Add agent info query to return nearest CPU agent. This can be used to
determine which CPU agent is in the same NUMA region as the GPU agent.
Change-Id: I5400b4347ffbf4d2a836df31c4de443a38b0ecd1
Aqlpacket:IsValid() function: Replaced bitwise AND operator (&) with the logical
AND operator (&&) when evaluating AQL packet type
Change-Id: I59980bc206cc7eff424023fff0bb92b618aa8c70
Support function to retain allocation handle for memory mappings.
The get allocation properties function will return the current
allocation properties for existing memory mappings.
This is part of patch series for Virtual Memory API.
Change-Id: I0a53a11b6efc2b5bf9d463512a489a2abd812551
Support exporting and importing dmabuf file descriptors for memory
mappings. The exported dmabuf file descriptors are shareable posix
file descriptors that can be used for cross-vendor, cross-device
and cross-process memory sharing.
This is part of patch series for Virtual Memory API.
Change-Id: I3673fc009f7e73bc26be8349e19f66e20d0607c5
Mapping memory handles to virtual memory addresses do not make them
accessible. The set access function is needed to make the memory
mappings accessible to specific agents. The get access function
returns current access properties for individual agents.
This is part of patch series for Virtual Memory API.
Change-Id: I152ba0557fd2a802eb9d840568b68cdd1911b72c
Add support for mapping and unmapping memory handles to virtual
address ranges.
This is part of patch series for Virtual Memory API.
Change-Id: If512d49ff4211e68f2064249add607a3200e458a
Add support for creating and releasing memory handles. Memory
handles are memory allocations on device memory without a virtual
address.
This is part of patch series for Virtual Memory API.
Change-Id: I5dfb162eb1661621cce171b2870a3c93b24d840e
Add support for reserving virtual address ranges. Virtual address
ranges are addresses without any memory backing. These address ranges
need to be mapped to memory handles later.
This is part of patch series for Virtual Memory API.
Change-Id: I5d066e7421d6896f933f524312afc230a13d594e
Change initialize libdrm device and file descriptor initialization
to use new APIs from Thunk. Libdrm recommends that we re-use the same
file descriptor thoughout the life of a process instead of re-creating
new one each time.
This is part of patch series for Virtual Memory API.
Change-Id: I1c0b8d1bd660cd25478b5f94c84071b90d93fc6c
Checks whether version of libdrm library installed on current
system supports the amdgpu_device_get_fd API. This API is
required to support the virtual memory API functions. The
amdgpu_device_get_fd function was introduced in libdrm-2.4.109.
Using a runtime check test instead of static dependency to be
able to support previous APIs on older versions of libdrm.
Add query for virtual memory API support.
This is part of patch series for Virtual Memory API.
Change-Id: Iec831eb24b5d1689c392e50ae86f4d52d4870ac4
Add new query for recommended granularity size. This is the
internal blocksize used. While the existing query for granularity
size returns the minimum size possible, it is recommended that
allocations and mappings are multiple of the recommended granularity
size to minimise internal memory fragmentation.
This is part of patch series for Virtual Memory API.
Change-Id: Ia82c8f073b2a2c47ecd26fbb0aba27b8b7cd965f
The access type for extended scope fine grained memory was being returned as never
allowed by default
Change-Id: I0167ea0e5931053f22f2d2755bf426d43d2bb8e5
On gfx11, with a sequence such as
s_trap 2
s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
s_endpgm
the s_sendmsg does deallocate registers while the wave is supposed to be
stopped. As a result, the wave cannot do the expected context save
operations, and cannot context save.
To avoid this problem, park the wave in the trap handler for gfx11.
Note that gfx11 has implemented an instruction cache prefetch. When
parked, the prefetch tries to access memory past the end of trap handler
which causes memory violation exceptions to be reported. To avoid this,
we need to add padding at the end of the trap handler. The padding
consists of `s_code_end` instructions Given that the trap handler is
loaded at a 0x1000 aligned address the maximum prefetch amount (in
bytes) is given by `256 - (trap_handler_size % 64)`.
Change-Id: I5446da54a965a64f21cb0fd3ce3caa4b6137a933
Thread yield doesn't drop the scoped acquired mutex so drop it around
yield to prevent a multithread deadlock.
Change-Id: Ie21f3bff89f6f9e4c57e5b3ccf17968f253fa23a
Fix a condition where we can get a divide-by-zero in the
TranslateTime(tick) function if the GPU tick predates HSA
startup and we did not do a SyncClocks since initialization.
Change-Id: I0dcec8553ccb8f01211928991f4b3ed3cb4a1ebb
In ASAN builds, the compiler used is clang. The initialization of
variable sized array using assignment operator is causing compilation
failure in ASAN builds. Used memset to fix the same.
Change-Id: Ifc748291a41a9886243e0fb1ba576d2760f5e15e
I've just reverted some code what it was in 5.5 by wrapping new x86
specific bits with #if's, e.g.:
- CPUID is x86 specific
- mwait is x86 specific
Change-Id: I6cefae34282c777c7340daf3f934d2a11742502e
Signed-off-by: Jeremy Newton <Jeremy.Newton@amd.com>
This fixes a segfault error in cases where the linking order of
compilation unit varies. Reason behind the segfault is that one
global variable in one compilation unit depends on another global
variable in another compilation unit, but there is no guarantee that
this other compilation unit is initialized first. The fix forces a
reinitialization at the first invocation of the library.
Change-Id: I1428592c6898bca13a330c4588941de260ff0370
Unless SDMA blits have actually been used for copies, prevent the DMA
copy status from querying the blit's pending byte status to avoid
creating an unnecessary HW queue.
Change-Id: Ied1fbed73c08f0408f0e3583f9b56f2768c71708
Querying pending bytes on a blit kernel is unnecessary when runtime
runs out of SDMA resource since we are returning an SDMA availabilty
mask.
Change-Id: I347efba0c85b70ea3ba8749d76a499afc23909e8
Added HSA_AMD_MEMORY_POOL_GLOBAL_FLAG_EXT_SCOPE_FINE_GRAINED flag to enable extended scope memory region
where the device-scope atomics act as system-scope atomics
Change-Id: I79fc3207cb630dfc68bed2f8aabd75f35fe80b12
Enable sleep for all waiters with event age tracking support kernel.
Change-Id: Icd4e1e8d83b4a54e9f6aaa99691a6573211b3337
Signed-off-by: James Zhu <James.Zhu@amd.com>
KFD kernel version 1.13 starts to support event age
tracking which help elimating unncessary busy wait.
Change-Id: Ib447ed6e0350f3110a4d6b9b80a0388000dd0e72
Signed-off-by: James Zhu <James.Zhu@amd.com>
Earlier, hsa-runtime was unable to find symbols from a stripped ELF-image becasue
no support to find symbols from ".dynsym" section.
Looking for symbols in .dynsym is enabled by LOADER_USE_DYNSYM=1
environment variable
Change-Id: I4f0e8dd0eb053a6066d4d49b670c52e51149531a
GFX11 and up including some GFX9 devices will not support
old trap handling without the new exception handling.
Instead of a hard assert failure that runs into a core dump,
let ROCr initialization continue instead.
Change-Id: I309becdc72ef4fb2fafd118c1faf0801407e658e
status.priv may be read after returning from the trap handler, which
causes sq_interrupt_word_wave.priv to be 0 even though the s_sendmsg
instruction was initiated when status.priv was 1.
To work around this, added a s_waitcnt lgkmcnt(0) after s_sendmsg
to make sure the message is sent before continuing.
Signed-off-by: Jay Cornwall <Jay.Cornwall@amd.com>
Signed-off-by: Laurent Morichetti <Laurent.Morichetti@amd.com>
Change-Id: Ieb75005ca1559ef03d0efac80e966f521e41fcb7
Removing this definition as this should already be defined by compiler.
This is causing compile errors on newer versions of llvm because the
macro is being redefined.
Change-Id: Ica6a06f46a14e16d3f52e83b9b5ee8cfd7359510
Adding support for gfx941 and gfx942 ISAs.
gfx940 ISA will use sc0:1 sc1:1 on load/store operations
gfx942 ISA will use default load/store operations
Change-Id: If1efbef86f59e2cf2d48fe359cd4166405a0a579
When compiling in ASAN mode, remap the first page of device allocations
to system memory. ASAN's memory allocator uses a small amount of extra
memory to store data for housekeeping purpose. But because this memory
is from the GPU memory pool, it might have uncommon memory type for host
to access. Mapping this section of memory to the host makes this memory
accessible to ASAN.
Change-Id: I36f659d616a4d15558372592439a8723c5c84a69
Signed-off-by: Bing Ma <Bing.Ma@amd.com>
Add support for HSA_ENABLE_PEER_SDMA env variable that can be used to
disable use of SDMA engines for device-to-device transfers. Note that
setting HSA_ENABLE_SDMA=0 will disable all SDMA transfers and override
HSA_ENABLE_PEER_SDMA values.
Change-Id: I737b3c2b2efcf3ff237f98bc748f49b8252ed24a
Update documentation for hsa_amd_pointer_info to clarify which fields
are invalid when the allocation type is HSA_EXT_POINTER_TYPE_UNKNOWN.
Change-Id: Idaed985962c4a98d281ebe01bef8ec2459da3985
Some workloads running on multi-GPU create 1 process per GPU. So each
process creates a GPU agent on every GPU, but will only create queues on
one GPU. This would cause un-necessary scratch reservation.
Change-Id: I50a216f0bcc0b5f707f3943147390b0ecec1ac22
If the required scratch allocation is too large, ROCr will attempt to
reduce it by lowering the dispatch's targeted occupancy. The reduction
loop however was prone to overflow if waves_per_cu was not a multiple of
waves_per_group. Ensure no overflow by aligning waves_per_cu to
waves_per_group.
On GC 9.4.3 dGPU, dispatches with a large grid size and a
waves_per_group of e.g. 16 may require to reduce occupancy such that
waves_per_cu is less than waves_per_group to ensure the allocation size
is small enough. Allow this while also ensuring the tmpring scratch wave
count is kept divisible by the number of SEs per XCC.
Signed-off-by: Graham Sider <Graham.Sider@amd.com>
Change-Id: Ie4016dcd8166a9ae69e9decc26a3eec882b49480
David Yat Sin
Jonathan Kim
Lang Yu
Shweta Khatri
Lancelot SIX
Ranjith Ramakrishnan
Jeremy Newton
Philipp Knechtges
Konstantin Zhuravlyov
James Zhu
raghavmedicherla
Laurent Morichetti
Ammar ELWazir
Xiaomeng Hou
Graham Sider