[v2,0/5] CXL 3.0 Performance Monitoring Unit support

Message

Jonathan Cameron March 24, 2023, 5:13 p.m. UTC

Significant changes since v1:
- New patch 2 - allow a PMU device to have a parent. This helps identify
  the association of the CPMU with the CXL devices without needing to
  rely on messy naming of the PMU.
- Set the parent for each CPMU instance.
- Add missing cpmu.h from patch 3.
- Fix a bug due to bad ordering of ida allocation wrt to device_initialize()
- Split the event list in two as handling for fixed and configurable
  counters is different.  Add util functions to search each list.
- Lots of other improvements in patch 4 called out in that patch description.

Thanks to all those who looked at it, and Kan Liang in particular for
all the feedback on the main driver.

Updated cover letter.

The CXL rev 3.0 specification introduces a CXL Performance Monitoring
Unit definition. CXL components may have any number of these blocks. The
definition is highly flexible, but that does bring complexity in the
driver.

Notes.

1) The QEMU model against which this was developed needs tidying up.
   Latest tree at https://gitlab.com/jic23/qemu branch cxl-2023-02-28
   It's backed up behind other series that I plan to upstream first.
2) There are quite a lot of corner cases that will need working through
   with variants of the model, or I'll have to design a pathological
   set of CPMUs to hit all the corner cases in one go. So whilst I believe
   the driver should be fine for what it supports we may find issues
   as those corners of what is allowed are explored.
3) I'm not sure it makes sense to hang this of the cxl/pci driver but
   couldn't really figure out where else in the current structure we could
   make it fit cleanly.
4) Driver location. In past perf maintainers have requested perf drivers
   for PCI etc be under drivers/perf. That would require moving some
   CXL headers to be more generally visible, but is certainly possible
   if there is agreement between CXL and perf maintainers on the correct
   location.
5) Patch 1 led to a discussion of how to handle the self describing
   and extensible nature of the CPMU counters.  That is likely to involve
   a description in the "caps" sysfs directory and perf tool code that is
   aware of the different event combinations that make sense and can
   establish which sets are available on a given device.
   That is likely to take a while to converge on, so as the driver is useful
   in the current state, I'm looking to upstream this first and deal with
   the more complex handling later.
6) There is a lot of other functionality that can be added in future
   include allowing for simpler hardware implementations that may not
   support the minimum level of features currently required by the driver
   (freeze, overflow interrupts etc).

CXL rev 3.0 specification available from https://www.computeexpresslink.org

Jonathan Cameron (5):
  cxl: Add function to count regblocks of a given type.
  perf: Allow a PMU to have a parent.
  cxl/pci: Find and register CXL PMU devices
  cxl: CXL Performance Monitoring Unit driver
  docs: perf: Minimal introduction the the CXL PMU device and driver.

 Documentation/admin-guide/perf/cxl.rst   |  65 ++
 Documentation/admin-guide/perf/index.rst |   1 +
 drivers/cxl/Kconfig                      |  12 +
 drivers/cxl/Makefile                     |   1 +
 drivers/cxl/core/Makefile                |   1 +
 drivers/cxl/core/core.h                  |   3 +
 drivers/cxl/core/cpmu.c                  |  72 ++
 drivers/cxl/core/pci.c                   |   2 +-
 drivers/cxl/core/port.c                  |   4 +-
 drivers/cxl/core/regs.c                  |  50 +-
 drivers/cxl/cpmu.c                       | 937 +++++++++++++++++++++++
 drivers/cxl/cpmu.h                       |  56 ++
 drivers/cxl/cxl.h                        |  17 +-
 drivers/cxl/cxlpci.h                     |   1 +
 drivers/cxl/pci.c                        |  27 +-
 include/linux/perf_event.h               |   1 +
 kernel/events/core.c                     |   1 +
 17 files changed, 1243 insertions(+), 8 deletions(-)
 create mode 100644 Documentation/admin-guide/perf/cxl.rst
 create mode 100644 drivers/cxl/core/cpmu.c
 create mode 100644 drivers/cxl/cpmu.c
 create mode 100644 drivers/cxl/cpmu.h