[0/3] perf: Add Arm Network-on-Chip PMU driver

Message

Robin Murphy April 25, 2024, 12:29 p.m. UTC

Hi all,

Somewhat later than hoped (sorry!), now that it's finally had the chance
for enough testing to flush out all the silly bugs, here's the NI-700
driver I've had kicking around in various forms for the last 18 months.
I squashed in NI-710AE support since that appeared in the meantime and
the differences are minimal, however I have not felt compelled to go
back and genericise all the original NI-700 references. I don't think
I'm giving too much away to say that there will be further new product
support to add in future, so I will be back with more soon enough.

Note that the driver patch depends on the the cpumask_any_and_but()
addition queued in arm64/for-next/perf.

Thanks,
Robin.


Robin Murphy (3):
  dt-bindings/perf: Add Arm NI-700 PMU
  perf: Add driver for Arm NI-700 interconnect PMU
  MAINTAINERS: List Arm interconnect PMUs as supported

 .../devicetree/bindings/perf/arm,ni.yaml      |  30 +
 MAINTAINERS                                   |  10 +
 drivers/perf/Kconfig                          |   7 +
 drivers/perf/Makefile                         |   1 +
 drivers/perf/arm-ni.c                         | 767 ++++++++++++++++++
 5 files changed, 815 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/perf/arm,ni.yaml
 create mode 100644 drivers/perf/arm-ni.c

Comments

Will Deacon July 1, 2024, 1:36 p.m. UTC | #1

Hi Robin,

On Thu, Apr 25, 2024 at 01:29:51PM +0100, Robin Murphy wrote:
> 
> Somewhat later than hoped (sorry!), now that it's finally had the chance
> for enough testing to flush out all the silly bugs, here's the NI-700
> driver I've had kicking around in various forms for the last 18 months.
> I squashed in NI-710AE support since that appeared in the meantime and
> the differences are minimal, however I have not felt compelled to go
> back and genericise all the original NI-700 references. I don't think
> I'm giving too much away to say that there will be further new product
> support to add in future, so I will be back with more soon enough.
> 
> Note that the driver patch depends on the the cpumask_any_and_but()
> addition queued in arm64/for-next/perf.

I left some minor comments on the driver code, but I think this probably
needs a respin anyway as the binding is still pending an Ack from the
DT folks.

Will