| Message ID | 20230902165216.1721082-1-syoshida@redhat.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net] kcm: Fix memory leak in error path of kcm_sendmsg() | expand |
Hello, On Sun, 2023-09-03 at 01:52 +0900, Shigeru Yoshida wrote: > syzbot reported a memory leak like below [1]: > > BUG: memory leak > unreferenced object 0xffff88810b088c00 (size 240): > comm "syz-executor186", pid 5012, jiffies 4294943306 (age 13.680s) > hex dump (first 32 bytes): > 00 89 08 0b 81 88 ff ff 00 00 00 00 00 00 00 00 ................ > 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > backtrace: > [<ffffffff83e5d5ff>] __alloc_skb+0x1ef/0x230 net/core/skbuff.c:634 > [<ffffffff84606e59>] alloc_skb include/linux/skbuff.h:1289 [inline] > [<ffffffff84606e59>] kcm_sendmsg+0x269/0x1050 net/kcm/kcmsock.c:815 > [<ffffffff83e479c6>] sock_sendmsg_nosec net/socket.c:725 [inline] > [<ffffffff83e479c6>] sock_sendmsg+0x56/0xb0 net/socket.c:748 > [<ffffffff83e47f55>] ____sys_sendmsg+0x365/0x470 net/socket.c:2494 > [<ffffffff83e4c389>] ___sys_sendmsg+0xc9/0x130 net/socket.c:2548 > [<ffffffff83e4c536>] __sys_sendmsg+0xa6/0x120 net/socket.c:2577 > [<ffffffff84ad7bb8>] do_syscall_x64 arch/x86/entry/common.c:50 [inline] > [<ffffffff84ad7bb8>] do_syscall_64+0x38/0xb0 arch/x86/entry/common.c:80 > [<ffffffff84c0008b>] entry_SYSCALL_64_after_hwframe+0x63/0xcd > > In kcm_sendmsg(), newly allocated socket buffers can be added to skb->next. If > an error occurred and jumped to out_error label, those newly allocated socket > buffers can be leaked. This patch fixes this issue by remembering the last > allocated socket buffer in kcm_tx_msg(head)->last_skb. I think the root cause should be clarified a little more. When the 'head' skb will be freed, all the frag_list skbs will be released, too. AFAICS the issue is that in case of error after copying some bytes, kcm_tx_msg(head)->last_skb is left unmodified and such reference is used as the cursor to append newly allocated skbs to 'head'. A later kcm_sendmsg will use an obsoleted 'last_skb' reference, corrupting the 'head' frag_list and causing the leak. The fix looks correct, but could you please send a v2 with an updated/more extensive commit message? > Link: https://syzkaller.appspot.com/bug?extid=6f98de741f7dbbfc4ccb [1] > Reported-by: syzbot+6f98de741f7dbbfc4ccb@syzkaller.appspotmail.com > Fixes: ab7ac4eb9832 ("kcm: Kernel Connection Multiplexor module") > Signed-off-by: Shigeru Yoshida <syoshida@redhat.com> Additionally please feed the patch to syzbot, so it can verify the fix, and add the relevant tag. Thanks, Paolo
Hi Paolo, On 9/5/23 17:09, Paolo Abeni wrote: > Hello, > > On Sun, 2023-09-03 at 01:52 +0900, Shigeru Yoshida wrote: >> syzbot reported a memory leak like below [1]: >> >> BUG: memory leak >> unreferenced object 0xffff88810b088c00 (size 240): >> comm "syz-executor186", pid 5012, jiffies 4294943306 (age 13.680s) >> hex dump (first 32 bytes): >> 00 89 08 0b 81 88 ff ff 00 00 00 00 00 00 00 00 ................ >> 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ >> backtrace: >> [<ffffffff83e5d5ff>] __alloc_skb+0x1ef/0x230 net/core/skbuff.c:634 >> [<ffffffff84606e59>] alloc_skb include/linux/skbuff.h:1289 [inline] >> [<ffffffff84606e59>] kcm_sendmsg+0x269/0x1050 net/kcm/kcmsock.c:815 >> [<ffffffff83e479c6>] sock_sendmsg_nosec net/socket.c:725 [inline] >> [<ffffffff83e479c6>] sock_sendmsg+0x56/0xb0 net/socket.c:748 >> [<ffffffff83e47f55>] ____sys_sendmsg+0x365/0x470 net/socket.c:2494 >> [<ffffffff83e4c389>] ___sys_sendmsg+0xc9/0x130 net/socket.c:2548 >> [<ffffffff83e4c536>] __sys_sendmsg+0xa6/0x120 net/socket.c:2577 >> [<ffffffff84ad7bb8>] do_syscall_x64 arch/x86/entry/common.c:50 [inline] >> [<ffffffff84ad7bb8>] do_syscall_64+0x38/0xb0 arch/x86/entry/common.c:80 >> [<ffffffff84c0008b>] entry_SYSCALL_64_after_hwframe+0x63/0xcd >> >> In kcm_sendmsg(), newly allocated socket buffers can be added to skb->next. If >> an error occurred and jumped to out_error label, those newly allocated socket >> buffers can be leaked. This patch fixes this issue by remembering the last >> allocated socket buffer in kcm_tx_msg(head)->last_skb. > > I think the root cause should be clarified a little more. When the > 'head' skb will be freed, all the frag_list skbs will be released, too. > > AFAICS the issue is that in case of error after copying some bytes, > kcm_tx_msg(head)->last_skb is left unmodified and such reference is > used as the cursor to append newly allocated skbs to 'head'. > > A later kcm_sendmsg will use an obsoleted 'last_skb' reference, > corrupting the 'head' frag_list and causing the leak. > > The fix looks correct, but could you please send a v2 with an > updated/more extensive commit message? Thank you so much for your feedback. Yes, I'll send a v2 patch with the detailed root cause. > >> Link: https://syzkaller.appspot.com/bug?extid=6f98de741f7dbbfc4ccb [1] >> Reported-by: syzbot+6f98de741f7dbbfc4ccb@syzkaller.appspotmail.com >> Fixes: ab7ac4eb9832 ("kcm: Kernel Connection Multiplexor module") >> Signed-off-by: Shigeru Yoshida <syoshida@redhat.com> > > Additionally please feed the patch to syzbot, so it can verify the fix, > and add the relevant tag. I got it. I'll try syzbot for the v2 patch. Thanks, Shigeru > > Thanks, > > Paolo >
diff --git a/net/kcm/kcmsock.c b/net/kcm/kcmsock.c index 393f01b2a7e6..34d4062f639a 100644 --- a/net/kcm/kcmsock.c +++ b/net/kcm/kcmsock.c @@ -939,6 +939,8 @@ static int kcm_sendmsg(struct socket *sock, struct msghdr *msg, size_t len) if (head != kcm->seq_skb) kfree_skb(head); + else if (copied) + kcm_tx_msg(head)->last_skb = skb; err = sk_stream_error(sk, msg->msg_flags, err);
syzbot reported a memory leak like below [1]: BUG: memory leak unreferenced object 0xffff88810b088c00 (size 240): comm "syz-executor186", pid 5012, jiffies 4294943306 (age 13.680s) hex dump (first 32 bytes): 00 89 08 0b 81 88 ff ff 00 00 00 00 00 00 00 00 ................ 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ backtrace: [<ffffffff83e5d5ff>] __alloc_skb+0x1ef/0x230 net/core/skbuff.c:634 [<ffffffff84606e59>] alloc_skb include/linux/skbuff.h:1289 [inline] [<ffffffff84606e59>] kcm_sendmsg+0x269/0x1050 net/kcm/kcmsock.c:815 [<ffffffff83e479c6>] sock_sendmsg_nosec net/socket.c:725 [inline] [<ffffffff83e479c6>] sock_sendmsg+0x56/0xb0 net/socket.c:748 [<ffffffff83e47f55>] ____sys_sendmsg+0x365/0x470 net/socket.c:2494 [<ffffffff83e4c389>] ___sys_sendmsg+0xc9/0x130 net/socket.c:2548 [<ffffffff83e4c536>] __sys_sendmsg+0xa6/0x120 net/socket.c:2577 [<ffffffff84ad7bb8>] do_syscall_x64 arch/x86/entry/common.c:50 [inline] [<ffffffff84ad7bb8>] do_syscall_64+0x38/0xb0 arch/x86/entry/common.c:80 [<ffffffff84c0008b>] entry_SYSCALL_64_after_hwframe+0x63/0xcd In kcm_sendmsg(), newly allocated socket buffers can be added to skb->next. If an error occurred and jumped to out_error label, those newly allocated socket buffers can be leaked. This patch fixes this issue by remembering the last allocated socket buffer in kcm_tx_msg(head)->last_skb. Link: https://syzkaller.appspot.com/bug?extid=6f98de741f7dbbfc4ccb [1] Reported-by: syzbot+6f98de741f7dbbfc4ccb@syzkaller.appspotmail.com Fixes: ab7ac4eb9832 ("kcm: Kernel Connection Multiplexor module") Signed-off-by: Shigeru Yoshida <syoshida@redhat.com> --- net/kcm/kcmsock.c | 2 ++ 1 file changed, 2 insertions(+)