* [PATCH] Remove some PMU events for FUJITSU-MONAKA
@ 2025-02-27 5:40 Yoshihiro Furudera
2025-02-27 10:53 ` James Clark
0 siblings, 1 reply; 7+ messages in thread
From: Yoshihiro Furudera @ 2025-02-27 5:40 UTC (permalink / raw)
To: John Garry, Will Deacon, James Clark, Mike Leach, Leo Yan,
Peter Zijlstra, Ingo Molnar, Arnaldo Carvalho de Melo,
Namhyung Kim, Mark Rutland, Alexander Shishkin, Jiri Olsa,
Ian Rogers, Adrian Hunter, Liang, Kan, linux-arm-kernel,
linux-perf-users, linux-kernel, Akio Kakuno, Yoshihiro Furudera
The following events are not counted properly:
0x0037 LL_CACHE_MISS_RD
0x400B L3D_CACHE_LMISS_RD
0x0396 L2D_CACHE_REFILL_L3D_MISS
0x039A L2D_CACHE_REFILL_L3D_MISS_PRF
0x039B L2D_CACHE_REFILL_L3D_MISS_HWPRF
0x039C L2D_CACHE_REFILL_L3D_HIT
0x03A0 L2D_CACHE_REFILL_L3D_HIT_PRF
0x03A1 L2D_CACHE_REFILL_L3D_HIT_HWPRF
Specifically, these events are always counted as misses,
regardless of whether the L3 prefetch is a hit or a miss.
So I remove these events in l3_cache.json, ll_cache.json
Signed-off-by: Yoshihiro Furudera <fj5100bi@fujitsu.com>
---
.../arch/arm64/fujitsu/monaka/l3_cache.json | 34 -------------------
.../arch/arm64/fujitsu/monaka/ll_cache.json | 4 ---
2 files changed, 38 deletions(-)
diff --git a/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/l3_cache.json b/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/l3_cache.json
index 3f3e0d22ac68..917b9b5bf8bb 100644
--- a/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/l3_cache.json
+++ b/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/l3_cache.json
@@ -37,11 +37,6 @@
"EventName": "L2D_CACHE_REFILL_L3D_CACHE_HWPRF",
"BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_CACHE caused by hardware prefetch access."
},
- {
- "EventCode": "0x0396",
- "EventName": "L2D_CACHE_REFILL_L3D_MISS",
- "BriefDescription": "This event counts operations that cause a miss of the L3 cache."
- },
{
"EventCode": "0x0397",
"EventName": "L2D_CACHE_REFILL_L3D_MISS_DM",
@@ -57,21 +52,6 @@
"EventName": "L2D_CACHE_REFILL_L3D_MISS_DM_WR",
"BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_MISS caused by demand write access."
},
- {
- "EventCode": "0x039A",
- "EventName": "L2D_CACHE_REFILL_L3D_MISS_PRF",
- "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_MISS caused by prefetch access."
- },
- {
- "EventCode": "0x039B",
- "EventName": "L2D_CACHE_REFILL_L3D_MISS_HWPRF",
- "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_MISS caused by hardware prefetch access."
- },
- {
- "EventCode": "0x039C",
- "EventName": "L2D_CACHE_REFILL_L3D_HIT",
- "BriefDescription": "This event counts operations that cause a hit of the L3 cache."
- },
{
"EventCode": "0x039D",
"EventName": "L2D_CACHE_REFILL_L3D_HIT_DM",
@@ -87,16 +67,6 @@
"EventName": "L2D_CACHE_REFILL_L3D_HIT_DM_WR",
"BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_HIT caused by demand write access."
},
- {
- "EventCode": "0x03A0",
- "EventName": "L2D_CACHE_REFILL_L3D_HIT_PRF",
- "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_HIT caused by prefetch access."
- },
- {
- "EventCode": "0x03A1",
- "EventName": "L2D_CACHE_REFILL_L3D_HIT_HWPRF",
- "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_HIT caused by hardware prefetch access."
- },
{
"EventCode": "0x03A2",
"EventName": "L2D_CACHE_REFILL_L3D_MISS_PFTGT_HIT",
@@ -151,9 +121,5 @@
"EventCode": "0x03AC",
"EventName": "L2D_CACHE_REFILL_L3D_MISS_FR_L3",
"BriefDescription": "This event counts the number of L3 cache misses where the requests access L3 cache in the different socket from the requests."
- },
- {
- "ArchStdEvent": "L3D_CACHE_LMISS_RD",
- "BriefDescription": "This event counts access counted by L3D_CACHE that is not completed by the L3D cache, and a Memory-read operation, as defined by the L2D_CACHE_REFILL_L3D_MISS events."
}
]
diff --git a/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/ll_cache.json b/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/ll_cache.json
index a441b84729ab..c2325855eda3 100644
--- a/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/ll_cache.json
+++ b/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/ll_cache.json
@@ -2,9 +2,5 @@
{
"ArchStdEvent": "LL_CACHE_RD",
"BriefDescription": "This event counts access counted by L3D_CACHE that is a Memory-read operation, as defined by the L2D_CACHE_REFILL_L3D_CACHE events."
- },
- {
- "ArchStdEvent": "LL_CACHE_MISS_RD",
- "BriefDescription": "This event counts access counted by L3D_CACHE that is not completed by the L3D cache, and a Memory-read operation, as defined by the L2D_CACHE_REFILL_L3D_MISS events."
}
]
--
2.43.0
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH] Remove some PMU events for FUJITSU-MONAKA
2025-02-27 5:40 [PATCH] Remove some PMU events for FUJITSU-MONAKA Yoshihiro Furudera
@ 2025-02-27 10:53 ` James Clark
2025-02-28 9:55 ` Yoshihiro Furudera (Fujitsu)
0 siblings, 1 reply; 7+ messages in thread
From: James Clark @ 2025-02-27 10:53 UTC (permalink / raw)
To: Yoshihiro Furudera
Cc: John Garry, Will Deacon, Mike Leach, Leo Yan, Peter Zijlstra,
Ingo Molnar, Arnaldo Carvalho de Melo, Namhyung Kim,
Mark Rutland, Alexander Shishkin, Jiri Olsa, Ian Rogers,
Adrian Hunter, Liang, Kan, linux-arm-kernel, linux-perf-users,
linux-kernel, Akio Kakuno
On 27/02/2025 5:40 am, Yoshihiro Furudera wrote:
> The following events are not counted properly:
>
> 0x0037 LL_CACHE_MISS_RD
> 0x400B L3D_CACHE_LMISS_RD
These two are discoverable so will still appear in
/sys/bus/event_source/devices/armv8_pmuv3_0/events/ if the hardware says
they exist. It might be better to change the json strings of these two
to warn that they don't work if that's the case, otherwise Perf will
still list them and you'll be worse off.
> 0x0396 L2D_CACHE_REFILL_L3D_MISS
> 0x039A L2D_CACHE_REFILL_L3D_MISS_PRF
> 0x039B L2D_CACHE_REFILL_L3D_MISS_HWPRF
> 0x039C L2D_CACHE_REFILL_L3D_HIT
> 0x03A0 L2D_CACHE_REFILL_L3D_HIT_PRF
> 0x03A1 L2D_CACHE_REFILL_L3D_HIT_HWPRF
>
> Specifically, these events are always counted as misses,
> regardless of whether the L3 prefetch is a hit or a miss.
> So I remove these events in l3_cache.json, ll_cache.json
>
> Signed-off-by: Yoshihiro Furudera <fj5100bi@fujitsu.com>
> ---
> .../arch/arm64/fujitsu/monaka/l3_cache.json | 34 -------------------
> .../arch/arm64/fujitsu/monaka/ll_cache.json | 4 ---
> 2 files changed, 38 deletions(-)
>
> diff --git a/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/l3_cache.json b/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/l3_cache.json
> index 3f3e0d22ac68..917b9b5bf8bb 100644
> --- a/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/l3_cache.json
> +++ b/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/l3_cache.json
> @@ -37,11 +37,6 @@
> "EventName": "L2D_CACHE_REFILL_L3D_CACHE_HWPRF",
> "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_CACHE caused by hardware prefetch access."
> },
> - {
> - "EventCode": "0x0396",
> - "EventName": "L2D_CACHE_REFILL_L3D_MISS",
> - "BriefDescription": "This event counts operations that cause a miss of the L3 cache."
> - },
> {
> "EventCode": "0x0397",
> "EventName": "L2D_CACHE_REFILL_L3D_MISS_DM",
> @@ -57,21 +52,6 @@
> "EventName": "L2D_CACHE_REFILL_L3D_MISS_DM_WR",
> "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_MISS caused by demand write access."
> },
> - {
> - "EventCode": "0x039A",
> - "EventName": "L2D_CACHE_REFILL_L3D_MISS_PRF",
> - "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_MISS caused by prefetch access."
> - },
> - {
> - "EventCode": "0x039B",
> - "EventName": "L2D_CACHE_REFILL_L3D_MISS_HWPRF",
> - "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_MISS caused by hardware prefetch access."
> - },
> - {
> - "EventCode": "0x039C",
> - "EventName": "L2D_CACHE_REFILL_L3D_HIT",
> - "BriefDescription": "This event counts operations that cause a hit of the L3 cache."
> - },
> {
> "EventCode": "0x039D",
> "EventName": "L2D_CACHE_REFILL_L3D_HIT_DM",
> @@ -87,16 +67,6 @@
> "EventName": "L2D_CACHE_REFILL_L3D_HIT_DM_WR",
> "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_HIT caused by demand write access."
> },
> - {
> - "EventCode": "0x03A0",
> - "EventName": "L2D_CACHE_REFILL_L3D_HIT_PRF",
> - "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_HIT caused by prefetch access."
> - },
> - {
> - "EventCode": "0x03A1",
> - "EventName": "L2D_CACHE_REFILL_L3D_HIT_HWPRF",
> - "BriefDescription": "This event counts L2D_CACHE_REFILL_L3D_HIT caused by hardware prefetch access."
> - },
> {
> "EventCode": "0x03A2",
> "EventName": "L2D_CACHE_REFILL_L3D_MISS_PFTGT_HIT",
> @@ -151,9 +121,5 @@
> "EventCode": "0x03AC",
> "EventName": "L2D_CACHE_REFILL_L3D_MISS_FR_L3",
> "BriefDescription": "This event counts the number of L3 cache misses where the requests access L3 cache in the different socket from the requests."
> - },
> - {
> - "ArchStdEvent": "L3D_CACHE_LMISS_RD",
> - "BriefDescription": "This event counts access counted by L3D_CACHE that is not completed by the L3D cache, and a Memory-read operation, as defined by the L2D_CACHE_REFILL_L3D_MISS events."
> }
> ]
> diff --git a/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/ll_cache.json b/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/ll_cache.json
> index a441b84729ab..c2325855eda3 100644
> --- a/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/ll_cache.json
> +++ b/tools/perf/pmu-events/arch/arm64/fujitsu/monaka/ll_cache.json
> @@ -2,9 +2,5 @@
> {
> "ArchStdEvent": "LL_CACHE_RD",
> "BriefDescription": "This event counts access counted by L3D_CACHE that is a Memory-read operation, as defined by the L2D_CACHE_REFILL_L3D_CACHE events."
> - },
> - {
> - "ArchStdEvent": "LL_CACHE_MISS_RD",
> - "BriefDescription": "This event counts access counted by L3D_CACHE that is not completed by the L3D cache, and a Memory-read operation, as defined by the L2D_CACHE_REFILL_L3D_MISS events."
> }
> ]
^ permalink raw reply [flat|nested] 7+ messages in thread* RE: [PATCH] Remove some PMU events for FUJITSU-MONAKA
2025-02-27 10:53 ` James Clark
@ 2025-02-28 9:55 ` Yoshihiro Furudera (Fujitsu)
2025-03-05 6:40 ` Yoshihiro Furudera (Fujitsu)
0 siblings, 1 reply; 7+ messages in thread
From: Yoshihiro Furudera (Fujitsu) @ 2025-02-28 9:55 UTC (permalink / raw)
To: 'James Clark'
Cc: John Garry, Will Deacon, Mike Leach, Leo Yan, Peter Zijlstra,
Ingo Molnar, Arnaldo Carvalho de Melo, Namhyung Kim,
Mark Rutland, Alexander Shishkin, Jiri Olsa, Ian Rogers,
Adrian Hunter, Liang, Kan, linux-arm-kernel, linux-perf-users,
linux-kernel, Akio Kakuno (Fujitsu)
Hi, James
Thank you for your comment.
> On 27/02/2025 5:40 am, Yoshihiro Furudera wrote:
> > The following events are not counted properly:
> >
> > 0x0037 LL_CACHE_MISS_RD
> > 0x400B L3D_CACHE_LMISS_RD
>
> These two are discoverable so will still appear in
> /sys/bus/event_source/devices/armv8_pmuv3_0/events/ if the hardware says
> they exist. It might be better to change the json strings of these two to warn that
> they don't work if that's the case, otherwise Perf will still list them and you'll be
> worse off.
I will leave these 2 events and
add a warning message to the description in the JSON file.
I will handle other events in the same way as these 2 events.
>
> > 0x0396 L2D_CACHE_REFILL_L3D_MISS
> > 0x039A L2D_CACHE_REFILL_L3D_MISS_PRF
> > 0x039B L2D_CACHE_REFILL_L3D_MISS_HWPRF 0x039C
> L2D_CACHE_REFILL_L3D_HIT
> > 0x03A0 L2D_CACHE_REFILL_L3D_HIT_PRF
> > 0x03A1 L2D_CACHE_REFILL_L3D_HIT_HWPRF
> >
Best Regards,
Yoshihiro Furudera
^ permalink raw reply [flat|nested] 7+ messages in thread
* RE: [PATCH] Remove some PMU events for FUJITSU-MONAKA
2025-02-28 9:55 ` Yoshihiro Furudera (Fujitsu)
@ 2025-03-05 6:40 ` Yoshihiro Furudera (Fujitsu)
2025-03-05 9:37 ` James Clark
0 siblings, 1 reply; 7+ messages in thread
From: Yoshihiro Furudera (Fujitsu) @ 2025-03-05 6:40 UTC (permalink / raw)
To: 'James Clark'
Cc: 'John Garry', 'Will Deacon', 'Mike Leach',
'Leo Yan', 'Peter Zijlstra',
'Ingo Molnar', 'Arnaldo Carvalho de Melo',
'Namhyung Kim', 'Mark Rutland',
'Alexander Shishkin', 'Jiri Olsa',
'Ian Rogers', 'Adrian Hunter',
'Liang, Kan',
'linux-arm-kernel@lists.infradead.org',
'linux-perf-users@vger.kernel.org',
'linux-kernel@vger.kernel.org', Akio Kakuno (Fujitsu)
Hi, James
>
> Hi, James
> Thank you for your comment.
>
> > On 27/02/2025 5:40 am, Yoshihiro Furudera wrote:
> > > The following events are not counted properly:
> > >
> > > 0x0037 LL_CACHE_MISS_RD
> > > 0x400B L3D_CACHE_LMISS_RD
> >
> > These two are discoverable so will still appear in
> > /sys/bus/event_source/devices/armv8_pmuv3_0/events/ if the hardware
> > says they exist. It might be better to change the json strings of
> > these two to warn that they don't work if that's the case, otherwise
> > Perf will still list them and you'll be worse off.
>
> I will leave these 2 events and
> add a warning message to the description in the JSON file.
> I will handle other events in the same way as these 2 events.
I'm thinking of adding one of the following warnings
to the description of events where inaccurate counts
occur on FUJITSU-MONAKA.
Is this okay?
1.Simple version
"Note: This event does not count accurately."
2.Detailed version
"Note: This event does not count accurately because it counts as a miss regardless of whether the L3 prefetch is a hit or miss."
I think "2.Detailed version" is better.
example:
{
"EventCode": "0x0396",
"EventName": "L2D_CACHE_REFILL_L3D_MISS",
"BriefDescription": "This event counts operations that cause a miss of the L3 cache. Note: This event does not count accurately because it counts as a miss regardless of whether the L3 prefetch is a hit or miss."
}
Best Regards,
Yoshihiro Furudera
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH] Remove some PMU events for FUJITSU-MONAKA
2025-03-05 6:40 ` Yoshihiro Furudera (Fujitsu)
@ 2025-03-05 9:37 ` James Clark
2025-03-07 3:10 ` Yoshihiro Furudera (Fujitsu)
0 siblings, 1 reply; 7+ messages in thread
From: James Clark @ 2025-03-05 9:37 UTC (permalink / raw)
To: Yoshihiro Furudera (Fujitsu)
Cc: 'John Garry', 'Will Deacon', 'Mike Leach',
'Leo Yan', 'Peter Zijlstra',
'Ingo Molnar', 'Arnaldo Carvalho de Melo',
'Namhyung Kim', 'Mark Rutland',
'Alexander Shishkin', 'Jiri Olsa',
'Ian Rogers', 'Adrian Hunter',
'Liang, Kan',
'linux-arm-kernel@lists.infradead.org',
'linux-perf-users@vger.kernel.org',
'linux-kernel@vger.kernel.org', Akio Kakuno (Fujitsu)
On 05/03/2025 6:40 am, Yoshihiro Furudera (Fujitsu) wrote:
> Hi, James
>
>>
>> Hi, James
>> Thank you for your comment.
>>
>>> On 27/02/2025 5:40 am, Yoshihiro Furudera wrote:
>>>> The following events are not counted properly:
>>>>
>>>> 0x0037 LL_CACHE_MISS_RD
>>>> 0x400B L3D_CACHE_LMISS_RD
>>>
>>> These two are discoverable so will still appear in
>>> /sys/bus/event_source/devices/armv8_pmuv3_0/events/ if the hardware
>>> says they exist. It might be better to change the json strings of
>>> these two to warn that they don't work if that's the case, otherwise
>>> Perf will still list them and you'll be worse off.
>>
>> I will leave these 2 events and
>> add a warning message to the description in the JSON file.
>> I will handle other events in the same way as these 2 events.
>
> I'm thinking of adding one of the following warnings
> to the description of events where inaccurate counts
> occur on FUJITSU-MONAKA.
> Is this okay?
>
> 1.Simple version
> "Note: This event does not count accurately."
>
> 2.Detailed version
> "Note: This event does not count accurately because it counts as a miss regardless of whether the L3 prefetch is a hit or miss."
>
> I think "2.Detailed version" is better.
>
> example:
> {
> "EventCode": "0x0396",
> "EventName": "L2D_CACHE_REFILL_L3D_MISS",
> "BriefDescription": "This event counts operations that cause a miss of the L3 cache. Note: This event does not count accurately because it counts as a miss regardless of whether the L3 prefetch is a hit or miss."
> }
>
> Best Regards,
> Yoshihiro Furudera
You could have both by using BriefDescription and PublicDescription. The
longer one being available with 'perf list -v'. I think that's what that
feature is for.
Either way you should probably also update the description before the
note, so add 'hit or miss' to the first sentence:
"EventCode": "0x0396",
"EventName": "L2D_CACHE_REFILL_L3D_MISS",
"BriefDescription": "This event counts operations that cause a hit or
miss of the L3 cache. Note that this incorrectly counts both hits _and_
misses."
^ permalink raw reply [flat|nested] 7+ messages in thread* RE: [PATCH] Remove some PMU events for FUJITSU-MONAKA
2025-03-05 9:37 ` James Clark
@ 2025-03-07 3:10 ` Yoshihiro Furudera (Fujitsu)
2025-03-27 4:24 ` Koichi Okuno (Fujitsu)
0 siblings, 1 reply; 7+ messages in thread
From: Yoshihiro Furudera (Fujitsu) @ 2025-03-07 3:10 UTC (permalink / raw)
To: 'James Clark'
Cc: 'John Garry', 'Will Deacon', 'Mike Leach',
'Leo Yan', 'Peter Zijlstra',
'Ingo Molnar', 'Arnaldo Carvalho de Melo',
'Namhyung Kim', 'Mark Rutland',
'Alexander Shishkin', 'Jiri Olsa',
'Ian Rogers', 'Adrian Hunter',
'Liang, Kan',
'linux-arm-kernel@lists.infradead.org',
'linux-perf-users@vger.kernel.org',
'linux-kernel@vger.kernel.org', Akio Kakuno (Fujitsu)
Hi, James
Thank you for your comment.
> On 05/03/2025 6:40 am, Yoshihiro Furudera (Fujitsu) wrote:
> > Hi, James
> >
> >>
> >> Hi, James
> >> Thank you for your comment.
> >>
> >>> On 27/02/2025 5:40 am, Yoshihiro Furudera wrote:
> >>>> The following events are not counted properly:
> >>>>
> >>>> 0x0037 LL_CACHE_MISS_RD
> >>>> 0x400B L3D_CACHE_LMISS_RD
> >>>
> >>> These two are discoverable so will still appear in
> >>> /sys/bus/event_source/devices/armv8_pmuv3_0/events/ if the hardware
> >>> says they exist. It might be better to change the json strings of
> >>> these two to warn that they don't work if that's the case, otherwise
> >>> Perf will still list them and you'll be worse off.
> >>
> >> I will leave these 2 events and
> >> add a warning message to the description in the JSON file.
> >> I will handle other events in the same way as these 2 events.
> >
> > I'm thinking of adding one of the following warnings to the
> > description of events where inaccurate counts occur on FUJITSU-MONAKA.
> > Is this okay?
> >
> > 1.Simple version
> > "Note: This event does not count accurately."
> >
> > 2.Detailed version
> > "Note: This event does not count accurately because it counts as a miss
> regardless of whether the L3 prefetch is a hit or miss."
> >
> > I think "2.Detailed version" is better.
> >
> > example:
> > {
> > "EventCode": "0x0396",
> > "EventName": "L2D_CACHE_REFILL_L3D_MISS",
> > "BriefDescription": "This event counts operations that cause a miss of the
> L3 cache. Note: This event does not count accurately because it counts as a miss
> regardless of whether the L3 prefetch is a hit or miss."
> > }
> >
> > Best Regards,
> > Yoshihiro Furudera
>
> You could have both by using BriefDescription and PublicDescription. The longer
> one being available with 'perf list -v'. I think that's what that feature is for.
Thank you for your suggestion.
After discussing this with the OS team,
we would like to move forward with the policy of
defining "2.Detailed version" in the Brief Description.
Please let me make a final decision after discussing it
with the hardware team.
>
> Either way you should probably also update the description before the note, so
> add 'hit or miss' to the first sentence:
>
> "EventCode": "0x0396",
> "EventName": "L2D_CACHE_REFILL_L3D_MISS",
> "BriefDescription": "This event counts operations that cause a hit or miss of the L3
> cache. Note that this incorrectly counts both hits _and_ misses."
We will also discuss with the hardware team on this matter.
Best Regards,
Yoshihiro Furudera
^ permalink raw reply [flat|nested] 7+ messages in thread* RE: [PATCH] Remove some PMU events for FUJITSU-MONAKA
2025-03-07 3:10 ` Yoshihiro Furudera (Fujitsu)
@ 2025-03-27 4:24 ` Koichi Okuno (Fujitsu)
0 siblings, 0 replies; 7+ messages in thread
From: Koichi Okuno (Fujitsu) @ 2025-03-27 4:24 UTC (permalink / raw)
To: 'James Clark'
Cc: 'John Garry', 'Will Deacon', 'Mike Leach',
'Leo Yan', 'Peter Zijlstra',
'Ingo Molnar', 'Arnaldo Carvalho de Melo',
'Namhyung Kim', 'Mark Rutland',
'Alexander Shishkin', 'Jiri Olsa',
'Ian Rogers', 'Adrian Hunter',
'Liang, Kan',
'linux-arm-kernel@lists.infradead.org',
'linux-perf-users@vger.kernel.org',
'linux-kernel@vger.kernel.org', Akio Kakuno (Fujitsu),
Yoshihiro Furudera (Fujitsu)
Hi, James
Sorry for the late reply.
Also, the person in charge here has changed from Furudera to Okuno.
> Hi, James
> Thank you for your comment.
>
> > On 05/03/2025 6:40 am, Yoshihiro Furudera (Fujitsu) wrote:
> > > Hi, James
> > >
> > >>
> > >> Hi, James
> > >> Thank you for your comment.
> > >>
> > >>> On 27/02/2025 5:40 am, Yoshihiro Furudera wrote:
> > >>>> The following events are not counted properly:
> > >>>>
> > >>>> 0x0037 LL_CACHE_MISS_RD
> > >>>> 0x400B L3D_CACHE_LMISS_RD
> > >>>
> > >>> These two are discoverable so will still appear in
> > >>> /sys/bus/event_source/devices/armv8_pmuv3_0/events/ if the
> > >>> hardware says they exist. It might be better to change the json
> > >>> strings of these two to warn that they don't work if that's the
> > >>> case, otherwise Perf will still list them and you'll be worse off.
> > >>
> > >> I will leave these 2 events and
> > >> add a warning message to the description in the JSON file.
> > >> I will handle other events in the same way as these 2 events.
> > >
> > > I'm thinking of adding one of the following warnings to the
> > > description of events where inaccurate counts occur on
> FUJITSU-MONAKA.
> > > Is this okay?
> > >
> > > 1.Simple version
> > > "Note: This event does not count accurately."
> > >
> > > 2.Detailed version
> > > "Note: This event does not count accurately because it counts as a
> > > miss
> > regardless of whether the L3 prefetch is a hit or miss."
> > >
> > > I think "2.Detailed version" is better.
> > >
> > > example:
> > > {
> > > "EventCode": "0x0396",
> > > "EventName": "L2D_CACHE_REFILL_L3D_MISS",
> > > "BriefDescription": "This event counts operations that cause a
> > > miss of the
> > L3 cache. Note: This event does not count accurately because it counts
> > as a miss regardless of whether the L3 prefetch is a hit or miss."
> > > }
> > >
> > > Best Regards,
> > > Yoshihiro Furudera
> >
> > You could have both by using BriefDescription and PublicDescription.
> > The longer one being available with 'perf list -v'. I think that's what that
> feature is for.
>
> Thank you for your suggestion.
> After discussing this with the OS team,
> we would like to move forward with the policy of defining "2.Detailed version" in
> the Brief Description.
> Please let me make a final decision after discussing it with the hardware team.
>
> >
> > Either way you should probably also update the description before the
> > note, so add 'hit or miss' to the first sentence:
> >
> > "EventCode": "0x0396",
> > "EventName": "L2D_CACHE_REFILL_L3D_MISS",
> > "BriefDescription": "This event counts operations that cause a hit or
> > miss of the L3 cache. Note that this incorrectly counts both hits _and_
> misses."
>
> We will also discuss with the hardware team on this matter.
After discussing with the hardware team,
for all events where the count is incorrect,
we have decided to add the following text to
the end of the "BriefDescription":
"Note: This event does not count accurately."
I will repost the patch later.
>
> Best Regards,
> Yoshihiro Furudera
Best Regards,
Koichi Okuno
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2025-03-27 4:24 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-02-27 5:40 [PATCH] Remove some PMU events for FUJITSU-MONAKA Yoshihiro Furudera
2025-02-27 10:53 ` James Clark
2025-02-28 9:55 ` Yoshihiro Furudera (Fujitsu)
2025-03-05 6:40 ` Yoshihiro Furudera (Fujitsu)
2025-03-05 9:37 ` James Clark
2025-03-07 3:10 ` Yoshihiro Furudera (Fujitsu)
2025-03-27 4:24 ` Koichi Okuno (Fujitsu)
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®