Support of optional PMC classes.

[ray]

Summary

Some PMU controllers are standalone and not bound to CPU. So classes of such controllers can be optional and in case when optional class in the middle of machine-dependent class list it must be skipped.

Sponsored By: Ampere Computing

[ray]

Update classes base ri to handle optional classes out of order.

[mhorne]

It seems to me that only the changes to pmc_arm64_initialize() should be necessary, because it handles optional classes in the same way that pmc_intel_initialize() does, by passing the correct nclasses value to pmc_mdep_alloc().

sys/dev/hwpmc/hwpmc_mod.c

5770–5771

How can such an optional class appear in this list? It seems like it shouldn't happen, because you take care to call pmc_mdep_alloc() with the correct nclasses value.

[ray]

In D32316#737876, @mhorne wrote:

It seems to me that only the changes to pmc_arm64_initialize() should be necessary, because it handles optional classes in the same way that pmc_intel_initialize() does, by passing the correct nclasses value to pmc_mdep_alloc().

Problem here is in static machdep class numbers. If classes will be initialized in incorrect order, adjusted ri will be incorrect. So that modification may save some time on debugging such issue for new optional classes with just little time in hwpmc(4) init.

[mhorne]

In D32316#737883, @ray wrote:

In D32316#737876, @mhorne wrote:

It seems to me that only the changes to pmc_arm64_initialize() should be necessary, because it handles optional classes in the same way that pmc_intel_initialize() does, by passing the correct nclasses value to pmc_mdep_alloc().

Problem here is in static machdep class numbers. If classes will be initialized in incorrect order, adjusted ri will be incorrect. So that modification may save some time on debugging such issue for new optional classes with just little time in hwpmc(4) init.

Ahh right, so it handles the case where you have a dmc620 but not a cmn600. This looks good to me then.

[emaste]

@ray I assume you can go ahead with committing this one?

[ray]

In D32316#741213, @emaste wrote:

@ray I assume you can go ahead with committing this one?

Fortunately I did not execute your assumption :)
Because here is not committed class identifiers.
But thanks for reminder!

sys/dev/hwpmc/hwpmc_mod.c

5770–5771

For example, one of classes fail to initialize. It have to be present in a list, otherwise class number and array position will not match. But to not process that class we can check number of counters it add (0 here).