[PATCH] bpf_lru_list: Read double-checked variable once without lock

For double-checked locking in bpf_common_lru_push_free(), node->type is
read outside the critical section and then re-checked under the lock.
However, concurrent writes to node->type result in data races.

For example, the following concurrent access was observed by KCSAN:

  write to 0xffff88801521bc22 of 1 bytes by task 10038 on cpu 1:
   __bpf_lru_node_move_in        kernel/bpf/bpf_lru_list.c:91
   __local_list_flush            kernel/bpf/bpf_lru_list.c:298
   ...
  read to 0xffff88801521bc22 of 1 bytes by task 10043 on cpu 0:
   bpf_common_lru_push_free      kernel/bpf/bpf_lru_list.c:507
   bpf_lru_push_free             kernel/bpf/bpf_lru_list.c:555
   ...

Fix the data races where node->type is read outside the critical section
(for double-checked locking) by marking the access with READ_ONCE() as
well as ensuring the variable is only accessed once.

Reported-by: syzbot+3536db46dfa58c573458@syzkaller.appspotmail.com
Reported-by: syzbot+516acdb03d3e27d91bcd@syzkaller.appspotmail.com
Signed-off-by: Marco Elver <elver@google.com>
---
Detailed reports:
	https://groups.google.com/g/syzkaller-upstream-moderation/c/PwsoQ7bfi8k/m/NH9Ni2WxAQAJ
	https://groups.google.com/g/syzkaller-upstream-moderation/c/-fXQO9ehxSM/m/RmQEcI2oAQAJ
---
 kernel/bpf/bpf_lru_list.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/kernel/bpf/bpf_lru_list.c b/kernel/bpf/bpf_lru_list.c
index 1b6b9349cb85..d99e89f113c4 100644
--- a/kernel/bpf/bpf_lru_list.c
+++ b/kernel/bpf/bpf_lru_list.c

@@ -502,13 +502,14 @@ struct bpf_lru_node *bpf_lru_pop_free(struct bpf_lru *lru, u32 hash)
 static void bpf_common_lru_push_free(struct bpf_lru *lru,
 				     struct bpf_lru_node *node)
 {
+	u8 node_type = READ_ONCE(node->type);
 	unsigned long flags;
 
-	if (WARN_ON_ONCE(node->type == BPF_LRU_LIST_T_FREE) ||
-	    WARN_ON_ONCE(node->type == BPF_LRU_LOCAL_LIST_T_FREE))
+	if (WARN_ON_ONCE(node_type == BPF_LRU_LIST_T_FREE) ||
+	    WARN_ON_ONCE(node_type == BPF_LRU_LOCAL_LIST_T_FREE))
 		return;
 
-	if (node->type == BPF_LRU_LOCAL_LIST_T_PENDING) {
+	if (node_type == BPF_LRU_LOCAL_LIST_T_PENDING) {
 		struct bpf_lru_locallist *loc_l;
 
 		loc_l = per_cpu_ptr(lru->common_lru.local_list, node->cpu);

-- 
2.30.0.478.g8a0d178c01-goog

On Tue, Feb 09, 2021 at 12:27:01PM +0100, Marco Elver wrote:

I think this can be bpf-next.

Acked-by: Martin KaFai Lau <redacted>

On Tue, Feb 9, 2021 at 10:00 PM Martin KaFai Lau [off-list ref] wrote:

On Tue, Feb 09, 2021 at 12:27:01PM +0100, Marco Elver wrote:

quoted

For double-checked locking in bpf_common_lru_push_free(), node->type is
read outside the critical section and then re-checked under the lock.
However, concurrent writes to node->type result in data races.

For example, the following concurrent access was observed by KCSAN:

  write to 0xffff88801521bc22 of 1 bytes by task 10038 on cpu 1:
   __bpf_lru_node_move_in        kernel/bpf/bpf_lru_list.c:91
   __local_list_flush            kernel/bpf/bpf_lru_list.c:298
   ...
  read to 0xffff88801521bc22 of 1 bytes by task 10043 on cpu 0:
   bpf_common_lru_push_free      kernel/bpf/bpf_lru_list.c:507
   bpf_lru_push_free             kernel/bpf/bpf_lru_list.c:555
   ...

Fix the data races where node->type is read outside the critical section
(for double-checked locking) by marking the access with READ_ONCE() as
well as ensuring the variable is only accessed once.

Reported-by: syzbot+3536db46dfa58c573458@syzkaller.appspotmail.com
Reported-by: syzbot+516acdb03d3e27d91bcd@syzkaller.appspotmail.com
Signed-off-by: Marco Elver <elver@google.com>
---
Detailed reports:
      https://groups.google.com/g/syzkaller-upstream-moderation/c/PwsoQ7bfi8k/m/NH9Ni2WxAQAJ
      https://groups.google.com/g/syzkaller-upstream-moderation/c/-fXQO9ehxSM/m/RmQEcI2oAQAJ
---
 kernel/bpf/bpf_lru_list.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/kernel/bpf/bpf_lru_list.c b/kernel/bpf/bpf_lru_list.c
index 1b6b9349cb85..d99e89f113c4 100644
--- a/kernel/bpf/bpf_lru_list.c
+++ b/kernel/bpf/bpf_lru_list.c

@@ -502,13 +502,14 @@ struct bpf_lru_node *bpf_lru_pop_free(struct bpf_lru *lru, u32 hash)
 static void bpf_common_lru_push_free(struct bpf_lru *lru,
                                   struct bpf_lru_node *node)
 {
+     u8 node_type = READ_ONCE(node->type);
      unsigned long flags;

-     if (WARN_ON_ONCE(node->type == BPF_LRU_LIST_T_FREE) ||
-         WARN_ON_ONCE(node->type == BPF_LRU_LOCAL_LIST_T_FREE))
+     if (WARN_ON_ONCE(node_type == BPF_LRU_LIST_T_FREE) ||
+         WARN_ON_ONCE(node_type == BPF_LRU_LOCAL_LIST_T_FREE))
              return;

-     if (node->type == BPF_LRU_LOCAL_LIST_T_PENDING) {
+     if (node_type == BPF_LRU_LOCAL_LIST_T_PENDING) {

I think this can be bpf-next.

Acked-by: Martin KaFai Lau <redacted>

Added Fixes: 3a08c2fd7634 ("bpf: LRU List") and applied to bpf-next, thanks.

Hello:

This patch was applied to bpf/bpf-next.git (refs/heads/master):

On Tue,  9 Feb 2021 12:27:01 +0100 you wrote:

For double-checked locking in bpf_common_lru_push_free(), node->type is
read outside the critical section and then re-checked under the lock.
However, concurrent writes to node->type result in data races.

For example, the following concurrent access was observed by KCSAN:

  write to 0xffff88801521bc22 of 1 bytes by task 10038 on cpu 1:
   __bpf_lru_node_move_in        kernel/bpf/bpf_lru_list.c:91
   __local_list_flush            kernel/bpf/bpf_lru_list.c:298
   ...
  read to 0xffff88801521bc22 of 1 bytes by task 10043 on cpu 0:
   bpf_common_lru_push_free      kernel/bpf/bpf_lru_list.c:507
   bpf_lru_push_free             kernel/bpf/bpf_lru_list.c:555
   ...

[...]

Here is the summary with links:
  - bpf_lru_list: Read double-checked variable once without lock
    https://git.kernel.org/bpf/bpf-next/c/6df8fb83301d

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html