All of lore.kernel.org
 help / color / mirror / Atom feed
From: David Gibson <david@gibson.dropbear.id.au>
To: Alexey Kardashevskiy <aik@ozlabs.ru>
Cc: qemu-devel@nongnu.org, qemu-ppc@nongnu.org,
	Alex Williamson <alex.williamson@redhat.com>,
	Paolo Bonzini <pbonzini@redhat.com>
Subject: Re: [Qemu-devel] [RFC PATCH qemu] spapr_iommu, vfio, memory: Notify IOMMU about starting/stopping listening
Date: Fri, 27 May 2016 10:43:48 +1000	[thread overview]
Message-ID: <20160527004348.GL17226@voom.fritz.box> (raw)
In-Reply-To: <1463731482-688-1-git-send-email-aik@ozlabs.ru>

[-- Attachment #1: Type: text/plain, Size: 8003 bytes --]

On Fri, May 20, 2016 at 06:04:42PM +1000, Alexey Kardashevskiy wrote:
> The sPAPR TCE tables manage 2 copies when VFIO is using an IOMMU -
> a guest view of the table and a hardware TCE table. If there is no VFIO
> presense in the address space, then just the guest view is used, if

Nit: s/presense/presence/

> this is the case, it is allocated in the KVM. However since there is no
> support yet for VFIO in KVM TCE hypercalls, when we start using VFIO,
> we need to move the guest view from KVM to the userspace; and we need
> to do this for every IOMMU on a bus with VFIO devices.
> 
> This adds notify_started/notify_stopped callbacks in MemoryRegionIOMMUOps
> to notify IOMMU that listeners were set/removed. This allows IOMMU to
> take necessary steps before actual notifications happen and do proper
> cleanup when the last notifier is removed.
> 
> This implements the callbacks for the sPAPR IOMMU - notify_started()
> reallocated the guest view to the user space, notify_stopped() does
> the opposite.
> 
> This removes explicit spapr_tce_set_need_vfio() call from PCI hotplug
> path as the new callbacks do this better - they notify IOMMU at
> the exact moment when the configuration is changed, and this also
> includes the case of PCI hot unplug.
> 
> This adds MemoryRegion* to memory_region_unregister_iommu_notifier()
> as we need iommu_ops to call notify_stopped() and Notifier* does not
> store the owner.
> 
> Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
> ---
> 
> 
> Is this any better? If so, I'll repost as a part of "v17". Thanks.

I agree with Alex that this is a much better approach.  I'm sad I
didn't think of it earlier.

Reviewed-by: David Gibson <david@gibson.dropbear.id.au>

> ---
> Changes:
> v17:
> * replaced IOMMU users counting with simple QLIST_EMPTY()
> * renamed the callbacks
> * removed requirement for region_del() to be called on memory_listener_unregister()
> 
> v16:
> * added a use counter in VFIOAddressSpace->VFIOIOMMUMR
> 
> v15:
> * s/need_vfio/vfio-Users/g
> ---
>  hw/ppc/spapr_iommu.c  | 12 ++++++++++++
>  hw/ppc/spapr_pci.c    |  6 ------
>  hw/vfio/common.c      |  5 +++--
>  include/exec/memory.h |  8 +++++++-
>  memory.c              | 10 +++++++++-
>  5 files changed, 31 insertions(+), 10 deletions(-)
> 
> diff --git a/hw/ppc/spapr_iommu.c b/hw/ppc/spapr_iommu.c
> index 73bc26b..fd38006 100644
> --- a/hw/ppc/spapr_iommu.c
> +++ b/hw/ppc/spapr_iommu.c
> @@ -156,6 +156,16 @@ static uint64_t spapr_tce_get_page_sizes(MemoryRegion *iommu)
>      return 1ULL << tcet->page_shift;
>  }
>  
> +static void spapr_tce_notify_started(MemoryRegion *iommu)
> +{
> +    spapr_tce_set_need_vfio(container_of(iommu, sPAPRTCETable, iommu), true);
> +}
> +
> +static void spapr_tce_notify_stopped(MemoryRegion *iommu)
> +{
> +    spapr_tce_set_need_vfio(container_of(iommu, sPAPRTCETable, iommu), false);
> +}
> +

At some point we probably want to use this cleanup to remove the
"need_vfio" names inside the spapr code, but I don't think that's
reasonably within the scope of this patch.

>  static void spapr_tce_table_do_enable(sPAPRTCETable *tcet);
>  static void spapr_tce_table_do_disable(sPAPRTCETable *tcet);
>  
> @@ -240,6 +250,8 @@ static const VMStateDescription vmstate_spapr_tce_table = {
>  static MemoryRegionIOMMUOps spapr_iommu_ops = {
>      .translate = spapr_tce_translate_iommu,
>      .get_page_sizes = spapr_tce_get_page_sizes,
> +    .notify_started = spapr_tce_notify_started,
> +    .notify_stopped = spapr_tce_notify_stopped,
>  };
>  
>  static int spapr_tce_table_realize(DeviceState *dev)
> diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
> index 68e77b0..7c2c622 100644
> --- a/hw/ppc/spapr_pci.c
> +++ b/hw/ppc/spapr_pci.c
> @@ -1089,12 +1089,6 @@ static void spapr_phb_add_pci_device(sPAPRDRConnector *drc,
>      void *fdt = NULL;
>      int fdt_start_offset = 0, fdt_size;
>  
> -    if (object_dynamic_cast(OBJECT(pdev), "vfio-pci")) {
> -        sPAPRTCETable *tcet = spapr_tce_find_by_liobn(phb->dma_liobn);
> -
> -        spapr_tce_set_need_vfio(tcet, true);
> -    }
> -
>      if (dev->hotplugged) {
>          fdt = create_device_tree(&fdt_size);
>          fdt_start_offset = spapr_create_pci_child_dt(phb, pdev, fdt, 0);
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index 2e4f703..d1fa9ab 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -523,7 +523,8 @@ static void vfio_listener_region_del(MemoryListener *listener,
>  
>          QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
>              if (giommu->iommu == section->mr) {
> -                memory_region_unregister_iommu_notifier(&giommu->n);
> +                memory_region_unregister_iommu_notifier(giommu->iommu,
> +                                                        &giommu->n);
>                  QLIST_REMOVE(giommu, giommu_next);
>                  g_free(giommu);
>                  break;
> @@ -1040,7 +1041,7 @@ static void vfio_disconnect_container(VFIOGroup *group)
>          QLIST_REMOVE(container, next);
>  
>          QLIST_FOREACH_SAFE(giommu, &container->giommu_list, giommu_next, tmp) {
> -            memory_region_unregister_iommu_notifier(&giommu->n);
> +            memory_region_unregister_iommu_notifier(giommu->iommu, &giommu->n);
>              QLIST_REMOVE(giommu, giommu_next);
>              g_free(giommu);
>          }
> diff --git a/include/exec/memory.h b/include/exec/memory.h
> index bfb08d4..1c41eb6 100644
> --- a/include/exec/memory.h
> +++ b/include/exec/memory.h
> @@ -151,6 +151,10 @@ struct MemoryRegionIOMMUOps {
>      IOMMUTLBEntry (*translate)(MemoryRegion *iommu, hwaddr addr, bool is_write);
>      /* Returns supported page sizes */
>      uint64_t (*get_page_sizes)(MemoryRegion *iommu);
> +    /* Called when the first notifier is set */
> +    void (*notify_started)(MemoryRegion *iommu);
> +    /* Called when the last notifier is removed */
> +    void (*notify_stopped)(MemoryRegion *iommu);
>  };
>  
>  typedef struct CoalescedMemoryRange CoalescedMemoryRange;
> @@ -620,9 +624,11 @@ void memory_region_iommu_replay(MemoryRegion *mr, Notifier *n, bool is_write);
>   * memory_region_unregister_iommu_notifier: unregister a notifier for
>   * changes to IOMMU translation entries.
>   *
> + * @mr: the memory region which was observed and for which notity_stopped()
> + *      needs to be called
>   * @n: the notifier to be removed.
>   */
> -void memory_region_unregister_iommu_notifier(Notifier *n);
> +void memory_region_unregister_iommu_notifier(MemoryRegion *mr, Notifier *n);
>  
>  /**
>   * memory_region_name: get a memory region's name
> diff --git a/memory.c b/memory.c
> index d22cf5e..fcf978a 100644
> --- a/memory.c
> +++ b/memory.c
> @@ -1512,6 +1512,10 @@ bool memory_region_is_logging(MemoryRegion *mr, uint8_t client)
>  
>  void memory_region_register_iommu_notifier(MemoryRegion *mr, Notifier *n)
>  {
> +    if (mr->iommu_ops->notify_started &&
> +        QLIST_EMPTY(&mr->iommu_notify.notifiers)) {
> +        mr->iommu_ops->notify_started(mr);
> +    }
>      notifier_list_add(&mr->iommu_notify, n);
>  }
>  
> @@ -1545,9 +1549,13 @@ void memory_region_iommu_replay(MemoryRegion *mr, Notifier *n, bool is_write)
>      }
>  }
>  
> -void memory_region_unregister_iommu_notifier(Notifier *n)
> +void memory_region_unregister_iommu_notifier(MemoryRegion *mr, Notifier *n)
>  {
>      notifier_remove(n);
> +    if (mr->iommu_ops->notify_stopped &&
> +        QLIST_EMPTY(&mr->iommu_notify.notifiers)) {
> +        mr->iommu_ops->notify_stopped(mr);
> +    }
>  }
>  
>  void memory_region_notify_iommu(MemoryRegion *mr,

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

  parent reply	other threads:[~2016-05-27  0:44 UTC|newest]

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-05-04  6:52 [Qemu-devel] [PATCH qemu v16 00/19] spapr: vfio: Enable Dynamic DMA windows (DDW) Alexey Kardashevskiy
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 01/19] vfio: Delay DMA address space listener release Alexey Kardashevskiy
2016-05-05 22:39   ` Alex Williamson
2016-05-13  7:16     ` Alexey Kardashevskiy
2016-05-13 22:24       ` Alex Williamson
2016-05-25  6:34         ` David Gibson
2016-05-25 13:59           ` Alex Williamson
2016-05-26  1:00             ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 02/19] memory: Call region_del() callbacks on memory listener unregistering Alexey Kardashevskiy
2016-05-05 22:45   ` Alex Williamson
2016-05-26  1:48     ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 03/19] memory: Fix IOMMU replay base address Alexey Kardashevskiy
2016-05-26  1:50   ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 04/19] vmstate: Define VARRAY with VMS_ALLOC Alexey Kardashevskiy
2016-05-27  7:54   ` Alexey Kardashevskiy
2016-06-01  2:29     ` Alexey Kardashevskiy
2016-06-01  8:11       ` Paolo Bonzini
2016-06-02  0:43         ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 05/19] vfio: Check that IOMMU MR translates to system address space Alexey Kardashevskiy
2016-05-26  1:51   ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 06/19] spapr_pci: Use correct DMA LIOBN when composing the device tree Alexey Kardashevskiy
2016-05-26  3:17   ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 07/19] spapr_iommu: Move table allocation to helpers Alexey Kardashevskiy
2016-05-26  3:32   ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 08/19] spapr_iommu: Introduce "enabled" state for TCE table Alexey Kardashevskiy
2016-05-26  3:39   ` David Gibson
2016-05-27  8:01     ` Alexey Kardashevskiy
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 09/19] spapr_iommu: Finish renaming vfio_accel to need_vfio Alexey Kardashevskiy
2016-05-26  3:18   ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 10/19] spapr_iommu: Migrate full state Alexey Kardashevskiy
2016-05-26  4:01   ` David Gibson
2016-05-31  8:19     ` Alexey Kardashevskiy
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 11/19] spapr_iommu: Add root memory region Alexey Kardashevskiy
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 12/19] spapr_pci: Reset DMA config on PHB reset Alexey Kardashevskiy
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 13/19] memory: Add reporting of supported page sizes Alexey Kardashevskiy
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 14/19] vfio: spapr: Add DMA memory preregistering (SPAPR IOMMU v2) Alexey Kardashevskiy
2016-05-13 22:25   ` Alex Williamson
2016-05-16  1:10     ` Alexey Kardashevskiy
2016-05-16 20:20       ` Alex Williamson
2016-05-26  4:53         ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 15/19] spapr_pci: Add and export DMA resetting helper Alexey Kardashevskiy
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 16/19] vfio: Add host side DMA window capabilities Alexey Kardashevskiy
2016-05-13 22:25   ` Alex Williamson
2016-05-27  0:36     ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 17/19] spapr_iommu, vfio, memory: Notify IOMMU about starting/stopping being used by VFIO Alexey Kardashevskiy
2016-05-13 22:26   ` Alex Williamson
2016-05-16  8:35     ` Alexey Kardashevskiy
2016-05-16 20:13       ` Alex Williamson
2016-05-20  8:04         ` [Qemu-devel] [RFC PATCH qemu] spapr_iommu, vfio, memory: Notify IOMMU about starting/stopping listening Alexey Kardashevskiy
2016-05-20 15:19           ` Alex Williamson
2016-05-27  0:43           ` David Gibson [this message]
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 18/19] vfio/spapr: Create DMA window dynamically (SPAPR IOMMU v2) Alexey Kardashevskiy
2016-05-13 22:26   ` Alex Williamson
2016-05-16  4:52     ` Alexey Kardashevskiy
2016-05-16 20:20       ` Alex Williamson
2016-05-27  0:50         ` David Gibson
2016-05-27  3:49         ` Alexey Kardashevskiy
2016-05-27  4:05           ` David Gibson
2016-05-04  6:52 ` [Qemu-devel] [PATCH qemu v16 19/19] spapr_pci/spapr_pci_vfio: Support Dynamic DMA Windows (DDW) Alexey Kardashevskiy
2016-05-13  8:41   ` Bharata B Rao
2016-05-13  8:49     ` Bharata B Rao
2016-05-16  6:25     ` Alexey Kardashevskiy
2016-05-17  5:32       ` Bharata B Rao
2016-05-27  4:44         ` David Gibson
2016-05-27  5:49           ` Bharata B Rao
2016-06-01  3:32             ` Bharata B Rao
2016-05-27  4:42     ` David Gibson
2016-05-13  4:54 ` [Qemu-devel] [PATCH qemu v16 00/19] spapr: vfio: Enable Dynamic DMA windows (DDW) Alexey Kardashevskiy
2016-05-13  5:36   ` Alex Williamson

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=20160527004348.GL17226@voom.fritz.box \
    --to=david@gibson.dropbear.id.au \
    --cc=aik@ozlabs.ru \
    --cc=alex.williamson@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-ppc@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.