| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Bump KFD ioctl after adding SMI events support
Signed-off-by: Amber Lin <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
When the compute is malfunctioning or performance drops, the system admin
will use SMI (System Management Interface) tool to monitor/diagnostic what
went wrong. This patch provides an event watch interface for the user
space to register devices and subscribe events they are interested. After
registered, the user can use annoymous file descriptor's poll function
with wait-time specified and wait for events to happen. Once an event
happens, the user can use read() to retrieve information related to the
event.
VM fault event is done in this patch.
v2: - remove UNREGISTER and add event ENABLE/DISABLE
- correct kfifo usage
- move event message API to kfd_ioctl.h
v3: send the event msg in text than in binary
v4: support multiple clients
v5: move events enablement from ioctl to fd write
v6: sparse fix
Signed-off-by: Amber Lin <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Corrected two function names. Added a missing space.
Signed-off-by: Felix Kuehling <[email protected]>
Reviewed-by: Kent Russell <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Add a new kfd ioctl to allocate queue GWS. Queue
GWS is released on queue destroy.
v2: re-introduce this API with the following fixes squashed in:
- drm/amdkfd: fix null pointer dereference on dev
- drm/amdkfd: Return proper error code for gws alloc API
- drm/amdkfd: Remove GPU ID in GWS queue creation
Signed-off-by: Oak Zeng <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This reverts commit 1a058c3376765ee31d65e28cbbb9d4ff15120056.
This interface is still in too much flux. Revert until
it's sorted out.
Acked-by: Oak Zeng <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Add a new kfd ioctl to allocate queue GWS. Queue
GWS is released on queue destroy.
Signed-off-by: Oak Zeng <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Existing QUEUE_TYPE_SDMA means PCIe optimized SDMA queues.
Introduce a new QUEUE_TYPE_SDMA_XGMI, which is optimized
for non-PCIe transfer such as XGMI.
Signed-off-by: Oak Zeng <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Introduce a new memory type (KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP) and
expose mmio page of HDP registers to user space through this new
memory type.
v2: moved remapped hdp regs to adev struct
v3: rename the new memory type to ALLOC_MEM_FLAGS_MMIO_REMAP
v4: use more generic function name
v5: Fail remapped mmio allocation for asics before gfx9
Signed-off-by: Oak Zeng <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Remap HDP_MEM_COHERENCY_FLUSH_CNTL and HDP_REG_COHERENCY_FLUSH_CNTL
to an empty page in mmio space. We will later map this page to process
space so application can flush hdp. This can't be done properly at
those registers' original location because it will expose more than
desired registers to process space.
v2: Use explicit register hole location
v3: Moved remapped hdp registers into adev struct
v4: Use more generic name for remapped page
Expose register offset in kfd_ioctl.h
v5: Move hdp register remap function to nbio ip function
v6: Fixed operator precedence issue and other bugs
Signed-off-by: Oak Zeng <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
RAS ECC event will combine with GPU reset event, due to
ECC interrupts are caused by uncorrectable error that triggers
GPU reset.
v2: Fix misleading-indentation warning
v3: fix build with CONFIG_HSA_AMD disabled
Signed-off-by: Eric Huang <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This is used for interoperability between ROCm compute and graphics
APIs. It allows importing graphics driver BOs into the ROCm SVM
address space for zero-copy GPU access.
The API is split into two steps (query and import) to allow user mode
to manage the virtual address space allocation for the imported buffer.
Acked-by: Alex Deucher <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Consistently use types provided by <linux/types.h> via <drm/drm.h>
to fix struct kfd_ioctl_get_queue_wave_state_args userspace compilation errors.
Fixes: 5df099e8bc83f ("drm/amdkfd: Add wavefront context save state retrieval ioctl")
Signed-off-by: Dmitry V. Levin <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Consistently use types provided by <linux/types.h> via <drm/drm.h>
to fix the following linux/kfd_ioctl.h userspace compilation errors:
/usr/include/linux/kfd_ioctl.h:250:2: error: unknown type name 'uint32_t'
uint32_t reset_type;
/usr/include/linux/kfd_ioctl.h:251:2: error: unknown type name 'uint32_t'
uint32_t reset_cause;
/usr/include/linux/kfd_ioctl.h:252:2: error: unknown type name 'uint32_t'
uint32_t memory_lost;
/usr/include/linux/kfd_ioctl.h:253:2: error: unknown type name 'uint32_t'
uint32_t gpu_id;
Fixes: 0c119abad7f0d ("drm/amd: Add kfd ioctl defines for hw_exception event")
Cc: <[email protected]> # v4.19
Signed-off-by: Dmitry V. Levin <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Wavefront context save data is of interest to userspace clients for
debugging static wavefront state. The MQD contains two parameters
required to parse the control stack and the control stack itself
is kept in the MQD from gfx9 onwards.
Add an ioctl to fetch the context save area and control stack offsets
and to copy the control stack to a userspace address if it is kept in
the MQD.
Signed-off-by: Jay Cornwall <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
Acked-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Signed-off-by: Felix Kuehling <[email protected]>
Acked-by: Oded Gabbay <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
Signed-off-by: Shaoyun Liu <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
Acked-by: Christian König <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
1. Pre-GFX9 the amdgpu ISR saves the vm-fault status and address per
per-vmid. amdkfd needs to get the information from amdgpu through the
new get_vm_fault_info interface. On GFX9 and later, all the required
information is in the IH ring
2. amdkfd unmaps all queues from the faulting process and create new
run-list without the guilty process
3. amdkfd notifies the runtime of the vm fault trap via EVENT_TYPE_MEMORY
Signed-off-by: shaoyun liu <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
Acked-by: Christian König <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
v2:
* Fix error handling after kfd_bind_process_to_device in
kfd_ioctl_map_memory_to_gpu
v3:
* Add ioctl to acquire VM from a DRM FD
v4:
* Return number of successful map/unmap operations in failure cases
* Facilitate partial retry after failed map/unmap
* Added comments with parameter descriptions to new APIs
* Defined AMDKFD_IOC_FREE_MEMORY_OF_GPU write-only
Signed-off-by: Felix Kuehling <[email protected]>
Acked-by: Christian König <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
Currently the number of GPUs is limited by aperture placement options
available on GFX7 and GFX8 hardware. This limitation is not necessary.
Scratch and LDS represent per-work-item and per-work-group storage
respectively. Different work-items and work-groups use the same virtual
address to access their own data. Work running on different GPUs is by
definition in different work-groups (different dispatches, in fact).
That means the same virtual addresses can be used for these apertures
on different GPUs.
Add a new AMDKFD_IOC_GET_PROCESS_APERTURES_NEW ioctl that removes the
artificial limitation on the number of GPUs that can be supported. The
new ioctl allows user mode to query the number of GPUs to allocate
enough memory for all GPUs to be reported.
This deprecates AMDKFD_IOC_GET_PROCESS_APERTURES.
Signed-off-by: Felix Kuehling <[email protected]>
Acked-by: Christian König <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
Use __u32 and __u64 instead of POSIX types that may not be defined
in user mode builds.
Signed-off-by: Felix Kuehling <[email protected]>
Reviewed-by: Oded Gabbay <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
A second-level user mode trap handler can be installed. The CWSR trap
handler jumps to the secondary trap handler conditionally for any
conditions not handled by it. This can be used e.g. for debugging or
catching math exceptions.
When CWSR is disabled, the user mode trap handler is installed as
first level trap handler.
Signed-off-by: Shaoyun.liu <[email protected]>
Signed-off-by: Jay Cornwall <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
This hardware feature allows the GPU to preempt shader execution in
the middle of a compute wave, save the state and restore it later
to resume execution.
Memory for saving the state is allocated per queue in user mode and
the address and size passed to the create_queue ioctl. The size
depends on the number of waves that can be in flight simultaneously
on a given ASIC.
Signed-off-by: Shaoyun.liu <[email protected]>
Signed-off-by: Yong Zhao <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
Consistently use types provided by <linux/types.h> via <drm/drm.h>
to fix the following linux/kfd_ioctl.h userspace compilation errors:
/usr/include/linux/kfd_ioctl.h:236:2: error: unknown type name 'uint64_t'
uint64_t va_addr; /* to KFD */
/usr/include/linux/kfd_ioctl.h:237:2: error: unknown type name 'uint32_t'
uint32_t gpu_id; /* to KFD */
/usr/include/linux/kfd_ioctl.h:238:2: error: unknown type name 'uint32_t'
uint32_t pad;
/usr/include/linux/kfd_ioctl.h:243:2: error: unknown type name 'uint64_t'
uint64_t tile_config_ptr;
/usr/include/linux/kfd_ioctl.h:245:2: error: unknown type name 'uint64_t'
uint64_t macro_tile_config_ptr;
/usr/include/linux/kfd_ioctl.h:249:2: error: unknown type name 'uint32_t'
uint32_t num_tile_configs;
/usr/include/linux/kfd_ioctl.h:253:2: error: unknown type name 'uint32_t'
uint32_t num_macro_tile_configs;
/usr/include/linux/kfd_ioctl.h:255:2: error: unknown type name 'uint32_t'
uint32_t gpu_id; /* to KFD */
/usr/include/linux/kfd_ioctl.h:256:2: error: unknown type name 'uint32_t'
uint32_t gb_addr_config; /* from KFD */
/usr/include/linux/kfd_ioctl.h:257:2: error: unknown type name 'uint32_t'
uint32_t num_banks; /* from KFD */
/usr/include/linux/kfd_ioctl.h:258:2: error: unknown type name 'uint32_t'
uint32_t num_ranks; /* from KFD */
Fixes: 6a1c9510694fe ("drm/amdkfd: Adding new IOCTL for scratch memory v2")
Fixes: 5d71dbc3a5886 ("drm/amdkfd: Implement image tiling mode support v2")
Signed-off-by: Dmitry V. Levin <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
Signed-off-by: Oded Gabbay <[email protected]>
Reviewed-by: Ben Goz <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
|
|
Include <drm/drm.h> instead of <linux/types.h> which on Linux includes
<linux/types.h> and on non-Linux platforms defines __u32 etc types.
Fixes user space compilation errors like:
linux/kfd_ioctl.h:33:2: error: unknown type name ‘uint32_t’
uint32_t major_version; /* from KFD */
^~~~~~~~
Signed-off-by: Mikko Rapeli <[email protected]>
Acked-by: Arnd Bergmann <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
v2: Removed hole in ioctl number space
Signed-off-by: Yong Zhao <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
Reviewed-by: Oded Gabbay <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
v2:
* Renamed ALLOC_MEMORY_OF_SCRATCH to SET_SCRATCH_BACKING_VA
* Removed size parameter from the ioctl, it was unused
* Removed hole in ioctl number space
* No more call to write_config_static_mem
* Return correct error code from ioctl
Signed-off-by: Moses Reuben <[email protected]>
Signed-off-by: Ben Goz <[email protected]>
Signed-off-by: Felix Kuehling <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
This patch adds four new IOCTLs to amdkfd. These IOCTLs expose a H/W
debugger functionality to the userspace.
The IOCTLs are:
- AMDKFD_IOC_DBG_REGISTER:
The purpose of this IOCTL is to notify amdkfd that a process wants to use
GPU debugging facilities on itself only.
It is expected that this IOCTL would be called before any other H/W
debugger requests are sent to amdkfd and for each GPU where the H/W
debugging needs to be enabled. The use of this IOCTL ensures that only
one instance of a debugger is active in the system.
- AMDKFD_IOC_DBG_UNREGISTER:
This IOCTL detaches the debugger/debugged process from the H/W
Debug which was established by the AMDKFD_IOC_DBG_REGISTER IOCTL.
- AMDKFD_IOC_DBG_ADDRESS_WATCH:
This IOCTL allows to set different watchpoints with various conditions as
indicated by the IOCTL's arguments. The available number of watchpoints
is retrieved from topology. This operation is confined to the current
debugged process, which was registered through AMDKFD_IOC_DBG_REGISTER.
- AMDKFD_IOC_DBG_WAVE_CONTROL:
This IOCTL allows to control a wavefront as indicated by the IOCTL's
arguments. For example, you can halt/resume or kill either a
single wavefront or a set of wavefronts. This operation is confined to
the current debugged process, which was registered through
AMDKFD_IOC_DBG_REGISTER.
Because the arguments for the address watch IOCTL and wave control IOCTL
are dynamic, meaning that they could vary in size, the userspace passes a
pointer to a structure (in userspace) that contains the value of the
arguments. The kernel driver is responsible to parse this structure and
validate its contents.
v2: change void* to uint64_t inside ioctl arguments
Signed-off-by: Yair Shachar <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
- AMDKFD_IOC_CREATE_EVENT:
Creates a new event of a specified type
- AMDKFD_IOC_DESTROY_EVENT:
Destroys an existing event
- AMDKFD_IOC_SET_EVENT:
Signal an existing event
- AMDKFD_IOC_RESET_EVENT:
Reset an existing event
- AMDKFD_IOC_WAIT_EVENTS:
Wait on event(s) until they are signaled
v2:
- Move the limit of the signal events to kfd_ioctl.h so it
can be used by userspace
v3:
- Change all bool fields in struct kfd_memory_exception_failure
to uint32_t
Signed-off-by: Andrew Lewycky <[email protected]>
Signed-off-by: Oded Gabbay <[email protected]>
|
|
This patch reformats the ioctl definitions in kfd_ioctl.h to be similar to the
drm ioctls definition style.
v2: Renamed KFD_COMMAND_(START|END) to AMDKFD_...
Signed-off-by: Oded Gabbay <[email protected]>
Acked-by: Christian König <[email protected]>
|
|
- KFD_IOC_GET_VERSION:
Retrieves the interface version of amdkfd
- KFD_IOC_CREATE_QUEUE:
Creates a usermode queue that runs on a specific GPU device
- KFD_IOC_DESTROY_QUEUE:
Destroys an existing usermode queue
- KFD_IOC_SET_MEMORY_POLICY:
Sets the memory policy of the default and alternate aperture of the
calling process
- KFD_IOC_GET_CLOCK_COUNTERS:
Retrieves counters (timestamps) of CPU and GPU
- KFD_IOC_GET_PROCESS_APERTURES:
Retrieves information about process apertures that were initialized
during the open() call of the amdkfd device
- KFD_IOC_UPDATE_QUEUE:
Updates configuration of an existing usermode queue
v3: Remove pragma pack and pmc ioctls. Added parameter for doorbell offset and
a comment on counters
v5:
Add define for AQL queues.
Fix arguments of Get Version IOCTL
Make IOCTL's structures to be the same size on 32/64 bit
v6: Change the version of the amdkfd-thunk interface
Signed-off-by: Oded Gabbay <[email protected]>
|