From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933957AbaE3QVX (ORCPT ); Fri, 30 May 2014 12:21:23 -0400 Received: from mx1.redhat.com ([209.132.183.28]:52727 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933202AbaE3QVW (ORCPT ); Fri, 30 May 2014 12:21:22 -0400 Date: Fri, 30 May 2014 12:20:42 -0400 From: Don Zickus To: Arnaldo Carvalho de Melo Cc: jolsa@redhat.com, Peter Zijlstra , LKML , namhyung@gmail.com, eranian@google.com, Andi Kleen Subject: Re: [PATCH 6/7 V2] perf: Add support to dynamically get cacheline size Message-ID: <20140530162042.GI191965@redhat.com> References: <1401208087-181977-7-git-send-email-dzickus@redhat.com> <1401461425-209679-1-git-send-email-dzickus@redhat.com> <20140530152830.GB1202@kernel.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20140530152830.GB1202@kernel.org> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Fri, May 30, 2014 at 12:28:30PM -0300, Arnaldo Carvalho de Melo wrote: > Em Fri, May 30, 2014 at 10:50:25AM -0400, Don Zickus escreveu: > > Different arches may have different cacheline sizes. Look it up and set > > a global variable for reference. > > [acme@zoo linux]$ strings `which getconf` | grep LINESIZE > LEVEL1_ICACHE_LINESIZE > LEVEL1_DCACHE_LINESIZE > LEVEL2_CACHE_LINESIZE > LEVEL3_CACHE_LINESIZE > LEVEL4_CACHE_LINESIZE > [acme@zoo linux]$ > > man sysconf > > That is what I use in pahole, for instance: > > [acme@zoo pahole]$ grep sysconf *.c > dwarves_fprintf.c: long sys_cacheline_size = > sysconf(_SC_LEVEL1_DCACHE_LINESIZE); > [acme@zoo pahole]$ Well that does look easier. I'll hack that up to see if it gives me the same answer. Thanks! Cheers, Don > > - Arnaldo > > > Signed-off-by: Don Zickus > > > > --- > > V3: remove unneeded cpumap.h (Namhyung Kim) > > V2: change to be global and setup in perf.c > > use filename__read_int for setup > > --- > > tools/perf/perf.c | 5 +++++ > > tools/perf/util/cpumap.c | 27 +++++++++++++++++++++++++++ > > tools/perf/util/cpumap.h | 3 +++ > > tools/perf/util/util.c | 1 + > > tools/perf/util/util.h | 1 + > > 5 files changed, 37 insertions(+) > > > > diff --git a/tools/perf/perf.c b/tools/perf/perf.c > > index 431798a..dabf08b 100644 > > --- a/tools/perf/perf.c > > +++ b/tools/perf/perf.c > > @@ -13,6 +13,7 @@ > > #include "util/quote.h" > > #include "util/run-command.h" > > #include "util/parse-events.h" > > +#include "util/cpumap.h" > > #include > > #include > > > > @@ -459,6 +460,10 @@ int main(int argc, const char **argv) > > /* The page_size is placed in util object. */ > > page_size = sysconf(_SC_PAGE_SIZE); > > > > + /* The cacheline_size is placed in util objet */ > > + if (cpu__setup_cacheline_size() < 0) > > + goto out; > > + > > cmd = perf_extract_argv0_path(argv[0]); > > if (!cmd) > > cmd = "perf-help"; > > diff --git a/tools/perf/util/cpumap.c b/tools/perf/util/cpumap.c > > index c4e55b7..1029982 100644 > > --- a/tools/perf/util/cpumap.c > > +++ b/tools/perf/util/cpumap.c > > @@ -477,3 +477,30 @@ int cpu__setup_cpunode_map(void) > > closedir(dir1); > > return 0; > > } > > + > > +int cpu__setup_cacheline_size(void) > > +{ > > + const char *mnt; > > + char path[PATH_MAX]; > > + int n, size; > > + > > + > > + mnt = sysfs__mountpoint(); > > + if (!mnt) > > + return -1; > > + > > + n = snprintf(path, PATH_MAX, "%s/devices/system/cpu/cpu0/cache/index0/coherency_line_size", mnt); > > + if (n == PATH_MAX) { > > + pr_err("sysfs path crossed PATH_MAX(%d) size\n", PATH_MAX); > > + return -1; > > + } > > + > > + if (filename__read_int(path, &size)) { > > + pr_err("Can not read cacheline size\n"); > > + return -1; > > + } > > + > > + cacheline_size = size; > > + > > + return 0; > > +} > > diff --git a/tools/perf/util/cpumap.h b/tools/perf/util/cpumap.h > > index 61a6548..507d7fd 100644 > > --- a/tools/perf/util/cpumap.h > > +++ b/tools/perf/util/cpumap.h > > @@ -5,6 +5,7 @@ > > #include > > > > #include "perf.h" > > +#include "util/util.h" > > #include "util/debug.h" > > > > struct cpu_map { > > @@ -81,4 +82,6 @@ static inline int cpu__get_node(int cpu) > > return cpunode_map[cpu]; > > } > > > > +int cpu__setup_cacheline_size(void); > > + > > #endif /* __PERF_CPUMAP_H */ > > diff --git a/tools/perf/util/util.c b/tools/perf/util/util.c > > index 7fff6be..95aefa7 100644 > > --- a/tools/perf/util/util.c > > +++ b/tools/perf/util/util.c > > @@ -17,6 +17,7 @@ > > * XXX We need to find a better place for these things... > > */ > > unsigned int page_size; > > +int cacheline_size; > > > > bool test_attr__enabled; > > > > diff --git a/tools/perf/util/util.h b/tools/perf/util/util.h > > index b03da44..6686436 100644 > > --- a/tools/perf/util/util.h > > +++ b/tools/perf/util/util.h > > @@ -304,6 +304,7 @@ char *rtrim(char *s); > > void dump_stack(void); > > > > extern unsigned int page_size; > > +extern int cacheline_size; > > > > void get_term_dimensions(struct winsize *ws); > > > > -- > > 1.7.11.7