| Message ID | 649a2281233f193d59240b13ed91b57337c77b32.1611839564.git.robin.murphy@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [1/2] perf/arm-cmn: Fix PMU instance naming | expand |
On Thu, Jan 28, 2021 at 01:12:43PM +0000, Robin Murphy wrote: > Although it's neat to avoid the suffix for the typical case of a > single PMU, it means systems with multiple CMN instances end up with > inconsistent naming. I think it also breaks perf tool's "uncore alias" > logic if the common instance prefix is also the full name of one. > > Avoid any surprises by not trying to be clever and simply numbering > every instance, even when it might technically prove redundant. > > Fixes: 0ba64770a2f2 ("perf: Add Arm CMN-600 PMU driver") > Signed-off-by: Robin Murphy <robin.murphy@arm.com> > --- > Documentation/admin-guide/perf/arm-cmn.rst | 2 +- > drivers/perf/arm-cmn.c | 13 ++++--------- > 2 files changed, 5 insertions(+), 10 deletions(-) Hmm, so usually I'd push-back on changes like this because it's a user-visible change, however in this case we're talking about a pretty niche PMU and a relatively new driver so I think we should proceed. However, if anybody complains then we'll need to support them in one way or another. Will
On Thu, 28 Jan 2021 13:12:43 +0000, Robin Murphy wrote: > Although it's neat to avoid the suffix for the typical case of a > single PMU, it means systems with multiple CMN instances end up with > inconsistent naming. I think it also breaks perf tool's "uncore alias" > logic if the common instance prefix is also the full name of one. > > Avoid any surprises by not trying to be clever and simply numbering > every instance, even when it might technically prove redundant. Applied to will (for-next/perf), thanks! [1/2] perf/arm-cmn: Fix PMU instance naming https://git.kernel.org/will/c/79d7c3dca99f [2/2] perf/arm-cmn: Move IRQs when migrating context https://git.kernel.org/will/c/1c8147ea89c8 Cheers,
On 2021-01-28 20:14, Will Deacon wrote: > On Thu, Jan 28, 2021 at 01:12:43PM +0000, Robin Murphy wrote: >> Although it's neat to avoid the suffix for the typical case of a >> single PMU, it means systems with multiple CMN instances end up with >> inconsistent naming. I think it also breaks perf tool's "uncore alias" >> logic if the common instance prefix is also the full name of one. >> >> Avoid any surprises by not trying to be clever and simply numbering >> every instance, even when it might technically prove redundant. >> >> Fixes: 0ba64770a2f2 ("perf: Add Arm CMN-600 PMU driver") >> Signed-off-by: Robin Murphy <robin.murphy@arm.com> >> --- >> Documentation/admin-guide/perf/arm-cmn.rst | 2 +- >> drivers/perf/arm-cmn.c | 13 ++++--------- >> 2 files changed, 5 insertions(+), 10 deletions(-) > > Hmm, so usually I'd push-back on changes like this because it's a > user-visible change, however in this case we're talking about a pretty > niche PMU and a relatively new driver so I think we should proceed. However, > if anybody complains then we'll need to support them in one way or another. Yeah, I'm kicking myself for not remembering about the aliasing thing from the start - it came up tangentially from conversation with some interested users the other day - so hopefully the practicality will outweigh any possible early-adopter inconvenience. Thanks, Robin.
diff --git a/Documentation/admin-guide/perf/arm-cmn.rst b/Documentation/admin-guide/perf/arm-cmn.rst index 0e4809346014..796e25b7027b 100644 --- a/Documentation/admin-guide/perf/arm-cmn.rst +++ b/Documentation/admin-guide/perf/arm-cmn.rst @@ -17,7 +17,7 @@ PMU events ---------- The PMU driver registers a single PMU device for the whole interconnect, -see /sys/bus/event_source/devices/arm_cmn. Multi-chip systems may link +see /sys/bus/event_source/devices/arm_cmn_0. Multi-chip systems may link more than one CMN together via external CCIX links - in this situation, each mesh counts its own events entirely independently, and additional PMU devices will be named arm_cmn_{1..n}. diff --git a/drivers/perf/arm-cmn.c b/drivers/perf/arm-cmn.c index a76ff594f3ca..f3071b5ddaae 100644 --- a/drivers/perf/arm-cmn.c +++ b/drivers/perf/arm-cmn.c @@ -1502,7 +1502,7 @@ static int arm_cmn_probe(struct platform_device *pdev) struct arm_cmn *cmn; const char *name; static atomic_t id; - int err, rootnode, this_id; + int err, rootnode; cmn = devm_kzalloc(&pdev->dev, sizeof(*cmn), GFP_KERNEL); if (!cmn) @@ -1549,14 +1549,9 @@ static int arm_cmn_probe(struct platform_device *pdev) .cancel_txn = arm_cmn_end_txn, }; - this_id = atomic_fetch_inc(&id); - if (this_id == 0) { - name = "arm_cmn"; - } else { - name = devm_kasprintf(cmn->dev, GFP_KERNEL, "arm_cmn_%d", this_id); - if (!name) - return -ENOMEM; - } + name = devm_kasprintf(cmn->dev, GFP_KERNEL, "arm_cmn_%d", atomic_fetch_inc(&id)); + if (!name) + return -ENOMEM; err = cpuhp_state_add_instance(arm_cmn_hp_state, &cmn->cpuhp_node); if (err)
Although it's neat to avoid the suffix for the typical case of a single PMU, it means systems with multiple CMN instances end up with inconsistent naming. I think it also breaks perf tool's "uncore alias" logic if the common instance prefix is also the full name of one. Avoid any surprises by not trying to be clever and simply numbering every instance, even when it might technically prove redundant. Fixes: 0ba64770a2f2 ("perf: Add Arm CMN-600 PMU driver") Signed-off-by: Robin Murphy <robin.murphy@arm.com> --- Documentation/admin-guide/perf/arm-cmn.rst | 2 +- drivers/perf/arm-cmn.c | 13 ++++--------- 2 files changed, 5 insertions(+), 10 deletions(-)