bpf: add schedule points in percpu arrays management

syszbot managed to trigger RCU detected stalls in
bpf_array_free_percpu()

It takes time to allocate a huge percpu map, but even more time to free
it.

Since we run in process context, use cond_resched() to yield cpu if
needed.

Fixes: a10423b87a7e ("bpf: introduce BPF_MAP_TYPE_PERCPU_ARRAY map")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c
index a364c40..14750e7 100644 (file)
--- a/kernel/bpf/arraymap.c
+++ b/kernel/bpf/arraymap.c
@@ -26,8 +26,10 @@ static void bpf_array_free_percpu(struct bpf_array *array)
 {
        int i;
 
-       for (i = 0; i < array->map.max_entries; i++)
+       for (i = 0; i < array->map.max_entries; i++) {
                free_percpu(array->pptrs[i]);
+               cond_resched();
+       }
 }
 
 static int bpf_array_alloc_percpu(struct bpf_array *array)
@@ -43,6 +45,7 @@ static int bpf_array_alloc_percpu(struct bpf_array *array)
                        return -ENOMEM;
                }
                array->pptrs[i] = ptr;
+               cond_resched();
        }
 
        return 0;