From: Eric Dumazet <edumazet@google.com>
Date: Thu, 22 Feb 2018 08:33:24 -0800
Subject: bpf: add schedule points in percpu arrays management
Patch-mainline: v4.16-rc3
Git-commit: 32fff239de37ef226d5b66329dd133f64d63b22d
References: bsc#1083647

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>
Acked-by: Gary Lin <glin@suse.com>
---
 kernel/bpf/arraymap.c |    5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

--- a/kernel/bpf/arraymap.c
+++ b/kernel/bpf/arraymap.c
@@ -23,8 +23,10 @@ static void bpf_array_free_percpu(struct
 {
 	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)
@@ -40,6 +42,7 @@ static int bpf_array_alloc_percpu(struct
 			return -ENOMEM;
 		}
 		array->pptrs[i] = ptr;
+		cond_resched();
 	}
 
 	return 0;