[RH7] ms/tracing: Fix race in trace_open and buffer resize call

Submitted by Vasily Averin on Nov. 20, 2020, 6:21 a.m.

Details

Message ID dd8d4480-3090-bc30-1068-21d923f25db0@virtuozzo.com
State New
Series "ms/tracing: Fix race in trace_open and buffer resize call"
Headers show

Commit Message

Vasily Averin Nov. 20, 2020, 6:21 a.m. 
Author: Gaurav Kohli <gkohli@codeaurora.org>
backported upstream commit bbeb97464eefc65f506084fd9f18f21653e01137

    Below race can come, if trace_open and resize of
    cpu buffer is running parallely on different cpus
    CPUX                                CPUY
                                    ring_buffer_resize
                                    atomic_read(&buffer->resize_disabled)
    tracing_open
    tracing_reset_online_cpus
    ring_buffer_reset_cpu
    rb_reset_cpu
                                    rb_update_pages
                                    remove/insert pages
    resetting pointer

    This race can cause data abort or some times infinte loop in
    rb_remove_pages and rb_insert_pages while checking pages
    for sanity.

    Take buffer lock to fix this.

    Signed-off-by: Gaurav Kohli <gkohli@codeaurora.org>
    Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>

backport changes: ring_buffer_reset_cpu() fixed only
ring_buffer_reset_online_cpus() is not present in this kernel
https://jira.sw.ru/browse/PSBM-122343
Signed-off-by: Vasily Averin <vvs@virtuozzo.com>
---
 kernel/trace/ring_buffer.c | 5 +++++
 1 file changed, 5 insertions(+)

Patch hide | download patch | download mbox 

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 8b3df28..7b1afd1 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -4170,6 +4170,9 @@  void ring_buffer_reset_cpu(struct ring_buffer *buffer, int cpu)
 	if (!cpumask_test_cpu(cpu, buffer->cpumask))
 		return;
 
+	/* prevent another thread from changing buffer sizes */
+	mutex_lock(&buffer->mutex);
+
 	atomic_inc(&buffer->resize_disabled);
 	atomic_inc(&cpu_buffer->record_disabled);
 
@@ -4192,6 +4195,8 @@  void ring_buffer_reset_cpu(struct ring_buffer *buffer, int cpu)
 
 	atomic_dec(&cpu_buffer->record_disabled);
 	atomic_dec(&buffer->resize_disabled);
+
+	mutex_unlock(&buffer->mutex);
 }
 EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu);