mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Sandipan Das <sandipan.das@amd.com>
To: <linux-perf-users@vger.kernel.org>, <linux-kernel@vger.kernel.org>
Cc: <peterz@infradead.org>, <mingo@redhat.com>, <acme@kernel.org>,
	<namhyung@kernel.org>, <mark.rutland@arm.com>,
	<alexander.shishkin@linux.intel.com>, <jolsa@kernel.org>,
	<irogers@google.com>, <adrian.hunter@intel.com>,
	<kan.liang@linux.intel.com>, <tglx@linutronix.de>, <bp@alien8.de>,
	<dave.hansen@linux.intel.com>, <x86@kernel.org>, <hpa@zytor.com>,
	<eranian@google.com>, <songliubraving@meta.com>,
	<ravi.bangoria@amd.com>, <ananth.narayan@amd.com>,
	<sandipan.das@amd.com>
Subject: [PATCH 4/4] perf/x86/amd/uncore: Prevent UMC counters from saturating
Date: Wed, 9 Apr 2025 13:27:09 +0530	[thread overview]
Message-ID: <c8e5808145985bb558d8daaefe42586b74d4085d.1744184837.git.sandipan.das@amd.com> (raw)
In-Reply-To: <cover.1744184837.git.sandipan.das@amd.com>

Unlike L3 and DF counters, UMC counters (PERF_CTRs) set the Overflow bit
(bit 48) and saturate on overflow. A subsequent pmu->read() of the event
reports an incorrect accumulated count as there is no difference between
the previous and the current values of the counter.

To avoid this, inspect the current counter value and proactively reset
the corresponding PERF_CTR register on every pmu->read(). Combined with
the periodic reads initiated by the hrtimer, the counters never get a
chance saturate but the resolution reduces to 47 bits.

Fixes: 25e56847821f ("perf/x86/amd/uncore: Add memory controller support")
Signed-off-by: Sandipan Das <sandipan.das@amd.com>
---
 arch/x86/events/amd/uncore.c | 32 +++++++++++++++++++++++++++++++-
 1 file changed, 31 insertions(+), 1 deletion(-)

diff --git a/arch/x86/events/amd/uncore.c b/arch/x86/events/amd/uncore.c
index 8135dd60668c..fe746d803a5d 100644
--- a/arch/x86/events/amd/uncore.c
+++ b/arch/x86/events/amd/uncore.c
@@ -965,6 +965,36 @@ static void amd_uncore_umc_start(struct perf_event *event, int flags)
 	perf_event_update_userpage(event);
 }
 
+static void amd_uncore_umc_read(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	u64 prev, new, shift;
+	s64 delta;
+
+	shift = COUNTER_SHIFT + 1;
+	prev = local64_read(&hwc->prev_count);
+
+	/*
+	 * UMC counters do not have RDPMC assignments. Read counts directly
+	 * from the corresponding PERF_CTR.
+	 */
+	rdmsrl(hwc->event_base, new);
+
+	/*
+	 * Unlike the other uncore counters, UMC counters saturate and set the
+	 * Overflow bit (bit 48) on overflow. Since they do not roll over,
+	 * proactively reset the corresponding PERF_CTR when bit 47 is set so
+	 * that the counter never gets a chance to saturate.
+	 */
+	if (new & BIT_ULL(63 - COUNTER_SHIFT))
+		wrmsrl(hwc->event_base, 0);
+
+	local64_set(&hwc->prev_count, new);
+	delta = (new << shift) - (prev << shift);
+	delta >>= shift;
+	local64_add(delta, &event->count);
+}
+
 static
 void amd_uncore_umc_ctx_scan(struct amd_uncore *uncore, unsigned int cpu)
 {
@@ -1043,7 +1073,7 @@ int amd_uncore_umc_ctx_init(struct amd_uncore *uncore, unsigned int cpu)
 				.del		= amd_uncore_del,
 				.start		= amd_uncore_umc_start,
 				.stop		= amd_uncore_stop,
-				.read		= amd_uncore_read,
+				.read		= amd_uncore_umc_read,
 				.capabilities	= PERF_PMU_CAP_NO_EXCLUDE | PERF_PMU_CAP_NO_INTERRUPT,
 				.module		= THIS_MODULE,
 			};
-- 
2.43.0


  parent reply	other threads:[~2025-04-09  7:58 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-04-09  7:57 [PATCH 0/4] perf/x86/amd/uncore: Overflow handling enhancements Sandipan Das
2025-04-09  7:57 ` [PATCH 1/4] perf/x86/amd/uncore: Remove unused member from amd_uncore_ctx Sandipan Das
2025-04-09  7:57 ` [PATCH 2/4] perf/x86/amd/uncore: Use hrtimer for handling overflows Sandipan Das
2025-04-10  8:20   ` Peter Zijlstra
2025-04-10  8:40     ` Sandipan Das
2025-04-09  7:57 ` [PATCH 3/4] perf/x86/amd/uncore: Add parameter to configure hrtimer Sandipan Das
2025-04-09  7:57 ` Sandipan Das [this message]
2025-04-16 19:15   ` [PATCH 4/4] perf/x86/amd/uncore: Prevent UMC counters from saturating Song Liu

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=c8e5808145985bb558d8daaefe42586b74d4085d.1744184837.git.sandipan.das@amd.com \
    --to=sandipan.das@amd.com \
    --cc=acme@kernel.org \
    --cc=adrian.hunter@intel.com \
    --cc=alexander.shishkin@linux.intel.com \
    --cc=ananth.narayan@amd.com \
    --cc=bp@alien8.de \
    --cc=dave.hansen@linux.intel.com \
    --cc=eranian@google.com \
    --cc=hpa@zytor.com \
    --cc=irogers@google.com \
    --cc=jolsa@kernel.org \
    --cc=kan.liang@linux.intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-perf-users@vger.kernel.org \
    --cc=mark.rutland@arm.com \
    --cc=mingo@redhat.com \
    --cc=namhyung@kernel.org \
    --cc=peterz@infradead.org \
    --cc=ravi.bangoria@amd.com \
    --cc=songliubraving@meta.com \
    --cc=tglx@linutronix.de \
    --cc=x86@kernel.org \
    /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

all inboxes | Powered by JetHome®