Set the KFD_IOC_ALLOC_MEM_FLAGS_COHERENT flag and
KFD_IOC_ALLOC_MEM_FLAGS_UNCACHED flag to allocate
uncached coherent memory when HSA_DISABLE_CACHE
environment variable is set. At KFD driver,
Single KFD_IOC_ALLOC_MEM_FLAGS_UNCACHED flag is
not sufficient to allocate uncached memory. We
have to use both two flags to allocate uncached
memory.
Change-Id: Ie490f37b2e696314e60048f5b1b57442431696e9
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
[ROCm/ROCR-Runtime commit: ae0e74095e]
Before gfx90a, coherent memory is uncached. So it was reasonable
when environment variable HSA_DISABLE_CACHE is set, memory is mapped
as coherent. On gfx90a, coherent memory can be cached, so mapping
memory as coherent can't guarantee memory is uncached. When
HSA_DISABLE_CACHE is set, we have to map memory as uncached.
Change-Id: Ia5ed4cf0ad6aef5644dc8c9e6632b52d606f06f4
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
[ROCm/ROCR-Runtime commit: f132fb2cd0]
This change might be redundant if ROCr takes care of it
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Change-Id: I7b67143a8ad21baa61b7eda7b8e5fe0ac1e33830
[ROCm/ROCR-Runtime commit: 10674916e4]
This change is for the A+A bring-up branch as it needs to made more
generic to handle all ASICs.
For A+A all the system buffers are mapped as NC (non coherent) unless
explicitly marked as UC (uncached). The coherency is then expected to be
handled by shader by explicitly using acquire/release instructions.
However, CP doesn't have same feature. The buffers used by CP thus have
to UC. For now queue buffer and Signal handler memory is marked as UC.
This change shouldn't affect other ASICs since Uncached flag is not used
in those. However, this change still need to be made more generic.
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Change-Id: I56c37a809913f7f08c94d01b0572d0f4864939aa
[ROCm/ROCR-Runtime commit: 7c05c5240f]
On gfx9, the maximum number of wavefronts per queue is the minimum of
40 waves per compute units, or 512 waves per shader engine. On gfx10,
there can only be 32 waves per compute units.
Signed-off-by: Laurent Morichetti <laurent.morichetti@amd.com>
Change-Id: I148d1a4fe6c07cdbfaa1f77939eb29311c81c008
[ROCm/ROCR-Runtime commit: 4cf11c3a7e]
Reserve some space in the context save area for the debugger's
use. There should be 32 bytes per wave for a given queue.
Change-Id: I65ddb6123d0f6afd3149844617ad19023009101d
[ROCm/ROCR-Runtime commit: a83f9b67ce]
XNACK API for GPUs that support this mode. This API
makes calls to amdgpu driver to configure xnack mode.
It supports set xnack mode and query the current mode used.
Change-Id: If865fd0e3f900f008243dc49504e1a0694e1791a
Signed-off-by: Alex Sierra <alex.sierra@amd.com>
[ROCm/ROCR-Runtime commit: 3f45f602d4]
strlen(src) should not be used as the length in strncpy. Use memcpy
since we know the length of the string, and ensure that we
NULL-terminate regardless of length
Signed-off-by: Kent Russell <kent.russell@amd.com>
Change-Id: I21cc6d106510c69464e7ac9d3fc7da3a1e6d1a68
[ROCm/ROCR-Runtime commit: 731a06c704]
It is to provide an option to map specific memory as
uncached on A+A HW platform.
Signed-off-by: Eric Huang <JinhuiEric.Huang@amd.com>
Change-Id: Ib665cb306a0e78aba3ea5ee2f0e46cb62ae139f8
[ROCm/ROCR-Runtime commit: 2464bfc714]
The purpose of this patch is to add a missing device ID for gfx1030.
The missing ID "0x73A1" is now added to the "topology.c" file.
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: I05a8a55e2c46f941a039fa72a6a5e76bf2a52736
[ROCm/ROCR-Runtime commit: 46c94662b7]
Inside Docker, when limit GPU number to one, it may cause node
numIOLinks bigger than total node number.
Signed-off-by: Gang Ba <gaba@amd.com>
Change-Id: Ib84f2f05f8e0c70e48b9043b79aec02b5a214bbe
[ROCm/ROCR-Runtime commit: 7652932c38]
This patch is to add Cezanne/Lucienne support on thunk.
Change-Id: Icd9b9913fa87bbfe6c71b36a2892d6ddb73e3ddd
Signed-off-by: Prike.Liang <Prike.Liang@amd.com>
[ROCm/ROCR-Runtime commit: 7e184ebb3a]
This patch make get_block_properties() function work on gfx1033 platform
Change-Id: Ie5be7dfb38575eec8b39b91f3ee5b3a31abe8bd1
Signed-off-by: Chen Gong <curry.gong@amd.com>
[ROCm/ROCR-Runtime commit: 4cf50fdeaa]
This patch is to add Van Gogh support on thunk.
Change-Id: I75819329b865e4c38c097e83e3a0cb4e4f566fa2
Signed-off-by: Huang Rui <ray.huang@amd.com>
[ROCm/ROCR-Runtime commit: 9600760ff7]
1. Create P2P links
2. Determine FRAMEBUFFER_PUBLIC/PRIVATE only based
host-accessibility, not peer-accesssibility
Signed-off-by: Gang Ba <gaba@amd.com>
Change-Id: I15fccdc60386b453e2a47849a16df15157324b21
[ROCm/ROCR-Runtime commit: bedecc5957]
This reverts commit 35b07e1e28.
Reason for revert: This commit caused a regression rocrtst memory
subtest: Maximum Single Allocation in Memory Pools failed.
Change-Id: I15330625603f893200a08cd8b5b097f9bf95361f
[ROCm/ROCR-Runtime commit: e515fd818b]
While the ternary is nice to read, strlen in general is an expensive
call, so call it once and check if the value is greater than our maximum
allowable string length and adjust accordingly
Signed-off-by: Kent Russell <kent.russell@amd.com>
Change-Id: Id744f2ba0eb81bb2b3c52eb69f38a615398a655d
[ROCm/ROCR-Runtime commit: 025036a662]
Don't update the vm_object if GPU mapping failed. Print an error message
to help diagnose underlying problems.
Change-Id: I801ab6fe6c155bd25e6c0358007c106a4a019480
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
[ROCm/ROCR-Runtime commit: c7e6f5a274]
Use MAP_POPULATE when allocating anonymous system memory for later
GPU mapping as a userptr. This can speed up large allocations by
more than factor 2. I suspect populating pages in this way is more
efficient than the CPU page fault code path triggered by
get_user_pages in the kernel.
Change-Id: I188bbc1462ccb650d48cbfb1080dbb8eb7ada8b5
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
[ROCm/ROCR-Runtime commit: 8f26c0c40c]
On gfx9, the maximum number of wavefronts per queue is the minimum of
40 waves per compute units, or 512 waves per shader engine. On gfx10,
there can only be 32 waves per compute units.
Signed-off-by: Laurent Morichetti <laurent.morichetti@amd.com>
Change-Id: I148d1a4fe6c07cdbfaa1f77939eb29311c81c008
[ROCm/ROCR-Runtime commit: 783e346777]
Reserve some space in the context save area for the debugger's
use. There should be 32 bytes per wave for a given queue.
Change-Id: I65ddb6123d0f6afd3149844617ad19023009101d
[ROCm/ROCR-Runtime commit: 2ed2e46b9b]
The queue control stack size cannot exceed 0x7000 on ASICs
gfx1010 through gfx1031. The lower limit is not achievable
with AQL so this should have no practical effect.
Fixes control stack size overflow on large ASICs.
Change-Id: Ib78cf6e4c5f096044bf8de24debe211689891caa
Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
[ROCm/ROCR-Runtime commit: 44f80d170d]
This patch is the hot fix to fix the param number checking after remove
dgpu input.
Signed-off-by: Huang Rui <ray.huang@amd.com>
Change-Id: Ic980588f78616f99076de742af580afb4273fb2f
[ROCm/ROCR-Runtime commit: 8fc816affe]
gfx90c should use GFX902 which is the same with gfx902.
Signed-off-by: Huang Rui <ray.huang@amd.com>
Change-Id: Id24dc2c85c9f49f36b00889c3b8b1b19cce34e09
[ROCm/ROCR-Runtime commit: 8ea0d49337]
Whether use dgpu path will check the props which exposed from kernel.
We won't need hard code in the ASIC table.
Signed-off-by: Huang Rui <ray.huang@amd.com>
Change-Id: I0c018a26b219914a41197ff36dbec7a75945d452
[ROCm/ROCR-Runtime commit: ad87f38dad]
KFD already implemented the fallback path for APU. Thunk will use flag
which exposed by kfd to configure is_dgpu instead of hardcode before.
Signed-off-by: Huang Rui <ray.huang@amd.com>
Change-Id: I445f6cf668f9484dd06cd9ae1bb3cfe7428ec7eb
[ROCm/ROCR-Runtime commit: 12813691a2]
Gfx10 need 12bytes/wave control stack
Change-Id: I6c6f2819572e6b43aa3140d4dbe79d930e4c1c9c
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
Signed-off-by: Jay Cornwall <Jay.Cornwall@amd.com>
[ROCm/ROCR-Runtime commit: 3d3b28b670]
This is needed to avoid additional references to mapped BOs in child
processes that can prevent freeing memory in the parent process and lead
to out-of-memory conditions.
Change-Id: I25c90510a14dde515cc23ea5dc1f68e8d7e37a66
Signed-off-by: Philip Cox <Philip.Cox@amd.com>
[ROCm/ROCR-Runtime commit: f7a3427c99]
strlen(src) should not be used as the length in strncpy. Use memcpy
since we know the length of the string, and ensure that we
NULL-terminate regardless of length
Signed-off-by: Kent Russell <kent.russell@amd.com>
Change-Id: I21cc6d106510c69464e7ac9d3fc7da3a1e6d1a68
[ROCm/ROCR-Runtime commit: 04f6b9e16b]
The option to use kfd_fd for cpu mapping is for very old broken KFD
version, it is not used in upstreaming process. This causes issue when
multiple process uses shared system memory because the GTT address is
over 40 bits.
Change to always use render node fd to create CPU mapping.
Change-Id: Id7e7b2a2e2f13c6e62c5de170589abfff4d456b0
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
[ROCm/ROCR-Runtime commit: 9e9771a7d9]
This reverts commit c0a0ada18b.
Reason for revert: Change was submitted by accident
Change-Id: If05c705e22296fd3ca789f269737d379a933361d
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
[ROCm/ROCR-Runtime commit: fec3780c1a]
Signed-off-by: Gang Ba <gaba@amd.com>
Change-Id: I5c23a8dacf9bc50c740908aabe391432f2c7112e
Signed-off-by: Gang Ba <gaba@amd.com>
[ROCm/ROCR-Runtime commit: d675d1cce1]
Kent Russell
Oak Zeng
Chengming Gui
Harish Kasiviswanathan
Laurent Morichetti
Alex Sierra
Eric Huang
Jonathan Kim
Amber Lin
Jay Cornwall
Ori Messinger
Gang Ba
Prike.Liang
Chen Gong
Huang Rui
Tianci.Yin
Felix Kuehling
Philip Cox
Philip Yang
Yong Zhao