On Mon, Nov 14, 2022 at 10:21:50AM -0500, Matthew Rosato wrote:
> On 11/14/22 9:59 AM, Jason Gunthorpe wrote:
> > On Mon, Nov 14, 2022 at 09:55:21AM -0500, Matthew Rosato wrote:
> >>>> AFAICT there is no equivalent means to specify
> >>>> vfio_iommu_type1.dm
On Mon, Nov 14, 2022 at 09:55:21AM -0500, Matthew Rosato wrote:
> >> AFAICT there is no equivalent means to specify
> >> vfio_iommu_type1.dma_entry_limit when using iommufd; looks like
> >> we'll just always get the default 65535.
> >
> > No, there is no arbitary limit on iommufd
>
> Yeah,
On Fri, Nov 11, 2022 at 12:16:02PM +0800, Yi Liu wrote:
> > +#if IS_ENABLED(CONFIG_IOMMUFD_VFIO_CONTAINER)
> > +MODULE_ALIAS_MISCDEV(VFIO_MINOR);
> > +MODULE_ALIAS("devname:vfio/vfio");
>
> will this line also result in systemd to create this devnodes at boot
> based on the module info even if
On Fri, Nov 11, 2022 at 12:12:36PM +0800, Yi Liu wrote:
> > +int vfio_iommufd_bind(struct vfio_device *vdev, struct iommufd_ctx *ictx)
> > +{
> > + u32 ioas_id;
> > + u32 device_id;
> > + int ret;
> > +
> > + lockdep_assert_held(>dev_set->lock);
> > +
> > + /*
> > +* If the driver
On Mon, Nov 14, 2022 at 08:51:58PM +0800, Yi Liu wrote:
> Our side, Yu He, Lixiao Yang has done below tests on Intel platform with
> the above kernel, results are:
>
> 1) GVT-g test suit passed, Intel iGFx passthrough passed.
>
> 2) NIC passthrough test with different guest memory (1G/4G),
On Thu, Nov 10, 2022 at 10:01:13PM -0500, Matthew Rosato wrote:
> On 11/7/22 7:52 PM, Jason Gunthorpe wrote:
> > This series provides an alternative container layer for VFIO implemented
> > using iommufd. This is optional, if CONFIG_IOMMUFD is not set then it will
>
On Wed, Nov 09, 2022 at 11:28:22AM -0700, Alex Williamson wrote:
> > > > I'd be much more comfortable with this as a system wide iommufd flag
> > > > if we also tied it to do some demonstration of privilege - eg a
> > > > requirement to open iommufd with CAP_SYS_RAWIO for instance.
> > >
> > >
On Thu, Nov 10, 2022 at 06:57:57AM +, Tian, Kevin wrote:
> > + /*
> > +* Emulation for NOIMMU is imperfect in that VFIO blocks almost all
> > +* other ioctls. We let them keep working but they mostly fail since no
> > +* IOAS should exist.
> > +*/
> > + if
On Thu, Nov 10, 2022 at 03:11:16AM +, Tian, Kevin wrote:
> > From: Jason Gunthorpe
> > Sent: Tuesday, November 8, 2022 8:53 AM
> >
> > +
> > +int vfio_iommufd_bind(struct vfio_device *vdev, struct iommufd_ctx *ictx)
> > +{
> > + u32 ioas_
On Wed, Nov 09, 2022 at 10:18:09AM -0700, Alex Williamson wrote:
> DPDK supports no-iommu mode.
Er? Huh? How? I thought no-iommu was for applications that didn't do
DMA? How is DPDK getting packets in/out without DMA? I guess it snoops
in /proc/ or something to learn PFNs of mlock'd memory?
>
On Tue, Nov 08, 2022 at 11:18:03PM +0800, Yi Liu wrote:
> On 2022/11/8 17:19, Nicolin Chen wrote:
> > On Mon, Nov 07, 2022 at 08:52:44PM -0400, Jason Gunthorpe wrote:
> >
> > > This is on github:
> > > https://github.com/jgunthorpe/linux/commits/vfio_iommufd
>
On Wed, Nov 09, 2022 at 03:21:29AM +, Tian, Kevin wrote:
> > From: Jason Gunthorpe
> > Sent: Wednesday, November 9, 2022 9:05 AM
> >
> > On Tue, Nov 08, 2022 at 03:55:20PM -0700, Alex Williamson wrote:
> >
> > > > > So why exactly isn't thi
On Wed, Nov 09, 2022 at 09:03:52AM +, Tian, Kevin wrote:
> every mail in this series is shown thrice in lore:
>
> https://lore.kernel.org/all/0-v2-65016290f146+33e-vfio_iommufd_...@nvidia.com/
>
> not sure what caused it but it's annoying to check the conversation there.
It is sort of a
On Tue, Nov 08, 2022 at 03:55:20PM -0700, Alex Williamson wrote:
> > > So why exactly isn't this an issue for VDPA? Are we just burying our
> > > head in the sand that such platforms exists and can still be useful
> > > given the appropriate risk vs reward trade-off?
> >
> > Simply that
On Tue, Nov 08, 2022 at 03:28:31PM -0700, Alex Williamson wrote:
> Perhaps this should have been obvious, but I'm realizing that
> vfio-noiommu mode is completely missing without VFIO_CONTAINER, which
> seems a barrier to deprecating VFIO_CONTAINER and perhaps makes it a
Yes, it is the same as
On Tue, Nov 08, 2022 at 03:41:25PM +0800, Yi Liu wrote:
> On 2022/11/8 14:10, Nicolin Chen wrote:
> > On Mon, Nov 07, 2022 at 08:52:51PM -0400, Jason Gunthorpe wrote:
> >
> > > @@ -795,6 +800,10 @@ static int vfio_device_first_open(struct vfio_device
> > > *de
On Mon, Nov 07, 2022 at 10:10:59PM -0800, Nicolin Chen wrote:
> > @@ -812,6 +821,7 @@ static int vfio_device_first_open(struct vfio_device
> > *device)
> > device->kvm = NULL;
> > if (device->group->container)
> > vfio_group_unuse_container(device->group);
> > +
On Tue, Nov 08, 2022 at 06:33:23AM +1000, Dave Airlie wrote:
> At plumbers we decided a direction, I think the direction is good, if
> there is refactoring to be done, I'd rather it was done in tree with a
> clear direction.
>
> Coming in now and saying we should go down a different path isn't
>
- New patch to fix vfio modaliases when vfio container is disabled
- Add a dmesg once when the iommufd provided /dev/vfio/vfio is opened
to signal that iommufd is providing this
v1: https://lore.kernel.org/r/0-v1-4991695894d8+211-vfio_iommufd_...@nvidia.com
Jason Gunthorpe (11):
vfio: Move
This error unwind is getting complicated. Move all the code into two
pair'd function. The functions should be called when the open_count == 1
after incrementing/before decrementing.
Reviewed-by: Kevin Tian
Reviewed-by: Yi Liu
Tested-by: Nicolin Chen
Signed-off-by: Jason Gunthorpe
---
drivers
ill be
connected to enforcing domains.
Tested-by: Nicolin Chen
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/container.c | 5 +++--
drivers/vfio/vfio.h | 2 --
drivers/vfio/vfio_main.c | 29 -
3 files changed, 19 insertions(+), 17 deletions(-)
diff --gi
by providing bind_iommufd/unbind_iommufd/attach_ioas callbacks to
implement as required. The core code provides three default callbacks for
physical mode using a real iommu_domain. This is suitable for drivers
using vfio_register_group_dev()
Tested-by: Nicolin Chen
Signed-off-by: Jason Gunthorpe
the device/group and container have been created.
Tested-by: Nicolin Chen
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/Kconfig | 35 +++
drivers/vfio/Makefile | 4 +--
drivers/vfio/vfio.h | 65 +++
3 files changed, 91 insertions
be
closed.
Thus we do not need to check this on every device FD open, just the
first. Reorganize the code so that only the first open and last close
manages the container.
Reviewed-by: Kevin Tian
Reviewed-by: Yi Liu
Tested-by: Nicolin Chen
Signed-off-by: Jason Gunthorpe
---
drivers/vfio
-by: Jason Gunthorpe
---
drivers/gpu/drm/i915/gvt/kvmgt.c | 3 +
drivers/s390/cio/vfio_ccw_ops.c | 3 +
drivers/s390/crypto/vfio_ap_ops.c | 3 +
drivers/vfio/container.c | 110 +---
drivers/vfio/iommufd.c| 58 +++
drivers/vfio/vfio.h
The miscdev is in container.c, so should these related MODULE_ALIAS
statements. This is necessary for the next patch to be able to fully
disable /dev/vfio/vfio.
Fixes: cdc71fe4ecbf ("vfio: Move container code into drivers/vfio/container.c")
Reported-by: "Liu, Yi L"
Signed-off
allow_unsafe_interrupts is in the current state of the world. If vfio
container is disabled then this option will not be available to the user.
Tested-by: Nicolin Chen
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/vfio.h | 2 ++
drivers/vfio/vfio_iommu_type1.c | 5 ++---
drivers/vfio/vfio_main.c
These functions don't really assign anything anymore, they just increment
some refcounts and do a sanity check. Call them
vfio_group_[un]use_container()
Reviewed-by: Kevin Tian
Reviewed-by: Yi Liu
Tested-by: Nicolin Chen
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/container.c | 14
.
Tested-by: Nicolin Chen
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/Kconfig | 1 +
drivers/vfio/container.c | 7 +++-
drivers/vfio/vfio.h | 2 +
drivers/vfio/vfio_main.c | 86 +---
4 files changed, 80 insertions(+), 16 deletions(-)
diff
-by: Nicolin Chen
Signed-off-by: Jason Gunthorpe
---
drivers/iommu/iommufd/Kconfig | 12
drivers/iommu/iommufd/main.c | 36 +++
2 files changed, 48 insertions(+)
diff --git a/drivers/iommu/iommufd/Kconfig b/drivers/iommu/iommufd/Kconfig
index 399a2edeaef6de
On Wed, Nov 02, 2022 at 03:28:20PM +0800, Yi Liu wrote:
> On 2022/10/26 02:50, Jason Gunthorpe wrote:
> > This makes VFIO_GROUP_SET_CONTAINER accept both a vfio container FD and an
> > iommufd.
> >
> > In iommufd mode an IOAS will exist after the SET_CONTAINER, but
On Mon, Nov 07, 2022 at 11:05:08AM -0700, Alex Williamson wrote:
> After further consideration... I don't think the option on vfio-main
> makes sense, basically for the same reason that the original option
> existed on the IOMMU backend rather than vfio-core. The option
> describes a means to
On Mon, Nov 07, 2022 at 05:53:55PM +0200, Oded Gabbay wrote:
> On Mon, Nov 7, 2022 at 4:10 PM Jason Gunthorpe wrote:
> >
> > On Mon, Nov 07, 2022 at 04:02:01PM +0200, Oded Gabbay wrote:
> > > On Mon, Nov 7, 2022 at 3:10 PM Jason Gunthorpe wrote:
> > > >
> &
On Sun, Nov 06, 2022 at 11:02:22PM +0200, Oded Gabbay wrote:
> Another important change is that I have reverted back to use IDR for minor
> handling instead of xarray. This is because I have found that xarray doesn't
> handle well the scenario where you allocate a NULL entry and then exchange it
>
On Mon, Nov 07, 2022 at 08:18:53AM -0700, Alex Williamson wrote:
> On Mon, 7 Nov 2022 09:19:43 -0400
> Jason Gunthorpe wrote:
>
> > On Mon, Oct 31, 2022 at 04:45:26PM -0600, Alex Williamson wrote:
> >
> > > > It is one idea, it depends how literal you want to
On Mon, Nov 07, 2022 at 04:02:01PM +0200, Oded Gabbay wrote:
> On Mon, Nov 7, 2022 at 3:10 PM Jason Gunthorpe wrote:
> >
> > On Mon, Nov 07, 2022 at 03:01:08PM +0200, Oded Gabbay wrote:
> > > I don't agree with your statement that it should be "a layer over top o
On Mon, Oct 31, 2022 at 04:53:11PM -0600, Alex Williamson wrote:
> On Fri, 28 Oct 2022 15:44:36 -0300
> Jason Gunthorpe wrote:
>
> > On Wed, Oct 26, 2022 at 03:31:33PM -0600, Alex Williamson wrote:
> > > On Tue, 25 Oct 2022 15:50:45 -0300
>
On Mon, Oct 31, 2022 at 04:45:26PM -0600, Alex Williamson wrote:
> > It is one idea, it depends how literal you want to be on "module
> > parameters are ABI". IMHO it is a weak form of ABI and the need of
> > this paramter in particular is not that common in modern times, AFAIK.
> >
> > So
On Mon, Nov 07, 2022 at 03:01:08PM +0200, Oded Gabbay wrote:
> I don't agree with your statement that it should be "a layer over top of DRM".
> Anything on top of DRM is a device driver.
> Accel is not a device driver, it is a new type of drm minor / drm driver.
Yeah, I still think this is not
On Thu, Nov 03, 2022 at 10:39:36PM +0200, Oded Gabbay wrote:
> On Thu, Nov 3, 2022 at 3:31 PM Oded Gabbay wrote:
> >
> > On Thu, Nov 3, 2022 at 2:31 AM Greg Kroah-Hartman
> > wrote:
> > >
> > > On Wed, Nov 02, 2022 at 10:34:03PM +0200, Oded Gabbay wrote:
> > > > --- /dev/null
> > > > +++
On Tue, Nov 01, 2022 at 08:21:20AM +, Tian, Kevin wrote:
> > From: Jason Gunthorpe
> > Sent: Wednesday, October 26, 2022 2:51 AM
> >
> > +int vfio_iommufd_bind(struct vfio_device *vdev, struct iommufd_ctx *ictx)
> > +{
> > + u32 ioas_id;
&g
On Thu, Nov 03, 2022 at 04:38:16AM +, Tian, Kevin wrote:
> > From: Jason Gunthorpe
> > Sent: Tuesday, November 1, 2022 8:26 PM
> > And this:
> >
> > /*
> > * If the device does not have
> > IOMMU_CAP_ENFORCE_CACHE_COHERENCY then
> >
On Tue, Nov 01, 2022 at 10:37:14PM +0800, Yi Liu wrote:
> > @@ -784,12 +791,6 @@ static struct file *vfio_device_open(struct
> > vfio_device *device)
> > struct file *filep;
> > int ret;
> > - mutex_lock(>group->group_lock);
> > - ret = vfio_device_assign_container(device);
> > -
On Tue, Nov 01, 2022 at 08:41:56AM +, Tian, Kevin wrote:
> > From: Jason Gunthorpe
> > Sent: Wednesday, October 26, 2022 2:51 AM
> >
> > if VFIO
> > +config VFIO_CONTAINER
> > + bool "Support for the VFIO container /dev/vfio/vfio"
> >
On Tue, Nov 01, 2022 at 08:37:39AM +, Tian, Kevin wrote:
> > From: Jason Gunthorpe
> > Sent: Wednesday, October 26, 2022 2:51 AM
> >
> > Emulated VFIO devices are calling vfio_register_emulated_iommu_dev() and
> > consist of all the mdev drivers.
> >
&g
On Tue, Nov 01, 2022 at 08:09:52AM +, Tian, Kevin wrote:
> > From: Jason Gunthorpe
> > Sent: Wednesday, October 26, 2022 2:51 AM
> >
> > menuconfig VFIO
> > tristate "VFIO Non-Privileged userspace driver framework"
> > select IOMMU_API
&
On Tue, Nov 01, 2022 at 07:52:23AM +, Tian, Kevin wrote:
> > IOMMU_CAP_ENFORCE_CACHE_COHERENCY we know what the
> > iommu_domain will be
> > capable of without having to create it. Use this to compute
>
> it's worth noting that the prerequisite is that vfio always enforces
> cache coherency
On Tue, Nov 01, 2022 at 07:38:47AM +, Tian, Kevin wrote:
> > From: Jason Gunthorpe
> > Sent: Wednesday, October 26, 2022 2:17 AM
> >
> > +err_container:
> > + vfio_device_unassign_container(device);
> > err_module_put:
> > device->kvm = NU
On Tue, Nov 01, 2022 at 07:33:30AM +, Tian, Kevin wrote:
> > + /*
> > +* Here we pass the KVM pointer with the group under the read lock.
>
> Now the read lock is replaced by mutex. Let's correct it when moving this
> piece of code.
Done, thanks
Jason
On Tue, Nov 01, 2022 at 02:19:04AM -0700, Nicolin Chen wrote:
> On Tue, Nov 01, 2022 at 08:09:52AM +, Tian, Kevin wrote:
>
> > > From: Jason Gunthorpe
> > > Sent: Wednesday, October 26, 2022 2:51 AM
> > >
> > > menuconfig VFIO
> > >
On Tue, Nov 01, 2022 at 11:04:38AM +0800, Yi Liu wrote:
> On 2022/11/1 07:24, Jason Gunthorpe wrote:
> > On Mon, Oct 31, 2022 at 08:25:39PM +0800, Yi Liu wrote:
> > > > There is something wrong with the test suite that it isn't covering
> > > > the above,
On Mon, Oct 31, 2022 at 08:25:39PM +0800, Yi Liu wrote:
> > There is something wrong with the test suite that it isn't covering
> > the above, I'm going to look into that today.
>
> sounds to be the cause. I didn't see any significant change in vfio_main.c
> that may fail gvt. So should the
On Mon, Oct 31, 2022 at 06:38:45PM +0800, Yi Liu wrote:
> Hi Jason,
>
> On 2022/10/26 02:17, Jason Gunthorpe wrote:
> > This series provides an alternative container layer for VFIO implemented
> > using iommufd. This is optional, if CONFIG_IOMMUFD is not set then it wil
On Wed, Oct 26, 2022 at 03:31:33PM -0600, Alex Williamson wrote:
> On Tue, 25 Oct 2022 15:50:45 -0300
> Jason Gunthorpe wrote:
>
> > If the VFIO container is compiled out, give a kconfig option for iommufd
> > to provide the miscdev node with the same name and permiss
On Wed, Oct 26, 2022 at 03:24:42PM -0600, Alex Williamson wrote:
> On Tue, 25 Oct 2022 15:17:10 -0300
> Jason Gunthorpe wrote:
>
> > This legacy module knob has become uAPI, when set on the vfio_iommu_type1
> > it disables some security protections in the iommu drivers.
On Thu, Oct 20, 2022 at 02:56:39PM -0700, Dan Williams wrote:
> A 'struct dev_pagemap' (pgmap) represents a collection of ZONE_DEVICE
> pages. The pgmap is a reference counted object that serves a similar
> role as a 'struct request_queue'. Live references are obtained for each
> in flight request
the device/group and container have been created.
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/Kconfig | 37
drivers/vfio/Makefile | 4 +--
drivers/vfio/vfio.h | 65 +++
3 files changed, 92 insertions(+), 14 deletions(-)
diff
by providing bind_iommufd/unbind_iommufd/attach_ioas callbacks to
implement as required. The core code provides three default callbacks for
physical mode using a real iommu_domain. This is suitable for drivers
using vfio_register_group_dev()
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/Makefile
lly present on the
group.
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/Kconfig | 1 +
drivers/vfio/container.c | 7 ++--
drivers/vfio/vfio.h | 2 ++
drivers/vfio/vfio_main.c | 76
4 files changed, 69 insertions(+), 17 deletions(-)
diff --
vfio_register_emulated_iommu_dev() does.
Emulated drivers are where it is more likely to see variation in the
iommfd support ops. For instance IDXD will probably need to setup both a
iommfd_device context linked to a PASID and an iommufd_access context to
support all their mdev operations.
Signed-off-by: Jason Gunthorpe
-by: Jason Gunthorpe
---
drivers/iommu/iommufd/Kconfig | 12
drivers/iommu/iommufd/main.c | 35 ---
2 files changed, 44 insertions(+), 3 deletions(-)
diff --git a/drivers/iommu/iommufd/Kconfig b/drivers/iommu/iommufd/Kconfig
index f0a2012234fa09
This error unwind is getting complicated. Move all the code into two
pair'd function. The functions should be called when the open_count == 1
after incrementing/before decrementing.
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/vfio_main.c | 95 ++--
1 file
be
closed.
Thus we do not need to check this on evey device FD open, just the
first. Reorganize the code so that only the first open and last close
manages the container.
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/container.c | 4 ++--
drivers/vfio/vfio_main.c | 18 --
2 files
lve the ordering problems.
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/container.c | 5 +++--
drivers/vfio/vfio.h | 2 --
drivers/vfio/vfio_main.c | 27 ++-
3 files changed, 17 insertions(+), 17 deletions(-)
diff --git a/drivers/vfio/container.c b/drivers/vfio/c
://github.com/jgunthorpe/linux/commits/vfio_iommufd
It requires the iommufd series:
https://lore.kernel.org/r/0-v3-402a7d6459de+24b-iommufd_...@nvidia.com
Jason Gunthorpe (10):
vfio: Move vfio_device driver open/close code to a function
vfio: Move vfio_device_assign_container
These functions don't really assign anything anymore, they just increment
some refcounts and do a sanity check. Call them
vfio_group_[un]use_container()
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/container.c | 14 ++
drivers/vfio/vfio.h | 4 ++--
drivers/vfio/vfio_main.c
This legacy module knob has become uAPI, when set on the vfio_iommu_type1
it disables some security protections in the iommu drivers. Move the
storage for this knob to vfio_main.c so that iommufd can access it too.
Signed-off-by: Jason Gunthorpe
---
drivers/vfio/vfio.h | 2
On Tue, Oct 25, 2022 at 10:21:34AM -0400, Alex Deucher wrote:
> E.g., the kfd node provides platform level compute
> topology information; e.g., the NUMA details for connected GPUs and
> CPUs, non-GPU compute node information, cache level topologies, etc.
See, this is exactly what I'm talking
On Tue, Oct 25, 2022 at 12:27:11PM +1000, Dave Airlie wrote:
> The userspace for those is normally bespoke like ROCm, which uses
> amdkfd, and amdkfd doesn't operate like most device files from what I
> know, so I'm not sure we'd want it to operate as an accel device.
I intensely dislike this
On Wed, Oct 19, 2022 at 01:57:41PM -0400, Eric Farman wrote:
> On Wed, 2022-10-19 at 14:15 -0300, Jason Gunthorpe wrote:
> > On Wed, Oct 19, 2022 at 06:21:34PM +0200, Eric Farman wrote:
> >
> > > /*
> > > * Initialize a vfio_device so it can be registered to vf
On Wed, Oct 19, 2022 at 06:21:28PM +0200, Eric Farman wrote:
> This series attempts to address the oddities/shortcomings of vfio-ccw, such
> that vfio-ccw can use the same helpers as everyone else, and the tricks that
> were implemented by the other drivers can be removed. It is built on 6.1-rc1,
em from driver release callbacks.
>
> Signed-off-by: Eric Farman
Reviewed-by: Jason Gunthorpe
Jason
On Wed, Oct 19, 2022 at 06:21:34PM +0200, Eric Farman wrote:
> /*
> * Initialize a vfio_device so it can be registered to vfio core.
> - *
> - * Only vfio-ccw driver should call this interface.
> */
> int vfio_init_device(struct vfio_device *device, struct device *dev,
>
> > - dma_buf_unmap_attachment(umem_dmabuf->attach, umem_dmabuf->sgt,
> > -DMA_BIDIRECTIONAL);
> > + dma_buf_unmap_attachment_unlocked(umem_dmabuf->attach, umem_dmabuf->sgt,
> > + DMA_BIDIRECTIONAL);
> >
> > umem_dmabuf->sgt = NULL;
> > }
>
> Jason / Leon,
>
> Could you please ack this patch?
You probably don't need it, for something so simple, but sure
Acked-by: Jason Gunthorpe
Jason
On Thu, Oct 06, 2022 at 06:45:25AM -0600, Jason A. Donenfeld wrote:
> Hi Kees,
>
> On Wed, Oct 05, 2022 at 09:16:50PM -0700, Kees Cook wrote:
> > On Wed, Oct 05, 2022 at 11:48:40PM +0200, Jason A. Donenfeld wrote:
> > > Rather than incurring a division or requesting too many random bytes for
> >
On Wed, Oct 05, 2022 at 11:48:42PM +0200, Jason A. Donenfeld wrote:
> index 14392c942f49..499a425a3379 100644
> --- a/drivers/infiniband/hw/cxgb4/cm.c
> +++ b/drivers/infiniband/hw/cxgb4/cm.c
> @@ -734,7 +734,7 @@ static int send_connect(struct c4iw_ep *ep)
>
On Thu, Oct 06, 2022 at 07:05:48AM -0600, Jason A. Donenfeld wrote:
> > > diff --git a/drivers/infiniband/ulp/ipoib/ipoib_cm.c
> > > b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
> > > index fd9d7f2c4d64..a605cf66b83e 100644
> > > --- a/drivers/infiniband/ulp/ipoib/ipoib_cm.c
> > > +++
On Wed, Oct 05, 2022 at 04:03:56PM -0600, Alex Williamson wrote:
> We can't have a .remove callback that does nothing, this breaks
> removing the device while it's in use. Once we have the
> vfio_unregister_group_dev() fix below, we'll block until the device is
> unused, at which point
On Wed, Oct 05, 2022 at 02:17:17PM -0600, Alex Williamson wrote:
> > diff --git a/drivers/gpu/drm/i915/gvt/kvmgt.c
> > b/drivers/gpu/drm/i915/gvt/kvmgt.c
> > index 41bba40feef8f4..9003145adb5a93 100644
> > --- a/drivers/gpu/drm/i915/gvt/kvmgt.c
> > +++ b/drivers/gpu/drm/i915/gvt/kvmgt.c
> > @@
On Thu, Sep 29, 2022 at 10:55:19AM -0600, Alex Williamson wrote:
> Hi Kevin,
>
> This introduced the regression discovered here:
>
> https://lore.kernel.org/all/20220928125650.0a2ea297.alex.william...@redhat.com/
>
> Seems we're not releasing the resources when removing an mdev. This is
> a
nvert to use
vfio_register_emulated_iommu_dev")
Reported-by: Alex Williamson
Signed-off-by: Jason Gunthorpe
---
drivers/gpu/drm/i915/gvt/kvmgt.c | 1 +
1 file changed, 1 insertion(+)
Should go through Alex's tree.
diff --git a/drivers/gpu/drm/i915/gvt/kvmgt.c b/drivers/gpu/drm/i915/gvt/kvm
On Thu, Sep 01, 2022 at 11:20:52AM -0300, Jason Gunthorpe wrote:
> This series adds support for DMABUF when creating a devx umem. devx umems
> are quite similar to MR's execpt they cannot be revoked, so this uses the
> dmabuf pinned memory flow. Several mlx5dv flows require umem and cann
On Mon, Sep 26, 2022 at 04:03:06PM +1000, Alistair Popple wrote:
> Since 27674ef6c73f ("mm: remove the extra ZONE_DEVICE struct page
> refcount") device private pages have no longer had an extra reference
> count when the page is in use. However before handing them back to the
> owning device
On Tue, Sep 20, 2022 at 10:55:40PM +, Tian, Kevin wrote:
> > From: Alex Williamson
> > Sent: Wednesday, September 21, 2022 4:27 AM
> >
> > On Fri, 9 Sep 2022 18:22:47 +0800
> > Kevin Tian wrote:
> >
> > > From: Yi Liu
> > >
> > > and replace kref. With it a 'vfio-dev/vfioX' node is
On Fri, Sep 09, 2022 at 08:42:25AM +, Tian, Kevin wrote:
> I think it's quite common to have an alloc() helper initialize refcount, e.g.
> vfio_group_alloc() both initialize its user refcount and also call
> device_initialize() to gets kref initialized. Similar example in
>
On Fri, Sep 09, 2022 at 06:24:35AM -0700, Christoph Hellwig wrote:
> On Wed, Sep 07, 2022 at 01:12:52PM -0300, Jason Gunthorpe wrote:
> > The PCI offset is some embedded thing - I've never seen it in a server
> > platform.
>
> That's not actually true, e.g. some power sy
On Thu, Sep 08, 2022 at 11:39:07AM +0200, Eric Auger wrote:
> >> I am not totally clear about remaining 'struct device *dev;' in
> >> vfio_device struct. I see it used in some places. Is it supposed to
> >> disappear at some point?
> >
> > no, Eric. *dev will not disappear, it stores the dev
On Wed, Sep 07, 2022 at 05:31:14PM +0100, Robin Murphy wrote:
> The only trouble is that it's not geared for *PCI* P2P when that may or may
> not happen entirely upstream of IOMMU translation.
This is why PCI users have to call the pci_distance stuff before using
dma_map_resource(), it ensures
On Wed, Sep 07, 2022 at 08:32:23AM -0700, Christoph Hellwig wrote:
> On Wed, Sep 07, 2022 at 12:23:28PM -0300, Jason Gunthorpe wrote:
> > 2) DMABUF abuses dma_map_resource() for P2P and thus doesn't work in
> > certain special cases.
>
> Not just certain special cases
On Wed, Sep 07, 2022 at 07:29:58AM -0700, Christoph Hellwig wrote:
> On Wed, Sep 07, 2022 at 09:33:11AM -0300, Jason Gunthorpe wrote:
> > Yes, you said that, and I said that when the AMD driver first merged
> > it - but it went in anyhow and now people are using it in a bunc
On Wed, Sep 07, 2022 at 05:05:57AM -0700, Christoph Hellwig wrote:
> On Tue, Sep 06, 2022 at 08:48:28AM -0300, Jason Gunthorpe wrote:
> > Right, this whole thing is the "standard" that dmabuf has adopted
> > instead of the struct pages. Once the AMD GPU driver started doi
On Wed, Sep 07, 2022 at 04:55:18AM -0700, Christoph Hellwig wrote:
> On Wed, Sep 07, 2022 at 12:43:30AM +, Tian, Kevin wrote:
> > > From: Christoph Hellwig
> > > Sent: Tuesday, September 6, 2022 5:42 PM
> > >
> > > What is the point? This adds indirect calls, and actually creates
> > > more
On Tue, Sep 06, 2022 at 03:34:02PM +0300, Oded Gabbay wrote:
> > > > > + /*
> > > > > + * Since the memory being mapped is a device memory it could never
> > > > > be in
> > > > > + * CPU caches.
> > > > > + */
> > > > DMA_ATTR_SKIP_CPU_SYNC doesn't even apply to dma_map_resource, not sure
>
On Thu, Sep 01, 2022 at 09:55:08AM +0200, Christian König wrote:
> Am 01.09.22 um 01:12 schrieb Jason Gunthorpe:
> > Used to increment the refcount of the dma buf's struct file, only if the
> > refcount is not zero. Useful to allow the struct file's lifetime to
> >
On Tue, Sep 06, 2022 at 12:38:44PM +0200, Christian König wrote:
> Am 06.09.22 um 11:51 schrieb Christoph Hellwig:
> > > +{
> > > + struct vfio_pci_dma_buf *priv = dmabuf->priv;
> > > + int rc;
> > > +
> > > + rc = pci_p2pdma_distance_many(priv->vdev->pdev, >dev, 1,
> > > +
redirect setting to be bypassed in the switch.
Have mlx5 automatically use ATS in places where it makes sense.
Jason Gunthorpe (4):
net/mlx5: Add IFC bits for mkey ATS
RDMA/core: Add UVERBS_ATTR_RAW_FD
RDMA/mlx5: Add support for dmabuf to devx umem
RDMA/mlx5: Enable ATS support for MRs
Allows telling a mkey to use PCI ATS for DMA that flows through it.
Signed-off-by: Jason Gunthorpe
---
include/linux/mlx5/mlx5_ifc.h | 11 ---
1 file changed, 8 insertions(+), 3 deletions(-)
diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
index 4acd5610e96bc0
of these cases revolve around optimizing PCI P2P
transfers and avoiding bad cases where the bus just doesn't work.
Signed-off-by: Jason Gunthorpe
---
drivers/infiniband/hw/mlx5/devx.c| 37
drivers/infiniband/hw/mlx5/mlx5_ib.h | 36 +++
drivers
else stays the same.
Signed-off-by: Jason Gunthorpe
---
drivers/infiniband/hw/mlx5/devx.c| 24 +---
include/uapi/rdma/mlx5_user_ioctl_cmds.h | 1 +
2 files changed, 22 insertions(+), 3 deletions(-)
diff --git a/drivers/infiniband/hw/mlx5/devx.c
b/drivers/infiniba
This uses the same passing protocol as UVERBS_ATTR_FD (eg len = 0 data_s64
= fd), except that the FD is not required to be a uverbs object and the
core code does not covert the FD to an object handle automatically.
Access to the int fd is provided by uverbs_get_raw_fd().
Signed-off-by: Jason
201 - 300 of 1605 matches
Mail list logo