[PATCH] perf: Fix the address filterig fix

* [PATCH] perf: Fix the address filterig fix
@ 2019-03-29  9:12 Alexander Shishkin
  2019-04-16 11:31 ` [tip:perf/core] perf/core: Fix the address filtering fix tip-bot for Alexander Shishkin
  0 siblings, 1 reply; 2+ messages in thread
From: Alexander Shishkin @ 2019-03-29  9:12 UTC (permalink / raw)
  To: Peter Zijlstra
  Cc: Arnaldo Carvalho de Melo, Ingo Molnar, linux-kernel, jolsa,
	Alexander Shishkin, Adrian Hunter

Commit c60f83b813e5 ("perf, pt, coresight: Fix address filters for vmas
with non-zero offset") changes the address filtering logic to communicate
filter ranges to the PMU driver via a single address range object, instead
of having the driver do the final bit of math. That change forgets to take
into account kernel filters, which are not calculated the same way as DSO
based filters.

Fix that by passing the kernel filters the same way as file-based filters.
This doesn't require any additional changes in the drivers.

Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Fixes: c60f83b813e5 ("perf, pt, coresight: Fix address filters for vmas with non-zero offset")
Reported-by: Adrian Hunter <adrian.hunter@intel.com>
---
 kernel/events/core.c | 37 +++++++++++++++++++++----------------
 1 file changed, 21 insertions(+), 16 deletions(-)

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 5f59d848171e..53e7188fa854 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -9041,26 +9041,29 @@ static void perf_event_addr_filters_apply(struct perf_event *event)
 	if (task == TASK_TOMBSTONE)
 		return;
 
-	if (!ifh->nr_file_filters)
-		return;
-
-	mm = get_task_mm(event->ctx->task);
-	if (!mm)
-		goto restart;
+	if (ifh->nr_file_filters) {
+		mm = get_task_mm(event->ctx->task);
+		if (!mm)
+			goto restart;
 
-	down_read(&mm->mmap_sem);
+		down_read(&mm->mmap_sem);
+	}
 
 	raw_spin_lock_irqsave(&ifh->lock, flags);
 	list_for_each_entry(filter, &ifh->list, entry) {
-		event->addr_filter_ranges[count].start = 0;
-		event->addr_filter_ranges[count].size = 0;
+		if (filter->path.dentry) {
+			/*
+			 * Adjust base offset if the filter is associated to a
+			 * binary that needs to be mapped:
+			 */
+			event->addr_filter_ranges[count].start = 0;
+			event->addr_filter_ranges[count].size = 0;
 
-		/*
-		 * Adjust base offset if the filter is associated to a binary
-		 * that needs to be mapped:
-		 */
-		if (filter->path.dentry)
 			perf_addr_filter_apply(filter, mm, &event->addr_filter_ranges[count]);
+		} else {
+			event->addr_filter_ranges[count].start = filter->offset;
+			event->addr_filter_ranges[count].size  = filter->size;
+		}
 
 		count++;
 	}
@@ -9068,9 +9071,11 @@ static void perf_event_addr_filters_apply(struct perf_event *event)
 	event->addr_filters_gen++;
 	raw_spin_unlock_irqrestore(&ifh->lock, flags);
 
-	up_read(&mm->mmap_sem);
+	if (ifh->nr_file_filters) {
+		up_read(&mm->mmap_sem);
 
-	mmput(mm);
+		mmput(mm);
+	}
 
 restart:
 	perf_event_stop(event, 1);
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 2+ messages in thread