| Age | Commit message (Collapse) | Author |
|---|
|
iommu_sva_bind_device() should return either a sva bond handle or an
ERR_PTR value in error cases. Existing drivers (idxd and uacce) only
check the return value with IS_ERR(). This could potentially lead to
a kernel NULL pointer dereference issue if the function returns NULL
instead of an error pointer.
In reality, this doesn't cause any problems because iommu_sva_bind_device()
only returns NULL when the kernel is not configured with CONFIG_IOMMU_SVA.
In this case, iommu_dev_enable_feature(dev, IOMMU_DEV_FEAT_SVA) will
return an error, and the device drivers won't call iommu_sva_bind_device()
at all.
Fixes: 26b25a2b98e4 ("iommu: Bind process address spaces to devices")
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Link: https://lore.kernel.org/r/20240528042528.71396-1-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
into next
|
|
Make a new op that receives the device and the mm_struct that the SVA
domain should be created for. Unlike domain_alloc_paging() the dev
argument is never NULL here.
This allows drivers to fully initialize the SVA domain and allocate the
mmu_notifier during allocation. It allows the notifier lifetime to follow
the lifetime of the iommu_domain.
Since we have only one call site, upgrade the new op to return ERR_PTR
instead of NULL.
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
[Removed smmu3 related changes - Vasant]
Signed-off-by: Vasant Hegde <vasant.hegde@amd.com>
Reviewed-by: Tina Zhang <tina.zhang@intel.com>
Link: https://lore.kernel.org/r/20240418103400.6229-15-vasant.hegde@amd.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
It's somewhat hard to see, but arm64's arch_setup_dma_ops() should only
ever call iommu_setup_dma_ops() after a successful iommu_probe_device(),
which means there should be no harm in achieving the same order of
operations by running it off the back of iommu_probe_device() itself.
This then puts it in line with the x86 and s390 .probe_finalize bodges,
letting us pull it all into the main flow properly. As a bonus this lets
us fold in and de-scope the PCI workaround setup as well.
At this point we can also then pull the call up inside the group mutex,
and avoid having to think about whether iommu_group_store_type() could
theoretically race and free the domain if iommu_setup_dma_ops() ran just
*before* iommu_device_use_default_domain() claims it... Furthermore we
replace one .probe_finalize call completely, since the only remaining
implementations are now one which only needs to run once for the initial
boot-time probe, and two which themselves render that path unreachable.
This leaves us a big step closer to realistically being able to unpick
the variety of different things that iommu_setup_dma_ops() has been
muddling together, and further streamline iommu-dma into core API flows
in future.
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com> # For Intel IOMMU
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Tested-by: Hanjun Guo <guohanjun@huawei.com>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Link: https://lore.kernel.org/r/bebea331c1d688b34d9862eefd5ede47503961b8.1713523152.git.robin.murphy@arm.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Make a new op that receives the device and the mm_struct that the SVA
domain should be created for. Unlike domain_alloc_paging() the dev
argument is never NULL here.
This allows drivers to fully initialize the SVA domain and allocate the
mmu_notifier during allocation. It allows the notifier lifetime to follow
the lifetime of the iommu_domain.
Since we have only one call site, upgrade the new op to return ERR_PTR
instead of NULL.
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Vasant Hegde <vasant.hegde@amd.com>
Reviewed-by: Tina Zhang <tina.zhang@intel.com>
Link: https://lore.kernel.org/r/20240311090843.133455-15-vasant.hegde@amd.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20240416080656.60968-12-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
According to Intel VT-d specification revision 4.0, "Private Data"
field has been removed from Page Request/Response.
Since the private data field is not used in fault message, remove the
related definitions in page request descriptor and remove the related
code in page request/response handler, as Intel hasn't shipped any
products which support private data in the page request message.
Signed-off-by: Jingqi Liu <Jingqi.liu@intel.com>
Link: https://lore.kernel.org/r/20240308103811.76744-3-Jingqi.liu@intel.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Existing remove_dev_pasid() callbacks of the underlying iommu drivers
get the attached domain from the group->pasid_array. However, the domain
stored in group->pasid_array is not always correct in all scenarios.
A wrong domain may result in failure in remove_dev_pasid() callback.
To avoid such problems, it is more reliable to pass the domain to the
remove_dev_pasid() op.
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20240328122958.83332-3-yi.l.liu@intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
'x86/amd' and 'core' into next
|
|
The current device_release callback for individual iommu drivers does the
following:
1) Silent IOMMU DMA translation: It detaches any existing domain from the
device and puts it into a blocking state (some drivers might use the
identity state).
2) Resource release: It releases resources allocated during the
device_probe callback and restores the device to its pre-probe state.
Step 1 is challenging for individual iommu drivers because each must check
if a domain is already attached to the device. Additionally, if a deferred
attach never occurred, the device_release should avoid modifying hardware
configuration regardless of the reason for its call.
To simplify this process, introduce a static release_domain within the
iommu_ops structure. It can be either a blocking or identity domain
depending on the iommu hardware. The iommu core will decide whether to
attach this domain before the device_release callback, eliminating the
need for repetitive code in various drivers.
Consequently, the device_release callback can focus solely on the opposite
operations of device_probe, including releasing all resources allocated
during that callback.
Co-developed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20240305013305.204605-2-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Make pointer to fwnode_handle a pointer to const for code safety.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20240216144027.185959-3-krzysztof.kozlowski@linaro.org
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
The xlate callbacks are supposed to translate of_phandle_args to proper
provider without modifying the of_phandle_args. Make the argument
pointer to const for code safety and readability.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20240216144027.185959-2-krzysztof.kozlowski@linaro.org
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Prior to commit 092edaddb660 ("iommu: Support mm PASID 1:n with sva
domains") the code allowed a SVA handle to be bound multiple times to the
same (mm, device) pair. This was alluded to in the kdoc comment, but we
had understood this to be more a remark about allowing multiple devices,
not a literal same-driver re-opening the same SVA.
It turns out uacce and idxd were both relying on the core code to handle
reference counting for same-device same-mm scenarios. As this looks hard
to resolve in the drivers bring it back to the core code.
The new design has changed the meaning of the domain->users refcount to
refer to the number of devices that are sharing that domain for the same
mm. This is part of the design to lift the SVA domain de-duplication out
of the drivers.
Return the old behavior by explicitly de-duplicating the struct iommu_sva
handle. The same (mm, device) will return the same handle pointer and the
core code will handle tracking this. The last unbind of the handle will
destroy it.
Fixes: 092edaddb660 ("iommu: Support mm PASID 1:n with sva domains")
Reported-by: Zhangfei Gao <zhangfei.gao@linaro.org>
Closes: https://lore.kernel.org/all/20240221110658.529-1-zhangfei.gao@linaro.org/
Tested-by: Zhangfei Gao <zhangfei.gao@linaro.org>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Link: https://lore.kernel.org/r/0-v1-9455fc497a6f+3b4-iommu_sva_sharing_jgg@nvidia.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
As the iommu_report_device_fault() has been converted to auto-respond a
page fault if it fails to enqueue it, there's no need to return a code
in any case. Make it return void.
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20240212012227.119381-17-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
The iopf_group_response() should return void, as nothing can do anything
with the failure. This implies that ops->page_response() must also return
void; this is consistent with what the drivers do. The failure paths,
which are all integrity validations of the fault, should be WARN_ON'd,
not return codes.
If the iommu core fails to enqueue the fault, it should respond the fault
directly by calling ops->page_response() instead of returning an error
number and relying on the iommu drivers to do so. Consolidate the error
fault handling code in the core.
Co-developed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20240212012227.119381-16-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Previously, before a group of page faults was passed to the domain's iopf
handler, the last page fault of the group was kept in the list of
iommu_fault_param::faults. In the page fault response path, the group's
last page fault was used to look up the list, and the page faults were
responded to device only if there was a matched fault.
The previous approach seems unnecessarily complex and not performance
friendly. Put the page fault group itself to the outstanding fault list.
It can be removed in the page fault response path or in the
iopf_queue_remove_device() path. The pending list is protected by
iommu_fault_param::lock. To allow checking for the group's presence in
the list using list_empty(), the iopf group should be removed from the
list with list_del_init().
IOMMU_PAGE_RESP_PASID_VALID is set in the code but not used anywhere.
Remove it to make the code clean. IOMMU_PAGE_RESP_PASID_VALID is set
in the response message indicating that the response message includes
a valid PASID value. Actually, we should keep this hardware detail in
the individual driver. When the page fault handling framework in IOMMU
and IOMMUFD subsystems includes a valid PASID in the fault message, the
response message should always contain the same PASID value. Individual
drivers should be responsible for deciding whether to include the PASID
in the messages they provide for the hardware.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Link: https://lore.kernel.org/r/20240212012227.119381-15-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Convert iopf_queue_remove_device() to return void instead of an error code,
as the return value is never used. This removal helper is designed to be
never-failed, so there's no need for error handling.
Ack all outstanding page requests from the device with the response code of
IOMMU_PAGE_RESP_INVALID, indicating device should not attempt any retry.
Add comments to this helper explaining the steps involved in removing a
device from the iopf queue and disabling its PRI. The individual drivers
are expected to be adjusted accordingly. Here we just define the expected
behaviors of the individual iommu driver from the core's perspective.
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Link: https://lore.kernel.org/r/20240212012227.119381-14-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
The per-device fault data structure stores information about faults
occurring on a device. Its lifetime spans from IOPF enablement to
disablement. Multiple paths, including IOPF reporting, handling, and
responding, may access it concurrently.
Previously, a mutex protected the fault data from use after free. But
this is not performance friendly due to the critical nature of IOPF
handling paths.
Refine this with a refcount-based approach. The fault data pointer is
obtained within an RCU read region with a refcount. The fault data
pointer is returned for usage only when the pointer is valid and a
refcount is successfully obtained. The fault data is freed with
kfree_rcu(), ensuring data is only freed after all RCU critical regions
complete.
An iopf handling work starts once an iopf group is created. The handling
work continues until iommu_page_response() is called to respond to the
iopf and the iopf group is freed. During this time, the device fault
parameter should always be available. Add a pointer to the device fault
parameter in the iopf_group structure and hold the reference until the
iopf_group is freed.
Make iommu_page_response() static as it is only used in io-pgfault.c.
Co-developed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Link: https://lore.kernel.org/r/20240212012227.119381-13-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Add CONFIG_IOMMU_IOPF for page fault handling framework and select it
from its real consumer. Move iopf function declaration from iommu-sva.h
to iommu.h and remove iommu-sva.h as it's empty now.
Consolidate all SVA related code into iommu-sva.c:
- Move iommu_sva_domain_alloc() from iommu.c to iommu-sva.c.
- Move sva iopf handling code from io-pgfault.c to iommu-sva.c.
Consolidate iommu_report_device_fault() and iommu_page_response() into
io-pgfault.c.
Export iopf_free_group() and iopf_group_response() for iopf handlers
implemented in modules. Some functions are renamed with more meaningful
names. No other intentional functionality changes.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Tested-by: Longfang Liu <liulongfang@huawei.com>
Link: https://lore.kernel.org/r/20240212012227.119381-11-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Make iommu_queue_iopf() more generic by making the iopf_group a minimal
set of iopf's that an iopf handler of domain should handle and respond
to. Add domain parameter to struct iopf_group so that the handler can
retrieve and use it directly.
Change iommu_queue_iopf() to forward groups of iopf's to the domain's
iopf handler. This is also a necessary step to decouple the sva iopf
handling code from this interface.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Tested-by: Longfang Liu <liulongfang@huawei.com>
Link: https://lore.kernel.org/r/20240212012227.119381-10-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Move iopf_group data structure to iommu.h to make it a minimal set of
faults that a domain's page fault handler should handle.
Add a new function, iopf_free_group(), to free a fault group after all
faults in the group are handled. This function will be made global so
that it can be called from other files, such as iommu-sva.c.
Move iopf_queue data structure to iommu.h to allow the workqueue to be
scheduled out of this file.
This will simplify the sequential patches.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Tested-by: Longfang Liu <liulongfang@huawei.com>
Link: https://lore.kernel.org/r/20240212012227.119381-9-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
The iommu_fault_event and iopf_fault data structures store the same
information about an iopf fault. They are also used in the same way.
Merge these two data structures into a single one to make the code
more concise and easier to maintain.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Tested-by: Longfang Liu <liulongfang@huawei.com>
Link: https://lore.kernel.org/r/20240212012227.119381-8-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
The individual iommu driver reports the iommu page faults by calling
iommu_report_device_fault(), where a pre-registered device fault handler
is called to route the fault to another fault handler installed on the
corresponding iommu domain.
The pre-registered device fault handler is static and won't be dynamic
as the fault handler is eventually per iommu domain. Replace calling
device fault handler with iommu_queue_iopf().
After this replacement, the registering and unregistering fault handler
interfaces are not needed anywhere. Remove the interfaces and the related
data structures to avoid dead code.
Convert cookie parameter of iommu_queue_iopf() into a device pointer that
is really passed.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Tested-by: Longfang Liu <liulongfang@huawei.com>
Link: https://lore.kernel.org/r/20240212012227.119381-7-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
The struct dev_iommu contains two pointers, fault_param and iopf_param.
The fault_param pointer points to a data structure that is used to store
pending faults that are awaiting responses. The iopf_param pointer points
to a data structure that is used to store partial faults that are part of
a Page Request Group.
The fault_param and iopf_param pointers are essentially duplicate. This
causes memory waste. Merge the iopf_device_param pointer into the
iommu_fault_param pointer to consolidate the code and save memory. The
consolidated pointer would be allocated on demand when the device driver
enables the iopf on device, and would be freed after iopf is disabled.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Tested-by: Longfang Liu <liulongfang@huawei.com>
Link: https://lore.kernel.org/r/20240212012227.119381-6-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
struct iommu_fault_page_request and struct iommu_page_response are not
part of uAPI anymore. Convert them to data structures for kAPI.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Tested-by: Longfang Liu <liulongfang@huawei.com>
Link: https://lore.kernel.org/r/20240212012227.119381-5-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
The unrecoverable fault data is not used anywhere. Remove it to avoid
dead code.
Suggested-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Tested-by: Longfang Liu <liulongfang@huawei.com>
Link: https://lore.kernel.org/r/20240212012227.119381-4-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
The iommu fault data is currently defined in uapi/linux/iommu.h, but is
only used inside the iommu subsystem. Move it to linux/iommu.h, where it
will be more accessible to kernel drivers.
With this done, uapi/linux/iommu.h becomes empty and can be removed from
the tree.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Tested-by: Yan Zhao <yan.y.zhao@intel.com>
Tested-by: Longfang Liu <liulongfang@huawei.com>
Link: https://lore.kernel.org/r/20240212012227.119381-2-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Add function to check iommu group mutex lock. So that device drivers can
rely on group mutex lock instead of adding another driver level lock
before modifying driver specific device data structure.
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Vasant Hegde <vasant.hegde@amd.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20240205115615.6053-10-vasant.hegde@amd.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Introduce get_amd_iommu_from_dev() and get_amd_iommu_from_dev_data().
And replace rlookup_amd_iommu() with the new helper function where
applicable to avoid unnecessary loop to look up struct amd_iommu from
struct device.
Suggested-by: Jason Gunthorpe <jgg@ziepe.ca>
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Vasant Hegde <vasant.hegde@amd.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20240205115615.6053-4-vasant.hegde@amd.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/jgg/iommufd
Pull iommufd updates from Jason Gunthorpe:
"This brings the first of three planned user IO page table invalidation
operations:
- IOMMU_HWPT_INVALIDATE allows invalidating the IOTLB integrated into
the iommu itself. The Intel implementation will also generate an
ATC invalidation to flush the device IOTLB as it unambiguously
knows the device, but other HW will not.
It goes along with the prior PR to implement userspace IO page tables
(aka nested translation for VMs) to allow Intel to have full
functionality for simple cases. An Intel implementation of the
operation is provided.
Also fix a small bug in the selftest mock iommu driver probe"
* tag 'for-linus-iommufd' of git://git.kernel.org/pub/scm/linux/kernel/git/jgg/iommufd:
iommufd/selftest: Check the bus type during probe
iommu/vt-d: Add iotlb flush for nested domain
iommufd: Add data structure for Intel VT-d stage-1 cache invalidation
iommufd/selftest: Add coverage for IOMMU_HWPT_INVALIDATE ioctl
iommufd/selftest: Add IOMMU_TEST_OP_MD_CHECK_IOTLB test op
iommufd/selftest: Add mock_domain_cache_invalidate_user support
iommu: Add iommu_copy_struct_from_user_array helper
iommufd: Add IOMMU_HWPT_INVALIDATE
iommu: Add cache_invalidate_user op
|
|
Wrap up the data pointer/num sanity and __iommu_copy_struct_from_user()
call for iommu drivers to copy driver specific data at a specific location
in the struct iommu_user_data_array.
And expect it to be used in cache_invalidate_user ops for example.
Link: https://lore.kernel.org/r/20240111041015.47920-4-yi.l.liu@intel.com
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Co-developed-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
|
|
The updates of the PTEs in the nested page table will be propagated to the
hardware caches.
Add a new domain op cache_invalidate_user() for the userspace to flush the
hardware caches for a nested domain through iommufd. No wrapper for it, as
it's only supposed to be used by iommufd. Then, pass in invalidation
requests in form of a user data array containing a number of invalidation
data entries.
Link: https://lore.kernel.org/r/20240111041015.47920-2-yi.l.liu@intel.com
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
|
|
A perfect driver would only call dev_iommu_priv_set() from its probe
callback. We've made it functionally correct to call it from the of_xlate
by adding a lock around that call.
lockdep assert that iommu_probe_device_lock is held to discourage misuse.
Exclude PPC kernels with CONFIG_FSL_PAMU turned on because FSL_PAMU uses a
global static for its priv and abuses priv for its domain.
Remove the pointless stores of NULL, all these are on paths where the core
code will free dev->iommu after the op returns.
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jerry Snitselaar <jsnitsel@redhat.com>
Tested-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/5-v2-16e4def25ebb+820-iommu_fwspec_p1_jgg@nvidia.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Each mm bound to devices gets a PASID and corresponding sva domains
allocated in iommu_sva_bind_device(), which are referenced by iommu_mm
field of the mm. The PASID is released in __mmdrop(), while a sva domain
is released when no one is using it (the reference count is decremented
in iommu_sva_unbind_device()). However, although sva domains and their
PASID are separate objects such that their own life cycles could be
handled independently, an enqcmd use case may require releasing the
PASID in releasing the mm (i.e., once a PASID is allocated for a mm, it
will be permanently used by the mm and won't be released until the end
of mm) and only allows to drop the PASID after the sva domains are
released. To this end, mmgrab() is called in iommu_sva_domain_alloc() to
increment the mm reference count and mmdrop() is invoked in
iommu_domain_free() to decrement the mm reference count.
Since the required info of PASID and sva domains is kept in struct
iommu_mm_data of a mm, use mm->iommu_mm field instead of the old pasid
field in mm struct. The sva domain list is protected by iommu_sva_lock.
Besides, this patch removes mm_pasid_init(), as with the introduced
iommu_mm structure, initializing mm pasid in mm_init() is unnecessary.
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Tina Zhang <tina.zhang@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20231027000525.1278806-6-tina.zhang@intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Introduce iommu_mm_data structure to keep sva information (pasid and the
related sva domains). Add iommu_mm pointer, pointing to an instance of
iommu_mm_data structure, to mm.
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Tina Zhang <tina.zhang@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20231027000525.1278806-5-tina.zhang@intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
mm_get_enqcmd_pasid() should be used by architecture code and closely
related to learn the PASID value that the x86 ENQCMD operation should
use for the mm.
For the moment SMMUv3 uses this without any connection to ENQCMD, it
will be cleaned up similar to how the prior patch made VT-d use the
PASID argument of set_dev_pasid().
The motivation is to replace mm->pasid with an iommu private data
structure that is introduced in a later patch.
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Tina Zhang <tina.zhang@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20231027000525.1278806-4-tina.zhang@intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Linus suggested that the kconfig here is confusing:
https://lore.kernel.org/all/CAHk-=wgUiAtiszwseM1p2fCJ+sC4XWQ+YN4TanFhUgvUqjr9Xw@mail.gmail.com/
Let's break it into three kconfigs controlling distinct things:
- CONFIG_IOMMU_MM_DATA controls if the mm_struct has the additional
fields for the IOMMU. Currently only PASID, but later patches store
a struct iommu_mm_data *
- CONFIG_ARCH_HAS_CPU_PASID controls if the arch needs the scheduling bit
for keeping track of the ENQCMD instruction. x86 will select this if
IOMMU_SVA is enabled
- IOMMU_SVA controls if the IOMMU core compiles in the SVA support code
for iommu driver use and the IOMMU exported API
This way ARM will not enable CONFIG_ARCH_HAS_CPU_PASID
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Link: https://lore.kernel.org/r/20231027000525.1278806-2-tina.zhang@intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
|
|
Before we can allow drivers to coexist, we need to make sure that one
driver's domain ops can't misinterpret another driver's dev_iommu_priv
data. To that end, add a token to the domain so we can remember how it
was allocated - for now this may as well be the device ops, since they
still correlate 1:1 with drivers. We can trust ourselves for internal
default domain attachment, so add checks to cover all the public attach
interfaces.
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Jerry Snitselaar <jsnitsel@redhat.com>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Link: https://lore.kernel.org/r/097c6f30480e4efe12195d00ba0e84ea4837fb4c.1700589539.git.robin.murphy@arm.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
It turns out there are more subtle races beyond just the main part of
__iommu_probe_device() itself running in parallel - the dev_iommu_free()
on the way out of an unsuccessful probe can still manage to trip up
concurrent accesses to a device's fwspec. Thus, extend the scope of
iommu_probe_device_lock() to also serialise fwspec creation and initial
retrieval.
Reported-by: Zhenhua Huang <quic_zhenhuah@quicinc.com>
Link: https://lore.kernel.org/linux-iommu/e2e20e1c-6450-4ac5-9804-b0000acdf7de@quicinc.com/
Fixes: 01657bc14a39 ("iommu: Avoid races around device probe")
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: André Draszik <andre.draszik@linaro.org>
Tested-by: André Draszik <andre.draszik@linaro.org>
Link: https://lore.kernel.org/r/16f433658661d7cadfea51e7c65da95826112a2b.1700071477.git.robin.murphy@arm.com
Cc: stable@vger.kernel.org
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu
Pull iommu updates from Joerg Roedel:
"Core changes:
- Make default-domains mandatory for all IOMMU drivers
- Remove group refcounting
- Add generic_single_device_group() helper and consolidate drivers
- Cleanup map/unmap ops
- Scaling improvements for the IOVA rcache depot
- Convert dart & iommufd to the new domain_alloc_paging()
ARM-SMMU:
- Device-tree binding update:
- Add qcom,sm7150-smmu-v2 for Adreno on SM7150 SoC
- SMMUv2:
- Support for Qualcomm SDM670 (MDSS) and SM7150 SoCs
- SMMUv3:
- Large refactoring of the context descriptor code to move the CD
table into the master, paving the way for '->set_dev_pasid()'
support on non-SVA domains
- Minor cleanups to the SVA code
Intel VT-d:
- Enable debugfs to dump domain attached to a pasid
- Remove an unnecessary inline function
AMD IOMMU:
- Initial patches for SVA support (not complete yet)
S390 IOMMU:
- DMA-API conversion and optimized IOTLB flushing
And some smaller fixes and improvements"
* tag 'iommu-updates-v6.7' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (102 commits)
iommu/dart: Remove the force_bypass variable
iommu/dart: Call apple_dart_finalize_domain() as part of alloc_paging()
iommu/dart: Convert to domain_alloc_paging()
iommu/dart: Move the blocked domain support to a global static
iommu/dart: Use static global identity domains
iommufd: Convert to alloc_domain_paging()
iommu/vt-d: Use ops->blocked_domain
iommu/vt-d: Update the definition of the blocking domain
iommu: Move IOMMU_DOMAIN_BLOCKED global statics to ops->blocked_domain
Revert "iommu/vt-d: Remove unused function"
iommu/amd: Remove DMA_FQ type from domain allocation path
iommu: change iommu_map_sgtable to return signed values
iommu/virtio: Add __counted_by for struct viommu_request and use struct_size()
iommu/vt-d: debugfs: Support dumping a specified page table
iommu/vt-d: debugfs: Create/remove debugfs file per {device, pasid}
iommu/vt-d: debugfs: Dump entry pointing to huge page
iommu/vt-d: Remove unused function
iommu/arm-smmu-v3-sva: Remove bond refcount
iommu/arm-smmu-v3-sva: Remove unused iommu_sva handle
iommu/arm-smmu-v3: Rename cdcfg to cd_table
...
|
|
'x86/amd', 'core' and 's390' into next
|
|
Following the pattern of identity domains, just assign the BLOCKED domain
global statics to a value in ops. Update the core code to use the global
static directly.
Update powerpc to use the new scheme and remove its empty domain_alloc
callback.
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Acked-by: Sven Peter <sven@svenpeter.dev>
Link: https://lore.kernel.org/r/1-v2-bff223cf6409+282-dart_paging_jgg@nvidia.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Wrap up the data type/pointer/len sanity and a copy_struct_from_user call
for iommu drivers to copy driver specific data via struct iommu_user_data.
And expect it to be used in the domain_alloc_user op for example.
Link: https://lore.kernel.org/r/20231026043938.63898-9-yi.l.liu@intel.com
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Co-developed-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
|
|
domain_alloc_user op already accepts user flags for domain allocation, add
a parent domain pointer and a driver specific user data support as well.
The user data would be tagged with a type for iommu drivers to add their
own driver specific user data per hw_pagetable.
Add a struct iommu_user_data as a bundle of data_ptr/data_len/type from an
iommufd core uAPI structure. Make the user data opaque to the core, since
a userspace driver must match the kernel driver. In the future, if drivers
share some common parameter, there would be a generic parameter as well.
Link: https://lore.kernel.org/r/20231026043938.63898-7-yi.l.liu@intel.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Co-developed-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
|
|
Introduce a new domain type for a user I/O page table, which is nested on
top of another user space address represented by a PAGING domain. This
new domain can be allocated by the domain_alloc_user op, and attached to
a device through the existing iommu_attach_device/group() interfaces.
The mappings of a nested domain are managed by user space software, so it
is not necessary to have map/unmap callbacks.
Link: https://lore.kernel.org/r/20231026043938.63898-2-yi.l.liu@intel.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
|
|
Add to iommu domain operations a set of callbacks to perform dirty
tracking, particulary to start and stop tracking and to read and clear the
dirty data.
Drivers are generally expected to dynamically change its translation
structures to toggle the tracking and flush some form of control state
structure that stands in the IOVA translation path. Though it's not
mandatory, as drivers can also enable dirty tracking at boot, and just
clear the dirty bits before setting dirty tracking. For each of the newly
added IOMMU core APIs:
iommu_cap::IOMMU_CAP_DIRTY_TRACKING: new device iommu_capable value when
probing for capabilities of the device.
.set_dirty_tracking(): an iommu driver is expected to change its
translation structures and enable dirty tracking for the devices in the
iommu_domain. For drivers making dirty tracking always-enabled, it should
just return 0.
.read_and_clear_dirty(): an iommu driver is expected to walk the pagetables
for the iova range passed in and use iommu_dirty_bitmap_record() to record
dirty info per IOVA. When detecting that a given IOVA is dirty it should
also clear its dirty state from the PTE, *unless* the flag
IOMMU_DIRTY_NO_CLEAR is passed in -- flushing is steered from the caller of
the domain_op via iotlb_gather. The iommu core APIs use the same data
structure in use for dirty tracking for VFIO device dirty (struct
iova_bitmap) abstracted by iommu_dirty_bitmap_record() helper function.
domain::dirty_ops: IOMMU domains will store the dirty ops depending on
whether the iommu device supports dirty tracking or not. iommu drivers can
then use this field to figure if the dirty tracking is supported+enforced
on attach. The enforcement is enable via domain_alloc_user() which is done
via IOMMUFD hwpt flag introduced later.
Link: https://lore.kernel.org/r/20231024135109.73787-5-joao.m.martins@oracle.com
Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
|
|
The iommu_map_sgtable() function returns ssize_t and negative error
codes but it's declared as size_t instead. I think that static checkers
would have complained if this caused a bug, but even though it doesn't
cause a bug, it's definitely worth fixing.
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Link: https://lore.kernel.org/r/06672b96-23fd-424c-8880-1626e7bf119c@moroto.mountain
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
Introduce a new iommu_domain op to create domains owned by userspace,
e.g. through IOMMUFD. These domains have a few different properties
compares to kernel owned domains:
- They may be PAGING domains, but created with special parameters.
For instance aperture size changes/number of levels, different
IOPTE formats, or other things necessary to make a vIOMMU work
- We have to track all the memory allocations with GFP_KERNEL_ACCOUNT
to make the cgroup sandbox stronger
- Device-specialty domains, such as NESTED domains can be created by
IOMMUFD.
The new op clearly says the domain is being created by IOMMUFD, that the
domain is intended for userspace use, and it provides a way to pass user
flags or a driver specific uAPI structure to customize the created domain
to exactly what the vIOMMU userspace driver requires.
iommu drivers that cannot support VFIO/IOMMUFD should not support this
op. This includes any driver that cannot provide a fully functional PAGING
domain.
This new op for now is only supposed to be used by IOMMUFD, hence no
wrapper for it. IOMMUFD would call the callback directly. As for domain
free, IOMMUFD would use iommu_domain_free().
Link: https://lore.kernel.org/r/20230928071528.26258-2-yi.l.liu@intel.com
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Co-developed-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
|
|
In some virtualized environments, including s390 paged memory guests,
IOTLB flushes are used to update IOMMU shadow tables. Due to this, they
are much more expensive than in typical bare metal environments or
non-paged s390 guests. In addition they may parallelize poorly in
virtualized environments. This changes the trade off for flushing IOVAs
such that minimizing the number of IOTLB flushes trumps any benefit of
cheaper queuing operations or increased paralellism.
In this scenario per-CPU flush queues pose several problems. Firstly
per-CPU memory is often quite limited prohibiting larger queues.
Secondly collecting IOVAs per-CPU but flushing via a global timeout
reduces the number of IOVAs flushed for each timeout especially on s390
where PCI interrupts may not be bound to a specific CPU.
Let's introduce a single flush queue mode that reuses the same queue
logic but only allocates a single global queue. This mode is selected by
dma-iommu if a newly introduced .shadow_on_flush flag is set in struct
dev_iommu. As a first user the s390 IOMMU driver sets this flag during
probe_device. With the unchanged small FQ size and timeouts this setting
is worse than per-CPU queues but a follow up patch will make the FQ size
and timeout variable. Together this allows the common IOVA flushing code
to more closely resemble the global flush behavior used on s390's
previous internal DMA API implementation.
Link: https://lore.kernel.org/all/9a466109-01c5-96b0-bf03-304123f435ee@arm.com/
Acked-by: Robin Murphy <robin.murphy@arm.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com> #s390
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Link: https://lore.kernel.org/r/20230928-dma_iommu-v13-5-9e5fc4dacc36@linux.ibm.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
|
On s390 when using a paging hypervisor, .iotlb_sync_map is used to sync
mappings by letting the hypervisor inspect the synced IOVA range and
updating a shadow table. This however means that .iotlb_sync_map can
fail as the hypervisor may run out of resources while doing the sync.
This can be due to the hypervisor being unable to pin guest pages, due
to a limit on mapped addresses such as vfio_iommu_type1.dma_entry_limit
or lack of other resources. Either way such a failure to sync a mapping
should result in a DMA_MAPPING_ERROR.
Now especially when running with batched IOTLB flushes for unmap it may
be that some IOVAs have already been invalidated but not yet synced via
.iotlb_sync_map. Thus if the hypervisor indicates running out of
resources, first do a global flush allowing the hypervisor to free
resources associated with these mappings as well a retry creating the
new mappings and only if that also fails report this error to callers.
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com> # sun50i
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Link: https://lore.kernel.org/r/20230928-dma_iommu-v13-1-9e5fc4dacc36@linux.ibm.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
|
