Message-ID: <DM8PR11MB5751FE5DCB6AF3EB6AAB6EEAB8A42@DM8PR11MB5751.namprd11.prod.outlook.com>
Date: Wed, 10 Jul 2024 06:03:02 +0000
From: "Wang, Xiao W" <xiao.w.wang@...el.com>
To: Atish Patra <atishp@...shpatra.org>, Conor Dooley <conor@...nel.org>
CC: Charlie Jenkins <charlie@...osinc.com>, Alexandre Ghiti
	<alexghiti@...osinc.com>, "paul.walmsley@...ive.com"
	<paul.walmsley@...ive.com>, "palmer@...belt.com" <palmer@...belt.com>,
	"aou@...s.berkeley.edu" <aou@...s.berkeley.edu>, "anup@...infault.org"
	<anup@...infault.org>, "linux-riscv@...ts.infradead.org"
	<linux-riscv@...ts.infradead.org>, "linux-kernel@...r.kernel.org"
	<linux-kernel@...r.kernel.org>
Subject: RE: [PATCH] drivers/perf: riscv: Remove redundant macro check



> -----Original Message-----
> From: Atish Patra <atishp@...shpatra.org>
> Sent: Wednesday, July 10, 2024 5:44 AM
> To: Conor Dooley <conor@...nel.org>
> Cc: Charlie Jenkins <charlie@...osinc.com>; Wang, Xiao W
> <xiao.w.wang@...el.com>; Alexandre Ghiti <alexghiti@...osinc.com>;
> paul.walmsley@...ive.com; palmer@...belt.com; aou@...s.berkeley.edu;
> anup@...infault.org; linux-riscv@...ts.infradead.org; linux-
> kernel@...r.kernel.org
> Subject: Re: [PATCH] drivers/perf: riscv: Remove redundant macro check
> 
> On Tue, Jul 9, 2024 at 2:05 PM Conor Dooley <conor@...nel.org> wrote:
> >
> > On Tue, Jul 09, 2024 at 01:57:47PM -0700, Charlie Jenkins wrote:
> > > On Tue, Jul 09, 2024 at 09:44:17PM +0100, Conor Dooley wrote:
> > > > On Tue, Jul 09, 2024 at 01:29:42PM -0700, Charlie Jenkins wrote:
> > > > > On Mon, Jul 08, 2024 at 01:22:11PM +0100, Conor Dooley wrote:
> > > > > > On Mon, Jul 08, 2024 at 08:12:24PM +0800, Xiao Wang wrote:
> > > > > > > The macro CONFIG_RISCV_PMU must have been defined when
> riscv_pmu.c gets
> > > > > > > compiled, so this patch removes the redundant check.
> > > > > >
> > > > > > Did you investigate why this define was added? Why do you think
> that it
> > > > > > is redundant, rather than checking the incorrect config option?
> > > > >
> > > > > This file is only compiled with CONFIG_RISCV_PMU:
> > > >
> > > > I might be ill, but I can still read. I was not disagreeing with Xiao
> > > > that the condition is redundant as written - I want to know whether they
> > > > made sure that this check was intentionally using CONFIG_RISCV_PMU in
> the
> > > > first place, or if another option should have been here instead.
> > >
> > > Makes sense! Looking through the lists I see this RFC from Atish where
> > > he introduced a different config option for this
> > > "CONFIG_RISCV_PMU_COMMON"[1]. I wonder if something got
> confused in the
> > > development of these two patches.
> >
> 
> Nope. That's not related as the above series is a complete revamp of
> PMU infrastructure
> and far from being merged.
> 
> > Perhaps.. What I was worried about was the wrong option being here
> > (maybe that it should have been RISCV_PMU_SBI or similar) and depending
> > on how the kernel is configured, userspace would get the wrong info
> > here. But maybe it is innocuous your theory would suggest, and there's
> > nothing to worry about. But that's for someone with a functioning brain
> > to figure out ;)
> >
> This macro has been there since v3 of the original series.
> https://lore.kernel.org/lkml/20230802080328.1213905-8-
> alexghiti@...osinc.com/
> 
> Ideally, this should have been RISCV_PMU_SBI as ctr_get_width was not
> defined earlier for legacy.
> However, that was fixed here
> https://lore.kernel.org/lkml/20240227170002.188671-3-
> vadim.shakirov@...tacore.com/
> 
> So we can remove the macro check here.
> 
> Reviewed-by: Atish Patra <atishp@...osinc.com>
> 

Thanks for all the comments from Conor, Charlie and Atish.
From userspace perspective (e.g. tools/lib/perf/mmap.c), if cap_user_rdpmc is enabled,
then pmc_width should be set. It's independent of arch and config options.

BRs,
Xiao

> > Cheers,
> > Conor.
> >
> > >
> > > Link:
> > > https://lore.kernel.org/lkml/20240217005738.3744121-12-
> atishp@...osinc.com/
> > > [1]
> > >
> > > > > # drivers/perf/Makefile
> > > > > obj-$(CONFIG_RISCV_PMU) += riscv_pmu.o
> > > > >
> > > > > So having this check does seem redundant. I am copying Alex as it looks
> > > > > like he wrote this.
> > > > >
> > > > > > >
> > > > > > > Signed-off-by: Xiao Wang <xiao.w.wang@...el.com>
> > > > > > > ---
> > > > > > >  drivers/perf/riscv_pmu.c | 2 --
> > > > > > >  1 file changed, 2 deletions(-)
> > > > > > >
> > > > > > > diff --git a/drivers/perf/riscv_pmu.c b/drivers/perf/riscv_pmu.c
> > > > > > > index 0a02e85a8951..7644147d50b4 100644
> > > > > > > --- a/drivers/perf/riscv_pmu.c
> > > > > > > +++ b/drivers/perf/riscv_pmu.c
> > > > > > > @@ -39,7 +39,6 @@ void arch_perf_update_userpage(struct
> perf_event *event,
> > > > > > >       userpg->cap_user_time_short = 0;
> > > > > > >       userpg->cap_user_rdpmc = riscv_perf_user_access(event);
> > > > > > >
> > > > > > > -#ifdef CONFIG_RISCV_PMU
> > > > > > >       /*
> > > > > > >        * The counters are 64-bit but the priv spec doesn't mandate all
> the
> > > > > > >        * bits to be implemented: that's why, counter width can vary
> based on
> > > > > > > @@ -47,7 +46,6 @@ void arch_perf_update_userpage(struct
> perf_event *event,
> > > > > > >        */
> > > > > > >       if (userpg->cap_user_rdpmc)
> > > > > > >               userpg->pmc_width = to_riscv_pmu(event->pmu)-
> >ctr_get_width(event->hw.idx) + 1;
> > > > > > > -#endif
> > > > > > >
> > > > > > >       do {
> > > > > > >               rd = sched_clock_read_begin(&seq);
> > > > > > > --
> > > > > > > 2.25.1
> > > > > > >
> > > > > > >
> > > > > > > _______________________________________________
> > > > > > > linux-riscv mailing list
> > > > > > > linux-riscv@...ts.infradead.org
> > > > > > > http://lists.infradead.org/mailman/listinfo/linux-riscv
> > > > >
> > >
> > >
> 
> 
> 
> --
> Regards,
> Atish

Powered by blists - more mailing lists