qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Alex Williamson <alex.williamson@redhat.com>
To: David Gibson <david@gibson.dropbear.id.au>
Cc: aik@ozlabs.ru, pbonzini@redhat.com, qemu-devel@nongnu.org,
	agraf@suse.de, mst@redhat.com
Subject: Re: [Qemu-devel] [PATCH 8/8] vfio: Create VFIOAddressSpace objects as needed
Date: Mon, 13 May 2013 15:33:06 -0600	[thread overview]
Message-ID: <1368480786.5520.81.camel@ul30vt.home> (raw)
In-Reply-To: <1368442465-14363-9-git-send-email-david@gibson.dropbear.id.au>

On Mon, 2013-05-13 at 20:54 +1000, David Gibson wrote:
> So far, VFIO has a notion of different logical DMA address spaces, but
> only ever uses one (system memory).  This patch extends this, creating
> new VFIOAddressSpace objects as necessary, according to the AddressSpace
> reported by the PCI subsystem for this device's DMAs.
> 
> This isn't enough yet to support guest side IOMMUs with VFIO, but it does
> mean we could now support VFIO devices on, for example, a guest side PCI
> host bridge which maps system memory at somewhere other than 0 in PCI
> space.
> 
> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> ---
>  hw/misc/vfio.c |   36 +++++++++++++++++++++++++++++-------
>  1 file changed, 29 insertions(+), 7 deletions(-)
> 
> diff --git a/hw/misc/vfio.c b/hw/misc/vfio.c
> index b1e9220..3850d39 100644
> --- a/hw/misc/vfio.c
> +++ b/hw/misc/vfio.c
> @@ -116,9 +116,10 @@ enum {
>  typedef struct VFIOAddressSpace {
>      AddressSpace *as;
>      QLIST_HEAD(, VFIOContainer) containers;
> +    QLIST_ENTRY(VFIOAddressSpace) list;
>  } VFIOAddressSpace;
>  
> -static VFIOAddressSpace vfio_address_space_memory;
> +QLIST_HEAD(, VFIOAddressSpace) vfio_address_spaces;
>  
>  struct VFIOGroup;
>  
> @@ -2635,6 +2636,23 @@ static void vfio_address_space_init(VFIOAddressSpace *vas, AddressSpace *as)
>      QLIST_INIT(&vas->containers);
>  }
>  
> +static VFIOAddressSpace *vfio_address_space_get(AddressSpace *as)

vfio_get_address_space is a better match for the rest of the code.

> +{
> +    VFIOAddressSpace *vas;
> +
> +    QLIST_FOREACH(vas, &vfio_address_spaces, list) {
> +        if (vas->as == as)
> +            return vas;
> +    }
> +
> +    /* No suitable VFIOAddressSpace, create a new one */
> +    vas = g_malloc0(sizeof(*vas));
> +    vfio_address_space_init(vas, as);
> +    QLIST_INSERT_HEAD(&vfio_address_spaces, vas, list);

Do we still need vfio_address_space_init?  Seems like it should be
rolled in here.

> +
> +    return vas;
> +}
> +
>  static int vfio_connect(VFIOGroup *group, VFIOAddressSpace *vas)
>  {
>      VFIOContainer *container;
> @@ -2727,6 +2745,8 @@ static void vfio_disconnect(VFIOGroup *group)
>      group->container = NULL;
>  
>      if (QLIST_EMPTY(&container->group_list)) {
> +        VFIOAddressSpace *vas = container->vas;
> +
>          if (container->iommu_data.release) {
>              container->iommu_data.release(container);
>          }
> @@ -2734,6 +2754,11 @@ static void vfio_disconnect(VFIOGroup *group)
>          DPRINTF("vfio_disconnect: close container->fd\n");
>          close(container->fd);
>          g_free(container);
> +
> +        if (QLIST_EMPTY(&vas->containers)) {
> +            QLIST_REMOVE(vas, list);
> +            g_free(vas);
> +        }

vfio_put_address_space?  Where there's a get...

>      }
>  }
>  
> @@ -2984,6 +3009,7 @@ static int vfio_initfn(PCIDevice *pdev)
>  {
>      VFIODevice *pvdev, *vdev = DO_UPCAST(VFIODevice, pdev, pdev);
>      VFIOGroup *group;
> +    VFIOAddressSpace *vas;
>      char path[PATH_MAX], iommu_group_path[PATH_MAX], *group_name;
>      ssize_t len;
>      struct stat st;
> @@ -3019,12 +3045,9 @@ static int vfio_initfn(PCIDevice *pdev)
>      DPRINTF("%s(%04x:%02x:%02x.%x) group %d\n", __func__, vdev->host.domain,
>              vdev->host.bus, vdev->host.slot, vdev->host.function, groupid);
>  
> -    if (pci_iommu_as(pdev) != &address_space_memory) {
> -        error_report("vfio: DMA address space must be system memory");
> -        return -ENXIO;
> -    }
> +    vas = vfio_address_space_get(pci_iommu_as(pdev));

I don't think the structure malloc'd here will get cleaned up in all
cases on error.  Thanks,

Alex

>  
> -    group = vfio_get_group(groupid, &vfio_address_space_memory);
> +    group = vfio_get_group(groupid, vas);
>      if (!group) {
>          error_report("vfio: failed to get group %d", groupid);
>          return -ENOENT;
> @@ -3244,7 +3267,6 @@ static const TypeInfo vfio_pci_dev_info = {
>  
>  static void register_vfio_pci_dev_type(void)
>  {
> -    vfio_address_space_init(&vfio_address_space_memory, &address_space_memory);
>      type_register_static(&vfio_pci_dev_info);
>  }
>  

  reply	other threads:[~2013-05-13 21:33 UTC|newest]

Thread overview: 28+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-05-13 10:54 [Qemu-devel] [0/8] RFC: VFIO and guest side IOMMUs, revisited David Gibson
2013-05-13 10:54 ` [Qemu-devel] [PATCH 1/8] iommu: Fix compile error in ioapic.c David Gibson
2013-05-13 12:14   ` Paolo Bonzini
2013-05-13 10:54 ` [Qemu-devel] [PATCH 2/8] pci: Don't del_subgregion on a non subregion David Gibson
2013-05-13 12:14   ` Paolo Bonzini
2013-05-13 10:54 ` [Qemu-devel] [PATCH 3/8] pci: Rework PCI iommu lifetime assumptions David Gibson
2013-05-13 12:15   ` Paolo Bonzini
2013-05-13 10:54 ` [Qemu-devel] [PATCH 4/8] pci: Use AddressSpace rather than MemoryRegion to represent PCI DMA space David Gibson
2013-05-13 12:16   ` Paolo Bonzini
2013-05-13 10:54 ` [Qemu-devel] [PATCH 5/8] pci: Introduce helper to retrieve a PCI device's DMA address space David Gibson
2013-05-13 10:54 ` [Qemu-devel] [PATCH 6/8] memory: Sanity check that no listeners remain on a destroyed AddressSpace David Gibson
2013-05-13 11:10   ` Peter Maydell
2013-05-13 11:48     ` David Gibson
2013-05-13 12:07       ` Peter Maydell
2013-05-13 12:19         ` Paolo Bonzini
2013-05-13 12:57           ` David Gibson
2013-05-13 10:54 ` [Qemu-devel] [PATCH 7/8] vfio: Introduce VFIO address spaces David Gibson
2013-05-13 21:32   ` Alex Williamson
2013-05-14  1:00     ` David Gibson
2013-05-13 10:54 ` [Qemu-devel] [PATCH 8/8] vfio: Create VFIOAddressSpace objects as needed David Gibson
2013-05-13 21:33   ` Alex Williamson [this message]
2013-05-14  1:58     ` David Gibson
2013-05-13 12:23 ` [Qemu-devel] [0/8] RFC: VFIO and guest side IOMMUs, revisited Paolo Bonzini
2013-05-13 13:13   ` David Gibson
2013-05-13 13:30     ` Paolo Bonzini
2013-05-14  2:39       ` David Gibson
2013-05-14  9:58         ` Paolo Bonzini
2013-05-15  3:55           ` David Gibson

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=1368480786.5520.81.camel@ul30vt.home \
    --to=alex.williamson@redhat.com \
    --cc=agraf@suse.de \
    --cc=aik@ozlabs.ru \
    --cc=david@gibson.dropbear.id.au \
    --cc=mst@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.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).