Re: [PATCH v21 4/4] perf: arm_pmuv3: Add support for the Branch Record Buffer Extension (BRBE)

[Will Deacon <will@kernel.org>]

On Wed, May 21, 2025 at 04:58:42PM +0100, James Clark wrote:
> 
> 
> On 19/05/2025 10:56 pm, Rob Herring wrote:
> > On Mon, May 19, 2025 at 04:06:22PM +0100, Will Deacon wrote:
> > > Hey Rob,
> > > 
> > > On Mon, Apr 07, 2025 at 12:41:33PM -0500, Rob Herring (Arm) wrote:
> > > > From: Anshuman Khandual <anshuman.khandual@arm.com>
> > > > 
> > > > The ARMv9.2 architecture introduces the optional Branch Record Buffer
> > > > Extension (BRBE), which records information about branches as they are
> > > > executed into set of branch record registers. BRBE is similar to x86's
> > > > Last Branch Record (LBR) and PowerPC's Branch History Rolling Buffer
> > > > (BHRB).
> > > 
> > > Since you picked this up from v19, the driver has changed considerably
> > > and I presume you will be continuing to extend it in future as the
> > > architecture progresses. Perhaps having you listed as Author (and
> > > crucially, in git blame :p) with Anshuman as a Co-developed-by: would be
> > > more appropriate?
> > 
> > Shrug.
> > 
> > > > ---
> > > >   drivers/perf/Kconfig         |  11 +
> > > >   drivers/perf/Makefile        |   1 +
> > > >   drivers/perf/arm_brbe.c      | 802 +++++++++++++++++++++++++++++++++++++++++++
> > > >   drivers/perf/arm_brbe.h      |  47 +++
> > > >   drivers/perf/arm_pmu.c       |  15 +-
> > > >   drivers/perf/arm_pmuv3.c     | 129 ++++++-
> > > >   include/linux/perf/arm_pmu.h |   8 +
> > > >   7 files changed, 1006 insertions(+), 7 deletions(-)
> > > 
> > > Do you know if James Clark's tests [1] are going to be respun for the
> > > perf tool? It would be handy to have some way to test this new
> > > functionality.
> > 
> > Yes. I dropped them here because I've been told by Arnaldo in the past
> > to send userspace stuff separately.
> > 
> 
> That version of the test was out of date so I've pushed the new version
> here: https://git.linaro.org/plugins/gitiles/people/james.clark/linux.git/+/16e4a18c2d5fc53736f05c9052b1d11d74909707
> 
> But I'll wait for the driver changes to be finalised before posting it. Or
> Rob can take it back into the patchset.

Thanks, James. I just wanted to make sure that there was still a plan
for upstreaming these separately from the driver. Have you had a chance
to run them against this v21?

Will