linux.git
17 months agovhost-vdpa: fix use after free in vhost_vdpa_probe()
Dan Carpenter [Fri, 27 Oct 2023 12:12:54 +0000 (15:12 +0300)]
vhost-vdpa: fix use after free in vhost_vdpa_probe()

The put_device() calls vhost_vdpa_release_dev() which calls
ida_simple_remove() and frees "v".  So this call to
ida_simple_remove() is a use after free and a double free.

Fixes: ebe6a354fa7e ("vhost-vdpa: Call ida_simple_remove() when failed")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Message-Id: <cf53cb61-0699-4e36-a980-94fd4268ff00@moroto.mountain>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
17 months agovirtio_pci: Switch away from deprecated irq_set_affinity_hint
Jakub Sitnicki [Wed, 25 Oct 2023 14:53:19 +0000 (16:53 +0200)]
virtio_pci: Switch away from deprecated irq_set_affinity_hint

Since commit 65c7cdedeb30 ("genirq: Provide new interfaces for affinity
hints") irq_set_affinity_hint is being phased out.

Switch to new interfaces for setting and applying irq affinity hints.

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Message-Id: <20231025145319.380775-1-jakub@cloudflare.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
17 months agoriscv, qemu_fw_cfg: Add support for RISC-V architecture
Björn Töpel [Thu, 12 Oct 2023 10:28:52 +0000 (12:28 +0200)]
riscv, qemu_fw_cfg: Add support for RISC-V architecture

Qemu fw_cfg support was missing for RISC-V, which made it hard to do
proper vmcore dumps from qemu.

Add the missing RISC-V arch-defines.

You can now do vmcore dumps from qemu. Add "-device vmcoreinfo" to the
qemu command-line. From the qemu monitor:
  (qemu) dump-guest-memory vmcore

The vmcore can now be used, e.g., with the "crash" utility.

Acked-by: "Michael S. Tsirkin" <mst@redhat.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Tested-by: Clément Léger <cleger@rivosinc.com>
Signed-off-by: Björn Töpel <bjorn@rivosinc.com>
Message-Id: <20231012102852.234442-1-bjorn@kernel.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
17 months agovdpa_sim_blk: allocate the buffer zeroed
Stefano Garzarella [Tue, 31 Oct 2023 14:43:39 +0000 (15:43 +0100)]
vdpa_sim_blk: allocate the buffer zeroed

Deleting and recreating a device can lead to having the same
content as the old device, so let's always allocate buffers
completely zeroed out.

Fixes: abebb16254b3 ("vdpa_sim_blk: support shared backend")
Suggested-by: Qing Wang <qinwang@redhat.com>
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Message-Id: <20231031144339.121453-1-sgarzare@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
17 months agovirtio_pci: move structure to a header
Michael S. Tsirkin [Tue, 31 Oct 2023 16:02:06 +0000 (12:02 -0400)]
virtio_pci: move structure to a header

These are guest/host interfaces, so they belong in the header where e.g.
qemu will know to find them.

Note: we added a new structure as opposed to extending existing one
because someone might be relying on the size of the existing structure
staying unchanged.  Add a warning to avoid using sizeof.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
17 months agovdpa_sim: implement .reset_map support
Si-Wei Liu [Sat, 21 Oct 2023 09:25:19 +0000 (02:25 -0700)]
vdpa_sim: implement .reset_map support

In order to reduce excessive memory mapping cost in live migration and
VM reboot, it is desirable to decouple the vhost-vdpa IOTLB abstraction
from the virtio device life cycle, i.e. mappings can be kept intact
across virtio device reset. Leverage the .reset_map callback, which is
meant to destroy the iotlb on the given ASID and recreate the 1:1
passthrough/identity mapping. To be consistent, the mapping on device
creation is initiailized to passthrough/identity with PA 1:1 mapped as
IOVA. With this the device .reset op doesn't have to maintain and clean
up memory mappings by itself.

Additionally, implement .compat_reset to cater for older userspace,
which may wish to see mapping to be cleared during reset.

Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Stefano Garzarella <sgarzare@redhat.com>
Message-Id: <1697880319-4937-8-git-send-email-si-wei.liu@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: implement .reset_map driver op
Si-Wei Liu [Sat, 21 Oct 2023 09:25:18 +0000 (02:25 -0700)]
vdpa/mlx5: implement .reset_map driver op

Since commit 6f5312f80183 ("vdpa/mlx5: Add support for running with
virtio_vdpa"), mlx5_vdpa starts with preallocate 1:1 DMA MR at device
creation time. This 1:1 DMA MR will be implicitly destroyed while the
first .set_map call is invoked, in which case callers like vhost-vdpa
will start to set up custom mappings. When the .reset callback is
invoked, the custom mappings will be cleared and the 1:1 DMA MR will be
re-created.

In order to reduce excessive memory mapping cost in live migration, it
is desirable to decouple the vhost-vdpa IOTLB abstraction from the
virtio device life cycle, i.e. mappings can be kept around intact across
virtio device reset. Leverage the .reset_map callback, which is meant to
destroy the regular MR (including cvq mapping) on the given ASID and
recreate the initial DMA mapping. That way, the device .reset op runs
free from having to maintain and clean up memory mappings by itself.

Additionally, implement .compat_reset to cater for older userspace,
which may wish to see mapping to be cleared during reset.

Co-developed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Message-Id: <1697880319-4937-7-git-send-email-si-wei.liu@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovhost-vdpa: clean iotlb map during reset for older userspace
Si-Wei Liu [Sat, 21 Oct 2023 09:25:17 +0000 (02:25 -0700)]
vhost-vdpa: clean iotlb map during reset for older userspace

Using .compat_reset op from the previous patch, the buggy .reset
behaviour can be kept as-is on older userspace apps, which don't ack the
IOTLB_PERSIST backend feature. As this compatibility quirk is limited to
those drivers that used to be buggy in the past, it won't affect change
the behaviour or affect ABI on the setups with API compliant driver.

The separation of .compat_reset from the regular .reset allows
vhost-vdpa able to know which driver had broken behaviour before, so it
can apply the corresponding compatibility quirk to the individual driver
whenever needed.  Compared to overloading the existing .reset with
flags, .compat_reset won't cause any extra burden to the implementation
of every compliant driver.

[mst: squashed in two fixup commits]

Message-Id: <1697880319-4937-6-git-send-email-si-wei.liu@oracle.com>
Message-Id: <1698102863-21122-1-git-send-email-si-wei.liu@oracle.com>
Reported-by: Dragos Tatulea <dtatulea@nvidia.com>
Tested-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <1698275594-19204-1-git-send-email-si-wei.liu@oracle.com>
Reported-by: Lei Yang <leiyang@redhat.com>
Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa: introduce .compat_reset operation callback
Si-Wei Liu [Sat, 21 Oct 2023 09:25:16 +0000 (02:25 -0700)]
vdpa: introduce .compat_reset operation callback

Some device specific IOMMU parent drivers have long standing bogus
behaviour that mistakenly clean up the maps during .reset. By
definition, this is violation to the on-chip IOMMU ops (i.e. .set_map,
or .dma_map & .dma_unmap) in those offending drivers, as the removal of
internal maps is completely agnostic to the upper layer, causing
inconsistent view between the userspace and the kernel. Some userspace
app like QEMU gets around of this brokenness by proactively removing and
adding back all the maps around vdpa device reset, but such workaround
actually penaltize other well-behaved driver setup, where vdpa reset
always comes with the associated mapping cost, especially for kernel
vDPA devices (use_va=false) that have high cost on pinning. It's
imperative to rectify this behaviour and remove the problematic code
from all those non-compliant parent drivers.

However, we cannot unconditionally remove the bogus map-cleaning code
from the buggy .reset implementation, as there might exist userspace
apps that already rely on the behaviour on some setup. Introduce a
.compat_reset driver op to keep compatibility with older userspace. New
and well behaved parent driver should not bother to implement such op,
but only those drivers that are doing or used to do non-compliant
map-cleaning reset will have to.

Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Message-Id: <1697880319-4937-5-git-send-email-si-wei.liu@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovhost-vdpa: introduce IOTLB_PERSIST backend feature bit
Si-Wei Liu [Sat, 21 Oct 2023 09:25:15 +0000 (02:25 -0700)]
vhost-vdpa: introduce IOTLB_PERSIST backend feature bit

Userspace needs this feature flag to distinguish if vhost-vdpa iotlb in
the kernel can be trusted to persist IOTLB mapping across vDPA reset.
Without it, userspace has no way to tell apart if it's running on an
older kernel, which could silently drop all iotlb mapping across vDPA
reset, especially with broken parent driver implementation for the
.reset driver op. The broken driver may incorrectly drop all mappings of
its own as part of .reset, which inadvertently ends up with corrupted
mapping state between vhost-vdpa userspace and the kernel. As a
workaround, to make the mapping behaviour predictable across reset,
userspace has to pro-actively remove all mappings before vDPA reset, and
then restore all the mappings afterwards. This workaround is done
unconditionally on top of all parent drivers today, due to the parent
driver implementation issue and no means to differentiate.  This
workaround had been utilized in QEMU since day one when the
corresponding vhost-vdpa userspace backend came to the world.

There are 3 cases that backend may claim this feature bit on for:

- parent device that has to work with platform IOMMU
- parent device with on-chip IOMMU that has the expected
  .reset_map support in driver
- parent device with vendor specific IOMMU implementation with
  persistent IOTLB mapping already that has to specifically
  declare this backend feature

The reason why .reset_map is being one of the pre-condition for
persistent iotlb is because without it, vhost-vdpa can't switch back
iotlb to the initial state later on, especially for the on-chip IOMMU
case which starts with identity mapping at device creation. virtio-vdpa
requires on-chip IOMMU to perform 1:1 passthrough translation from PA to
IOVA as-is to begin with, and .reset_map is the only means to turn back
iotlb to the identity mapping mode after vhost-vdpa is gone.

The difference in behavior did not matter as QEMU unmaps all the memory
unregistering the memory listener at vhost_vdpa_dev_start( started =
false), but the backend acknowledging this feature flag allows QEMU to
make sure it is safe to skip this unmap & map in the case of vhost stop
& start cycle.

In that sense, this feature flag is actually a signal for userspace to
know that the driver bug has been solved. Not offering it indicates that
userspace cannot trust the kernel will retain the maps.

Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <1697880319-4937-4-git-send-email-si-wei.liu@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovhost-vdpa: reset vendor specific mapping to initial state in .release
Si-Wei Liu [Sat, 21 Oct 2023 09:25:14 +0000 (02:25 -0700)]
vhost-vdpa: reset vendor specific mapping to initial state in .release

Devices with on-chip IOMMU or vendor specific IOTLB implementation may
need to restore iotlb mapping to the initial or default state using the
.reset_map op, as it's desirable for some parent devices to not work
with DMA ops and maintain a simple IOMMU model with .reset_map. In
particular, device reset should not cause mapping to go away on such
IOTLB model, so persistent mapping is implied across reset. Before the
userspace process using vhost-vdpa is gone, give it a chance to reset
iotlb back to the initial state in vhost_vdpa_cleanup().

Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <1697880319-4937-3-git-send-email-si-wei.liu@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa: introduce .reset_map operation callback
Si-Wei Liu [Sat, 21 Oct 2023 09:25:13 +0000 (02:25 -0700)]
vdpa: introduce .reset_map operation callback

Some device specific IOMMU parent drivers have long standing bogus
behavior that mistakenly clean up the maps during .reset. By definition,
this is violation to the on-chip IOMMU ops (i.e. .set_map, or .dma_map &
.dma_unmap) in those offending drivers, as the removal of internal maps
is completely agnostic to the upper layer, causing inconsistent view
between the userspace and the kernel. Some userspace app like QEMU gets
around of this brokenness by proactively removing and adding back all
the maps around vdpa device reset, but such workaround actually penalize
other well-behaved driver setup, where vdpa reset always comes with the
associated mapping cost, especially for kernel vDPA devices
(use_va=false) that have high cost on pinning. It's imperative to
rectify this behavior and remove the problematic code from all those
non-compliant parent drivers.

The reason why a separate .reset_map op is introduced is because this
allows a simple on-chip IOMMU model without exposing too much device
implementation detail to the upper vdpa layer. The .dma_map/unmap or
.set_map driver API is meant to be used to manipulate the IOTLB
mappings, and has been abstracted in a way similar to how a real IOMMU
device maps or unmaps pages for certain memory ranges. However, apart
from this there also exists other mapping needs, in which case 1:1
passthrough mapping has to be used by other users (read virtio-vdpa). To
ease parent/vendor driver implementation and to avoid abusing DMA ops in
an unexpacted way, these on-chip IOMMU devices can start with 1:1
passthrough mapping mode initially at the time of creation. Then the
.reset_map op can be used to switch iotlb back to this initial state
without having to expose a complex two-dimensional IOMMU device model.

The .reset_map is not a MUST for every parent that implements the
.dma_map or .set_map API, because device may work with DMA ops directly
by implement their own to manipulate system memory mappings, so don't
have to use .reset_map to achieve a simple IOMMU device model for 1:1
passthrough mapping.

Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <1697880319-4937-2-git-send-email-si-wei.liu@oracle.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovirtio_pci: add check for common cfg size
Xuan Zhuo [Thu, 19 Oct 2023 03:49:02 +0000 (11:49 +0800)]
virtio_pci: add check for common cfg size

Some buggy devices, the common cfg size may not match the features.

This patch checks the common cfg size for the
features(VIRTIO_F_NOTIF_CONFIG_DATA, VIRTIO_F_RING_RESET). When the
common cfg size does not match the corresponding feature, we fail the
probe and print error message.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20231019034902.7346-1-xuanzhuo@linux.alibaba.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
17 months agovirtio-blk: fix implicit overflow on virtio_max_dma_size
zhenwei pi [Mon, 4 Sep 2023 06:10:45 +0000 (14:10 +0800)]
virtio-blk: fix implicit overflow on virtio_max_dma_size

The following codes have an implicit conversion from size_t to u32:
(u32)max_size = (size_t)virtio_max_dma_size(vdev);

This may lead overflow, Ex (size_t)4G -> (u32)0. Once
virtio_max_dma_size() has a larger size than U32_MAX, use U32_MAX
instead.

Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230904061045.510460-1-pizhenwei@bytedance.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
17 months agovirtio_pci: add build offset check for the new common cfg items
Xuan Zhuo [Tue, 10 Oct 2023 03:11:20 +0000 (11:11 +0800)]
virtio_pci: add build offset check for the new common cfg items

Add checks to the check_offsets(void) for queue_notify_data and
queue_reset.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20231010031120.81272-5-xuanzhuo@linux.alibaba.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
17 months agovirtio: add definition of VIRTIO_F_NOTIF_CONFIG_DATA feature bit
Xuan Zhuo [Tue, 10 Oct 2023 03:11:17 +0000 (11:11 +0800)]
virtio: add definition of VIRTIO_F_NOTIF_CONFIG_DATA feature bit

This patch adds the definition of VIRTIO_F_NOTIF_CONFIG_DATA feature bit
in the relevant header file.

This feature indicates that the driver uses the data provided by the
device as a virtqueue identifier in available buffer notifications.

It comes from here:
    https://github.com/oasis-tcs/virtio-spec/issues/89

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Message-Id: <20231010031120.81272-2-xuanzhuo@linux.alibaba.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
17 months agovduse: make vduse_class constant
Greg Kroah-Hartman [Fri, 6 Oct 2023 14:30:44 +0000 (16:30 +0200)]
vduse: make vduse_class constant

Now that the driver core allows for struct class to be in read-only
memory, we should make all 'class' structures declared at build time
placing them into read-only memory, instead of having to be dynamically
allocated at runtime.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Cc: Xie Yongji <xieyongji@bytedance.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Message-Id: <2023100643-tricolor-citizen-6c2d@gregkh>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Xie Yongji <xieyongji@bytedance.com>
Acked-by: Jason Wang <jasowang@redhat.com>
17 months agovhost-scsi: Spelling s/preceeding/preceding/g
Geert Uytterhoeven [Thu, 28 Sep 2023 12:18:33 +0000 (14:18 +0200)]
vhost-scsi: Spelling s/preceeding/preceding/g

Fix a misspelling of "preceding".

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Message-Id: <b57b882675809f1f9dacbf42cf6b920b2bea9cba.1695903476.git.geert+renesas@glider.be>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
17 months agovirtio: kdoc for struct virtio_pci_modern_device
Shannon Nelson [Mon, 11 Sep 2023 21:31:04 +0000 (14:31 -0700)]
virtio: kdoc for struct virtio_pci_modern_device

Finally following up to Simon's suggestion for some kdoc attention
on struct virtio_pci_modern_device.

Link: https://lore.kernel.org/netdev/ZE%2FQS0lnUvxFacjf@corigine.com/
Cc: Simon Horman <simon.horman@corigine.com>
Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230911213104.14391-1-shannon.nelson@amd.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
17 months agovdpa: Update sysfs ABI documentation
Shawn.Shao [Mon, 21 Aug 2023 01:15:35 +0000 (09:15 +0800)]
vdpa: Update sysfs ABI documentation

Fix the wrong drivers_autoprobe path name in the document

Signed-off-by: Shawn.Shao <shawn.shao@jaguarmicro.com>
Message-Id: <20230821011535.1117-1-shawn.shao@jaguarmicro.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
17 months agoMAINTAINERS: Add myself as mlx5_vdpa driver
Dragos Tatulea [Mon, 25 Sep 2023 16:06:51 +0000 (19:06 +0300)]
MAINTAINERS: Add myself as mlx5_vdpa driver

As Eli Cohen moved to other work, I'll be the contact point for
mlx5_vdpa.

Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20230925160654.1558627-1-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
17 months agovirtio-balloon: correct the comment of virtballoon_migratepage()
Xueshi Hu [Sun, 13 Aug 2023 14:07:09 +0000 (22:07 +0800)]
virtio-balloon: correct the comment of virtballoon_migratepage()

After commit 68f2736a8583 ("mm: Convert all PageMovable users to
movable_operations"), the execution path has been changed to

move_to_new_folio
movable_operations->migrate_page
balloon_page_migrate
balloon_page_migrate->balloon_page_migrate
balloon_page_migrate

Correct the outdated comment.

Signed-off-by: Xueshi Hu <xueshi.hu@smartx.com>
Message-Id: <20230813140709.835536-1-xueshi.hu@smartx.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
17 months agomlx5_vdpa: offer VHOST_BACKEND_F_ENABLE_AFTER_DRIVER_OK
Eugenio Pérez [Mon, 3 Jul 2023 14:25:14 +0000 (16:25 +0200)]
mlx5_vdpa: offer VHOST_BACKEND_F_ENABLE_AFTER_DRIVER_OK

Offer this backend feature as mlx5 is compatible with it. It allows it
to do live migration with CVQ, dynamically switching between passthrough
and shadow virtqueue.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230703142514.363256-1-eperezma@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
17 months agovdpa/mlx5: Update cvq iotlb mapping on ASID change
Dragos Tatulea [Wed, 18 Oct 2023 17:14:55 +0000 (20:14 +0300)]
vdpa/mlx5: Update cvq iotlb mapping on ASID change

For the following sequence:
- cvq group is in ASID 0
- .set_map(1, cvq_iotlb)
- .set_group_asid(cvq_group, 1)

... the cvq mapping from ASID 0 will be used. This is not always correct
behaviour.

This patch adds support for the above mentioned flow by saving the iotlb
on each .set_map and updating the cvq iotlb with it on a cvq group change.

Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-18-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Make iotlb helper functions more generic
Dragos Tatulea [Wed, 18 Oct 2023 17:14:54 +0000 (20:14 +0300)]
vdpa/mlx5: Make iotlb helper functions more generic

They will be used in a follow-up patch.

For dup_iotlb, avoid the src == dst case. This is an error.

Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-17-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Enable hw support for vq descriptor mapping
Dragos Tatulea [Wed, 18 Oct 2023 17:14:53 +0000 (20:14 +0300)]
vdpa/mlx5: Enable hw support for vq descriptor mapping

Vq descriptor mappings are supported in hardware by filling in an
additional mkey which contains the descriptor mappings to the hw vq.

A previous patch in this series added support for hw mkey (mr) creation
for ASID 1.

This patch fills in both the vq data and vq descriptor mkeys based on
group ASID mapping.

The feature is signaled to the vdpa core through the presence of the
.get_vq_desc_group op.

Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-16-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Introduce mr for vq descriptor
Dragos Tatulea [Wed, 18 Oct 2023 17:14:52 +0000 (20:14 +0300)]
vdpa/mlx5: Introduce mr for vq descriptor

Introduce the vq descriptor group and mr per ASID. Until now
.set_map on ASID 1 was only updating the cvq iotlb. From now on it also
creates a mkey for it. The current patch doesn't use it but follow-up
patches will add hardware support for mapping the vq descriptors.

Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-15-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Improve mr update flow
Dragos Tatulea [Wed, 18 Oct 2023 17:14:51 +0000 (20:14 +0300)]
vdpa/mlx5: Improve mr update flow

The current flow for updating an mr works directly on mvdev->mr which
makes it cumbersome to handle multiple new mr structs.

This patch makes the flow more straightforward by having
mlx5_vdpa_create_mr return a new mr which will update the old mr (if
any). The old mr will be deleted and unlinked from mvdev. For the case
when the iotlb is empty (not NULL), the old mr will be cleared.

This change paves the way for adding mrs for different ASIDs.

The initialized bool is no longer needed as mr is now a pointer in the
mlx5_vdpa_dev struct which will be NULL when not initialized.

Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-14-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Move mr mutex out of mr struct
Dragos Tatulea [Wed, 18 Oct 2023 17:14:50 +0000 (20:14 +0300)]
vdpa/mlx5: Move mr mutex out of mr struct

The mutex is named like it is supposed to protect only the mkey but in
reality it is a global lock for all mr resources.

Shift the mutex to it's rightful location (struct mlx5_vdpa_dev) and
give it a more appropriate name.

Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20231018171456.1624030-13-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Allow creation/deletion of any given mr struct
Dragos Tatulea [Wed, 18 Oct 2023 17:14:49 +0000 (20:14 +0300)]
vdpa/mlx5: Allow creation/deletion of any given mr struct

This patch adapts the mr creation/deletion code to be able to work with
any given mr struct pointer. All the APIs are adapted to take an extra
parameter for the mr.

mlx5_vdpa_create/delete_mr doesn't need a ASID parameter anymore. The
check is done in the caller instead (mlx5_set_map).

This change is needed for a followup patch which will introduce an
additional mr for the vq descriptor data.

Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20231018171456.1624030-12-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Rename mr destroy functions
Dragos Tatulea [Wed, 18 Oct 2023 17:14:48 +0000 (20:14 +0300)]
vdpa/mlx5: Rename mr destroy functions

Make mlx5_destroy_mr symmetric to mlx5_create_mr.

Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-11-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Collapse "dvq" mr add/delete functions
Dragos Tatulea [Wed, 18 Oct 2023 17:14:47 +0000 (20:14 +0300)]
vdpa/mlx5: Collapse "dvq" mr add/delete functions

Now that the cvq code is out of mlx5_vdpa_create/destroy_mr, the "dvq"
functions can be folded into their callers.

Having "dvq" in the naming will no longer be accurate in the downstream
patches.

Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-10-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Take cvq iotlb lock during refresh
Dragos Tatulea [Wed, 18 Oct 2023 17:14:46 +0000 (20:14 +0300)]
vdpa/mlx5: Take cvq iotlb lock during refresh

The reslock is taken while refresh is called but iommu_lock is more
specific to this resource. So take the iommu_lock during cvq iotlb
refresh.

Based on Eugenio's patch [0].

[0] https://lore.kernel.org/lkml/20230112142218.725622-4-eperezma@redhat.com/

Acked-by: Jason Wang <jasowang@redhat.com>
Suggested-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-9-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Decouple cvq iotlb handling from hw mapping code
Dragos Tatulea [Wed, 18 Oct 2023 17:14:45 +0000 (20:14 +0300)]
vdpa/mlx5: Decouple cvq iotlb handling from hw mapping code

The handling of the cvq iotlb is currently coupled with the creation
and destruction of the hardware mkeys (mr).

This patch moves cvq iotlb handling into its own function and shifts it
to a scope that is not related to mr handling. As cvq handling is just a
prune_iotlb + dup_iotlb cycle, put it all in the same "update" function.
Finally, the destruction path is handled by directly pruning the iotlb.

After this move is done the ASID mr code can be collapsed into a single
function.

Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-8-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa/mlx5: Create helper function for dma mappings
Dragos Tatulea [Wed, 18 Oct 2023 17:14:44 +0000 (20:14 +0300)]
vdpa/mlx5: Create helper function for dma mappings

Necessary for upcoming cvq separation from mr allocation.

Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20231018171456.1624030-7-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovhost-vdpa: uAPI to get dedicated descriptor group id
Si-Wei Liu [Wed, 18 Oct 2023 17:14:43 +0000 (20:14 +0300)]
vhost-vdpa: uAPI to get dedicated descriptor group id

With _F_DESC_ASID backend feature, the device can now support the
VHOST_VDPA_GET_VRING_DESC_GROUP ioctl, and it may expose the descriptor
table (including avail and used ring) in a different group than the
buffers it contains. This new uAPI will fetch the group ID of the
descriptor table.

Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20231018171456.1624030-6-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovhost-vdpa: introduce descriptor group backend feature
Si-Wei Liu [Wed, 18 Oct 2023 17:14:42 +0000 (20:14 +0300)]
vhost-vdpa: introduce descriptor group backend feature

Userspace knows if the device has dedicated descriptor group or not
by checking this feature bit.

It's only exposed if the vdpa driver backend implements the
.get_vq_desc_group() operation callback. Userspace trying to negotiate
this feature when it or the dependent _F_IOTLB_ASID feature hasn't
been exposed will result in an error.

Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20231018171456.1624030-5-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agovdpa: introduce dedicated descriptor group for virtqueue
Si-Wei Liu [Wed, 18 Oct 2023 17:14:41 +0000 (20:14 +0300)]
vdpa: introduce dedicated descriptor group for virtqueue

In some cases, the access to the virtqueue's descriptor area, device
and driver areas (precluding indirect descriptor table in guest memory)
may have to be confined to a different address space than where its
buffers reside. Without loss of simplicity and generality with already
established terminology, let's fold up these 3 areas and call them
as a whole as descriptor table group, or descriptor group for short.
Specifically, in case of split virtqueues, descriptor group consists of
regions for Descriptor Table, Available Ring and Used Ring; for packed
virtqueues layout, descriptor group contains Descriptor Ring, Driver
and Device Event Suppression structures.

The group ID for a dedicated descriptor group can be obtained through a
new .get_vq_desc_group() op. If driver implements this op, it means that
the descriptor, device and driver areas of the virtqueue may reside
in a dedicated group than where its buffers reside, a.k.a the default
virtqueue group through the .get_vq_group() op.

In principle, the descriptor group may or may not have same group ID
as the default group. Even if the descriptor group has a different ID,
meaning the vq's descriptor group areas can optionally move to a
separate address space than where guest memory resides, the descriptor
group may still start from a default address space, same as where its
buffers reside. To move the descriptor group to a different address
space, .set_group_asid() has to be called to change the ASID binding
for the group, which is no different than what needs to be done on any
other virtqueue group. On the other hand, the .reset() semantics also
applies on descriptor table group, meaning the device reset will clear
all ASID bindings and move all virtqueue groups including descriptor
group back to the default address space, i.e. in ASID 0.

QEMU's shadow virtqueue is going to utilize dedicated descriptor group
to speed up map and unmap operations, yielding tremendous downtime
reduction by avoiding the full and slow remap cycle in SVQ switching.

Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Acked-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20231018171456.1624030-4-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Si-Wei Liu <si-wei.liu@oracle.com>
Tested-by: Lei Yang <leiyang@redhat.com>
17 months agoMerge branch 'mlx5-vhost' of https://git.kernel.org/pub/scm/linux/kernel/git/mellanox...
Michael S. Tsirkin [Wed, 1 Nov 2023 13:14:52 +0000 (09:14 -0400)]
Merge branch 'mlx5-vhost' of https://git./linux/kernel/git/mellanox/linux.git

This merges a single commit that contains changes to mlx5_ifc.h
It's required to support vq descriptor mappings in mlx5/vdpa

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
17 months agoLinux 6.6
Linus Torvalds [Mon, 30 Oct 2023 02:31:08 +0000 (16:31 -1000)]
Linux 6.6

17 months agoMerge tag 'x86-urgent-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 28 Oct 2023 18:15:07 +0000 (08:15 -1000)]
Merge tag 'x86-urgent-2023-10-28' of git://git./linux/kernel/git/tip/tip

Pull misc x86 fixes from Ingo Molnar:

 - Fix a possible CPU hotplug deadlock bug caused by the new TSC
   synchronization code

 - Fix a legacy PIC discovery bug that results in device troubles on
   affected systems, such as non-working keybards, etc

 - Add a new Intel CPU model number to <asm/intel-family.h>

* tag 'x86-urgent-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/tsc: Defer marking TSC unstable to a worker
  x86/i8259: Skip probing when ACPI/MADT advertises PCAT compatibility
  x86/cpu: Add model number for Intel Arrow Lake mobile processor

17 months agoMerge tag 'irq-urgent-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 28 Oct 2023 18:12:34 +0000 (08:12 -1000)]
Merge tag 'irq-urgent-2023-10-28' of git://git./linux/kernel/git/tip/tip

Pull irq fix from Ingo Molnar:
 "Restore unintentionally lost quirk settings in the GIC irqchip driver,
  which broke certain devices"

* tag 'irq-urgent-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/gic-v3-its: Don't override quirk settings with default values

17 months agoMerge tag 'perf-urgent-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 28 Oct 2023 18:10:47 +0000 (08:10 -1000)]
Merge tag 'perf-urgent-2023-10-28' of git://git./linux/kernel/git/tip/tip

Pull perf event fix from Ingo Molnar:
 "Fix a potential NULL dereference bug"

* tag 'perf-urgent-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/core: Fix potential NULL deref

17 months agoMerge tag 'probes-fixes-v6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 28 Oct 2023 18:04:56 +0000 (08:04 -1000)]
Merge tag 'probes-fixes-v6.6-rc7' of git://git./linux/kernel/git/trace/linux-trace

Pull probes fixes from Masami Hiramatsu:

 - tracing/kprobes: Fix kernel-doc warnings for the variable length
   arguments

 - tracing/kprobes: Fix to count the symbols in modules even if the
   module name is not specified so that user can probe the symbols in
   the modules without module name

* tag 'probes-fixes-v6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracing/kprobes: Fix symbol counting logic by looking at modules as well
  tracing/kprobes: Fix the description of variable length arguments

17 months agoMerge tag 'dma-mapping-6.6-2023-10-28' of git://git.infradead.org/users/hch/dma-mapping
Linus Torvalds [Sat, 28 Oct 2023 18:01:31 +0000 (08:01 -1000)]
Merge tag 'dma-mapping-6.6-2023-10-28' of git://git.infradead.org/users/hch/dma-mapping

Pull dma-mapping fix from Christoph Hellwig:

 - reduce the initialy dynamic swiotlb size to remove an annoying but
   harmless warning from the page allocator (Petr Tesarik)

* tag 'dma-mapping-6.6-2023-10-28' of git://git.infradead.org/users/hch/dma-mapping:
  swiotlb: do not try to allocate a TLB bigger than MAX_ORDER pages

17 months agoMerge tag 'char-misc-6.6-final' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 28 Oct 2023 17:51:27 +0000 (07:51 -1000)]
Merge tag 'char-misc-6.6-final' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc driver fixes from Greg KH:
 "Here are some very small driver fixes for 6.6-final that have shown up
  in the past two weeks. Included in here are:

   - tiny fastrpc bugfixes for reported errors

   - nvmem register fixes

   - iio driver fixes for some reported problems

   - fpga test fix

   - MAINTAINERS file update for fpga

  All of these have been in linux-next this week with no reported
  problems"

* tag 'char-misc-6.6-final' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  fpga: Fix memory leak for fpga_region_test_class_find()
  fpga: m10bmc-sec: Change contact for secure update driver
  fpga: disable KUnit test suites when module support is enabled
  iio: afe: rescale: Accept only offset channels
  nvmem: imx: correct nregs for i.MX6ULL
  nvmem: imx: correct nregs for i.MX6UL
  nvmem: imx: correct nregs for i.MX6SLL
  misc: fastrpc: Unmap only if buffer is unmapped from DSP
  misc: fastrpc: Clean buffers on remote invocation failures
  misc: fastrpc: Free DMA handles for RPC calls with no arguments
  misc: fastrpc: Reset metadata buffer to avoid incorrect free
  iio: exynos-adc: request second interupt only when touchscreen mode is used
  iio: adc: xilinx-xadc: Correct temperature offset/scale for UltraScale
  iio: adc: xilinx-xadc: Don't clobber preset voltage/temperature thresholds
  dt-bindings: iio: add missing reset-gpios constrain

17 months agoMerge tag 'i2c-for-6.6-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sat, 28 Oct 2023 17:48:37 +0000 (07:48 -1000)]
Merge tag 'i2c-for-6.6-rc8' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "Bugfixes for Axxia when it is a target and for PEC handling of
  stm32f7.

  Plus, fix an OF node leak pattern in the mux subsystem"

* tag 'i2c-for-6.6-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: stm32f7: Fix PEC handling in case of SMBUS transfers
  i2c: muxes: i2c-mux-gpmux: Use of_get_i2c_adapter_by_node()
  i2c: muxes: i2c-demux-pinctrl: Use of_get_i2c_adapter_by_node()
  i2c: muxes: i2c-mux-pinctrl: Use of_get_i2c_adapter_by_node()
  i2c: aspeed: Fix i2c bus hang in slave read

17 months agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 28 Oct 2023 02:52:51 +0000 (16:52 -1000)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "Three fixes, one for the clk framework and two for clk drivers:

   - Avoid an oops in possible_parent_show() by checking for no parent
     properly when a DT index based lookup is used

   - Handle errors returned from divider_ro_round_rate() in
     clk_stm32_composite_determine_rate()

   - Fix clk_ops::determine_rate() implementation of socfpga's
     gateclk_ops that was ruining uart output because the divider
     was forgotten about"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: stm32: Fix a signedness issue in clk_stm32_composite_determine_rate()
  clk: Sanitize possible_parent_show to Handle Return Value of of_clk_get_parent_name
  clk: socfpga: gate: Account for the divider in determine_rate

17 months agoMerge tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Sat, 28 Oct 2023 02:44:58 +0000 (16:44 -1000)]
Merge tag 'pull-fixes' of git://git./linux/kernel/git/viro/vfs

Pull misc filesystem fixes from Al Viro:
 "Assorted fixes all over the place: literally nothing in common, could
  have been three separate pull requests.

  All are simple regression fixes, but not for anything from this cycle"

* tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  ceph_wait_on_conflict_unlink(): grab reference before dropping ->d_lock
  io_uring: kiocb_done() should *not* trust ->ki_pos if ->{read,write}_iter() failed
  sparc32: fix a braino in fault handling in csum_and_copy_..._user()

17 months agotracing/kprobes: Fix symbol counting logic by looking at modules as well
Andrii Nakryiko [Fri, 27 Oct 2023 23:31:26 +0000 (16:31 -0700)]
tracing/kprobes: Fix symbol counting logic by looking at modules as well

Recent changes to count number of matching symbols when creating
a kprobe event failed to take into account kernel modules. As such, it
breaks kprobes on kernel module symbols, by assuming there is no match.

Fix this my calling module_kallsyms_on_each_symbol() in addition to
kallsyms_on_each_match_symbol() to perform a proper counting.

Link: https://lore.kernel.org/all/20231027233126.2073148-1-andrii@kernel.org/
Cc: Francis Laniel <flaniel@linux.microsoft.com>
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Fixes: b022f0c7e404 ("tracing/kprobes: Return EADDRNOTAVAIL when func matches several symbols")
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Song Liu <song@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
17 months agoceph_wait_on_conflict_unlink(): grab reference before dropping ->d_lock
Al Viro [Fri, 15 Sep 2023 01:55:29 +0000 (21:55 -0400)]
ceph_wait_on_conflict_unlink(): grab reference before dropping ->d_lock

Use of dget() after we'd dropped ->d_lock is too late - dentry might
be gone by that point.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
17 months agoio_uring: kiocb_done() should *not* trust ->ki_pos if ->{read,write}_iter() failed
Al Viro [Mon, 28 Aug 2023 22:47:31 +0000 (18:47 -0400)]
io_uring: kiocb_done() should *not* trust ->ki_pos if ->{read,write}_iter() failed

->ki_pos value is unreliable in such cases.  For an obvious example,
consider O_DSYNC write - we feed the data to page cache and start IO,
then we make sure it's completed.  Update of ->ki_pos is dealt with
by the first part; failure in the second ends up with negative value
returned _and_ ->ki_pos left advanced as if sync had been successful.
In the same situation write(2) does not advance the file position
at all.

Reviewed-by: Christian Brauner <brauner@kernel.org>
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
17 months agoMerge tag 'io_uring-6.6-2023-10-27' of git://git.kernel.dk/linux
Linus Torvalds [Sat, 28 Oct 2023 00:10:32 +0000 (14:10 -1000)]
Merge tag 'io_uring-6.6-2023-10-27' of git://git.kernel.dk/linux

Pull io_uring fixes from Jens Axboe:
 "Fix for an issue reported where reading fdinfo could find a NULL
  thread as we didn't properly synchronize, and then a disable for the
  IOCB_DIO_CALLER_COMP optimization as a recent reported highlighted how
  that could lead to deadlocks if the task issued async O_DIRECT writes
  and then proceeded to do sync fallocate() calls"

* tag 'io_uring-6.6-2023-10-27' of git://git.kernel.dk/linux:
  io_uring/rw: disable IOCB_DIO_CALLER_COMP
  io_uring/fdinfo: lock SQ thread while retrieving thread cpu/pid

17 months agosparc32: fix a braino in fault handling in csum_and_copy_..._user()
Al Viro [Sun, 22 Oct 2023 23:34:28 +0000 (19:34 -0400)]
sparc32: fix a braino in fault handling in csum_and_copy_..._user()

Fault handler used to make non-trivial calls, so it needed
to set a stack frame up.  Used to be
save ... - grab a stack frame, old %o... become %i...
....
ret - go back to address originally in %o7, currently %i7
 restore - switch to previous stack frame, in delay slot
Non-trivial calls had been gone since ab5e8b331244 and that code should
have become
retl - go back to address in %o7
 clr %o0 - have return value set to 0
What it had become instead was
ret - go back to address in %i7 - return address of *caller*
 clr %o0 - have return value set to 0
which is not good, to put it mildly - we forcibly return 0 from
csum_and_copy_{from,to}_iter() (which is what the call of that
thing had been inlined into) and do that without dropping the
stack frame of said csum_and_copy_..._iter().  Confuses the
hell out of the caller of csum_and_copy_..._iter(), obviously...

Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Fixes: ab5e8b331244 "sparc32: propagate the calling conventions change down to __csum_partial_copy_sparc_generic()"
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
17 months agoMerge tag 'block-6.6-2023-10-27' of git://git.kernel.dk/linux
Linus Torvalds [Sat, 28 Oct 2023 00:01:59 +0000 (14:01 -1000)]
Merge tag 'block-6.6-2023-10-27' of git://git.kernel.dk/linux

Pull block fix from Jens Axboe:
 "Just a single fix for a potential divide-by-zero, introduced in this
  cycle"

* tag 'block-6.6-2023-10-27' of git://git.kernel.dk/linux:
  blk-throttle: check for overflow in calculate_bytes_allowed

17 months agoMerge tag 'ata-6.6-final' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal...
Linus Torvalds [Fri, 27 Oct 2023 23:38:59 +0000 (13:38 -1000)]
Merge tag 'ata-6.6-final' of git://git./linux/kernel/git/dlemoal/libata

Pull ATA fix from Damien Le Moal:
 "A single patch to fix a regression introduced by the recent
  suspend/resume fixes.

  The regression is that ATA disks are not stopped on system shutdown,
  which is not recommended and increases the disks SMART counters for
  unclean power off events.

  This patch fixes this by refining the recent rework of the scsi device
  manage_xxx flags"

* tag 'ata-6.6-final' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/libata:
  scsi: sd: Introduce manage_shutdown device flag

17 months agoMerge tag 'platform-drivers-x86-v6.6-6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 27 Oct 2023 23:32:48 +0000 (13:32 -1000)]
Merge tag 'platform-drivers-x86-v6.6-6' of git://git./linux/kernel/git/pdx86/platform-drivers-x86

Pull x86 platform driver fix from Hans de Goede:
 "A single patch to extend the AMD PMC driver DMI quirk list
  for laptops which need special handling to avoid NVME s2idle
  suspend/resume errors"

* tag 'platform-drivers-x86-v6.6-6' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
  platform/x86: Add s2idle quirk for more Lenovo laptops

17 months agox86/tsc: Defer marking TSC unstable to a worker
Thomas Gleixner [Wed, 25 Oct 2023 21:31:35 +0000 (23:31 +0200)]
x86/tsc: Defer marking TSC unstable to a worker

Tetsuo reported the following lockdep splat when the TSC synchronization
fails during CPU hotplug:

   tsc: Marking TSC unstable due to check_tsc_sync_source failed

   WARNING: inconsistent lock state
   inconsistent {IN-HARDIRQ-W} -> {HARDIRQ-ON-W} usage.
   ffffffff8cfa1c78 (watchdog_lock){?.-.}-{2:2}, at: clocksource_watchdog+0x23/0x5a0
   {IN-HARDIRQ-W} state was registered at:
     _raw_spin_lock_irqsave+0x3f/0x60
     clocksource_mark_unstable+0x1b/0x90
     mark_tsc_unstable+0x41/0x50
     check_tsc_sync_source+0x14f/0x180
     sysvec_call_function_single+0x69/0x90

   Possible unsafe locking scenario:
     lock(watchdog_lock);
     <Interrupt>
       lock(watchdog_lock);

   stack backtrace:
    _raw_spin_lock+0x30/0x40
    clocksource_watchdog+0x23/0x5a0
    run_timer_softirq+0x2a/0x50
    sysvec_apic_timer_interrupt+0x6e/0x90

The reason is the recent conversion of the TSC synchronization function
during CPU hotplug on the control CPU to a SMP function call. In case
that the synchronization with the upcoming CPU fails, the TSC has to be
marked unstable via clocksource_mark_unstable().

clocksource_mark_unstable() acquires 'watchdog_lock', but that lock is
taken with interrupts enabled in the watchdog timer callback to minimize
interrupt disabled time. That's obviously a possible deadlock scenario,

Before that change the synchronization function was invoked in thread
context so this could not happen.

As it is not crucical whether the unstable marking happens slightly
delayed, defer the call to a worker thread which avoids the lock context
problem.

Fixes: 9d349d47f0e3 ("x86/smpboot: Make TSC synchronization function call based")
Reported-by: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/87zg064ceg.ffs@tglx
17 months agox86/i8259: Skip probing when ACPI/MADT advertises PCAT compatibility
Thomas Gleixner [Wed, 25 Oct 2023 21:04:15 +0000 (23:04 +0200)]
x86/i8259: Skip probing when ACPI/MADT advertises PCAT compatibility

David and a few others reported that on certain newer systems some legacy
interrupts fail to work correctly.

Debugging revealed that the BIOS of these systems leaves the legacy PIC in
uninitialized state which makes the PIC detection fail and the kernel
switches to a dummy implementation.

Unfortunately this fallback causes quite some code to fail as it depends on
checks for the number of legacy PIC interrupts or the availability of the
real PIC.

In theory there is no reason to use the PIC on any modern system when
IO/APIC is available, but the dependencies on the related checks cannot be
resolved trivially and on short notice. This needs lots of analysis and
rework.

The PIC detection has been added to avoid quirky checks and force selection
of the dummy implementation all over the place, especially in VM guest
scenarios. So it's not an option to revert the relevant commit as that
would break a lot of other scenarios.

One solution would be to try to initialize the PIC on detection fail and
retry the detection, but that puts the burden on everything which does not
have a PIC.

Fortunately the ACPI/MADT table header has a flag field, which advertises
in bit 0 that the system is PCAT compatible, which means it has a legacy
8259 PIC.

Evaluate that bit and if set avoid the detection routine and keep the real
PIC installed, which then gets initialized (for nothing) and makes the rest
of the code with all the dependencies work again.

Fixes: e179f6914152 ("x86, irq, pic: Probe for legacy PIC and set legacy_pic appropriately")
Reported-by: David Lazar <dlazar@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: David Lazar <dlazar@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Cc: stable@vger.kernel.org
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=218003
Link: https://lore.kernel.org/r/875y2u5s8g.ffs@tglx
17 months agox86/cpu: Add model number for Intel Arrow Lake mobile processor
Tony Luck [Wed, 25 Oct 2023 20:25:13 +0000 (13:25 -0700)]
x86/cpu: Add model number for Intel Arrow Lake mobile processor

For "reasons" Intel has code-named this CPU with a "_H" suffix.

[ dhansen: As usual, apply this and send it upstream quickly to
   make it easier for anyone who is doing work that
   consumes this. ]

Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/all/20231025202513.12358-1-tony.luck%40intel.com
17 months agoMerge tag 'iommu-fix-v6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/joro...
Linus Torvalds [Fri, 27 Oct 2023 15:43:05 +0000 (05:43 -1000)]
Merge tag 'iommu-fix-v6.6-rc7' of git://git./linux/kernel/git/joro/iommu

Pull iommu fix from Joerg Roedel:

 - Fix boot regression for Sapphire Rapids with Intel VT-d driver

* tag 'iommu-fix-v6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu: Avoid unnecessary cache invalidations

17 months agoMerge tag 'powerpc-6.6-6' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Fri, 27 Oct 2023 15:40:42 +0000 (05:40 -1000)]
Merge tag 'powerpc-6.6-6' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:

 - Fix boot crash with FLATMEM since set_ptes() introduction

 - Avoid calling arch_enter/leave_lazy_mmu() in set_ptes()

Thanks to Aneesh Kumar K.V and Erhard Furtner.

* tag 'powerpc-6.6-6' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/mm: Avoid calling arch_enter/leave_lazy_mmu() in set_ptes
  powerpc/mm: Fix boot crash with FLATMEM

17 months agoplatform/x86: Add s2idle quirk for more Lenovo laptops
David Lazar [Wed, 25 Oct 2023 19:30:16 +0000 (21:30 +0200)]
platform/x86: Add s2idle quirk for more Lenovo laptops

When suspending to idle and resuming on some Lenovo laptops using the
Mendocino APU, multiple NVME IOMMU page faults occur, showing up in
dmesg as repeated errors:

nvme 0000:01:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x000b
address=0xb6674000 flags=0x0000]

The system is unstable afterwards.

Applying the s2idle quirk introduced by commit 455cd867b85b ("platform/x86:
thinkpad_acpi: Add a s2idle resume quirk for a number of laptops")
allows these systems to work with the IOMMU enabled and s2idle
resume to work.

Closes: https://bugzilla.kernel.org/show_bug.cgi?id=218024
Suggested-by: Mario Limonciello <mario.limonciello@amd.com>
Suggested-by: Mark Pearson <mpearson-lenovo@squebb.ca>
Signed-off-by: David Lazar <dlazar@gmail.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Mark Pearson <mpearson-lenovo@squebb.ca>
Link: https://lore.kernel.org/r/ZTlsyOaFucF2pWrL@localhost
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
17 months agotracing/kprobes: Fix the description of variable length arguments
Yujie Liu [Fri, 27 Oct 2023 04:13:14 +0000 (12:13 +0800)]
tracing/kprobes: Fix the description of variable length arguments

Fix the following kernel-doc warnings:

kernel/trace/trace_kprobe.c:1029: warning: Excess function parameter 'args' description in '__kprobe_event_gen_cmd_start'
kernel/trace/trace_kprobe.c:1097: warning: Excess function parameter 'args' description in '__kprobe_event_add_fields'

Refer to the usage of variable length arguments elsewhere in the kernel
code, "@..." is the proper way to express it in the description.

Link: https://lore.kernel.org/all/20231027041315.2613166-1-yujie.liu@intel.com/
Fixes: 2a588dd1d5d6 ("tracing: Add kprobe event command generation functions")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202310190437.paI6LYJF-lkp@intel.com/
Signed-off-by: Yujie Liu <yujie.liu@intel.com>
Reviewed-by: Mukesh Ojha <quic_mojha@quicinc.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
17 months agoiommu: Avoid unnecessary cache invalidations
Lu Baolu [Thu, 26 Oct 2023 08:49:42 +0000 (16:49 +0800)]
iommu: Avoid unnecessary cache invalidations

The iommu_create_device_direct_mappings() only needs to flush the caches
when the mappings are changed in the affected domain. This is not true
for non-DMA domains, or for devices attached to the domain that have no
reserved regions. To avoid unnecessary cache invalidations, add a check
before iommu_flush_iotlb_all().

Fixes: a48ce36e2786 ("iommu: Prevent RESV_DIRECT devices from blocking domains")
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Tested-by: Henry Willard <henry.willard@oracle.com>
Link: https://lore.kernel.org/r/20231026084942.17387-1-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
17 months agoMerge tag 'drm-fixes-2023-10-27' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 27 Oct 2023 06:42:02 +0000 (20:42 -1000)]
Merge tag 'drm-fixes-2023-10-27' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "This is the final set of fixes for 6.6, just misc bits mainly in
  amdgpu and i915, nothing too noteworthy.

  amdgpu:
   - ignore duplicated BOs in CS parser
   - remove redundant call to amdgpu_ctx_priority_is_valid()
   - Extend VI APSM quirks to more platforms

  amdkfd:
   - reserve fence slot while locking BO

  dp_mst:
   - Fix NULL deref in get_mst_branch_device_by_guid_helper()

  logicvc:
   - Kconfig: Select REGMAP and REGMAP_MMIO

  ivpu:
   - Fix missing VPUIP interrupts

  i915:
   - Determine context valid in OA reports
   - Hold GT forcewake during steering operations
   - Check if PMU is closed before stopping event"

* tag 'drm-fixes-2023-10-27' of git://anongit.freedesktop.org/drm/drm:
  accel/ivpu/37xx: Fix missing VPUIP interrupts
  drm/amd: Disable ASPM for VI w/ all Intel systems
  drm/i915/pmu: Check if pmu is closed before stopping event
  drm/i915/mcr: Hold GT forcewake during steering operations
  drm/logicvc: Kconfig: select REGMAP and REGMAP_MMIO
  drm/i915/perf: Determine context valid in OA reports
  drm/amdkfd: reserve a fence slot while locking the BO
  drm/amdgpu: Remove redundant call to priority_is_valid()
  drm/dp_mst: Fix NULL deref in get_mst_branch_device_by_guid_helper()
  drm/amdgpu: ignore duplicate BOs again

17 months agoMerge tag 'amd-drm-fixes-6.6-2023-10-25' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Fri, 27 Oct 2023 02:13:29 +0000 (12:13 +1000)]
Merge tag 'amd-drm-fixes-6.6-2023-10-25' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.6-2023-10-25:

amdgpu:
- Extend VI APSM quirks to more platforms

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231026035452.14921-1-alexander.deucher@amd.com
17 months agoMerge tag 'drm-intel-fixes-2023-10-26' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Fri, 27 Oct 2023 01:58:28 +0000 (11:58 +1000)]
Merge tag 'drm-intel-fixes-2023-10-26' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

- Determine context valid in OA reports (Umesh)
- Hold GT forcewake during steering operations (Matt Roper)
- Check if PMU is closed before stopping event (Umesh)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZTp8IQ0wxzxVjN7J@intel.com
17 months agoMerge tag 'drm-misc-fixes-2023-10-26' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Fri, 27 Oct 2023 01:50:51 +0000 (11:50 +1000)]
Merge tag 'drm-misc-fixes-2023-10-26' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

Short summary of fixes pull:

amdgpu:
- ignore duplicated BOs in CS parser
- remove redundant call to amdgpu_ctx_priority_is_valid()

amdkfd:
- reserve fence slot while locking BO

dp_mst:
- Fix NULL deref in get_mst_branch_device_by_guid_helper()

logicvc:
- Kconfig: Select REGMAP and REGMAP_MMIO

ivpu:
- Fix missing VPUIP interrupts

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20231026110132.GA10591@linux-uq9g.fritz.box
17 months agoscsi: sd: Introduce manage_shutdown device flag
Damien Le Moal [Wed, 25 Oct 2023 06:46:12 +0000 (15:46 +0900)]
scsi: sd: Introduce manage_shutdown device flag

Commit aa3998dbeb3a ("ata: libata-scsi: Disable scsi device
manage_system_start_stop") change setting the manage_system_start_stop
flag to false for libata managed disks to enable libata internal
management of disk suspend/resume. However, a side effect of this change
is that on system shutdown, disks are no longer being stopped (set to
standby mode with the heads unloaded). While this is not a critical
issue, this unclean shutdown is not recommended and shows up with
increased smart counters (e.g. the unexpected power loss counter
"Unexpect_Power_Loss_Ct").

Instead of defining a shutdown driver method for all ATA adapter
drivers (not all of them define that operation), this patch resolves
this issue by further refining the sd driver start/stop control of disks
using the new flag manage_shutdown. If this new flag is set to true by
a low level driver, the function sd_shutdown() will issue a
START STOP UNIT command with the start argument set to 0 when a disk
needs to be powered off (suspended) on system power off, that is, when
system_state is equal to SYSTEM_POWER_OFF.

Similarly to the other manage_xxx flags, the new manage_shutdown flag is
exposed through sysfs as a read-write device attribute.

To avoid any confusion between manage_shutdown and
manage_system_start_stop, the comments describing these flags in
include/scsi/scsi.h are also improved.

Fixes: aa3998dbeb3a ("ata: libata-scsi: Disable scsi device manage_system_start_stop")
Cc: stable@vger.kernel.org
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=218038
Link: https://lore.kernel.org/all/cd397c88-bf53-4768-9ab8-9d107df9e613@gmail.com/
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Niklas Cassel <niklas.cassel@wdc.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Acked-by: Martin K. Petersen <martin.petersen@oracle.com>
17 months agoMerge tag 'soc-fixes-6.7-3' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Thu, 26 Oct 2023 18:17:26 +0000 (08:17 -1000)]
Merge tag 'soc-fixes-6.7-3' of git://git./linux/kernel/git/soc/soc

Pull ARM SoC fixes from Arnd Bergmann:
 "A couple of platforms have some last-minute fixes, in particular:

   - riscv gets some fixes for noncoherent DMA on the renesas and thead
     platforms and dts fix for SPI on the visionfive 2 board

   - Qualcomm Snapdragon gets three dts fixes to address board specific
     regressions on the pmic and gpio nodes

   - Rockchip platforms get multiple dts fixes to address issues on the
     recent rk3399 platform as well as the older rk3128 platform that
     apparently regressed a while ago.

   - TI OMAP gets some trivial code and dts fixes and a regression fix
     for the omap1 ams-delta modem

   - NXP i.MX firmware has one fix for a use-after-free but in its error
     handling"

* tag 'soc-fixes-6.7-3' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (25 commits)
  soc: renesas: ARCH_R9A07G043 depends on !RISCV_ISA_ZICBOM
  riscv: only select DMA_DIRECT_REMAP from RISCV_ISA_ZICBOM and ERRATA_THEAD_PBMT
  riscv: RISCV_NONSTANDARD_CACHE_OPS shouldn't depend on RISCV_DMA_NONCOHERENT
  riscv: dts: thead: set dma-noncoherent to soc bus
  arm64: dts: rockchip: Fix i2s0 pin conflict on ROCK Pi 4 boards
  arm64: dts: rockchip: Add i2s0-2ch-bus-bclk-off pins to RK3399
  clk: ti: Fix missing omap5 mcbsp functional clock and aliases
  clk: ti: Fix missing omap4 mcbsp functional clock and aliases
  ARM: OMAP1: ams-delta: Fix MODEM initialization failure
  soc: renesas: Make ARCH_R9A07G043 depend on required options
  riscv: dts: starfive: visionfive 2: correct spi's ss pin
  firmware/imx-dsp: Fix use_after_free in imx_dsp_setup_channels()
  ARM: OMAP: timer32K: fix all kernel-doc warnings
  ARM: omap2: fix a debug printk
  ARM: dts: rockchip: Fix timer clocks for RK3128
  ARM: dts: rockchip: Add missing quirk for RK3128's dma engine
  ARM: dts: rockchip: Add missing arm timer interrupt for RK3128
  ARM: dts: rockchip: Fix i2c0 register address for RK3128
  arm64: dts: rockchip: set codec system-clock-fixed on px30-ringneck-haikou
  arm64: dts: rockchip: use codec as clock master on px30-ringneck-haikou
  ...

17 months agoMerge tag 'net-6.6-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 26 Oct 2023 17:41:27 +0000 (07:41 -1000)]
Merge tag 'net-6.6-rc8' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Paolo Abeni:
 "Including fixes from WiFi and netfilter.

  Most regressions addressed here come from quite old versions, with the
  exceptions of the iavf one and the WiFi fixes. No known outstanding
  reports or investigation.

  Fixes to fixes:

   - eth: iavf: in iavf_down, disable queues when removing the driver

  Previous releases - regressions:

   - sched: act_ct: additional checks for outdated flows

   - tcp: do not leave an empty skb in write queue

   - tcp: fix wrong RTO timeout when received SACK reneging

   - wifi: cfg80211: pass correct pointer to rdev_inform_bss()

   - eth: i40e: sync next_to_clean and next_to_process for programming
     status desc

   - eth: iavf: initialize waitqueues before starting watchdog_task

  Previous releases - always broken:

   - eth: r8169: fix data-races

   - eth: igb: fix potential memory leak in igb_add_ethtool_nfc_entry

   - eth: r8152: avoid writing garbage to the adapter's registers

   - eth: gtp: fix fragmentation needed check with gso"

* tag 'net-6.6-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (43 commits)
  iavf: in iavf_down, disable queues when removing the driver
  vsock/virtio: initialize the_virtio_vsock before using VQs
  net: ipv6: fix typo in comments
  net: ipv4: fix typo in comments
  net/sched: act_ct: additional checks for outdated flows
  netfilter: flowtable: GC pushes back packets to classic path
  i40e: Fix wrong check for I40E_TXR_FLAGS_WB_ON_ITR
  gtp: fix fragmentation needed check with gso
  gtp: uapi: fix GTPA_MAX
  Fix NULL pointer dereference in cn_filter()
  sfc: cleanup and reduce netlink error messages
  net/handshake: fix file ref count in handshake_nl_accept_doit()
  wifi: mac80211: don't drop all unprotected public action frames
  wifi: cfg80211: fix assoc response warning on failed links
  wifi: cfg80211: pass correct pointer to rdev_inform_bss()
  isdn: mISDN: hfcsusb: Spelling fix in comment
  tcp: fix wrong RTO timeout when received SACK reneging
  r8152: Block future register access if register access fails
  r8152: Rename RTL8152_UNPLUG to RTL8152_INACCESSIBLE
  r8152: Check for unplug in r8153b_ups_en() / r8153c_ups_en()
  ...

17 months agoMerge tag 'renesas-fixes-for-v6.6-tag3' of git://git.kernel.org/pub/scm/linux/kernel...
Arnd Bergmann [Thu, 26 Oct 2023 15:06:37 +0000 (17:06 +0200)]
Merge tag 'renesas-fixes-for-v6.6-tag3' of git://git./linux/kernel/git/geert/renesas-devel into arm/fixes

Renesas fixes for v6.6 (take three)

  - Sort out a few Kconfig dependency issues for the rich set of RISC-V
    non-coherent DMA support.

* tag 'renesas-fixes-for-v6.6-tag3' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/renesas-devel:
  soc: renesas: ARCH_R9A07G043 depends on !RISCV_ISA_ZICBOM
  riscv: only select DMA_DIRECT_REMAP from RISCV_ISA_ZICBOM and ERRATA_THEAD_PBMT
  riscv: RISCV_NONSTANDARD_CACHE_OPS shouldn't depend on RISCV_DMA_NONCOHERENT

Link: https://lore.kernel.org/r/cover.1698312384.git.geert+renesas@glider.be
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
17 months agosoc: renesas: ARCH_R9A07G043 depends on !RISCV_ISA_ZICBOM
Christoph Hellwig [Wed, 18 Oct 2023 05:26:54 +0000 (07:26 +0200)]
soc: renesas: ARCH_R9A07G043 depends on !RISCV_ISA_ZICBOM

ARCH_R9A07G043 has its own non-standard global pool based DMA coherent
allocator, which conflicts with the remap based RISCV_ISA_ZICBOM version.
Add a proper dependency.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20231018052654.50074-4-hch@lst.de
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
17 months agoriscv: only select DMA_DIRECT_REMAP from RISCV_ISA_ZICBOM and ERRATA_THEAD_PBMT
Christoph Hellwig [Wed, 18 Oct 2023 05:26:53 +0000 (07:26 +0200)]
riscv: only select DMA_DIRECT_REMAP from RISCV_ISA_ZICBOM and ERRATA_THEAD_PBMT

RISCV_DMA_NONCOHERENT is also used for whacky non-standard
non-coherent ops that use different hooks in dma-direct.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Tested-by: Samuel Holland <samuel.holland@sifive.com>
Link: https://lore.kernel.org/r/20231018052654.50074-3-hch@lst.de
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
17 months agoriscv: RISCV_NONSTANDARD_CACHE_OPS shouldn't depend on RISCV_DMA_NONCOHERENT
Christoph Hellwig [Wed, 18 Oct 2023 05:26:52 +0000 (07:26 +0200)]
riscv: RISCV_NONSTANDARD_CACHE_OPS shouldn't depend on RISCV_DMA_NONCOHERENT

RISCV_NONSTANDARD_CACHE_OPS is also used for the pmem cache maintenance
helpers, which are built into the kernel unconditionally.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20231018052654.50074-2-hch@lst.de
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
17 months agoaccel/ivpu/37xx: Fix missing VPUIP interrupts
Karol Wachowski [Tue, 24 Oct 2023 16:19:52 +0000 (18:19 +0200)]
accel/ivpu/37xx: Fix missing VPUIP interrupts

Move sequence of masking and unmasking global interrupts from buttress
interrupt handler to generic one that handles both VPUIP and BTRS
interrupts. Unmasking global interrupts will re-trigger MSI for any
pending interrupts.

Lack of this sequence will cause the driver to miss any
VPUIP interrupt that comes after reading VPU_37XX_HOST_SS_ICB_STATUS_0
and before clearing all active interrupt sources.

Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU")
Cc: stable@vger.kernel.org
Signed-off-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231024161952.759914-1-stanislaw.gruszka@linux.intel.com
17 months agoiavf: in iavf_down, disable queues when removing the driver
Michal Schmidt [Wed, 25 Oct 2023 18:32:13 +0000 (11:32 -0700)]
iavf: in iavf_down, disable queues when removing the driver

In iavf_down, we're skipping the scheduling of certain operations if
the driver is being removed. However, the IAVF_FLAG_AQ_DISABLE_QUEUES
request must not be skipped in this case, because iavf_close waits
for the transition to the __IAVF_DOWN state, which happens in
iavf_virtchnl_completion after the queues are released.

Without this fix, "rmmod iavf" takes half a second per interface that's
up and prints the "Device resources not yet released" warning.

Fixes: c8de44b577eb ("iavf: do not process adminq tasks when __IAVF_IN_REMOVE_TASK is set")
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Tested-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20231025183213.874283-1-jacob.e.keller@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
17 months agoMerge tag 'nf-23-10-25' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf
Jakub Kicinski [Wed, 25 Oct 2023 23:02:06 +0000 (16:02 -0700)]
Merge tag 'nf-23-10-25' of git://git./linux/kernel/git/netfilter/nf

Pablo Neira Ayuso says:

====================
Netfilter fixes for net

This patch contains two late Netfilter's flowtable fixes for net:

1) Flowtable GC pushes back packets to classic path in every GC run,
   ie. every second. This is because NF_FLOW_HW_ESTABLISHED is only
   used by sched/act_ct (never set) and IPS_SEEN_REPLY might be unset
   by the time the flow is offloaded (this status bit is only reliable
   in the sched/act_ct datapath).

2) sched/act_ct logic to push back packets to classic path to reevaluate
   if UDP flow is unidirectional only applies if IPS_HW_OFFLOAD_BIT is
   set on and no hardware offload request is pending to be handled.
   From Vlad Buslov.

These two patches fixes two problems that were introduced in the
previous 6.5 development cycle.

* tag 'nf-23-10-25' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
  net/sched: act_ct: additional checks for outdated flows
  netfilter: flowtable: GC pushes back packets to classic path
====================

Link: https://lore.kernel.org/r/20231025100819.2664-1-pablo@netfilter.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
17 months agovsock/virtio: initialize the_virtio_vsock before using VQs
Alexandru Matei [Tue, 24 Oct 2023 19:17:42 +0000 (22:17 +0300)]
vsock/virtio: initialize the_virtio_vsock before using VQs

Once VQs are filled with empty buffers and we kick the host, it can send
connection requests. If the_virtio_vsock is not initialized before,
replies are silently dropped and do not reach the host.

virtio_transport_send_pkt() can queue packets once the_virtio_vsock is
set, but they won't be processed until vsock->tx_run is set to true. We
queue vsock->send_pkt_work when initialization finishes to send those
packets queued earlier.

Fixes: 0deab087b16a ("vsock/virtio: use RCU to avoid use-after-free on the_virtio_vsock")
Signed-off-by: Alexandru Matei <alexandru.matei@uipath.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Link: https://lore.kernel.org/r/20231024191742.14259-1-alexandru.matei@uipath.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
17 months agoirqchip/gic-v3-its: Don't override quirk settings with default values
Marc Zyngier [Tue, 24 Oct 2023 14:34:31 +0000 (15:34 +0100)]
irqchip/gic-v3-its: Don't override quirk settings with default values

When splitting the allocation of the ITS node from its configuration,
some of the default settings were kept in the latter instead of
being moved to the former.

This has the side effect of negating some of the quirk detections that
have happened in between, amongst which the dreaded Synquacer hack
(that also affect Dominic's TI platform).

Move the initialisation of these fields early, so that they can again be
overriden by the Synquacer quirk.

Fixes: 9585a495ac93 ("irqchip/gic-v3-its: Split allocation from initialisation of its_node")
Reported by: Dominic Rath <dominic.rath@ibv-augsburg.net>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Dominic Rath <dominic.rath@ibv-augsburg.net>
Link: https://lore.kernel.org/r/20231024084831.GA3788@JADEVM-DRA
Link: https://lore.kernel.org/r/20231024143431.2144579-1-maz@kernel.org
17 months agoMerge tag 'acpi-6.6-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Wed, 25 Oct 2023 17:51:56 +0000 (07:51 -1000)]
Merge tag 'acpi-6.6-rc8' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI fix from Rafael Wysocki:
 "Unbreak the ACPI NFIT driver after a recent change that inadvertently
  altered its behavior (Xiang Chen)"

* tag 'acpi-6.6-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI: NFIT: Install Notify() handler before getting NFIT table

17 months agoswiotlb: do not try to allocate a TLB bigger than MAX_ORDER pages
Petr Tesarik [Wed, 25 Oct 2023 08:44:25 +0000 (10:44 +0200)]
swiotlb: do not try to allocate a TLB bigger than MAX_ORDER pages

When allocating a new pool at runtime, reduce the number of slabs so
that the allocation order is at most MAX_ORDER.  This avoids a kernel
warning in __alloc_pages().

The warning is relatively benign, because the pool size is subsequently
reduced when allocation fails, but it is silly to start with a request
that is known to fail, especially since this is the default behavior if
the kernel is built with CONFIG_SWIOTLB_DYNAMIC=y and booted without any
swiotlb= parameter.

Reported-by: Ben Greear <greearb@candelatech.com>
Closes: https://lore.kernel.org/netdev/4f173dd2-324a-0240-ff8d-abf5c191be18@candelatech.com/
Fixes: 1aaa736815eb ("swiotlb: allocate a new memory pool when existing pools are full")
Signed-off-by: Petr Tesarik <petr.tesarik1@huawei-partners.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
17 months agoio_uring/rw: disable IOCB_DIO_CALLER_COMP
Jens Axboe [Tue, 24 Oct 2023 20:39:06 +0000 (14:39 -0600)]
io_uring/rw: disable IOCB_DIO_CALLER_COMP

If an application does O_DIRECT writes with io_uring and the file system
supports IOCB_DIO_CALLER_COMP, then completions of the dio write side is
done from the task_work that will post the completion event for said
write as well.

Whenever a dio write is done against a file, the inode i_dio_count is
elevated. This enables other callers to use inode_dio_wait() to wait for
previous writes to complete. If we defer the full dio completion to
task_work, we are dependent on that task_work being run before the
inode i_dio_count can be decremented.

If the same task that issues io_uring dio writes with
IOCB_DIO_CALLER_COMP performs a synchronous system call that calls
inode_dio_wait(), then we can deadlock as we're blocked sleeping on
the event to become true, but not processing the completions that will
result in the inode i_dio_count being decremented.

Until we can guarantee that this is the case, then disable the deferred
caller completions.

Fixes: 099ada2c8726 ("io_uring/rw: add write support for IOCB_DIO_CALLER_COMP")
Reported-by: Andres Freund <andres@anarazel.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
17 months agodrm/amd: Disable ASPM for VI w/ all Intel systems
Mario Limonciello [Fri, 20 Oct 2023 15:26:29 +0000 (10:26 -0500)]
drm/amd: Disable ASPM for VI w/ all Intel systems

Originally we were quirking ASPM disabled specifically for VI when
used with Alder Lake, but it appears to have problems with Rocket
Lake as well.

Like we've done in the case of dpm for newer platforms, disable
ASPM for all Intel systems.

Cc: stable@vger.kernel.org # 5.15+
Fixes: 0064b0ce85bb ("drm/amd/pm: enable ASPM by default")
Reported-and-tested-by: Paolo Gentili <paolo.gentili@canonical.com>
Closes: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2036742
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
17 months agoio_uring/fdinfo: lock SQ thread while retrieving thread cpu/pid
Jens Axboe [Sat, 21 Oct 2023 18:30:29 +0000 (12:30 -0600)]
io_uring/fdinfo: lock SQ thread while retrieving thread cpu/pid

We could race with SQ thread exit, and if we do, we'll hit a NULL pointer
dereference when the thread is cleared. Grab the SQPOLL data lock before
attempting to get the task cpu and pid for fdinfo, this ensures we have a
stable view of it.

Cc: stable@vger.kernel.org
Link: https://bugzilla.kernel.org/show_bug.cgi?id=218032
Reviewed-by: Gabriel Krisman Bertazi <krisman@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
17 months agodrm/i915/pmu: Check if pmu is closed before stopping event
Umesh Nerlige Ramappa [Fri, 20 Oct 2023 15:24:41 +0000 (08:24 -0700)]
drm/i915/pmu: Check if pmu is closed before stopping event

When the driver unbinds, pmu is unregistered and i915->uabi_engines is
set to RB_ROOT. Due to this, when i915 PMU tries to stop the engine
events, it issues a warn_on because engine lookup fails.

All perf hooks are taking care of this using a pmu->closed flag that is
set when PMU unregisters. The stop event seems to have been left out.

Check for pmu->closed in pmu_event_stop as well.

Based on discussion here -
https://patchwork.freedesktop.org/patch/492079/?series=105790&rev=2

v2: s/is/if/ in commit title
v3: Add fixes tag and cc stable

Cc: <stable@vger.kernel.org> # v5.11+
Fixes: b00bccb3f0bb ("drm/i915/pmu: Handle PCI unbind")
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231020152441.3764850-1-umesh.nerlige.ramappa@intel.com
(cherry picked from commit 31f6a06f0c543b43a38fab10f39e5fc45ad62aa2)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
17 months agodrm/i915/mcr: Hold GT forcewake during steering operations
Matt Roper [Thu, 19 Oct 2023 17:02:42 +0000 (10:02 -0700)]
drm/i915/mcr: Hold GT forcewake during steering operations

The steering control and semaphore registers are inside an "always on"
power domain with respect to RC6.  However there are some issues if
higher-level platform sleep states are entering/exiting at the same time
these registers are accessed.  Grabbing GT forcewake and holding it over
the entire lock/steer/unlock cycle ensures that those sleep states have
been fully exited before we access these registers.

This is expected to become a formally documented/numbered workaround
soon.

Note that this patch alone isn't expected to have an immediately
noticeable impact on MCR (mis)behavior; an upcoming pcode firmware
update will also be necessary to provide the other half of this
workaround.

v2:
 - Move the forcewake inside the Xe_LPG-specific IP version check.  This
   should only be necessary on platforms that have a steering semaphore.

Fixes: 3100240bf846 ("drm/i915/mtl: Add hardware-level lock for steering")
Cc: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Cc: Jonathan Cavitt <jonathan.cavitt@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231019170241.2102037-2-matthew.d.roper@intel.com
(cherry picked from commit 8fa1c7cd1fe9cdfc426a603e1f1eecd3f463c487)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
17 months agodrm/logicvc: Kconfig: select REGMAP and REGMAP_MMIO
Sui Jingfeng [Thu, 8 Jun 2023 02:42:07 +0000 (10:42 +0800)]
drm/logicvc: Kconfig: select REGMAP and REGMAP_MMIO

drm/logicvc driver is depend on REGMAP and REGMAP_MMIO, should select this
two kconfig option, otherwise the driver failed to compile on platform
without REGMAP_MMIO selected:

ERROR: modpost: "__devm_regmap_init_mmio_clk" [drivers/gpu/drm/logicvc/logicvc-drm.ko] undefined!
make[1]: *** [scripts/Makefile.modpost:136: Module.symvers] Error 1
make: *** [Makefile:1978: modpost] Error 2

Signed-off-by: Sui Jingfeng <suijingfeng@loongson.cn>
Acked-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Fixes: efeeaefe9be5 ("drm: Add support for the LogiCVC display controller")
Link: https://patchwork.freedesktop.org/patch/msgid/20230608024207.581401-1-suijingfeng@loongson.cn
Signed-off-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
17 months agonet: ipv6: fix typo in comments
Deming Wang [Wed, 25 Oct 2023 06:16:56 +0000 (02:16 -0400)]
net: ipv6: fix typo in comments

The word "advertize" should be replaced by "advertise".

Signed-off-by: Deming Wang <wangdeming@inspur.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 months agonet: ipv4: fix typo in comments
Deming Wang [Wed, 25 Oct 2023 06:14:34 +0000 (02:14 -0400)]
net: ipv4: fix typo in comments

The word "advertize" should be replaced by "advertise".

Signed-off-by: Deming Wang <wangdeming@inspur.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 months agonet/sched: act_ct: additional checks for outdated flows
Vlad Buslov [Tue, 24 Oct 2023 19:58:57 +0000 (21:58 +0200)]
net/sched: act_ct: additional checks for outdated flows

Current nf_flow_is_outdated() implementation considers any flow table flow
which state diverged from its underlying CT connection status for teardown
which can be problematic in the following cases:

- Flow has never been offloaded to hardware in the first place either
because flow table has hardware offload disabled (flag
NF_FLOWTABLE_HW_OFFLOAD is not set) or because it is still pending on 'add'
workqueue to be offloaded for the first time. The former is incorrect, the
later generates excessive deletions and additions of flows.

- Flow is already pending to be updated on the workqueue. Tearing down such
flows will also generate excessive removals from the flow table, especially
on highly loaded system where the latency to re-offload a flow via 'add'
workqueue can be quite high.

When considering a flow for teardown as outdated verify that it is both
offloaded to hardware and doesn't have any pending updates.

Fixes: 41f2c7c342d3 ("net/sched: act_ct: Fix promotion of offloaded unreplied tuple")
Reviewed-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Vlad Buslov <vladbu@nvidia.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
17 months agonetfilter: flowtable: GC pushes back packets to classic path
Pablo Neira Ayuso [Tue, 24 Oct 2023 19:09:47 +0000 (21:09 +0200)]
netfilter: flowtable: GC pushes back packets to classic path

Since 41f2c7c342d3 ("net/sched: act_ct: Fix promotion of offloaded
unreplied tuple"), flowtable GC pushes back flows with IPS_SEEN_REPLY
back to classic path in every run, ie. every second. This is because of
a new check for NF_FLOW_HW_ESTABLISHED which is specific of sched/act_ct.

In Netfilter's flowtable case, NF_FLOW_HW_ESTABLISHED never gets set on
and IPS_SEEN_REPLY is unreliable since users decide when to offload the
flow before, such bit might be set on at a later stage.

Fix it by adding a custom .gc handler that sched/act_ct can use to
deal with its NF_FLOW_HW_ESTABLISHED bit.

Fixes: 41f2c7c342d3 ("net/sched: act_ct: Fix promotion of offloaded unreplied tuple")
Reported-by: Vladimir Smelhaus <vl.sm@email.cz>
Reviewed-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
17 months agopowerpc/mm: Avoid calling arch_enter/leave_lazy_mmu() in set_ptes
Aneesh Kumar K.V [Tue, 24 Oct 2023 14:36:04 +0000 (20:06 +0530)]
powerpc/mm: Avoid calling arch_enter/leave_lazy_mmu() in set_ptes

With commit 9fee28baa601 ("powerpc: implement the new page table range
API") we added set_ptes to powerpc architecture. The implementation
included calling arch_enter/leave_lazy_mmu() calls.

The patch removes the usage of arch_enter/leave_lazy_mmu() because
set_pte is not supposed to be used when updating a pte entry. Powerpc
architecture uses this rule to skip the expensive tlb invalidate which
is not needed when you are setting up the pte for the first time. See
commit 56eecdb912b5 ("mm: Use ptep/pmdp_set_numa() for updating
_PAGE_NUMA bit") for more details

The patch also makes sure we are not using the interface to update a
valid/present pte entry by adding VM_WARN_ON check all the ptes we
are setting up. Furthermore, we add a comment to set_pte_filter to
clarify it can only update folio-related flags and cannot filter
pfn specific details in pte filtering.

Removal of arch_enter/leave_lazy_mmu() also will avoid nesting of
these functions that are not supported. For ex:

remap_pte_range()
  -> arch_enter_lazy_mmu()
  -> set_ptes()
      -> arch_enter_lazy_mmu()
      -> arch_leave_lazy_mmu()
  -> arch_leave_lazy_mmu()

Fixes: 9fee28baa601 ("powerpc: implement the new page table range API")
Signed-off-by: "Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20231024143604.16749-1-aneesh.kumar@linux.ibm.com
17 months agoi40e: Fix wrong check for I40E_TXR_FLAGS_WB_ON_ITR
Ivan Vecera [Mon, 23 Oct 2023 21:27:14 +0000 (14:27 -0700)]
i40e: Fix wrong check for I40E_TXR_FLAGS_WB_ON_ITR

The I40E_TXR_FLAGS_WB_ON_ITR is i40e_ring flag and not i40e_pf one.

Fixes: 8e0764b4d6be42 ("i40e/i40evf: Add support for writeback on ITR feature for X722")
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Tested-by: Pucha Himasekhar Reddy <himasekharx.reddy.pucha@intel.com> (A Contingent worker at Intel)
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20231023212714.178032-1-jacob.e.keller@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
17 months agoMerge tag 'wireless-2023-10-24' of git://git.kernel.org/pub/scm/linux/kernel/git...
Jakub Kicinski [Tue, 24 Oct 2023 20:10:53 +0000 (13:10 -0700)]
Merge tag 'wireless-2023-10-24' of git://git./linux/kernel/git/wireless/wireless

Johannes Berg says:

====================
Three more fixes:
 - don't drop all unprotected public action frames since
   some don't have a protected dual
 - fix pointer confusion in scanning code
 - fix warning in some connections with multiple links

* tag 'wireless-2023-10-24' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless:
  wifi: mac80211: don't drop all unprotected public action frames
  wifi: cfg80211: fix assoc response warning on failed links
  wifi: cfg80211: pass correct pointer to rdev_inform_bss()
====================

Link: https://lore.kernel.org/r/20231024103540.19198-2-johannes@sipsolutions.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
17 months agoMerge tag 'mm-hotfixes-stable-2023-10-24-09-40' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Tue, 24 Oct 2023 19:52:16 +0000 (09:52 -1000)]
Merge tag 'mm-hotfixes-stable-2023-10-24-09-40' of git://git./linux/kernel/git/akpm/mm

Pull misc fixes from Andrew Morton:
 "20 hotfixes. 12 are cc:stable and the remainder address post-6.5
  issues or aren't considered necessary for earlier kernel versions"

* tag 'mm-hotfixes-stable-2023-10-24-09-40' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm:
  maple_tree: add GFP_KERNEL to allocations in mas_expected_entries()
  selftests/mm: include mman header to access MREMAP_DONTUNMAP identifier
  mailmap: correct email aliasing for Oleksij Rempel
  mailmap: map Bartosz's old address to the current one
  mm/damon/sysfs: check DAMOS regions update progress from before_terminate()
  MAINTAINERS: Ondrej has moved
  kasan: disable kasan_non_canonical_hook() for HW tags
  kasan: print the original fault addr when access invalid shadow
  hugetlbfs: close race between MADV_DONTNEED and page fault
  hugetlbfs: extend hugetlb_vma_lock to private VMAs
  hugetlbfs: clear resv_map pointer if mmap fails
  mm: zswap: fix pool refcount bug around shrink_worker()
  mm/migrate: fix do_pages_move for compat pointers
  riscv: fix set_huge_pte_at() for NAPOT mappings when a swap entry is set
  riscv: handle VM_FAULT_[HWPOISON|HWPOISON_LARGE] faults instead of panicking
  mmap: fix error paths with dup_anon_vma()
  mmap: fix vma_iterator in error path of vma_merge()
  mm: fix vm_brk_flags() to not bail out while holding lock
  mm/mempolicy: fix set_mempolicy_home_node() previous VMA pointer
  mm/page_alloc: correct start page when guard page debug is enabled

17 months agofpga: Fix memory leak for fpga_region_test_class_find()
Jinjie Ruan [Mon, 23 Oct 2023 03:28:57 +0000 (11:28 +0800)]
fpga: Fix memory leak for fpga_region_test_class_find()

fpga_region_class_find() in fpga_region_test_class_find() will call
get_device() if the data is matched, which will increment refcount for
dev->kobj, so it should call put_device() to decrement refcount for
dev->kobj to free the region, because fpga_region_unregister() will call
fpga_region_dev_release() only when the refcount for dev->kobj is zero
but fpga_region_test_init() call device_register() in
fpga_region_register_full(), which also increment refcount.

So call put_device() after calling fpga_region_class_find() in
fpga_region_test_class_find(). After applying this patch, the following
memory leak is never detected.

unreferenced object 0xffff88810c8ef000 (size 1024):
  comm "kunit_try_catch", pid 1875, jiffies 4294715298 (age 836.836s)
  hex dump (first 32 bytes):
    b8 d1 fb 05 81 88 ff ff 08 f0 8e 0c 81 88 ff ff  ................
    08 f0 8e 0c 81 88 ff ff 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<ffffffff817ebad7>] kmalloc_trace+0x27/0xa0
    [<ffffffffa02385e1>] fpga_region_register_full+0x51/0x430 [fpga_region]
    [<ffffffffa0228e47>] 0xffffffffa0228e47
    [<ffffffff829c479d>] kunit_try_run_case+0xdd/0x250
    [<ffffffff829c9f2a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
    [<ffffffff81238b85>] kthread+0x2b5/0x380
    [<ffffffff81097ded>] ret_from_fork+0x2d/0x70
    [<ffffffff810034d1>] ret_from_fork_asm+0x11/0x20
unreferenced object 0xffff888105fbd1b8 (size 8):
  comm "kunit_try_catch", pid 1875, jiffies 4294715298 (age 836.836s)
  hex dump (first 8 bytes):
    72 65 67 69 6f 6e 30 00                          region0.
  backtrace:
    [<ffffffff817ec023>] __kmalloc_node_track_caller+0x53/0x150
    [<ffffffff82995590>] kvasprintf+0xb0/0x130
    [<ffffffff83f713b1>] kobject_set_name_vargs+0x41/0x110
    [<ffffffff8304ac1b>] dev_set_name+0xab/0xe0
    [<ffffffffa02388a2>] fpga_region_register_full+0x312/0x430 [fpga_region]
    [<ffffffffa0228e47>] 0xffffffffa0228e47
    [<ffffffff829c479d>] kunit_try_run_case+0xdd/0x250
    [<ffffffff829c9f2a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
    [<ffffffff81238b85>] kthread+0x2b5/0x380
    [<ffffffff81097ded>] ret_from_fork+0x2d/0x70
    [<ffffffff810034d1>] ret_from_fork_asm+0x11/0x20
unreferenced object 0xffff88810b3b8a00 (size 256):
  comm "kunit_try_catch", pid 1875, jiffies 4294715298 (age 836.836s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 08 8a 3b 0b 81 88 ff ff  ..........;.....
    08 8a 3b 0b 81 88 ff ff e0 ac 04 83 ff ff ff ff  ..;.............
  backtrace:
    [<ffffffff817ebad7>] kmalloc_trace+0x27/0xa0
    [<ffffffff83056d7a>] device_add+0xa2a/0x15e0
    [<ffffffffa02388b1>] fpga_region_register_full+0x321/0x430 [fpga_region]
    [<ffffffffa0228e47>] 0xffffffffa0228e47
    [<ffffffff829c479d>] kunit_try_run_case+0xdd/0x250
    [<ffffffff829c9f2a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
    [<ffffffff81238b85>] kthread+0x2b5/0x380
    [<ffffffff81097ded>] ret_from_fork+0x2d/0x70
    [<ffffffff810034d1>] ret_from_fork_asm+0x11/0x20

Fixes: 64a5f972c93d ("fpga: add an initial KUnit suite for the FPGA Region")
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Marco Pagani <marpagan@redhat.com>
Acked-by: Xu Yilun <yilun.xu@intel.com>
Link: https://lore.kernel.org/r/20231007094321.3447084-1-ruanjinjie@huawei.com
[yilun.xu@intel.com: slightly changes the commit message]
Signed-off-by: Xu Yilun <yilun.xu@linux.intel.com>
Link: https://lore.kernel.org/r/20231023032857.902699-3-yilun.xu@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agofpga: m10bmc-sec: Change contact for secure update driver
Russ Weight [Mon, 23 Oct 2023 03:28:56 +0000 (11:28 +0800)]
fpga: m10bmc-sec: Change contact for secure update driver

Change the maintainer for the Intel MAX10 BMC Secure Update driver from
Russ Weight to Peter Colberg. Update the ABI documentation contact
information as well.

Signed-off-by: Russ Weight <russell.h.weight@intel.com>
Acked-by: Peter Colberg <peter.colberg@intel.com>
Link: https://lore.kernel.org/r/20230928164753.278684-1-russell.h.weight@intel.com
Signed-off-by: Xu Yilun <yilun.xu@linux.intel.com>
Link: https://lore.kernel.org/r/20231023032857.902699-2-yilun.xu@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
17 months agodrm/i915/perf: Determine context valid in OA reports
Umesh Nerlige Ramappa [Wed, 2 Aug 2023 20:28:54 +0000 (13:28 -0700)]
drm/i915/perf: Determine context valid in OA reports

When supporting OA for TGL, it was seen that the context valid bit in
the report ID was not defined, however revisiting the spec seems to have
this bit defined. The bit is used to determine if a context is valid on
a context switch and is essential to determine active and idle periods
for a context. Re-enable the context valid bit for gen12 platforms.

BSpec: 52196 (description of report_id)

v2: Include BSpec reference (Ashutosh)

Fixes: 00a7f0d7155c ("drm/i915/tgl: Add perf support on TGL")
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230802202854.1224547-1-umesh.nerlige.ramappa@intel.com
(cherry picked from commit 7eeaedf79989a8f131939782832e21e9218ed2a0)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>