diff mbox series

powerpc/perf/hv-24x7: Dont create sysfs event files for dummy events

Message ID 20201217113230.1069882-1-kjain@linux.ibm.com (mailing list archive)
State Changes Requested
Headers show
Series powerpc/perf/hv-24x7: Dont create sysfs event files for dummy events | expand

Commit Message

Kajol Jain Dec. 17, 2020, 11:32 a.m. UTC
hv_24x7 performance monitoring unit creates list of supported events
from the event catalog obtained via HCALL. hv_24x7 catalog could also
contain invalid or dummy events (with names like FREE_  or CPM_FREE_ so
on). These events does not have any hardware counters backing them.
So patch adds a check to string compare the event names to filter
out them.

Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
---
 arch/powerpc/perf/hv-24x7.c | 8 ++++++++
 1 file changed, 8 insertions(+)

Comments

Madhavan Srinivasan Dec. 17, 2020, 11:40 a.m. UTC | #1
On 12/17/20 5:02 PM, Kajol Jain wrote:
> hv_24x7 performance monitoring unit creates list of supported events
> from the event catalog obtained via HCALL. hv_24x7 catalog could also
> contain invalid or dummy events (with names like FREE_  or CPM_FREE_ so


Can you also include " RESERVED_NEST*" as part of the check.

# ls /sys/devices/hv_24x7/events | grep RESERVED
RESERVED_NEST1
RESERVED_NEST10
RESERVED_NEST11
RESERVED_NEST12
...


Maddy


> on). These events does not have any hardware counters backing them.
> So patch adds a check to string compare the event names to filter
> out them.
>
> Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
> ---
>   arch/powerpc/perf/hv-24x7.c | 8 ++++++++
>   1 file changed, 8 insertions(+)
>
> diff --git a/arch/powerpc/perf/hv-24x7.c b/arch/powerpc/perf/hv-24x7.c
> index 6e7e820508df..c3252d8a7818 100644
> --- a/arch/powerpc/perf/hv-24x7.c
> +++ b/arch/powerpc/perf/hv-24x7.c
> @@ -894,6 +894,11 @@ static int create_events_from_catalog(struct attribute ***events_,
>
>   		name = event_name(event, &nl);
>
> +		if (strstr(name, "FREE_")) {
> +			pr_info("invalid event %zu (%.*s)\n", event_idx, nl, name);
> +			junk_events++;
> +			continue;
> +		}
>   		if (event->event_group_record_len == 0) {
>   			pr_devel("invalid event %zu (%.*s): group_record_len == 0, skipping\n",
>   					event_idx, nl, name);
> @@ -955,6 +960,9 @@ static int create_events_from_catalog(struct attribute ***events_,
>   			continue;
>
>   		name  = event_name(event, &nl);
> +		if (strstr(name, "FREE_"))
> +			continue;
> +
>   		nonce = event_uniq_add(&ev_uniq, name, nl, event->domain);
>   		ct    = event_data_to_attrs(event_idx, events + event_attr_ct,
>   					    event, nonce);
Kajol Jain Dec. 17, 2020, 11:51 a.m. UTC | #2
On 12/17/20 5:10 PM, Madhavan Srinivasan wrote:
> 
> On 12/17/20 5:02 PM, Kajol Jain wrote:
>> hv_24x7 performance monitoring unit creates list of supported events
>> from the event catalog obtained via HCALL. hv_24x7 catalog could also
>> contain invalid or dummy events (with names like FREE_  or CPM_FREE_ so
> 
> 
> Can you also include " RESERVED_NEST*" as part of the check.

Hi Maddy,
   Sure, I will add this check.

Thanks,
Kajol Jain

> 
> # ls /sys/devices/hv_24x7/events | grep RESERVED
> RESERVED_NEST1
> RESERVED_NEST10
> RESERVED_NEST11
> RESERVED_NEST12
> ...
>
> 
> Maddy
> 
> 
>> on). These events does not have any hardware counters backing them.
>> So patch adds a check to string compare the event names to filter
>> out them.
>>
>> Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
>> ---
>>   arch/powerpc/perf/hv-24x7.c | 8 ++++++++
>>   1 file changed, 8 insertions(+)
>>
>> diff --git a/arch/powerpc/perf/hv-24x7.c b/arch/powerpc/perf/hv-24x7.c
>> index 6e7e820508df..c3252d8a7818 100644
>> --- a/arch/powerpc/perf/hv-24x7.c
>> +++ b/arch/powerpc/perf/hv-24x7.c
>> @@ -894,6 +894,11 @@ static int create_events_from_catalog(struct attribute ***events_,
>>
>>           name = event_name(event, &nl);
>>
>> +        if (strstr(name, "FREE_")) {
>> +            pr_info("invalid event %zu (%.*s)\n", event_idx, nl, name);
>> +            junk_events++;
>> +            continue;
>> +        }
>>           if (event->event_group_record_len == 0) {
>>               pr_devel("invalid event %zu (%.*s): group_record_len == 0, skipping\n",
>>                       event_idx, nl, name);
>> @@ -955,6 +960,9 @@ static int create_events_from_catalog(struct attribute ***events_,
>>               continue;
>>
>>           name  = event_name(event, &nl);
>> +        if (strstr(name, "FREE_"))
>> +            continue;
>> +
>>           nonce = event_uniq_add(&ev_uniq, name, nl, event->domain);
>>           ct    = event_data_to_attrs(event_idx, events + event_attr_ct,
>>                           event, nonce);
Michael Ellerman Dec. 18, 2020, 12:56 a.m. UTC | #3
Kajol Jain <kjain@linux.ibm.com> writes:
> hv_24x7 performance monitoring unit creates list of supported events
> from the event catalog obtained via HCALL. hv_24x7 catalog could also
> contain invalid or dummy events (with names like FREE_  or CPM_FREE_ so
> on). These events does not have any hardware counters backing them.
> So patch adds a check to string compare the event names to filter
> out them.
>
> Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
> ---
>  arch/powerpc/perf/hv-24x7.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
>
> diff --git a/arch/powerpc/perf/hv-24x7.c b/arch/powerpc/perf/hv-24x7.c
> index 6e7e820508df..c3252d8a7818 100644
> --- a/arch/powerpc/perf/hv-24x7.c
> +++ b/arch/powerpc/perf/hv-24x7.c
> @@ -894,6 +894,11 @@ static int create_events_from_catalog(struct attribute ***events_,
>  
>  		name = event_name(event, &nl);
>  
> +		if (strstr(name, "FREE_")) {
> +			pr_info("invalid event %zu (%.*s)\n", event_idx, nl, name);
> +			junk_events++;
> +			continue;

I don't think we want a print for each event, just one at the end saying
"Dropped %d invalid events" would be preferable I think.


> +		}
>  		if (event->event_group_record_len == 0) {
>  			pr_devel("invalid event %zu (%.*s): group_record_len == 0, skipping\n",
>  					event_idx, nl, name);
> @@ -955,6 +960,9 @@ static int create_events_from_catalog(struct attribute ***events_,
>  			continue;
>  
>  		name  = event_name(event, &nl);
> +		if (strstr(name, "FREE_"))
> +			continue;

Would be nice if the string comparison was in a single place, ie. in a
helper function.

cheers
Kajol Jain Dec. 18, 2020, 7:51 a.m. UTC | #4
On 12/18/20 6:26 AM, Michael Ellerman wrote:
> Kajol Jain <kjain@linux.ibm.com> writes:
>> hv_24x7 performance monitoring unit creates list of supported events
>> from the event catalog obtained via HCALL. hv_24x7 catalog could also
>> contain invalid or dummy events (with names like FREE_  or CPM_FREE_ so
>> on). These events does not have any hardware counters backing them.
>> So patch adds a check to string compare the event names to filter
>> out them.
>>
>> Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
>> ---
>>  arch/powerpc/perf/hv-24x7.c | 8 ++++++++
>>  1 file changed, 8 insertions(+)
>>
>> diff --git a/arch/powerpc/perf/hv-24x7.c b/arch/powerpc/perf/hv-24x7.c
>> index 6e7e820508df..c3252d8a7818 100644
>> --- a/arch/powerpc/perf/hv-24x7.c
>> +++ b/arch/powerpc/perf/hv-24x7.c
>> @@ -894,6 +894,11 @@ static int create_events_from_catalog(struct attribute ***events_,
>>  
>>  		name = event_name(event, &nl);
>>  
>> +		if (strstr(name, "FREE_")) {
>> +			pr_info("invalid event %zu (%.*s)\n", event_idx, nl, name);
>> +			junk_events++;
>> +			continue;
> 
> I don't think we want a print for each event, just one at the end saying
> "Dropped %d invalid events" would be preferable I think.

Hi Michael,
  Sure I will remove prints for each event. Having one print for number of dropped
events may not be useful. So I will drop that too.

> 
> 
>> +		}
>>  		if (event->event_group_record_len == 0) {
>>  			pr_devel("invalid event %zu (%.*s): group_record_len == 0, skipping\n",
>>  					event_idx, nl, name);
>> @@ -955,6 +960,9 @@ static int create_events_from_catalog(struct attribute ***events_,
>>  			continue;
>>  
>>  		name  = event_name(event, &nl);
>> +		if (strstr(name, "FREE_"))
>> +			continue;
> 
> Would be nice if the string comparison was in a single place, ie. in a
> helper function.

Sure I will make that change.

Thanks,
Kajol Jain
> 
> cheers
>
diff mbox series

Patch

diff --git a/arch/powerpc/perf/hv-24x7.c b/arch/powerpc/perf/hv-24x7.c
index 6e7e820508df..c3252d8a7818 100644
--- a/arch/powerpc/perf/hv-24x7.c
+++ b/arch/powerpc/perf/hv-24x7.c
@@ -894,6 +894,11 @@  static int create_events_from_catalog(struct attribute ***events_,
 
 		name = event_name(event, &nl);
 
+		if (strstr(name, "FREE_")) {
+			pr_info("invalid event %zu (%.*s)\n", event_idx, nl, name);
+			junk_events++;
+			continue;
+		}
 		if (event->event_group_record_len == 0) {
 			pr_devel("invalid event %zu (%.*s): group_record_len == 0, skipping\n",
 					event_idx, nl, name);
@@ -955,6 +960,9 @@  static int create_events_from_catalog(struct attribute ***events_,
 			continue;
 
 		name  = event_name(event, &nl);
+		if (strstr(name, "FREE_"))
+			continue;
+
 		nonce = event_uniq_add(&ev_uniq, name, nl, event->domain);
 		ct    = event_data_to_attrs(event_idx, events + event_attr_ct,
 					    event, nonce);