[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c
From: |
Jan Kiszka |
Subject: |
Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c |
Date: |
Thu, 31 May 2012 12:40:11 +0200 |
User-agent: |
Mozilla/5.0 (X11; U; Linux i686 (x86_64); de; rv:1.8.1.12) Gecko/20080226 SUSE/2.0.0.12-1.1 Thunderbird/2.0.0.12 Mnenhy/0.7.5.666 |
On 2012-05-31 12:03, Paolo Bonzini wrote:
> Il 31/05/2012 10:23, Jan Kiszka ha scritto:
>>>> @@ -922,6 +923,7 @@ set_rdt(E1000State *s, int index, uint32_t val)
>>>> {
>>>> s->check_rxov = 0;
>>>> s->mac_reg[index] = val & 0xffff;
>>>> + qemu_notify_event();
>> This still looks like the wrong tool: Packets that can't be delivered
>> are queued.
>
> Packets that are read from the tap but can't be delivered are queued;
> packets that are left on the tap need qemu_notify_event to be flushed.
>
>> So we need to flush the queue and clear the blocked delivery
>> there. qemu_flush_queued_packets appears more appropriate for this.
>
> Right, and qemu_flush_queued_packets needs to call qemu_notify_event
> which makes the call in virtio-net unnecessary.
>
> Paolo
>
> diff --git a/hw/e1000.c b/hw/e1000.c
> index 4573f13..43d933a 100644
> --- a/hw/e1000.c
> +++ b/hw/e1000.c
> @@ -295,6 +295,7 @@ set_rx_control(E1000State *s, int index, uint32_t val)
> s->rxbuf_min_shift = ((val / E1000_RCTL_RDMTS_QUAT) & 3) + 1;
> DBGOUT(RX, "RCTL: %d, mac_reg[RCTL] = 0x%x\n", s->mac_reg[RDT],
> s->mac_reg[RCTL]);
> + qemu_flush_queued_packets(&s->nic->nc);
> }
>
> static void
> @@ -926,6 +927,9 @@ set_rdt(E1000State *s, int index, uint32_t val)
> {
> s->check_rxov = 0;
> s->mac_reg[index] = val & 0xffff;
> + if (e1000_has_rxbufs(s, 1)) {
> + qemu_flush_queued_packets(&s->nic->nc);
> + }
> }
>
> static void
> diff --git a/hw/virtio-net.c b/hw/virtio-net.c
> index 3f190d4..0974945 100644
> --- a/hw/virtio-net.c
> +++ b/hw/virtio-net.c
> @@ -447,10 +447,6 @@ static void virtio_net_handle_rx(VirtIODevice *vdev,
> VirtQueue *vq)
> VirtIONet *n = to_virtio_net(vdev);
>
> qemu_flush_queued_packets(&n->nic->nc);
> -
> - /* We now have RX buffers, signal to the IO thread to break out of the
> - * select to re-poll the tap file descriptor */
> - qemu_notify_event();
> }
>
> static int virtio_net_can_receive(VLANClientState *nc)
> diff --git a/net.c b/net.c
> index 1922d8a..fa846ae 100644
> --- a/net.c
> +++ b/net.c
> @@ -491,7 +491,12 @@ void qemu_flush_queued_packets(VLANClientState *vc)
> queue = vc->send_queue;
> }
>
> - qemu_net_queue_flush(queue);
> + if (qemu_net_queue_flush(queue)) {
> + /* We emptied the queue successfully, signal to the IO thread to
> repoll
> + * the file descriptor (for tap, for example).
> + */
> + qemu_notify_event();
> + }
> }
>
> static ssize_t qemu_send_packet_async_with_flags(VLANClientState *sender,
> diff --git a/net/queue.c b/net/queue.c
> index 1ab5247..fd1c7e6 100644
> --- a/net/queue.c
> +++ b/net/queue.c
> @@ -232,7 +232,7 @@ void qemu_net_queue_purge(NetQueue *queue,
> VLANClientState *from)
> }
> }
>
> -void qemu_net_queue_flush(NetQueue *queue)
> +bool qemu_net_queue_flush(NetQueue *queue)
> {
> while (!QTAILQ_EMPTY(&queue->packets)) {
> NetPacket *packet;
> @@ -248,7 +248,7 @@ void qemu_net_queue_flush(NetQueue *queue)
> packet->size);
> if (ret == 0) {
> QTAILQ_INSERT_HEAD(&queue->packets, packet, entry);
> - break;
> + return 0;
> }
>
> if (packet->sent_cb) {
> @@ -257,4 +257,5 @@ void qemu_net_queue_flush(NetQueue *queue)
>
> g_free(packet);
> }
> + return 1;
> }
> diff --git a/net/queue.h b/net/queue.h
> index a31958e..4bf6d3c 100644
> --- a/net/queue.h
> +++ b/net/queue.h
> @@ -66,6 +66,6 @@ ssize_t qemu_net_queue_send_iov(NetQueue *queue,
> NetPacketSent *sent_cb);
>
> void qemu_net_queue_purge(NetQueue *queue, VLANClientState *from);
> -void qemu_net_queue_flush(NetQueue *queue);
> +bool qemu_net_queue_flush(NetQueue *queue);
>
> #endif /* QEMU_NET_QUEUE_H */
Looks good.
Jan
--
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, (continued)
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Luigi Rizzo, 2012/05/30
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Jan Kiszka, 2012/05/30
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Luigi Rizzo, 2012/05/30
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Paolo Bonzini, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Luigi Rizzo, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Jan Kiszka, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Jan Kiszka, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Luigi Rizzo, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Jan Kiszka, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Paolo Bonzini, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c,
Jan Kiszka <=
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Paolo Bonzini, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Stefan Weil, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Stefano Stabellini, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Paolo Bonzini, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Stefano Stabellini, 2012/05/31
- Re: [Qemu-devel] Proposed patch: huge RX speedup for hw/e1000.c, Paolo Bonzini, 2012/05/31