public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: kan.liang@intel.com
To: a.p.zijlstra@chello.nl, eranian@google.com
Cc: linux-kernel@vger.kernel.org, mingo@redhat.com, paulus@samba.org,
	acme@kernel.org, ak@linux.intel.com, kan.liang@intel.com, "Yan,
	Zheng" <zheng.z.yan@intel.com>
Subject: [PATCH V5 09/16] perf, x86: Save/resotre LBR stack during context switch
Date: Wed, 10 Sep 2014 10:09:06 -0400	[thread overview]
Message-ID: <1410358153-421-10-git-send-email-kan.liang@intel.com> (raw)
In-Reply-To: <1410358153-421-1-git-send-email-kan.liang@intel.com>

From: Kan Liang <kan.liang@intel.com>

When the LBR call stack is enabled, it is necessary to save/restore
the LBR stack on context switch. The solution is saving/restoring
the LBR stack to/from task's perf event context.

The LBR stack is saved/restored only when there are events that use
the LBR call stack. If no event uses LBR call stack, the LBR stack
is reset when task is scheduled in.

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
---
 arch/x86/kernel/cpu/perf_event_intel_lbr.c | 88 ++++++++++++++++++++++++++----
 1 file changed, 76 insertions(+), 12 deletions(-)

diff --git a/arch/x86/kernel/cpu/perf_event_intel_lbr.c b/arch/x86/kernel/cpu/perf_event_intel_lbr.c
index 8c6da0f..6aabbb4 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_lbr.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_lbr.c
@@ -180,13 +180,89 @@ void intel_pmu_lbr_reset(void)
 		intel_pmu_lbr_reset_64();
 }
 
+/*
+ * TOS = most recently recorded branch
+ */
+static inline u64 intel_pmu_lbr_tos(void)
+{
+	u64 tos;
+
+	rdmsrl(x86_pmu.lbr_tos, tos);
+	return tos;
+}
+
+enum {
+	LBR_NONE,
+	LBR_VALID,
+};
+
+static void __intel_pmu_lbr_restore(struct x86_perf_task_context *task_ctx)
+{
+	int i;
+	unsigned lbr_idx, mask;
+	u64 tos;
+
+	if (task_ctx->lbr_callstack_users == 0 ||
+	    task_ctx->lbr_stack_state == LBR_NONE) {
+		intel_pmu_lbr_reset();
+		return;
+	}
+
+	mask = x86_pmu.lbr_nr - 1;
+	tos = intel_pmu_lbr_tos();
+	for (i = 0; i < x86_pmu.lbr_nr; i++) {
+		lbr_idx = (tos - i) & mask;
+		wrmsrl(x86_pmu.lbr_from + lbr_idx, task_ctx->lbr_from[i]);
+		wrmsrl(x86_pmu.lbr_to + lbr_idx, task_ctx->lbr_to[i]);
+	}
+	task_ctx->lbr_stack_state = LBR_NONE;
+}
+
+static void __intel_pmu_lbr_save(struct x86_perf_task_context *task_ctx)
+{
+	int i;
+	unsigned lbr_idx, mask;
+	u64 tos;
+
+	if (task_ctx->lbr_callstack_users == 0) {
+		task_ctx->lbr_stack_state = LBR_NONE;
+		return;
+	}
+
+	mask = x86_pmu.lbr_nr - 1;
+	tos = intel_pmu_lbr_tos();
+	for (i = 0; i < x86_pmu.lbr_nr; i++) {
+		lbr_idx = (tos - i) & mask;
+		rdmsrl(x86_pmu.lbr_from + lbr_idx, task_ctx->lbr_from[i]);
+		rdmsrl(x86_pmu.lbr_to + lbr_idx, task_ctx->lbr_to[i]);
+	}
+	task_ctx->lbr_stack_state = LBR_VALID;
+}
+
+
 void intel_pmu_lbr_sched_task(struct perf_event_context *ctx, bool sched_in)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct x86_perf_task_context *task_ctx;
 
 	if (!x86_pmu.lbr_nr)
 		return;
 	/*
+	 * If LBR callstack feature is enabled and the stack was saved when
+	 * the task was scheduled out, restore the stack. Otherwise flush
+	 * the LBR stack.
+	 */
+	task_ctx = ctx ? ctx->task_ctx_data : NULL;
+	if (task_ctx) {
+		if (sched_in) {
+			__intel_pmu_lbr_restore(task_ctx);
+			cpuc->lbr_context = ctx;
+		} else {
+			__intel_pmu_lbr_save(task_ctx);
+		}
+	}
+
+	/*
 	 * When sampling the branck stack in system-wide, it may be
 	 * necessary to flush the stack on context switch. This happens
 	 * when the branch stack does not tag its entries with the pid
@@ -276,18 +352,6 @@ void intel_pmu_lbr_disable_all(void)
 		__intel_pmu_lbr_disable();
 }
 
-/*
- * TOS = most recently recorded branch
- */
-static inline u64 intel_pmu_lbr_tos(void)
-{
-	u64 tos;
-
-	rdmsrl(x86_pmu.lbr_tos, tos);
-
-	return tos;
-}
-
 static void intel_pmu_lbr_read_32(struct cpu_hw_events *cpuc)
 {
 	unsigned long mask = x86_pmu.lbr_nr - 1;
-- 
1.8.3.2


  parent reply	other threads:[~2014-09-10 14:13 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-09-10 14:08 [PATCH V5 00/16] perf, x86: Haswell LBR call stack support kan.liang
2014-09-10 14:08 ` [PATCH V5 01/16] perf, x86: Reduce lbr_sel_map size kan.liang
2014-09-24 10:50   ` Peter Zijlstra
2014-09-10 14:08 ` [PATCH V5 02/16] perf, core: introduce pmu context switch callback kan.liang
2014-09-24 11:23   ` Peter Zijlstra
2014-09-24 13:13   ` Peter Zijlstra
2014-09-10 14:09 ` [PATCH V5 03/16] perf, x86: use context switch callback to flush LBR stack kan.liang
2014-09-10 14:09 ` [PATCH V5 04/16] perf, x86: Basic Haswell LBR call stack support kan.liang
2014-09-10 14:09 ` [PATCH V5 05/16] perf, core: pmu specific data for perf task context kan.liang
2014-09-10 14:09 ` [PATCH V5 06/16] perf, core: always switch pmu specific data during context switch kan.liang
2014-09-10 14:09 ` [PATCH V5 07/16] perf, x86: allocate space for storing LBR stack kan.liang
2014-09-10 14:09 ` [PATCH V5 08/16] perf, x86: track number of events that use LBR callstack kan.liang
2014-09-24 12:53   ` Peter Zijlstra
2014-10-07  2:59     ` Liang, Kan
2014-10-07 15:19       ` Peter Zijlstra
2014-09-10 14:09 ` kan.liang [this message]
2014-09-24 13:33   ` [PATCH V5 09/16] perf, x86: Save/resotre LBR stack during context switch Peter Zijlstra
2014-09-10 14:09 ` [PATCH V5 10/16] perf, core: simplify need branch stack check kan.liang
2014-09-24 13:55   ` Peter Zijlstra
2014-09-10 14:09 ` [PATCH V5 11/16] perf, core: Pass perf_sample_data to perf_callchain() kan.liang
2014-09-24 14:15   ` Peter Zijlstra
2014-10-07  3:00     ` Liang, Kan
2014-10-07 15:24       ` Peter Zijlstra
2014-10-07 15:50         ` Liang, Kan
2014-10-07 16:29           ` Peter Zijlstra
2014-09-10 14:09 ` [PATCH V5 12/16] perf, x86: use LBR call stack to get user callchain kan.liang
2014-09-10 14:09 ` [PATCH V5 13/16] perf, x86: re-organize code that implicitly enables LBR/PEBS kan.liang
2014-09-10 14:09 ` [PATCH V5 14/16] perf, x86: enable LBR callstack when recording callchain kan.liang
2014-09-24 14:21   ` Peter Zijlstra
2014-10-07  3:00     ` Liang, Kan
2014-10-07 15:25       ` Peter Zijlstra
2014-10-07 16:04         ` Liang, Kan
2014-09-10 14:09 ` [PATCH V5 15/16] perf, x86: disable FREEZE_LBRS_ON_PMI when LBR operates in callstack mode kan.liang
2014-09-10 14:09 ` [PATCH V5 16/16] perf, x86: Discard zero length call entries in LBR call stack kan.liang
  -- strict thread matches above, loose matches on Subject: below --
2014-07-07  6:28 [PATCH v5 00/16] perf, x86: Haswell LBR call stack support Yan, Zheng
2014-07-07  6:28 ` [PATCH v5 09/16] perf, x86: Save/resotre LBR stack during context switch Yan, Zheng
2001-01-08  2:29 [PATCH V5 00/16] perf, x86: Haswell LBR call stack support kan.liang
2001-01-08  2:29 ` [PATCH V5 09/16] perf, x86: Save/resotre LBR stack during context switch kan.liang

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=1410358153-421-10-git-send-email-kan.liang@intel.com \
    --to=kan.liang@intel.com \
    --cc=a.p.zijlstra@chello.nl \
    --cc=acme@kernel.org \
    --cc=ak@linux.intel.com \
    --cc=eranian@google.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=paulus@samba.org \
    --cc=zheng.z.yan@intel.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