Skip to contentSkip to conversation

perf: Fix AUX record suppression

PATCH In review Alexander Shishkin 4 participants 5 messages Mar 29, 2019 – Apr 16, 2019

Subsystems (2)
lore
Share & tools
lore

Alexander Shishkin

author reviewer
#0 lore raw

PATCH 1/1 perf: Fix AUX record suppression

Commit 1627314fb54a33e ("perf: Suppress AUX/OVERWRITE records") has an unintended side-effect of also suppressing all AUX records with no flags and non-zero size, so all the regular records in the full trace mode. This breaks some use cases for people.

Fix this by restoring "regular" AUX records.

Fixes 1627314fb54a33e ("perf: Suppress AUX/OVERWRITE records")
CC [email protected] # v4.20+
---
 kernel/events/ring_buffer.c | 33 +++++++++++++++------------------
 1 file changed, 15 insertions(+), 18 deletions(-)
kernel/events/ring_buffer.c +15 -18
index 678ccec60d8f..626256dc26c1 100644
@@ -455,24 +455,21 @@ void perf_aux_output_end(struct perf_output_handle *handle, unsigned long size)
 		rb->aux_head += size;
 	}
 
-	if (size || handle->aux_flags) {
-		/*
-		 * Only send RECORD_AUX if we have something useful to communicate
-		 *
-		 * Note: the OVERWRITE records by themselves are not considered
-		 * useful, as they don't communicate any *new* information,
-		 * aside from the short-lived offset, that becomes history at
-		 * the next event sched-in and therefore isn't useful.
-		 * The userspace that needs to copy out AUX data in overwrite
-		 * mode should know to use user_page::aux_head for the actual
-		 * offset. So, from now on we don't output AUX records that
-		 * have *only* OVERWRITE flag set.
-		 */
-
-		if (handle->aux_flags & ~(u64)PERF_AUX_FLAG_OVERWRITE)
-			perf_event_aux_event(handle->event, aux_head, size,
-			                     handle->aux_flags);
-	}
+	/*
+	 * Only send RECORD_AUX if we have something useful to communicate
+	 *
+	 * Note: the OVERWRITE records by themselves are not considered
+	 * useful, as they don't communicate any *new* information,
+	 * aside from the short-lived offset, that becomes history at
+	 * the next event sched-in and therefore isn't useful.
+	 * The userspace that needs to copy out AUX data in overwrite
+	 * mode should know to use user_page::aux_head for the actual
+	 * offset. So, from now on we don't output AUX records that
+	 * have *only* OVERWRITE flag set.
+	 */
+	if (size || (handle->aux_flags & ~(u64)PERF_AUX_FLAG_OVERWRITE))
+		perf_event_aux_event(handle->event, aux_head, size,
+				     handle->aux_flags);
 
 	rb->user_page->aux_head = rb->aux_head;
 	if (rb_need_aux_wakeup(rb))
Signature · 1 line
2.20.1
#1 lore raw

Hi Alex

I was just wondering if this was expected to make 5.1? (I could not see it in RC5)?

Thanks
Ben

Quoted 70 lines · Alexander Shishkin

Commit 1627314fb54a33e ("perf: Suppress AUX/OVERWRITE records") has an
unintended side-effect of also suppressing all AUX records with no flags
and non-zero size, so all the regular records in the full trace mode. This breaks some use cases for people.

Fix this by restoring "regular" AUX records.

Quoted 1 line
Fixes 1627314fb54a33e ("perf: Suppress AUX/OVERWRITE records")
CC [email protected] # v4.20+
---
 kernel/events/ring_buffer.c | 33 +++++++++++++++------------------
 1 file changed, 15 insertions(+), 18 deletions(-)
kernel/events/ring_buffer.c +0 -3
index 678ccec60d8f..626256dc26c1 100644
@@ -455,24 +455,21 @@ void perf_aux_output_end(struct
 perf_output_handle *handle, unsigned long size)
 rb->aux_head += size;
 }
 
-if (size || handle->aux_flags) {
-/*
- * Only send RECORD_AUX if we have something useful to

communicate
- *
- * Note: the OVERWRITE records by themselves are not considered
- * useful, as they don't communicate any *new* information,
- * aside from the short-lived offset, that becomes history at
- * the next event sched-in and therefore isn't useful.
- * The userspace that needs to copy out AUX data in overwrite
- * mode should know to use user_page::aux_head for the actual
- * offset. So, from now on we don't output AUX records that
- * have *only* OVERWRITE flag set.
- */

kernel/events/ring_buffer.c +0 -3
-
-if (handle->aux_flags & ~(u64)PERF_AUX_FLAG_OVERWRITE)
-perf_event_aux_event(handle->event, aux_head,

size,

kernel/events/ring_buffer.c +2 -2
-                     handle->aux_flags);
-}
+/*
+ * Only send RECORD_AUX if we have something useful to

communicate
+ *
+ * Note: the OVERWRITE records by themselves are not considered
+ * useful, as they don't communicate any *new* information,
+ * aside from the short-lived offset, that becomes history at
+ * the next event sched-in and therefore isn't useful.
+ * The userspace that needs to copy out AUX data in overwrite
+ * mode should know to use user_page::aux_head for the actual
+ * offset. So, from now on we don't output AUX records that
+ * have *only* OVERWRITE flag set.
+ */
+if (size || (handle->aux_flags & ~(u64)PERF_AUX_FLAG_OVERWRITE))

kernel/events/ring_buffer.c void perf_aux_output_end(struct perf_output_handle *handle, unsigned long size) · line 475

Quoted 5 lines
kernel/events/ring_buffer.c +2 -0
+perf_event_aux_event(handle->event, aux_head, size,
+     handle->aux_flags);
 
 rb->user_page->aux_head = rb->aux_head;
 if (rb_need_aux_wakeup(rb))

IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.

maintainer
#2 lore raw
Quoted 4 lines · Ben Gainey

Hi Alex

I was just wondering if this was expected to make 5.1? (I could not see it in RC5)?

Yikes, I forgot about this again, didn't I ?

At least, they appear to still apply, so I've queued them. Not sure they'll still make this release, but at least they should not go missing _again_.

#3 lore raw

Thanks Peter :-)

Quoted 13 lines · Peter Zijlstra
Quoted 5 lines · Ben Gainey

Hi Alex

I was just wondering if this was expected to make 5.1? (I could not see
it in RC5)?

Yikes, I forgot about this again, didn't I ?

At least, they appear to still apply, so I've queued them. Not sure they'll still make this release, but at least they should not go missing
_again_.

#4 lore raw

[tip:perf/core] perf/ring_buffer: Fix AUX record suppression

In reply to Alexander Shishkin
perf/ring_buffer: Fix AUX record suppression
Commit-ID:  339bc4183596e1f68c2c98a03b87aa124107c317
Gitweb:     https://git.kernel.org/tip/339bc4183596e1f68c2c98a03b87aa124107c317
Author:     Alexander Shishkin <[email protected]>
AuthorDate: Fri, 29 Mar 2019 11:13:38 +0200
Committer:  Ingo Molnar <[email protected]>
CommitDate: Tue, 16 Apr 2019 12:13:57 +0200

perf/ring_buffer: Fix AUX record suppression

The following commit:

1627314fb54a33e ("perf: Suppress AUX/OVERWRITE records")

has an unintended side-effect of also suppressing all AUX records with no flags and non-zero size, so all the regular records in the full trace mode. This breaks some use cases for people.

Fix this by restoring "regular" AUX records.

Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Linus Torvalds <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Stephane Eranian <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Vince Weaver <[email protected]>
Fixes 1627314fb54a33e ("perf: Suppress AUX/OVERWRITE records")
---
 kernel/events/ring_buffer.c | 33 +++++++++++++++------------------
 1 file changed, 15 insertions(+), 18 deletions(-)
kernel/events/ring_buffer.c +15 -18
index 2545ac08cc77..5eedb49a65ea 100644
@@ -455,24 +455,21 @@ void perf_aux_output_end(struct perf_output_handle *handle, unsigned long size)
 		rb->aux_head += size;
 	}
 
-	if (size || handle->aux_flags) {
-		/*
-		 * Only send RECORD_AUX if we have something useful to communicate
-		 *
-		 * Note: the OVERWRITE records by themselves are not considered
-		 * useful, as they don't communicate any *new* information,
-		 * aside from the short-lived offset, that becomes history at
-		 * the next event sched-in and therefore isn't useful.
-		 * The userspace that needs to copy out AUX data in overwrite
-		 * mode should know to use user_page::aux_head for the actual
-		 * offset. So, from now on we don't output AUX records that
-		 * have *only* OVERWRITE flag set.
-		 */
-
-		if (handle->aux_flags & ~(u64)PERF_AUX_FLAG_OVERWRITE)
-			perf_event_aux_event(handle->event, aux_head, size,
-			                     handle->aux_flags);
-	}
+	/*
+	 * Only send RECORD_AUX if we have something useful to communicate
+	 *
+	 * Note: the OVERWRITE records by themselves are not considered
+	 * useful, as they don't communicate any *new* information,
+	 * aside from the short-lived offset, that becomes history at
+	 * the next event sched-in and therefore isn't useful.
+	 * The userspace that needs to copy out AUX data in overwrite
+	 * mode should know to use user_page::aux_head for the actual
+	 * offset. So, from now on we don't output AUX records that
+	 * have *only* OVERWRITE flag set.
+	 */
+	if (size || (handle->aux_flags & ~(u64)PERF_AUX_FLAG_OVERWRITE))
+		perf_event_aux_event(handle->event, aux_head, size,
+				     handle->aux_flags);
 
 	rb->user_page->aux_head = rb->aux_head;
 	if (rb_need_aux_wakeup(rb))