* Introduce HsaKFDContext structure and infrastructure for multiple KFD contexts, enabling
independent contexts within a single process.
* Refactor core components (queue, event, FMM, topology) to be context-aware,
using explicit HsaKFDContext parameters instead of global state.
* Replace global hsakmt_kfd_fd with context-specific file descriptors, ensuring full context isolation.
* Maintain backward compatibility by redirecting legacy APIs to use the primary context.
This refactoring establishes a foundation for multi-context support while preserving existing functionality.
Signed-off-by: Junhua Shen <Junhua.Shen@amd.com>
* clr: Adjust call to ICmdBuffer::CmdCopyMemoryToImage for PAL >= 955
PAL starting versino 955 adds a new argument to
ICmdBuffer::CmdCopyMemoryToImage. Adjust teh callsite to account
fort his.
* clr: Handle new GpuUtil::TraceSessionState cases for PAL >= 939
Starting PAL API version 939, GpuUtil::TraceSessionState changes its
possible values. Adjust for it.
* clr: require PAL version 954
Bump the PAL required vesion to 954, as this is required for proper
debugger support.
* Support Windows HANDLE in interop_map_buffer
* Refactored Windows HANDLE in interop_map_buffer
* ROCr System Dependent Handle Type
* Fix for ROCr Handle Conversion Bug
* Remove Windows Header
Remove libamdhsacode/win32/elf.h due to license restrictions.
Separate Linux coredump implementation because we do not have the ELF
definitions on Windows.
Co-authored-by: JeniferC99 <150404595+JeniferC99@users.noreply.github.com>
* Fix for SWDEV-552584
Two calls to ompt_callback_task_scheduled were issued for the same
prior task. One of them was ompt_task_complete, which causes
internal storage to be release and a pointer zeroed. The other
was ompt_task_early_fulfill, which attempted to reference the
pointer. The callbacks could come in any order as they were
from different threads, thus causing a null pointer
dereference on occasion. The code was changed to do nothing
for the early_fulfill. Additional null pointer checks were
added.
* formatting
* Update ompt.cpp
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
---------
Co-authored-by: Jonathan R. Madsen <jrmadsen@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Add clean up of buffered_storage files
* Add step to workflows to test for remaining temp files after tests
* Applied suggestions from code review
* add deletion of all cache files
---------
Co-authored-by: David Galiffi <David.Galiffi@amd.com>
- Integrate rocprofiler-systems with rocprofiler-sdk-rocpd to fetch schema
- If rocprofiler-sdk-rocpd is not availabe, use embedded schema files. With this we provide rocpd format support even if ROCm is not available
- Include detection in CMake if rocprofiler-sdk-rocpd package is available (and valid), and build database class upon that
- Update embedded schema that is used as a fallback.
- Update some validation tests to account for schema changes.
Were not handling the case where the eval result is None e.g. some
columns have a peak value, but it is unused, so we use 'None', which
evaluates to the None object.
Return empty string in this case.
* Update rocprofiler_config_interfaces.cmake to use different elf naming
* try out conditional for libelf
* run cmake-format to fix formatting issue
* Remove libelf.patch file from therock-ci-windows.yml
* Remove libelf patch from therock-ci-linux.yml as well
- Update the pinned SHA for TheRock in CI workflows.
- Update the version for actions in those same workflows.
- Comment out the rm .patch line and provide details on its use.
* Changes:
- Modified attempting to open files to check
permissions -> check read access only.
Do not try to open all paths, may cause driver issues.
Read access is sufficient to check permissions.
Reason: GPUs which support partitioning (memory/compute),
logical devices will not be valid until configured.
See `sudo amd-smi set -h` or applicable APIs
to configure on supported hardware.
Example error dmesg output:
[965358.883112] amdgpu 0000:15:00.0: amdgpu: renderD153 partition 1 not valid!
[965358.883283] amdgpu 0000:15:00.0: amdgpu: renderD154 partition 2 not valid!
[965358.883438] amdgpu 0000:15:00.0: amdgpu: renderD155 partition 3 not valid!
[965358.883594] amdgpu 0000:15:00.0: amdgpu: renderD156 partition 4 not valid!
[965358.883749] amdgpu 0000:15:00.0: amdgpu: renderD157 partition 5 not valid!
[965358.883904] amdgpu 0000:15:00.0: amdgpu: renderD158 partition 6 not valid!
[965358.884060] amdgpu 0000:15:00.0: amdgpu: renderD159 partition 7 not valid!
---------
Signed-off-by: Charis Poag <Charis.Poag@amd.com>
* Changes:
- Modified attempting to open files to check
permissions -> check read access only.
Do not try to open all paths, may cause driver issues.
Read access is sufficient to check permissions.
Reason: GPUs which support partitioning (memory/compute),
logical devices will not be valid until configured.
See `sudo amd-smi set -h` or applicable APIs
to configure on supported hardware.
Example error dmesg output:
[965358.883112] amdgpu 0000:15:00.0: amdgpu: renderD153 partition 1 not valid!
[965358.883283] amdgpu 0000:15:00.0: amdgpu: renderD154 partition 2 not valid!
[965358.883438] amdgpu 0000:15:00.0: amdgpu: renderD155 partition 3 not valid!
[965358.883594] amdgpu 0000:15:00.0: amdgpu: renderD156 partition 4 not valid!
[965358.883749] amdgpu 0000:15:00.0: amdgpu: renderD157 partition 5 not valid!
[965358.883904] amdgpu 0000:15:00.0: amdgpu: renderD158 partition 6 not valid!
[965358.884060] amdgpu 0000:15:00.0: amdgpu: renderD159 partition 7 not valid!
---------
Signed-off-by: Charis Poag <Charis.Poag@amd.com>
[ROCm/amdsmi commit: d73726698b]
* clr: SWDEV-547890 - Maintain an MQD for the emulated AQL queue
To simplify the shader debugger implementation, maintain the relevant
parts of the emulated AQL queue's MQD (amd_queue_t): read_dispatch_id,
write_dispatch_id, compute_tmpring_size.
With this MQD, the shader debugger can handle the emulated AQL queue
the same way it does the real AQL queue, no specialization is required.
* clr: SWDEV-547890 - Conservatively update the MQD's read_dispatch_id
The read_dispatch_id cannot be smaller than the current aql_packet_id
- hsa_queue.size for the debugger to work correctly.
The read_dispatch_id really should be updated when the CmdBuf is marked
as complete. Left a FIXME to address it in a future commit.
---------
Co-authored-by: Laurent Morichetti <laurent.morichetti@amd.com>
Junhua Shen
Jin Jung
Victor Zhang
Gerardo Hernandez
SaleelK
ajanicijamd
Pengda Xie
lancesix
Pratik Basyal
Jimbo
David Yat Sin
Larry Meadows
Milan Radosavljevic
Ethan Trinh
Yiannis Papadopoulos
Ben Richard
Gopesh Bhardwaj
Jason Bonnell
habajpai-amd
David Galiffi
Sourabh U Betigeri
Kian Cossettini
MachineTom
Joseph Macaranas
Poag, Charis
jamessiddeley-amd
Galantsev, Dmitrii
marandje
Satyanvesh Dittakavi
systems-assistant[bot]
Xie, AlexBin
alexxu-amd
Rakesh Roy
Apurv Mishra