From: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
To: Peter Zijlstra <a.p.zijlstra@chello.nl>,
Paul Mackerras <paulus@samba.org>, Ingo Molnar <mingo@redhat.com>,
Arnaldo Carvalho de Melo <acme@ghostprotocols.net>,
Anton Blanchard <anton@au1.ibm.com>,
robert.richter@amd.com, eranian@google.com
Cc: linuxppc-dev@ozlabs.org, linux-kernel@vger.kernel.org
Subject: perf: POWER-event translation questions
Date: Wed, 7 Nov 2012 17:10:35 -0800 [thread overview]
Message-ID: <20121108011035.GA20670@us.ibm.com> (raw)
Looking for feedback on this prototype for making POWER-specific event
translations available in sysfs. It is based on the patchset:
https://lkml.org/lkml/2012/11/7/402
which makes the translations for _generic_ events in POWER available in sysfs:
Since this is in POWER7 specific code I am assigning the names given in the
POWER7 CPU spec for now.
I had earlier tried mapping these events to generic names outside sysfs:
Power7 name Generic name
cmpl-stall-fxu stalled-cycles-fixed-point
cmpl-stall-lsu stalled-cycles-load-store
cmpl-stall-ifu stalled-cycles-instruction-fetch
cmpl-stall-bru stalled-cycles-branch-unit
But like Stephane Eranian pointed out mapping such events across architectures
can be confusing.
Another challenge I suspect we will have is the extremely long generic names
we could end up with as the events get more specific.
1. Can we have more than one name for an event ? i.e two sysfs entries,
eg: 'cmpl-stall-fxu' and 'stalled-cycles-fixed-point' for an event ?
2. Can we allow hyphens in the {name} token (please see my change to
util/parse-events.l below). With this change, I can run:
perf stat -e cpu/cmplu-stall-bru /tmp/nop
without any changes to the user level tool (parse-events.l) I have
tested some common cases, not sure if it will break something :-)
If we are going to create generic or arch specific sysfs entries in
/sys/bus/event_source/devices/cpu/events, do we need to add corresponding
entry in tools/perf/util/parse-events.l ?
Sukadev
---
arch/powerpc/perf/power7-pmu.c | 13 +++++++++++++
tools/perf/util/parse-events.l | 2 +-
2 files changed, 14 insertions(+), 1 deletions(-)
diff --git a/arch/powerpc/perf/power7-pmu.c b/arch/powerpc/perf/power7-pmu.c
index aa9f588..9f46abc 100644
--- a/arch/powerpc/perf/power7-pmu.c
+++ b/arch/powerpc/perf/power7-pmu.c
@@ -303,6 +303,10 @@ static void power7_disable_pmc(unsigned int pmc, unsigned long mmcr[])
#define PM_LD_MISS_L1 0x400f0
#define PM_BRU_FIN 0x10068
#define PM_BRU_MPRED 0x400f6
+#define PM_CMPLU_STALL_FXU 0x20014
+#define PM_CMPLU_STALL_LSU 0x20012
+#define PM_CMPLU_STALL_IFU 0x4004c
+#define PM_CMPLU_STALL_BRU 0x4004e
static int power7_generic_events[] = {
[PERF_COUNT_HW_CPU_CYCLES] = PM_CYC,
@@ -369,6 +373,11 @@ EVENT_ATTR(cache-misses, LD_MISS_L1);
EVENT_ATTR(branch-instructions, BRU_FIN);
EVENT_ATTR(branch-misses, BRU_MPRED);
+EVENT_ATTR(cmplu-stall-fxu, CMPLU_STALL_FXU);
+EVENT_ATTR(cmplu-stall-lsu, CMPLU_STALL_LSU);
+EVENT_ATTR(cmplu-stall-ifu, CMPLU_STALL_IFU);
+EVENT_ATTR(cmplu-stall-bru, CMPLU_STALL_BRU);
+
static struct attribute *power7_events_attr[] = {
EVENT_PTR(CYC),
EVENT_PTR(GCT_NOSLOT_CYC),
@@ -378,6 +387,10 @@ static struct attribute *power7_events_attr[] = {
EVENT_PTR(LD_MISS_L1),
EVENT_PTR(BRU_FIN),
EVENT_PTR(BRU_MPRED),
+ EVENT_PTR(CMPLU_STALL_FXU),
+ EVENT_PTR(CMPLU_STALL_LSU),
+ EVENT_PTR(CMPLU_STALL_IFU),
+ EVENT_PTR(CMPLU_STALL_BRU),
NULL,
};
diff --git a/tools/perf/util/parse-events.l b/tools/perf/util/parse-events.l
index c87efc1..1967bb2 100644
--- a/tools/perf/util/parse-events.l
+++ b/tools/perf/util/parse-events.l
@@ -80,7 +80,7 @@ event [^,{}/]+
num_dec [0-9]+
num_hex 0x[a-fA-F0-9]+
num_raw_hex [a-fA-F0-9]+
-name [a-zA-Z_*?][a-zA-Z0-9_*?]*
+name [-a-zA-Z_*?][-a-zA-Z0-9_*?]*
modifier_event [ukhpGH]{1,8}
modifier_bp [rwx]{1,3}
--
1.7.1
next reply other threads:[~2012-11-08 1:36 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-11-08 1:10 Sukadev Bhattiprolu [this message]
2012-11-09 10:26 ` perf: POWER-event translation questions Stephane Eranian
2012-11-20 16:50 ` Robert Richter
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20121108011035.GA20670@us.ibm.com \
--to=sukadev@linux.vnet.ibm.com \
--cc=a.p.zijlstra@chello.nl \
--cc=acme@ghostprotocols.net \
--cc=anton@au1.ibm.com \
--cc=eranian@google.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxppc-dev@ozlabs.org \
--cc=mingo@redhat.com \
--cc=paulus@samba.org \
--cc=robert.richter@amd.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).