From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:60686) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Y9H0C-0004WW-5F for qemu-devel@nongnu.org; Thu, 08 Jan 2015 12:36:40 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Y9Gzr-0006os-Rw for qemu-devel@nongnu.org; Thu, 08 Jan 2015 12:36:08 -0500 Received: from e9.ny.us.ibm.com ([32.97.182.139]:46566) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Y9Gzr-0006oZ-Hk for qemu-devel@nongnu.org; Thu, 08 Jan 2015 12:35:47 -0500 Received: from /spool/local by e9.ny.us.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Thu, 8 Jan 2015 12:35:47 -0500 From: Michael Roth Date: Thu, 8 Jan 2015 11:34:04 -0600 Message-Id: <1420738472-23267-61-git-send-email-mdroth@linux.vnet.ibm.com> In-Reply-To: <1420738472-23267-1-git-send-email-mdroth@linux.vnet.ibm.com> References: <1420738472-23267-1-git-send-email-mdroth@linux.vnet.ibm.com> Subject: [Qemu-devel] [PATCH 60/88] virtio-scsi: work around bug in old BIOSes List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: qemu-stable@nongnu.org From: Paolo Bonzini Old BIOSes left some padding by mistake after the req_size/resp_size. New QEMU does not like it, thinking it is a bidirectional command. As a workaround, we can check if the ANY_LAYOUT bit is set; if not, we always consider the first buffer as the virtio-scsi request/response, because, back when QEMU did not support ANY_LAYOUT, it expected the payload to start at the second element of the iovec. This can show up during migration. Cc: qemu-stable@nongnu.org Signed-off-by: Paolo Bonzini (cherry picked from commit 55783a5521a3b1f93ee6a072e414a27c6cfa15f0) Signed-off-by: Michael Roth --- hw/scsi/virtio-scsi.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c index f041991..f7e77e9 100644 --- a/hw/scsi/virtio-scsi.c +++ b/hw/scsi/virtio-scsi.c @@ -135,6 +135,7 @@ static size_t qemu_sgl_concat(VirtIOSCSIReq *req, struct iovec *iov, static int virtio_scsi_parse_req(VirtIOSCSIReq *req, unsigned req_size, unsigned resp_size) { + VirtIODevice *vdev = (VirtIODevice *) req->dev; size_t in_size, out_size; if (iov_to_buf(req->elem.out_sg, req->elem.out_num, 0, @@ -147,8 +148,24 @@ static int virtio_scsi_parse_req(VirtIOSCSIReq *req, resp_size) < resp_size) { return -EINVAL; } + req->resp_size = resp_size; + /* Old BIOSes left some padding by mistake after the req_size/resp_size. + * As a workaround, always consider the first buffer as the virtio-scsi + * request/response, making the payload start at the second element + * of the iovec. + * + * The actual length of the response header, stored in req->resp_size, + * does not change. + * + * TODO: always disable this workaround for virtio 1.0 devices. + */ + if ((vdev->guest_features & VIRTIO_F_ANY_LAYOUT) == 0) { + req_size = req->elem.out_sg[0].iov_len; + resp_size = req->elem.in_sg[0].iov_len; + } + out_size = qemu_sgl_concat(req, req->elem.out_sg, &req->elem.out_addr[0], req->elem.out_num, req_size); -- 1.9.1