diff options
authorJesper Dangaard Brouer <brouer@redhat.com>2018-08-08 23:00:34 +0200
committerDaniel Borkmann <daniel@iogearbox.net>2018-08-09 21:50:44 +0200
commitad0ab027fc6da08cbd34070d816ff3b7986c64ae (patch)
parentbf9bae0ea6ec7013ef37b19fbbf29b62a35474fb (diff)
xdp: fix bug in cpumap teardown code path
When removing a cpumap entry, a number of syncronization steps happen. Eventually the teardown code __cpu_map_entry_free is invoked from/via call_rcu. The teardown code __cpu_map_entry_free() flushes remaining xdp_frames, by invoking bq_flush_to_queue, which calls xdp_return_frame_rx_napi(). The issues is that the teardown code is not running in the RX NAPI code path. Thus, it is not allowed to invoke the NAPI variant of xdp_return_frame. This bug was found and triggered by using the --stress-mode option to the samples/bpf program xdp_redirect_cpu. It is hard to trigger, because the ptr_ring have to be full and cpumap bulk queue max contains 8 packets, and a remote CPU is racing to empty the ptr_ring queue. Fixes: 389ab7f01af9 ("xdp: introduce xdp_return_frame_rx_napi") Tested-by: Jean-Tsung Hsiao <jhsiao@redhat.com> Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
1 files changed, 9 insertions, 6 deletions
diff --git a/kernel/bpf/cpumap.c b/kernel/bpf/cpumap.c
index e0918d180f08..46f5f29605d4 100644
--- a/kernel/bpf/cpumap.c
+++ b/kernel/bpf/cpumap.c
@@ -69,7 +69,7 @@ struct bpf_cpu_map {
static int bq_flush_to_queue(struct bpf_cpu_map_entry *rcpu,
- struct xdp_bulk_queue *bq);
+ struct xdp_bulk_queue *bq, bool in_napi_ctx);
static u64 cpu_map_bitmap_size(const union bpf_attr *attr)
@@ -375,7 +375,7 @@ static void __cpu_map_entry_free(struct rcu_head *rcu)
struct xdp_bulk_queue *bq = per_cpu_ptr(rcpu->bulkq, cpu);
/* No concurrent bq_enqueue can run at this point */
- bq_flush_to_queue(rcpu, bq);
+ bq_flush_to_queue(rcpu, bq, false);
/* Cannot kthread_stop() here, last put free rcpu resources */
@@ -558,7 +558,7 @@ const struct bpf_map_ops cpu_map_ops = {
static int bq_flush_to_queue(struct bpf_cpu_map_entry *rcpu,
- struct xdp_bulk_queue *bq)
+ struct xdp_bulk_queue *bq, bool in_napi_ctx)
unsigned int processed = 0, drops = 0;
const int to_cpu = rcpu->cpu;
@@ -578,7 +578,10 @@ static int bq_flush_to_queue(struct bpf_cpu_map_entry *rcpu,
err = __ptr_ring_produce(q, xdpf);
if (err) {
- xdp_return_frame_rx_napi(xdpf);
+ if (likely(in_napi_ctx))
+ xdp_return_frame_rx_napi(xdpf);
+ else
+ xdp_return_frame(xdpf);
@@ -598,7 +601,7 @@ static int bq_enqueue(struct bpf_cpu_map_entry *rcpu, struct xdp_frame *xdpf)
struct xdp_bulk_queue *bq = this_cpu_ptr(rcpu->bulkq);
if (unlikely(bq->count == CPU_MAP_BULK_SIZE))
- bq_flush_to_queue(rcpu, bq);
+ bq_flush_to_queue(rcpu, bq, true);
/* Notice, xdp_buff/page MUST be queued here, long enough for
* driver to code invoking us to finished, due to driver
@@ -661,7 +664,7 @@ void __cpu_map_flush(struct bpf_map *map)
/* Flush all frames in bulkq to real queue */
bq = this_cpu_ptr(rcpu->bulkq);
- bq_flush_to_queue(rcpu, bq);
+ bq_flush_to_queue(rcpu, bq, true);
/* If already running, costs spin_lock_irqsave + smb_mb */