From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:51349) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YkReo-0004ec-B2 for qemu-devel@nongnu.org; Tue, 21 Apr 2015 02:27:43 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YkRel-000898-3Q for qemu-devel@nongnu.org; Tue, 21 Apr 2015 02:27:42 -0400 Received: from mx1.redhat.com ([209.132.183.28]:60234) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YkRek-000892-Rg for qemu-devel@nongnu.org; Tue, 21 Apr 2015 02:27:39 -0400 Date: Tue, 21 Apr 2015 08:27:32 +0200 From: "Michael S. Tsirkin" Message-ID: <20150421081111-mutt-send-email-mst@redhat.com> References: <1429257573-7359-1-git-send-email-famz@redhat.com> <1429257573-7359-4-git-send-email-famz@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1429257573-7359-4-git-send-email-famz@redhat.com> Subject: Re: [Qemu-devel] [PATCH 03/18] virtio: Return error from virtqueue_get_head List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Fam Zheng Cc: Kevin Wolf , qemu-devel@nongnu.org, "Aneesh Kumar K.V" , Stefan Hajnoczi , Amit Shah , Paolo Bonzini On Fri, Apr 17, 2015 at 03:59:18PM +0800, Fam Zheng wrote: > Return type is changed to int. > > When data is invalid, return -EINVAL with an error. > > Signed-off-by: Fam Zheng Problem is, callers can't really handle -EINVAL here, so it's adding dead code if we just apply patches 1-9. How about the following refactoring? Your patch can go on top, it doesn't add dead code then. If you like this, feel free to include it. ---> virtio: get rid of virtqueue_num_heads We don't really care about # of heads, squash virtqueue_num_heads and virtqueue_get_head together, to get a simpler internal API. Memory barrier logic is also simplified a bit. Signed-off-by: Michael S. Tsirkin -- diff --git a/hw/virtio/virtio.c b/hw/virtio/virtio.c index 60d10cd..580effe 100644 --- a/hw/virtio/virtio.c +++ b/hw/virtio/virtio.c @@ -295,8 +295,9 @@ void virtqueue_push(VirtQueue *vq, const VirtQueueElement *elem, virtqueue_flush(vq, 1); } -static int virtqueue_num_heads(VirtQueue *vq, unsigned int idx) +static int virtqueue_get_head(VirtQueue *vq, unsigned int idx) { + unsigned int head; uint16_t num_heads = vring_avail_idx(vq) - idx; /* Check it isn't doing very strange things with descriptor numbers. */ @@ -305,18 +306,13 @@ static int virtqueue_num_heads(VirtQueue *vq, unsigned int idx) idx, vring_avail_idx(vq)); exit(1); } - /* On success, callers read a descriptor at vq->last_avail_idx. - * Make sure descriptor read does not bypass avail index read. */ - if (num_heads) { - smp_rmb(); - } - return num_heads; -} + if (!num_heads) { + return -EAGAIN; + } -static unsigned int virtqueue_get_head(VirtQueue *vq, unsigned int idx) -{ - unsigned int head; + /* Make sure head read below does not bypass avail index read. */ + smp_rmb(); /* Grab the next descriptor number they're advertising, and increment * the index we've seen. */ @@ -360,19 +356,19 @@ void virtqueue_get_avail_bytes(VirtQueue *vq, unsigned int *in_bytes, { unsigned int idx; unsigned int total_bufs, in_total, out_total; + int i; idx = vq->last_avail_idx; total_bufs = in_total = out_total = 0; - while (virtqueue_num_heads(vq, idx)) { + while ((i = virtqueue_get_head(vq, idx)) >= 0) { VirtIODevice *vdev = vq->vdev; unsigned int max, num_bufs, indirect = 0; hwaddr desc_pa; - int i; max = vq->vring.num; num_bufs = total_bufs; - i = virtqueue_get_head(vq, idx++); + idx++; desc_pa = vq->vring.desc; if (vring_desc_flags(vdev, desc_pa, i) & VRING_DESC_F_INDIRECT) { @@ -462,15 +458,17 @@ int virtqueue_pop(VirtQueue *vq, VirtQueueElement *elem) hwaddr desc_pa = vq->vring.desc; VirtIODevice *vdev = vq->vdev; - if (!virtqueue_num_heads(vq, vq->last_avail_idx)) + if ((i = virtqueue_get_head(vq, vq->last_avail_idx)) < 0) { return 0; + } /* When we start there are none of either input nor output. */ elem->out_num = elem->in_num = 0; max = vq->vring.num; - i = head = virtqueue_get_head(vq, vq->last_avail_idx++); + head = i; + vq->last_avail_idx++; if (virtio_has_feature(vdev, VIRTIO_RING_F_EVENT_IDX)) { vring_set_avail_event(vq, vq->last_avail_idx); }