[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH v8 3/9] memory: provide iommu_replay_all()
From: |
Peter Xu |
Subject: |
Re: [Qemu-devel] [PATCH v8 3/9] memory: provide iommu_replay_all() |
Date: |
Thu, 6 Apr 2017 19:46:47 +0800 |
User-agent: |
Mutt/1.5.24 (2015-08-30) |
On Thu, Apr 06, 2017 at 12:52:19PM +0200, Auger Eric wrote:
> Hi Peter,
>
> On 06/04/2017 09:08, Peter Xu wrote:
> > This is an "global" version of exising memory_region_iommu_replay() - we
> s/exising/existing
> > announce the translations to all the registered notifiers, instead of a
> > specific one.
> >
> > Reviewed-by: David Gibson <address@hidden>
> > Signed-off-by: Peter Xu <address@hidden>
> > ---
> > include/exec/memory.h | 8 ++++++++
> > memory.c | 9 +++++++++
> > 2 files changed, 17 insertions(+)
> >
> > diff --git a/include/exec/memory.h b/include/exec/memory.h
> > index 07e43da..fb7dff3 100644
> > --- a/include/exec/memory.h
> > +++ b/include/exec/memory.h
> > @@ -713,6 +713,14 @@ void memory_region_iommu_replay(MemoryRegion *mr,
> > IOMMUNotifier *n,
> > bool is_write);
> >
> > /**
> > + * memory_region_iommu_replay_all: replay existing IOMMU translations
> > + * to all the notifiers registered.
> > + *
> > + * @mr: the memory region to observe
> > + */
> > +void memory_region_iommu_replay_all(MemoryRegion *mr);
> > +
> > +/**
> > * memory_region_unregister_iommu_notifier: unregister a notifier for
> > * changes to IOMMU translation entries.
> > *
> > diff --git a/memory.c b/memory.c
> > index 7496b3d..b4ed67b 100644
> > --- a/memory.c
> > +++ b/memory.c
> > @@ -1642,6 +1642,15 @@ void memory_region_iommu_replay(MemoryRegion *mr,
> > IOMMUNotifier *n,
> > }
> > }
> >
> > +void memory_region_iommu_replay_all(MemoryRegion *mr)
> > +{
> > + IOMMUNotifier *notifier;
> > +
> > + IOMMU_NOTIFIER_FOREACH(notifier, mr) {
> > + memory_region_iommu_replay(mr, notifier, false);
> It is not fully clear to me what is the consequence of setting
> is_write=false always?
I am not quite sure about it either, on why we are having the is_write
parameter on memory_region_iommu_replay(). It's there since the first
commit that introduced the interface:
commit a788f227ef7bd2912fcaacdfe13d13ece2998149
Author: David Gibson <address@hidden>
Date: Wed Sep 30 12:13:55 2015 +1000
memory: Allow replay of IOMMU mapping notifications
However imho that should be a check against page RW permissions, which
seems unecessary during replay. Or say, not sure whether below patch
would be good (as a standalone one besides current patch/series):
------8<------
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 6b33b9f..d008a4b 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -488,7 +488,7 @@ static void vfio_listener_region_add(MemoryListener
*listener,
QLIST_INSERT_HEAD(&container->giommu_list, giommu, giommu_next);
memory_region_register_iommu_notifier(giommu->iommu, &giommu->n);
- memory_region_iommu_replay(giommu->iommu, &giommu->n, false);
+ memory_region_iommu_replay(giommu->iommu, &giommu->n);
return;
}
diff --git a/include/exec/memory.h b/include/exec/memory.h
index c4fc94d..5ab0151 100644
--- a/include/exec/memory.h
+++ b/include/exec/memory.h
@@ -725,11 +725,8 @@ void memory_region_register_iommu_notifier(MemoryRegion
*mr,
*
* @mr: the memory region to observe
* @n: the notifier to which to replay iommu mappings
- * @is_write: Whether to treat the replay as a translate "write"
- * through the iommu
*/
-void memory_region_iommu_replay(MemoryRegion *mr, IOMMUNotifier *n,
- bool is_write);
+void memory_region_iommu_replay(MemoryRegion *mr, IOMMUNotifier *n);
/**
* memory_region_iommu_replay_all: replay existing IOMMU translations
diff --git a/memory.c b/memory.c
index b782d5b..155ca1c 100644
--- a/memory.c
+++ b/memory.c
@@ -1620,8 +1620,7 @@ uint64_t
memory_region_iommu_get_min_page_size(MemoryRegion *mr)
return TARGET_PAGE_SIZE;
}
-void memory_region_iommu_replay(MemoryRegion *mr, IOMMUNotifier *n,
- bool is_write)
+void memory_region_iommu_replay(MemoryRegion *mr, IOMMUNotifier *n)
{
hwaddr addr, granularity;
IOMMUTLBEntry iotlb;
@@ -1635,7 +1634,13 @@ void memory_region_iommu_replay(MemoryRegion *mr,
IOMMUNotifier *n,
granularity = memory_region_iommu_get_min_page_size(mr);
for (addr = 0; addr < memory_region_size(mr); addr += granularity) {
- iotlb = mr->iommu_ops->translate(mr, addr, is_write);
+ /*
+ * For a replay, we don't need to do permission check.
+ * Assuming a "read" operation here to make sure we will pass
+ * the check (in case we don't have write permission on the
+ * page).
+ */
+ iotlb = mr->iommu_ops->translate(mr, addr, false);
if (iotlb.perm != IOMMU_NONE) {
n->notify(n, &iotlb);
}
@@ -1653,7 +1658,7 @@ void memory_region_iommu_replay_all(MemoryRegion *mr)
IOMMUNotifier *notifier;
IOMMU_NOTIFIER_FOREACH(notifier, mr) {
- memory_region_iommu_replay(mr, notifier, false);
+ memory_region_iommu_replay(mr, notifier);
}
}
----->8------
Thanks,
-- peterx
[Qemu-devel] [PATCH v8 4/9] memory: introduce memory_region_notify_one(), Peter Xu, 2017/04/06
[Qemu-devel] [PATCH v8 5/9] memory: add MemoryRegionIOMMUOps.replay() callback, Peter Xu, 2017/04/06
[Qemu-devel] [PATCH v8 6/9] intel_iommu: use the correct memory region for device IOTLB notification, Peter Xu, 2017/04/06