instead of one. AFAIR one array and one loop would just work because config_status wold report global ones as unsupported for gt > 0.
The idea was to add the legacy events into the global array. These
events will not have -gtN appeneded to it. Note that on a single gt
platform, my idea is to have both legacy as well as gt0 events.
ADLP:
actual-frequency
actual-frequency-gt0
MTL:
actual-frequency
actual-frequency-gt0
actual-frequency-gt1
>
>[Comes back later. It looked like this in my code:
>
> static const struct {
>- u64 config;
>+ unsigned int counter;
> const char *name;
> const char *unit;
>+ bool global;
> } events[] = {
>- __event(I915_PMU_ACTUAL_FREQUENCY, "actual-frequency", "M"),
>- __event(I915_PMU_REQUESTED_FREQUENCY, "requested-frequency", "M"),
>- __event(I915_PMU_INTERRUPTS, "interrupts", NULL),
>- __event(I915_PMU_RC6_RESIDENCY, "rc6-residency", "ns"),
>+ /*
>+ * #define __I915_PMU_ACTUAL_FREQUENCY(gt) ___I915_PMU_OTHER(gt, 0)
>+ * #define __I915_PMU_REQUESTED_FREQUENCY(gt) ___I915_PMU_OTHER(gt, 1)
>+ * #define __I915_PMU_INTERRUPTS(gt) ___I915_PMU_OTHER(gt, 2)
>+ * #define __I915_PMU_RC6_RESIDENCY(gt) ___I915_PMU_OTHER(gt, 3)
>+ */
>+ __event(0, "actual-frequency", "M"),
>+ __event(1, "requested-frequency", "M"),
>+ __global_event(2, "interrupts", NULL),
>+ __event(3, "rc6-residency", "ns"),
>
>...
>
> /* Count how many counters we will be exposing. */
>- for (i = 0; i < ARRAY_SIZE(events); i++) {
>- if (!config_status(i915, events[i].config))
>- count++;
>+ for_each_gt(i915, j, gt) {
>+ for (i = 0; i < ARRAY_SIZE(events); i++) {
>+ u64 config = ___I915_PMU_OTHER(j, events[i].counter);
>+
>+ if (!config_status(i915, config))
>+ count++;
>+ }
>
>So AFAICT it just worked.
If we decide to drop 9/9, then I would drop the 7/9 and 8/9 and just
move back to your original patches because they worked as is. The only
open then would be if we want to have the -gt0 events as well for single
gt platforms.
The idea is to make this similar to what's implemented for the sysfs
frequency/rc6 attribues in /sys/class/drm/card0. There is a root version
as well as a gt/gt0 version. fwiu, gt/gt0 attributes are used on a
single gt platform.
>
>]
>
>>+ for_each_gt(gt, i915, j) {
>>+ for (i = 0; i < ARRAY_SIZE(events); i++) {
>>+ u64 config = ___I915_PMU_OTHER(j, events[i].counter);
>>+
>>+ if (!config_status(i915, config))
>>+ count++;
>>+ }
>>+ }
>>+
>>+ /* global (per GPU) counters */
>>+ for (i = 0; i < ARRAY_SIZE(global_events); i++) {
>>+ u64 config = ___I915_PMU_OTHER(0, global_events[i].counter);
>>+
>>+ if (!config_status(i915, config))
>> count++;
>> }
>>+ /* per engine counters */
>> for_each_uabi_engine(engine, i915) {
>> for (i = 0; i < ARRAY_SIZE(engine_events); i++) {
>> if (!engine_event_status(engine,
>>@@ -1033,26 +1057,60 @@ create_event_attributes(struct i915_pmu *pmu)
>> attr_iter = attr;
>> /* Initialize supported non-engine counters. */
>>- for (i = 0; i < ARRAY_SIZE(events); i++) {
>>+ /* per gt counters */
>>+ for_each_gt(gt, i915, j) {
>>+ for (i = 0; i < ARRAY_SIZE(events); i++) {
>>+ u64 config = ___I915_PMU_OTHER(j, events[i].counter);
>>+ char *str;
>>+
>>+ if (config_status(i915, config))
>>+ continue;
>>+
>>+ str = kasprintf(GFP_KERNEL, "%s-gt%u",
>>+ events[i].name, j);
>
>So with this patch all old platforms change the event names. This is not how I wrote it, and more importantly, it breaks userspace. Why would we do it?
With this patch alone, yes, this would break the uapi. With the series,
not really because those events are added back in 9/9. Should we retain
uapi compatibility in each patch? If yes, then I need to change this.
>
>For reference I dug out my code from 2020 and it looked like this:
>
>+ if (events[i].global || !i915->remote_tiles)
>+ str = kstrdup(events[i].name, GFP_KERNEL);
>+ else
>+ str = kasprintf(GFP_KERNEL, "%s-gt%u",
>+ events[i].name, j);
>
>So on single tile platforms names remain the same.
The series still maintains the same idea, but also adds xxxx-gt0 events
for the old platforms.
Thanks,
Umesh
>
>Regards,
>
>Tvrtko
>
>>+ if (!str)
>>+ goto err;
>>+
>>+ *attr_iter++ = &i915_iter->attr.attr;
>>+ i915_iter = add_i915_attr(i915_iter, str, config);
>>+
>>+ if (events[i].unit) {
>>+ str = kasprintf(GFP_KERNEL, "%s-gt%u.unit",
>>+ events[i].name, j);
>>+ if (!str)
>>+ goto err;
>>+
>>+ *attr_iter++ = &pmu_iter->attr.attr;
>>+ pmu_iter = add_pmu_attr(pmu_iter, str,
>>+ events[i].unit);
>>+ }
>>+ }
>>+ }
>>+
>>+ /* global (per GPU) counters */
>>+ for (i = 0; i < ARRAY_SIZE(global_events); i++) {
>>+ u64 config = ___I915_PMU_OTHER(0, global_events[i].counter);
>> char *str;
>>- if (config_status(i915, events[i].config))
>>+ if (config_status(i915, config))
>> continue;
>>- str = kstrdup(events[i].name, GFP_KERNEL);
>>+ str = kstrdup(global_events[i].name, GFP_KERNEL);
>> if (!str)
>> goto err;
>> *attr_iter++ = &i915_iter->attr.attr;
>>- i915_iter = add_i915_attr(i915_iter, str, events[i].config);
>>+ i915_iter = add_i915_attr(i915_iter, str, config);
>>- if (events[i].unit) {
>>- str = kasprintf(GFP_KERNEL, "%s.unit", events[i].name);
>>+ if (global_events[i].unit) {
>>+ str = kasprintf(GFP_KERNEL, "%s.unit",
>>+ global_events[i].name);
>> if (!str)
>> goto err;
>> *attr_iter++ = &pmu_iter->attr.attr;
>>- pmu_iter = add_pmu_attr(pmu_iter, str, events[i].unit);
>>+ pmu_iter = add_pmu_attr(pmu_iter, str,
>>+ global_events[i].unit);
>> }
>> }
More information about the Intel-gfx
mailing list