qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Anthony Liguori <anthony@codemonkey.ws>
To: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Anthony Liguori <aliguori@us.ibm.com>,
	kvm@vger.kernel.org, Alexey Kardashevskiy <aik@ozlabs.ru>,
	rusty@rustcorp.com.au, qemu-devel@nongnu.org,
	virtualization@lists.linux-foundation.org, avi@redhat.com
Subject: Re: [Qemu-devel] [PATCH RFC] virtio-pci: add MMIO property
Date: Mon, 19 Mar 2012 14:19:33 -0500	[thread overview]
Message-ID: <4F6786C5.3080902@codemonkey.ws> (raw)
In-Reply-To: <20120319155650.GA6430@redhat.com>

On 03/19/2012 10:56 AM, Michael S. Tsirkin wrote:
> Currently virtio-pci is specified so that configuration of the device is
> done through a PCI IO space (via BAR 0 of the virtual PCI device).
> However, Linux guests happen to use ioread/iowrite/iomap primitives
> for access, and these work uniformly across memory/io BARs.
>
> While PCI IO accesses are faster than MMIO on x86 kvm,
> MMIO might be helpful on other systems which don't
> implement PIO or where PIO is slower than MMIO.
>
> Add a property to make it possible to tweak the BAR type.
>
> Signed-off-by: Michael S. Tsirkin<mst@redhat.com>
>
> This is harmless by default but causes segfaults in memory.c
> when enabled. Thus an RFC until I figure out what's wrong.

Doesn't this violate the virtio-pci spec?

Making the same vendor/device ID have different semantics depending on a magic 
flag in QEMU seems like a pretty bad idea to me.

Regards,

Anthony Liguori

>
> ---
>   hw/virtio-pci.c |   16 ++++++++++++++--
>   hw/virtio-pci.h |    4 ++++
>   2 files changed, 18 insertions(+), 2 deletions(-)
>
> diff --git a/hw/virtio-pci.c b/hw/virtio-pci.c
> index 28498ec..6f338d2 100644
> --- a/hw/virtio-pci.c
> +++ b/hw/virtio-pci.c
> @@ -655,6 +655,7 @@ void virtio_init_pci(VirtIOPCIProxy *proxy, VirtIODevice *vdev)
>   {
>       uint8_t *config;
>       uint32_t size;
> +    uint8_t bar0_type;
>
>       proxy->vdev = vdev;
>
> @@ -684,8 +685,14 @@ void virtio_init_pci(VirtIOPCIProxy *proxy, VirtIODevice *vdev)
>
>       memory_region_init_io(&proxy->bar,&virtio_pci_config_ops, proxy,
>                             "virtio-pci", size);
> -    pci_register_bar(&proxy->pci_dev, 0, PCI_BASE_ADDRESS_SPACE_IO,
> -&proxy->bar);
> +
> +    if (proxy->flags&  VIRTIO_PCI_FLAG_USE_MMIO) {
> +        bar0_type = PCI_BASE_ADDRESS_SPACE_MEMORY;
> +    } else {
> +        bar0_type = PCI_BASE_ADDRESS_SPACE_IO;
> +    }
> +
> +    pci_register_bar(&proxy->pci_dev, 0, bar0_type,&proxy->bar);
>
>       if (!kvm_has_many_ioeventfds()) {
>           proxy->flags&= ~VIRTIO_PCI_FLAG_USE_IOEVENTFD;
> @@ -823,6 +830,7 @@ static Property virtio_blk_properties[] = {
>       DEFINE_PROP_BIT("ioeventfd", VirtIOPCIProxy, flags, VIRTIO_PCI_FLAG_USE_IOEVENTFD_BIT, true),
>       DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, 2),
>       DEFINE_VIRTIO_BLK_FEATURES(VirtIOPCIProxy, host_features),
> +    DEFINE_PROP_BIT("mmio", VirtIOPCIProxy, flags, VIRTIO_PCI_FLAG_USE_MMIO_BIT, false),
>       DEFINE_PROP_END_OF_LIST(),
>   };
>
> @@ -856,6 +864,7 @@ static Property virtio_net_properties[] = {
>       DEFINE_PROP_UINT32("x-txtimer", VirtIOPCIProxy, net.txtimer, TX_TIMER_INTERVAL),
>       DEFINE_PROP_INT32("x-txburst", VirtIOPCIProxy, net.txburst, TX_BURST),
>       DEFINE_PROP_STRING("tx", VirtIOPCIProxy, net.tx),
> +    DEFINE_PROP_BIT("mmio", VirtIOPCIProxy, flags, VIRTIO_PCI_FLAG_USE_MMIO_BIT, false),
>       DEFINE_PROP_END_OF_LIST(),
>   };
>
> @@ -888,6 +897,7 @@ static Property virtio_serial_properties[] = {
>       DEFINE_PROP_HEX32("class", VirtIOPCIProxy, class_code, 0),
>       DEFINE_VIRTIO_COMMON_FEATURES(VirtIOPCIProxy, host_features),
>       DEFINE_PROP_UINT32("max_ports", VirtIOPCIProxy, serial.max_virtserial_ports, 31),
> +    DEFINE_PROP_BIT("mmio", VirtIOPCIProxy, flags, VIRTIO_PCI_FLAG_USE_MMIO_BIT, false),
>       DEFINE_PROP_END_OF_LIST(),
>   };
>
> @@ -915,6 +925,7 @@ static TypeInfo virtio_serial_info = {
>
>   static Property virtio_balloon_properties[] = {
>       DEFINE_VIRTIO_COMMON_FEATURES(VirtIOPCIProxy, host_features),
> +    DEFINE_PROP_BIT("mmio", VirtIOPCIProxy, flags, VIRTIO_PCI_FLAG_USE_MMIO_BIT, false),
>       DEFINE_PROP_END_OF_LIST(),
>   };
>
> @@ -969,6 +980,7 @@ static int virtio_scsi_exit_pci(PCIDevice *pci_dev)
>   static Property virtio_scsi_properties[] = {
>       DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, 2),
>       DEFINE_VIRTIO_SCSI_PROPERTIES(VirtIOPCIProxy, host_features, scsi),
> +    DEFINE_PROP_BIT("mmio", VirtIOPCIProxy, flags, VIRTIO_PCI_FLAG_USE_MMIO_BIT, false),
>       DEFINE_PROP_END_OF_LIST(),
>   };
>
> diff --git a/hw/virtio-pci.h b/hw/virtio-pci.h
> index e560428..e6a8861 100644
> --- a/hw/virtio-pci.h
> +++ b/hw/virtio-pci.h
> @@ -24,6 +24,10 @@
>   #define VIRTIO_PCI_FLAG_USE_IOEVENTFD_BIT 1
>   #define VIRTIO_PCI_FLAG_USE_IOEVENTFD   (1<<  VIRTIO_PCI_FLAG_USE_IOEVENTFD_BIT)
>
> +/* Some guests don't support port IO. Use MMIO instead. */
> +#define VIRTIO_PCI_FLAG_USE_MMIO_BIT 2
> +#define VIRTIO_PCI_FLAG_USE_MMIO   (1<<  VIRTIO_PCI_FLAG_USE_MMIO_BIT)
> +
>   typedef struct {
>       PCIDevice pci_dev;
>       VirtIODevice *vdev;

  parent reply	other threads:[~2012-03-19 19:19 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-03-19 15:56 [Qemu-devel] [PATCH RFC] virtio-pci: add MMIO property Michael S. Tsirkin
2012-03-19 17:58 ` Avi Kivity
2012-03-19 18:57   ` Michael S. Tsirkin
2012-03-20 14:49     ` Avi Kivity
2012-03-19 19:19 ` Anthony Liguori [this message]
2012-03-19 20:49   ` Michael S. Tsirkin
2012-03-19 21:07     ` Anthony Liguori
2012-03-19 21:20       ` Michael S. Tsirkin
2012-03-19 21:29       ` Michael S. Tsirkin
2012-03-19 22:13         ` Anthony Liguori
2012-03-19 23:52           ` Rusty Russell
2012-03-20  0:39             ` Anthony Liguori
2012-06-18 12:05             ` Michael S. Tsirkin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4F6786C5.3080902@codemonkey.ws \
    --to=anthony@codemonkey.ws \
    --cc=aik@ozlabs.ru \
    --cc=aliguori@us.ibm.com \
    --cc=avi@redhat.com \
    --cc=kvm@vger.kernel.org \
    --cc=mst@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=rusty@rustcorp.com.au \
    --cc=virtualization@lists.linux-foundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).