[PATCH 0/2] x86/perf: A CR4.PCE bugfix and clarification

[PATCH 0/2] x86/perf: A CR4.PCE bugfix and clarification

[Andy Lutomirski]

Hi all-

Vince found a bug in our CR4.PCE handling.  While investigating, I
found what looked like another bug but actually wasn't.

Here's a fix for the bug (cc'd to stable) and a comment and lockdep
annotation for why the other non-bug isn't actually a bug.

Andy Lutomirski (2):
  x86/perf: Fix CR4.PCE propagation to use active_mm instead of mm
  x86/perf: Clarify why x86_pmu_event_mapped() isn't racy

 arch/x86/events/core.c | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

-- 
2.9.3

[PATCH 1/2] x86/perf: Fix CR4.PCE propagation to use active_mm instead of mm

[Andy Lutomirski]

If one thread mmaps a perf event while another thread in the same mm
is in some context where active_mm != mm (which can happen in the
scheduler, for example), refresh_pce() would write the wrong value
to CR4.PCE.  This broke some PAPI tests.

Cc: stable@vger.kernel.org
Fixes: 7911d3f7af14 ("perf/x86: Only allow rdpmc if a perf_event is mapped")
Reported-and-tested-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Andy Lutomirski <luto@kernel.org>
---
 arch/x86/events/core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index 349d4d17aa7f..4f564df73b8f 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -2101,8 +2101,8 @@ static int x86_pmu_event_init(struct perf_event *event)
 
 static void refresh_pce(void *ignored)
 {
-	if (current->mm)
-		load_mm_cr4(current->mm);
+	if (current->active_mm)
+		load_mm_cr4(current->active_mm);
 }
 
 static void x86_pmu_event_mapped(struct perf_event *event)
-- 
2.9.3

[PATCH 2/2] x86/perf: Clarify why x86_pmu_event_mapped() isn't racy

[Andy Lutomirski]

Naively, it looks racy, but mmap_sem saves it.  Add a comment and a
lockdep assertion.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
---
 arch/x86/events/core.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index 4f564df73b8f..2aa1ad194db2 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -2110,6 +2110,18 @@ static void x86_pmu_event_mapped(struct perf_event *event)
 	if (!(event->hw.flags & PERF_X86_EVENT_RDPMC_ALLOWED))
 		return;
 
+	/*
+	 * This function relies on not being called concurrently in two
+	 * tasks in the same mm.  Otherwise one task could observe
+	 * perf_rdpmc_allowed > 1 and return all the way back to
+	 * userspace with CR4.PCE clear while another task is still
+	 * doing on_each_cpu_mask() to propagate CR4.PCE.
+	 *
+	 * For now, this can't happen because all callers hold mmap_sem
+	 * for write.  If this changes, we'll need a different solution.
+	 */
+	lockdep_assert_held_exclusive(&current->mm->mmap_sem);
+
 	if (atomic_inc_return(&current->mm->context.perf_rdpmc_allowed) == 1)
 		on_each_cpu_mask(mm_cpumask(current->mm), refresh_pce, NULL, 1);
 }
-- 
2.9.3

[tip:perf/urgent] x86/perf: Fix CR4.PCE propagation to use active_mm instead of mm

[tip-bot for Andy Lutomirski]

Commit-ID:  5dc855d44c2ad960a86f593c60461f1ae1566b6d
Gitweb:     http://git.kernel.org/tip/5dc855d44c2ad960a86f593c60461f1ae1566b6d
Author:     Andy Lutomirski <luto@kernel.org>
AuthorDate: Thu, 16 Mar 2017 12:59:39 -0700
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 17 Mar 2017 08:28:26 +0100

x86/perf: Fix CR4.PCE propagation to use active_mm instead of mm

If one thread mmaps a perf event while another thread in the same mm
is in some context where active_mm != mm (which can happen in the
scheduler, for example), refresh_pce() would write the wrong value
to CR4.PCE.  This broke some PAPI tests.

Reported-and-tested-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bpetkov@suse.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Fixes: 7911d3f7af14 ("perf/x86: Only allow rdpmc if a perf_event is mapped")
Link: http://lkml.kernel.org/r/0c5b38a76ea50e405f9abe07a13dfaef87c173a1.1489694270.git.luto@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index 1635c0c..e07b36c 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -2100,8 +2100,8 @@ static int x86_pmu_event_init(struct perf_event *event)
 
 static void refresh_pce(void *ignored)
 {
-	if (current->mm)
-		load_mm_cr4(current->mm);
+	if (current->active_mm)
+		load_mm_cr4(current->active_mm);
 }
 
 static void x86_pmu_event_mapped(struct perf_event *event)

[tip:perf/urgent] x86/perf: Clarify why x86_pmu_event_mapped() isn't racy

[tip-bot for Andy Lutomirski]

Commit-ID:  4b07372a32c0c1505a7634ad7e607d83340ef645
Gitweb:     http://git.kernel.org/tip/4b07372a32c0c1505a7634ad7e607d83340ef645
Author:     Andy Lutomirski <luto@kernel.org>
AuthorDate: Thu, 16 Mar 2017 12:59:40 -0700
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Fri, 17 Mar 2017 08:28:26 +0100

x86/perf: Clarify why x86_pmu_event_mapped() isn't racy

Naively, it looks racy, but ->mmap_sem saves it.  Add a comment and a
lockdep assertion.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Borislav Petkov <bpetkov@suse.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Link: http://lkml.kernel.org/r/03a1e629063899168dfc4707f3bb6e581e21f5c6.1489694270.git.luto@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/events/core.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index e07b36c..183a972 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -2109,6 +2109,18 @@ static void x86_pmu_event_mapped(struct perf_event *event)
 	if (!(event->hw.flags & PERF_X86_EVENT_RDPMC_ALLOWED))
 		return;
 
+	/*
+	 * This function relies on not being called concurrently in two
+	 * tasks in the same mm.  Otherwise one task could observe
+	 * perf_rdpmc_allowed > 1 and return all the way back to
+	 * userspace with CR4.PCE clear while another task is still
+	 * doing on_each_cpu_mask() to propagate CR4.PCE.
+	 *
+	 * For now, this can't happen because all callers hold mmap_sem
+	 * for write.  If this changes, we'll need a different solution.
+	 */
+	lockdep_assert_held_exclusive(&current->mm->mmap_sem);
+
 	if (atomic_inc_return(&current->mm->context.perf_rdpmc_allowed) == 1)
 		on_each_cpu_mask(mm_cpumask(current->mm), refresh_pce, NULL, 1);
 }