| Message ID | ef46a47fc4ab909093f14b2b4289a4835836ab6c.1738851844.git.robin.murphy@arm.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | perf/arm-cmn: Minor event type housekeeping | expand |
On Thu, 6 Feb 2025, Robin Murphy wrote: > While handling RN-D nodes under the functionally-identical RN-I type > works fine for perf tool users using the "rnid_" event aliases, and that > is the documented and expected ABI, there's little reason not to be > permissive and accept the actual RN-D type as an additional encoding for > the same events as well. This may be convenient for other tooling > generating event configs directly from its own topology data. > > In the RN-I event mood, it also seems as good a time as any to clean up > a forgotten macro for CCLA_RNI events which ended up being unnecessary. > > Signed-off-by: Robin Murphy <robin.murphy@arm.com> Looks good to me. Reviewed-by: Ilkka Koskinen <ilkka@os.amperecomputing.com> > --- > drivers/perf/arm-cmn.c | 5 +++-- > 1 file changed, 3 insertions(+), 2 deletions(-) > > diff --git a/drivers/perf/arm-cmn.c b/drivers/perf/arm-cmn.c > index ef959e66db7c..d4fe30ff225b 100644 > --- a/drivers/perf/arm-cmn.c > +++ b/drivers/perf/arm-cmn.c > @@ -802,8 +802,6 @@ static umode_t arm_cmn_event_attr_is_visible(struct kobject *kobj, > CMN_EVENT_ATTR(_model, ccha_##_name, CMN_TYPE_CCHA, _event) > #define CMN_EVENT_CCLA(_name, _event) \ > CMN_EVENT_ATTR(CMN_ANY, ccla_##_name, CMN_TYPE_CCLA, _event) > -#define CMN_EVENT_CCLA_RNI(_name, _event) \ > - CMN_EVENT_ATTR(CMN_ANY, ccla_rni_##_name, CMN_TYPE_CCLA_RNI, _event) > #define CMN_EVENT_HNS(_name, _event) \ > CMN_EVENT_ATTR(CMN_ANY, hns_##_name, CMN_TYPE_HNS, _event) > > @@ -1798,6 +1796,9 @@ static int arm_cmn_event_init(struct perf_event *event) > } else if (type == CMN_TYPE_XP && > (cmn->part == PART_CMN700 || cmn->part == PART_CMN_S3)) { > hw->wide_sel = true; > + } else if (type == CMN_TYPE_RND) { > + /* Secretly permit this as an alias for "rnid" events */ > + type = CMN_TYPE_RNI; > } > > /* This is sufficiently annoying to recalculate, so cache it */ > -- > 2.39.2.101.g768bb238c484.dirty > > >
On Thu, 06 Feb 2025 14:24:04 +0000, Robin Murphy wrote: > While handling RN-D nodes under the functionally-identical RN-I type > works fine for perf tool users using the "rnid_" event aliases, and that > is the documented and expected ABI, there's little reason not to be > permissive and accept the actual RN-D type as an additional encoding for > the same events as well. This may be convenient for other tooling > generating event configs directly from its own topology data. > > [...] Applied to will (for-next/perf), thanks! [1/1] perf/arm-cmn: Minor event type housekeeping https://git.kernel.org/will/c/678a5d3d6db6 Cheers,
diff --git a/drivers/perf/arm-cmn.c b/drivers/perf/arm-cmn.c index ef959e66db7c..d4fe30ff225b 100644 --- a/drivers/perf/arm-cmn.c +++ b/drivers/perf/arm-cmn.c @@ -802,8 +802,6 @@ static umode_t arm_cmn_event_attr_is_visible(struct kobject *kobj, CMN_EVENT_ATTR(_model, ccha_##_name, CMN_TYPE_CCHA, _event) #define CMN_EVENT_CCLA(_name, _event) \ CMN_EVENT_ATTR(CMN_ANY, ccla_##_name, CMN_TYPE_CCLA, _event) -#define CMN_EVENT_CCLA_RNI(_name, _event) \ - CMN_EVENT_ATTR(CMN_ANY, ccla_rni_##_name, CMN_TYPE_CCLA_RNI, _event) #define CMN_EVENT_HNS(_name, _event) \ CMN_EVENT_ATTR(CMN_ANY, hns_##_name, CMN_TYPE_HNS, _event) @@ -1798,6 +1796,9 @@ static int arm_cmn_event_init(struct perf_event *event) } else if (type == CMN_TYPE_XP && (cmn->part == PART_CMN700 || cmn->part == PART_CMN_S3)) { hw->wide_sel = true; + } else if (type == CMN_TYPE_RND) { + /* Secretly permit this as an alias for "rnid" events */ + type = CMN_TYPE_RNI; } /* This is sufficiently annoying to recalculate, so cache it */
While handling RN-D nodes under the functionally-identical RN-I type works fine for perf tool users using the "rnid_" event aliases, and that is the documented and expected ABI, there's little reason not to be permissive and accept the actual RN-D type as an additional encoding for the same events as well. This may be convenient for other tooling generating event configs directly from its own topology data. In the RN-I event mood, it also seems as good a time as any to clean up a forgotten macro for CCLA_RNI events which ended up being unnecessary. Signed-off-by: Robin Murphy <robin.murphy@arm.com> --- drivers/perf/arm-cmn.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)