| Message ID | 20120914170432.C8B8DC6405@unicorn.suse.cz |
|---|---|
| State | Accepted, archived |
| Delegated to: | David Miller |
| Headers | show |
From: Michal Kubecek <mkubecek@suse.cz> Date: Fri, 14 Sep 2012 16:59:52 +0200 > If recv() syscall is called for a TCP socket so that > - IOAT DMA is used > - MSG_WAITALL flag is used > - requested length is bigger than sk_rcvbuf > - enough data has already arrived to bring rcv_wnd to zero > then when tcp_recvmsg() gets to calling sk_wait_data(), receive > window can be still zero while sk_async_wait_queue exhausts > enough space to keep it zero. As this queue isn't cleaned until > the tcp_service_net_dma() call, sk_wait_data() cannot receive > any data and blocks forever. > > If zero receive window and non-empty sk_async_wait_queue is > detected before calling sk_wait_data(), process the queue first. > > Signed-off-by: Michal Kubecek <mkubecek@suse.cz> Applied and queued up for -stable, thanks. > Cc: <stable@vger.kernel.org> Please do not CC: stable on networking bug fixes like this. Simply ask me to add it to the networking -stable queue instead. I do not want fixes to propagate immedately to -stable right when they hit Linus's tree. Rather, I want them to soak upstream for a while before they get submitted to -stable. And that's why I maintain a special queue for networking fixes which should be submitted to -stable at some point in the future at: http://patchwork.ozlabs.org/user/bundle/2566/?state=* Thanks. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 2109ff4..bf9a8ab 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -1762,8 +1762,14 @@ int tcp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg, } #ifdef CONFIG_NET_DMA - if (tp->ucopy.dma_chan) - dma_async_memcpy_issue_pending(tp->ucopy.dma_chan); + if (tp->ucopy.dma_chan) { + if (tp->rcv_wnd == 0 && + !skb_queue_empty(&sk->sk_async_wait_queue)) { + tcp_service_net_dma(sk, true); + tcp_cleanup_rbuf(sk, copied); + } else + dma_async_memcpy_issue_pending(tp->ucopy.dma_chan); + } #endif if (copied >= target) { /* Do not sleep, just process backlog. */
If recv() syscall is called for a TCP socket so that - IOAT DMA is used - MSG_WAITALL flag is used - requested length is bigger than sk_rcvbuf - enough data has already arrived to bring rcv_wnd to zero then when tcp_recvmsg() gets to calling sk_wait_data(), receive window can be still zero while sk_async_wait_queue exhausts enough space to keep it zero. As this queue isn't cleaned until the tcp_service_net_dma() call, sk_wait_data() cannot receive any data and blocks forever. If zero receive window and non-empty sk_async_wait_queue is detected before calling sk_wait_data(), process the queue first. Signed-off-by: Michal Kubecek <mkubecek@suse.cz> Cc: <stable@vger.kernel.org> --- net/ipv4/tcp.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)