[PATCH 0/2] perf tools: Fix record failure when mixed with ARM SPE event

[PATCH 0/2] perf tools: Fix record failure when mixed with ARM SPE event

[Wei Li]

This patch set fixes perf record failure when we mix arm_spe_x event with
other events in specific order.

Wei Li (2):
  perf tools: ARM SPE code cleanup
  perf tools: Fix record failure when mixed with ARM SPE event

 tools/perf/arch/arm/util/auxtrace.c | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

-- 
2.17.1

[PATCH 1/2] perf tools: ARM SPE code cleanup

[Wei Li]

Remove the useless check code to make it clear.

Signed-off-by: Wei Li <liwei391@huawei.com>
---
 tools/perf/arch/arm/util/auxtrace.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/tools/perf/arch/arm/util/auxtrace.c b/tools/perf/arch/arm/util/auxtrace.c
index 0a6e75b8777a..62b7b03d691a 100644
--- a/tools/perf/arch/arm/util/auxtrace.c
+++ b/tools/perf/arch/arm/util/auxtrace.c
@@ -57,7 +57,7 @@ struct auxtrace_record
 	struct evsel *evsel;
 	bool found_etm = false;
 	bool found_spe = false;
-	static struct perf_pmu **arm_spe_pmus = NULL;
+	static struct perf_pmu **arm_spe_pmus;
 	static int nr_spes = 0;
 	int i = 0;
 
@@ -65,9 +65,7 @@ struct auxtrace_record
 		return NULL;
 
 	cs_etm_pmu = perf_pmu__find(CORESIGHT_ETM_PMU_NAME);
-
-	if (!arm_spe_pmus)
-		arm_spe_pmus = find_all_arm_spe_pmus(&nr_spes, err);
+	arm_spe_pmus = find_all_arm_spe_pmus(&nr_spes, err);
 
 	evlist__for_each_entry(evlist, evsel) {
 		if (cs_etm_pmu &&
-- 
2.17.1

[PATCH 2/2] perf tools: Fix record failure when mixed with ARM SPE event

[Wei Li]

When recording with cache-misses and arm_spe_x event, i found that
it will just fail without showing any error info if i put cache-misses
after arm_spe_x event.

[root@localhost 0620]# perf record -e cache-misses -e \
arm_spe_0/ts_enable=1,pct_enable=1,pa_enable=1,load_filter=1,\
jitter=1,store_filter=1,min_latency=0/ sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.067 MB perf.data ]
[root@localhost 0620]# perf record -e \
arm_spe_0/ts_enable=1,pct_enable=1,pa_enable=1,load_filter=1,jitter=1,\
store_filter=1,min_latency=0/ -e cache-misses sleep 1
[root@localhost 0620]#

Finally, i found the reason is that the parameter 'arm_spe_pmu' passed to
arm_spe_recording_init() in auxtrace_record__init() is wrong. When the
arm_spe_x event is not the last event, 'arm_spe_pmus[i]' will be out of
bounds.

It seems that the code can't support concurrent multiple different
arm_spe_x events currently. So add the code to check and record the
found 'arm_spe_pmu' to fix this issue.

In fact, we don't support concurrent multiple same arm_spe_x events either,
that is checked in arm_spe_recording_options(), and it will show the
relevant info.

Fixes: ffd3d18c20b8d ("perf tools: Add ARM Statistical Profiling Extensions (SPE) support")
Signed-off-by: Wei Li <liwei391@huawei.com>
---
 tools/perf/arch/arm/util/auxtrace.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/tools/perf/arch/arm/util/auxtrace.c b/tools/perf/arch/arm/util/auxtrace.c
index 62b7b03d691a..7bb6f29e766c 100644
--- a/tools/perf/arch/arm/util/auxtrace.c
+++ b/tools/perf/arch/arm/util/auxtrace.c
@@ -58,6 +58,7 @@ struct auxtrace_record
 	bool found_etm = false;
 	bool found_spe = false;
 	static struct perf_pmu **arm_spe_pmus;
+	static struct perf_pmu *arm_spe_pmu;
 	static int nr_spes = 0;
 	int i = 0;
 
@@ -77,6 +78,13 @@ struct auxtrace_record
 
 		for (i = 0; i < nr_spes; i++) {
 			if (evsel->core.attr.type == arm_spe_pmus[i]->type) {
+				if (found_spe && (arm_spe_pmu != arm_spe_pmus[i])) {
+					pr_err("Concurrent multiple SPE operation not currently supported\n");
+					*err = -EOPNOTSUPP;
+					return NULL;
+				}
+
+				arm_spe_pmu = arm_spe_pmus[i];
 				found_spe = true;
 				break;
 			}
@@ -94,7 +102,7 @@ struct auxtrace_record
 
 #if defined(__aarch64__)
 	if (found_spe)
-		return arm_spe_recording_init(err, arm_spe_pmus[i]);
+		return arm_spe_recording_init(err, arm_spe_pmu);
 #endif
 
 	/*
-- 
2.17.1

Re: [PATCH 1/2] perf tools: ARM SPE code cleanup

[Mathieu Poirier]

On Tue, Jun 23, 2020 at 08:31:40PM +0800, Wei Li wrote:
> Remove the useless check code to make it clear.
> 
> Signed-off-by: Wei Li <liwei391@huawei.com>
> ---
>  tools/perf/arch/arm/util/auxtrace.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/tools/perf/arch/arm/util/auxtrace.c b/tools/perf/arch/arm/util/auxtrace.c
> index 0a6e75b8777a..62b7b03d691a 100644
> --- a/tools/perf/arch/arm/util/auxtrace.c
> +++ b/tools/perf/arch/arm/util/auxtrace.c
> @@ -57,7 +57,7 @@ struct auxtrace_record
>  	struct evsel *evsel;
>  	bool found_etm = false;
>  	bool found_spe = false;
> -	static struct perf_pmu **arm_spe_pmus = NULL;
> +	static struct perf_pmu **arm_spe_pmus;
>  	static int nr_spes = 0;
>  	int i = 0;
>  
> @@ -65,9 +65,7 @@ struct auxtrace_record
>  		return NULL;
>  
>  	cs_etm_pmu = perf_pmu__find(CORESIGHT_ETM_PMU_NAME);
> -
> -	if (!arm_spe_pmus)
> -		arm_spe_pmus = find_all_arm_spe_pmus(&nr_spes, err);
> +	arm_spe_pmus = find_all_arm_spe_pmus(&nr_spes, err);

Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>

>  
>  	evlist__for_each_entry(evlist, evsel) {
>  		if (cs_etm_pmu &&
> -- 
> 2.17.1
> 

Re: [PATCH 2/2] perf tools: Fix record failure when mixed with ARM SPE event

[Mathieu Poirier]

Hi Li,

On Tue, Jun 23, 2020 at 08:31:41PM +0800, Wei Li wrote:
> When recording with cache-misses and arm_spe_x event, i found that
> it will just fail without showing any error info if i put cache-misses
> after arm_spe_x event.
> 
> [root@localhost 0620]# perf record -e cache-misses -e \
> arm_spe_0/ts_enable=1,pct_enable=1,pa_enable=1,load_filter=1,\
> jitter=1,store_filter=1,min_latency=0/ sleep 1
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.067 MB perf.data ]
> [root@localhost 0620]# perf record -e \
> arm_spe_0/ts_enable=1,pct_enable=1,pa_enable=1,load_filter=1,jitter=1,\
> store_filter=1,min_latency=0/ -e cache-misses sleep 1
> [root@localhost 0620]#
> 
> Finally, i found the reason is that the parameter 'arm_spe_pmu' passed to
> arm_spe_recording_init() in auxtrace_record__init() is wrong. When the
> arm_spe_x event is not the last event, 'arm_spe_pmus[i]' will be out of
> bounds.

Yes, this indeed broken.  

The current code can only work if the only event to be
traced is an arm_spe_X, or if it is the last event to be specified.
Otherwise the last event type will be checked against all the
arm_spe_pmus[i]->types, none will match and an out of bound i index will be
used in arm_spc_recording_init().

Since this problem is not easy to figure out please include the above
explanation in the changelog.

> 
> It seems that the code can't support concurrent multiple different
> arm_spe_x events currently. So add the code to check and record the
> found 'arm_spe_pmu' to fix this issue.
> 
> In fact, we don't support concurrent multiple same arm_spe_x events either,
> that is checked in arm_spe_recording_options(), and it will show the
> relevant info.
> 
> Fixes: ffd3d18c20b8d ("perf tools: Add ARM Statistical Profiling Extensions (SPE) support")
> Signed-off-by: Wei Li <liwei391@huawei.com>
> ---
>  tools/perf/arch/arm/util/auxtrace.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/perf/arch/arm/util/auxtrace.c b/tools/perf/arch/arm/util/auxtrace.c
> index 62b7b03d691a..7bb6f29e766c 100644
> --- a/tools/perf/arch/arm/util/auxtrace.c
> +++ b/tools/perf/arch/arm/util/auxtrace.c
> @@ -58,6 +58,7 @@ struct auxtrace_record
>  	bool found_etm = false;
>  	bool found_spe = false;
>  	static struct perf_pmu **arm_spe_pmus;
> +	static struct perf_pmu *arm_spe_pmu;

As far as I can tell the "static" doesn't do anything.

>  	static int nr_spes = 0;
>  	int i = 0;
>  
> @@ -77,6 +78,13 @@ struct auxtrace_record
>  
>  		for (i = 0; i < nr_spes; i++) {
>  			if (evsel->core.attr.type == arm_spe_pmus[i]->type) {
> +				if (found_spe && (arm_spe_pmu != arm_spe_pmus[i])) {
> +					pr_err("Concurrent multiple SPE operation not currently supported\n");
> +					*err = -EOPNOTSUPP;
> +					return NULL;
> +				}

Instead of the above, which as you rightly pointed out, is also done in
arm_spe_recording_options() it might be best to just fix the "if (!nr_spes)"
condition:
                if (!nr_spes || arm_spe_pmu)
                        continue

Furthermore, instead of having a new arm_spe_pmu variable you could simply make
found_spe a struct perf_pmu.  That would be one less variable to take care of.

> +
> +				arm_spe_pmu = arm_spe_pmus[i];
>  				found_spe = true;

Last but not least do you know where the memory allocated for array arm_spe_pmus
is released?  If you can't find it either then we have a memory leak and it
would be nice to have that fixed.

Regards
Mathieu

PS: Leo Yan has spent a fair amount of time in the SPE code - please CC him on
your next revision.


>  				break;
>  			}
> @@ -94,7 +102,7 @@ struct auxtrace_record
>  
>  #if defined(__aarch64__)
>  	if (found_spe)
> -		return arm_spe_recording_init(err, arm_spe_pmus[i]);
> +		return arm_spe_recording_init(err, arm_spe_pmu);
>  #endif
>  
>  	/*
> -- 
> 2.17.1
> 

Re: [PATCH 2/2] perf tools: Fix record failure when mixed with ARM SPE event

[liwei (GF)]

Hi Mathieu,

On 2020/7/3 7:03, Mathieu Poirier wrote:
> Hi Li,
> 
> On Tue, Jun 23, 2020 at 08:31:41PM +0800, Wei Li wrote:
>> When recording with cache-misses and arm_spe_x event, i found that
>> it will just fail without showing any error info if i put cache-misses
>> after arm_spe_x event.
>>
>> [root@localhost 0620]# perf record -e cache-misses -e \
>> arm_spe_0/ts_enable=1,pct_enable=1,pa_enable=1,load_filter=1,\
>> jitter=1,store_filter=1,min_latency=0/ sleep 1
>> [ perf record: Woken up 1 times to write data ]
>> [ perf record: Captured and wrote 0.067 MB perf.data ]
>> [root@localhost 0620]# perf record -e \
>> arm_spe_0/ts_enable=1,pct_enable=1,pa_enable=1,load_filter=1,jitter=1,\
>> store_filter=1,min_latency=0/ -e cache-misses sleep 1
>> [root@localhost 0620]#
>>
>> Finally, i found the reason is that the parameter 'arm_spe_pmu' passed to
>> arm_spe_recording_init() in auxtrace_record__init() is wrong. When the
>> arm_spe_x event is not the last event, 'arm_spe_pmus[i]' will be out of
>> bounds.
> 
> Yes, this indeed broken.  
> 
> The current code can only work if the only event to be
> traced is an arm_spe_X, or if it is the last event to be specified.
> Otherwise the last event type will be checked against all the
> arm_spe_pmus[i]->types, none will match and an out of bound i index will be
> used in arm_spc_recording_init().
> 
> Since this problem is not easy to figure out please include the above
> explanation in the changelog.

OK.
>>
>> It seems that the code can't support concurrent multiple different
>> arm_spe_x events currently. So add the code to check and record the
>> found 'arm_spe_pmu' to fix this issue.
>>
>> In fact, we don't support concurrent multiple same arm_spe_x events either,
>> that is checked in arm_spe_recording_options(), and it will show the
>> relevant info.
>>
>> Fixes: ffd3d18c20b8d ("perf tools: Add ARM Statistical Profiling Extensions (SPE) support")
>> Signed-off-by: Wei Li <liwei391@huawei.com>
>> ---
>>  tools/perf/arch/arm/util/auxtrace.c | 10 +++++++++-
>>  1 file changed, 9 insertions(+), 1 deletion(-)
>>
>> diff --git a/tools/perf/arch/arm/util/auxtrace.c b/tools/perf/arch/arm/util/auxtrace.c
>> index 62b7b03d691a..7bb6f29e766c 100644
>> --- a/tools/perf/arch/arm/util/auxtrace.c
>> +++ b/tools/perf/arch/arm/util/auxtrace.c
>> @@ -58,6 +58,7 @@ struct auxtrace_record
>>  	bool found_etm = false;
>>  	bool found_spe = false;
>>  	static struct perf_pmu **arm_spe_pmus;
>> +	static struct perf_pmu *arm_spe_pmu;
> 
> As far as I can tell the "static" doesn't do anything.
> 
I will remove that in v2.
>>  	static int nr_spes = 0;
>>  	int i = 0;
>>  
>> @@ -77,6 +78,13 @@ struct auxtrace_record
>>  
>>  		for (i = 0; i < nr_spes; i++) {
>>  			if (evsel->core.attr.type == arm_spe_pmus[i]->type) {
>> +				if (found_spe && (arm_spe_pmu != arm_spe_pmus[i])) {
>> +					pr_err("Concurrent multiple SPE operation not currently supported\n");
>> +					*err = -EOPNOTSUPP;
>> +					return NULL;
>> +				}
> 
> Instead of the above, which as you rightly pointed out, is also done in
> arm_spe_recording_options() it might be best to just fix the "if (!nr_spes)"
> condition:
>                 if (!nr_spes || arm_spe_pmu)
>                         continue

This is more brief, i will use 'found_spe' as 'arm_spe_pmu' is not initialized.
> Furthermore, instead of having a new arm_spe_pmu variable you could simply make
> found_spe a struct perf_pmu.  That would be one less variable to take care of.
> 
>> +
>> +				arm_spe_pmu = arm_spe_pmus[i];
>>  				found_spe = true;
> 
> Last but not least do you know where the memory allocated for array arm_spe_pmus
> is released?  If you can't find it either then we have a memory leak and it
> would be nice to have that fixed.
Yes, we have a memory leak here indeed, i forgot to free it in this function.
As 'arm_spe_pmus' is defined as static, i think the author meant to assign it only at the first call,
but this function is only called once when we executing 'record', should i go on fixing it
or just drop the patch 1?

> Regards
> Mathieu
> 
> PS: Leo Yan has spent a fair amount of time in the SPE code - please CC him on
> your next revision.
> 
Thanks,
Wei

Re: [PATCH 1/2] perf tools: ARM SPE code cleanup

[Leo Yan]

Hi Wei,

On Tue, Jun 23, 2020 at 08:31:40PM +0800, Wei Li wrote:
> Remove the useless check code to make it clear.
> 
> Signed-off-by: Wei Li <liwei391@huawei.com>
> ---
>  tools/perf/arch/arm/util/auxtrace.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/tools/perf/arch/arm/util/auxtrace.c b/tools/perf/arch/arm/util/auxtrace.c
> index 0a6e75b8777a..62b7b03d691a 100644
> --- a/tools/perf/arch/arm/util/auxtrace.c
> +++ b/tools/perf/arch/arm/util/auxtrace.c
> @@ -57,7 +57,7 @@ struct auxtrace_record
>  	struct evsel *evsel;
>  	bool found_etm = false;
>  	bool found_spe = false;
> -	static struct perf_pmu **arm_spe_pmus = NULL;
> +	static struct perf_pmu **arm_spe_pmus;

Here the 'static' should be removed as well.

Just for more complete background info, IIUC, at the beginning to
enable SPE's PMU event, since SPE is micro-architecture dependent
(though it's defined in ARMv8-ARM, but it might be different for
different ARM micro-architectures).  So this is why here it uses
'static' for varaible "arm_spe_pmus", it wants to initialize the
variable with finding all SPE PMU structure at the first time when
invoke the function auxtrace_record__init(), and afterwards we can
reuse the variable "arm_spe_pmus" and without calling
find_all_arm_spe_pmus() anymore.

So I struggled to figure out what's good thing to do with multiple SPE
PMU events, and your change is good thing to me.  The reason is:

- Firstly, the function auxtrace_record__init() will be invoked only
  once, the variable "arm_spe_pmus" will not be used afterwards, thus
  we don't need to check "arm_spe_pmus" is NULL or not;
- Another reason is, even though SPE is micro-architecture dependent,
  but so far it only supports "statistical-profiling-extension-v1" and
  we have no chance to use multiple SPE's PMU events in Perf command.


So after removing 'static' for varaible "arm_spe_pmus":
Reviewed-by: Leo Yan <leo.yan@linaro.org>


P.s. Sorry if it's my reason that James Clark's patch [1] has not been
merged in the mainline kernel and introduced duplicate efforts at here.
James's patch used similiar method to resolve this same issue.

[1] https://lkml.org/lkml/2019/12/20/293

>  	static int nr_spes = 0;
>  	int i = 0;
>  
> @@ -65,9 +65,7 @@ struct auxtrace_record
>  		return NULL;
>  
>  	cs_etm_pmu = perf_pmu__find(CORESIGHT_ETM_PMU_NAME);
> -
> -	if (!arm_spe_pmus)
> -		arm_spe_pmus = find_all_arm_spe_pmus(&nr_spes, err);
> +	arm_spe_pmus = find_all_arm_spe_pmus(&nr_spes, err);
>  
>  	evlist__for_each_entry(evlist, evsel) {
>  		if (cs_etm_pmu &&
> -- 
> 2.17.1
> 

Re: [PATCH 2/2] perf tools: Fix record failure when mixed with ARM SPE event

[Leo Yan]

On Fri, Jul 03, 2020 at 12:06:15PM +0800, liwei (GF) wrote:

[...]

Thanks for Mathieu's looping and agreed with his comments.

> > Last but not least do you know where the memory allocated for array arm_spe_pmus
> > is released?  If you can't find it either then we have a memory leak and it
> > would be nice to have that fixed.
>
> Yes, we have a memory leak here indeed, i forgot to free it in this function.
> As 'arm_spe_pmus' is defined as static, i think the author meant to assign it only at the first call,
> but this function is only called once when we executing 'record', should i go on fixing it
> or just drop the patch 1?

I personally think patch 1 is reasonable.  So for fixing memory leak,
I did a quick check, it's good to release the array "arm_spe_pmus" in
the function auxtrace_record__init(), since the array is only used in
this function.

Thanks,
Leo