[PATCH] [193/223] perf: Fix inherit vs. context rotation bug

From: Andi Kleen
Date: Sun Dec 12 2010 - 18:54:07 EST


2.6.35-longterm review patch. If anyone has any objections, please let me know.

------------------
From: Thomas Gleixner <tglx@xxxxxxxxxxxxx>

commit dddd3379a619a4cb8247bfd3c94ca9ae3797aa2e upstream.

It was found that sometimes children of tasks with inherited events had
one extra event. Eventually it turned out to be due to the list rotation
no being exclusive with the list iteration in the inheritance code.

Cure this by temporarily disabling the rotation while we inherit the events.

Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@xxxxxxxxx>
Signed-off-by: Andi Kleen <ak@xxxxxxxxxxxxxxx>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@xxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxx>

---
include/linux/perf_event.h | 1 +
kernel/perf_event.c | 22 ++++++++++++++++++++--
2 files changed, 21 insertions(+), 2 deletions(-)

Index: linux/include/linux/perf_event.h
===================================================================
--- linux.orig/include/linux/perf_event.h
+++ linux/include/linux/perf_event.h
@@ -764,6 +764,7 @@ struct perf_event_context {
int nr_active;
int is_active;
int nr_stat;
+ int rotate_disable;
atomic_t refcount;
struct task_struct *task;

Index: linux/kernel/perf_event.c
===================================================================
--- linux.orig/kernel/perf_event.c
+++ linux/kernel/perf_event.c
@@ -1609,8 +1609,12 @@ static void rotate_ctx(struct perf_event
{
raw_spin_lock(&ctx->lock);

- /* Rotate the first entry last of non-pinned groups */
- list_rotate_left(&ctx->flexible_groups);
+ /*
+ * Rotate the first entry last of non-pinned groups. Rotation might be
+ * disabled by the inheritance code.
+ */
+ if (!ctx->rotate_disable)
+ list_rotate_left(&ctx->flexible_groups);

raw_spin_unlock(&ctx->lock);
}
@@ -5596,6 +5600,7 @@ int perf_event_init_task(struct task_str
struct perf_event *event;
struct task_struct *parent = current;
int inherited_all = 1;
+ unsigned long flags;
int ret = 0;

child->perf_event_ctxp = NULL;
@@ -5636,6 +5641,15 @@ int perf_event_init_task(struct task_str
break;
}

+ /*
+ * We can't hold ctx->lock when iterating the ->flexible_group list due
+ * to allocations, but we need to prevent rotation because
+ * rotate_ctx() will change the list from interrupt context.
+ */
+ raw_spin_lock_irqsave(&parent_ctx->lock, flags);
+ parent_ctx->rotate_disable = 1;
+ raw_spin_unlock_irqrestore(&parent_ctx->lock, flags);
+
list_for_each_entry(event, &parent_ctx->flexible_groups, group_entry) {
ret = inherit_task_group(event, parent, parent_ctx, child,
&inherited_all);
@@ -5643,6 +5657,10 @@ int perf_event_init_task(struct task_str
break;
}

+ raw_spin_lock_irqsave(&parent_ctx->lock, flags);
+ parent_ctx->rotate_disable = 0;
+ raw_spin_unlock_irqrestore(&parent_ctx->lock, flags);
+
child_ctx = child->perf_event_ctxp;

if (child_ctx && inherited_all) {
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/