Re: [PATCH 2/3] perf parse-regs: Add generic support for non-gprs

[Liang, Kan]

On 5/14/2019 2:19 PM, Arnaldo Carvalho de Melo wrote:
> Em Tue, May 14, 2019 at 07:39:12AM -0700, kan.liang@xxxxxxxxxxxxxxx escreveu:
> > From: Kan Liang <kan.liang@xxxxxxxxxxxxxxx>
> >
> > Some non general purpose registers, e.g. XMM, can be collected on some
> > platforms, e.g. X86 Icelake.
> >
> > Add a weak function has_non_gprs_support() to check if the
> > kernel/hardware support non-gprs collection.
> > Add a weak function non_gprs_mask() to return non-gprs mask.
> >
> > By default, the non-gprs collection is not support. Specific platforms
> > should implement their own non-gprs functions if they can collect
> > non-gprs.
> >
> > Signed-off-by: Kan Liang <kan.liang@xxxxxxxxxxxxxxx>
> > ---
> >   tools/perf/util/parse-regs-options.c | 10 +++++++++-
> >   tools/perf/util/perf_regs.c          | 10 ++++++++++
> >   tools/perf/util/perf_regs.h          |  2 ++
> >   3 files changed, 21 insertions(+), 1 deletion(-)
> >
> > diff --git a/tools/perf/util/parse-regs-options.c b/tools/perf/util/parse-regs-options.c
> > index b21617f..2f90f31 100644
> > --- a/tools/perf/util/parse-regs-options.c
> > +++ b/tools/perf/util/parse-regs-options.c
> > @@ -12,6 +12,7 @@ __parse_regs(const struct option *opt, const char *str, int unset, bool intr)
> >   	const struct sample_reg *r;
> >   	char *s, *os = NULL, *p;
> >   	int ret = -1;
> > +	bool has_non_gprs = has_non_gprs_support(intr);
>
> This is generic code, what does "non gprs" means for !Intel? Can we come
> up with a better, not architecture specific jargon? Or you mean "general
> purpose registers"?

I mean "general purpose registers".

> Perhaps we can ask for a register mask for use with intr? I.e.:
>
> For the -I/--intr-regs
>
>          uint64_t mask = arch__intr_reg_mask();
>
> And for --user-regs
>
>          uint64_t mask = arch__user_reg_mask();
>
> And then on that loop do:
>
>    	for (r = sample_reg_masks; r->name; r++) {
> 		if (r->mask & mask))
>    			fprintf(stderr, "%s ", r->name);
>    	}
>
> ?

Yes, it looks like a little bit simpler than my implementation.
I will send out V2.

Thanks,
Kan

>    
> >   	if (unset)
> >   		return 0;
> > @@ -37,6 +38,8 @@ __parse_regs(const struct option *opt, const char *str, int unset, bool intr)
> >   			if (!strcmp(s, "?")) {
> >   				fprintf(stderr, "available registers: ");
> >   				for (r = sample_reg_masks; r->name; r++) {
> > +					if (!has_non_gprs && (r->mask & ~PERF_REGS_MASK))
> > +						break;
> >   					fprintf(stderr, "%s ", r->name);
> >   				}
> >   				fputc('\n', stderr);
> > @@ -44,6 +47,8 @@ __parse_regs(const struct option *opt, const char *str, int unset, bool intr)
> >   				return -1;
> >   			}
> >   			for (r = sample_reg_masks; r->name; r++) {
> > +				if (!has_non_gprs && (r->mask & ~PERF_REGS_MASK))
> > +					continue;
> >   				if (!strcasecmp(s, r->name))
> >   					break;
> >   			}
> > @@ -64,8 +69,11 @@ __parse_regs(const struct option *opt, const char *str, int unset, bool intr)
> >   	ret = 0;
> >   
> >   	/* default to all possible regs */
> > -	if (*mode == 0)
> > +	if (*mode == 0) {
> >   		*mode = PERF_REGS_MASK;
> > +		if (has_non_gprs)
> > +			*mode |= non_gprs_mask(intr);
> > +	}
> >   error:
> >   	free(os);
> >   	return ret;
> > diff --git a/tools/perf/util/perf_regs.c b/tools/perf/util/perf_regs.c
> > index 2acfcc5..0d278b6 100644
> > --- a/tools/perf/util/perf_regs.c
> > +++ b/tools/perf/util/perf_regs.c
> > @@ -13,6 +13,16 @@ int __weak arch_sdt_arg_parse_op(char *old_op __maybe_unused,
> >   	return SDT_ARG_SKIP;
> >   }
> >   
> > +bool __weak has_non_gprs_support(bool intr __maybe_unused)
> > +{
> > +	return false;
> > +}
> > +
> > +u64 __weak non_gprs_mask(bool intr __maybe_unused)
> > +{
> > +	return 0;
> > +}
> > +
> >   #ifdef HAVE_PERF_REGS_SUPPORT
> >   int perf_reg_value(u64 *valp, struct regs_dump *regs, int id)
> >   {
> > diff --git a/tools/perf/util/perf_regs.h b/tools/perf/util/perf_regs.h
> > index 1a15a4b..983b4e6 100644
> > --- a/tools/perf/util/perf_regs.h
> > +++ b/tools/perf/util/perf_regs.h
> > @@ -23,6 +23,8 @@ enum {
> >   };
> >   
> >   int arch_sdt_arg_parse_op(char *old_op, char **new_op);
> > +bool has_non_gprs_support(bool intr);
> > +u64 non_gprs_mask(bool intr);
> >   
> >   #ifdef HAVE_PERF_REGS_SUPPORT
> >   #include <perf_regs.h>
> > --
> > 2.7.4