On small bar multi-gpu system, hsaKmtMemoryMapToGPU will fail due to latest
kernel P2P sanity check. Swith to use hsaKmtMemoryMapToGPUNodes to fix
the failure
Change-Id: Id8b6329d1243df0e908cc9a171b5c7f9156f4a8b
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Map scratch memory to the GPU that specified when allocate the memory
Change-Id: I788f9ef0dccb63b894a75e75cac5f94a60d7ec48
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
HW has limited bits for wave scratch base address stride. Enforcement
prevents programs with larger than supported scratch allocations from
running and clobbering neighboring scratch space.
Change-Id: I574da888e9d1d5e290a9c0025ba13b5ef9f1e5c0
BaseQueue class has a member function GetQueueType so m_Type
is duplicated. m_Type is only used in one function. Move it to
a local variable.
Change-Id: Ice144cf723178dd628cb49261c23d10605f9ee7d
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
Those new types are used to create SDMA queue on specific engine
Change-Id: I91c3bcc14fef7404cf42b256a18651432e171091
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
EPERM means "operation not permitted" and is returned when CGroup
access checks fail. EACCES means "permission denied" and is returned
when the device file permission bits or access control list don't
allow access.
EPERM can fail silently, since we assume the administrator disabled
a device on purpose in the CGroup. EACCESS should produce an error
message and an info message to check the device file permissions.
Change-Id: Iee4c5584c5fdc4e113c3d760dede6661097b4341
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Also rename blit_agent to region_gpu and add comments to clarify
its role in deprecated region API support rather than to do blits.
Change-Id: I80b1043db2e1c5d40a58fc801eef70a688ea9169
During registration we must not call any function that depends on registered
data as the lists are not yet complete. This includes signal allocation since
allocating shared GPU mapped memory depends on the list of GPUs.
Change-Id: I94d59e847802c546c2a5a0d9f55fe5ac3fd1d878
Device whiltelist controller cgroup allows to track and enforce open and
mknod restrictions on device files. Tasks should works with
/dev/dri/renderN devices that are whitelisted for its cgroup. If a
certain node is not whitelisted it is not an error condition.
Change-Id: I0b997423ccdc00aee98df5b6f04ed6794549604e
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Add the numa libs to the thunk specs for DEB/RPM, so we can remove the
manual installation requirement
Change-Id: I5aadcf581b64e9a20aee9c1e1204af4715d1e990
Move debug trap support capabilities to their own
structure to fix thunk spec vs header mismatch.
Change-Id: I6694601bfa36097502c8ab932e082d7a4645d5b2
Signed-off-by: Philip Cox <Philip.Cox@amd.com>
Debug agent requires handles to internal queues for single step debugging.
Added tools only API hsa_amd_runtime_queue_create_register for reporting.
hsa_amd_runtime_queue_create_register sets a callback which is invoked
when internal queues are created.
Change-Id: Ia5190ae724fadba686c15f25b2cd085350eeff0e
Required for debug agent requires copy API and trap handler to be initalized
prior to loading. Existing tools do not make use of internal queue or scratch
memory intercept which is what PostToolsInit allows.
PostToolsInit() will be removed in a following cleanup change.
Change-Id: If43377843808e3eff0defd9204910a67a852902f
On gfx900+, the test sometimes timeout due to cp fw bug.
Blacklist it until we address the root cause and have a fix.
Change-Id: Iff600a6f6dbd86c56e034f530484205520bced32
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
We observe this test fails on gfx900+. Looks like the sdma packets are not
executed at all after we submit sometimes.
Run it with timeout 2s on gfx900.
[ RUN ] KFDQMTest.SdmaEventInterrupt
[----------] SDMACopyData FAIL! 1485262707170 VS 1485262747814
[----------] Event On Queue 1:0 Timeout, try to resubmit packets!
[----------] The timeout event is signaled!
[ ] Time Consumption (ns)
[ ] 1: 1859427148
[ ] 2: 680148
[ ] 3: 6370
[ ] 4: 5481
/home/pp/code/compute/libhsakmt/tests/kfdtest/src/KFDQMTest.cpp:1670: Failure
Value of: (ret)
Actual: 31
Expected: HSAKMT_STATUS_SUCCESS
Which is: 0
[----------] SDMACopyData FAIL! 1485367669958 VS 1485367750022
[----------] Event On Queue 2:1 Timeout, try to resubmit packets!
[----------] The timeout event is signaled!
[ ] Time Consumption (ns)
[ ] 1: 1881615148
[ ] 2: 673629
[ ] 3: 6074
[ ] 4: 5481
/home/pp/code/compute/libhsakmt/tests/kfdtest/src/KFDQMTest.cpp:1670: Failure
Value of: (ret)
Actual: 31
Expected: HSAKMT_STATUS_SUCCESS
Which is: 0
[----------] SDMACopyData FAIL! 1485427671250 VS 1485427751238
[----------] Event On Queue 2:1 Timeout, try to resubmit packets!
[----------] The timeout event is signaled!
[ ] Time Consumption (ns)
[ ] 1: 1881508777
[ ] 2: 741629
[ ] 3: 6074
[ ] 4: 5481
/home/pp/code/compute/libhsakmt/tests/kfdtest/src/KFDQMTest.cpp:1670: Failure
Value of: (ret)
Actual: 31
Expected: HSAKMT_STATUS_SUCCESS
Which is: 0
[ FAILED ] KFDQMTest.SdmaEventInterrupt (23675 ms)
Change-Id: I7c1b752537d89782570df20838bf976578614f75
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Create an extra event so that the event id to test is non zero. That
way we can be sure the context id received in kernel ISR is non zero, which
is different from the default value 0 when context id is not set at all.
Change-Id: I7e261d1bbb783d5afd15558c7ac00493b1218cef
Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Apertures now overlap with the change to 48bit addressing which
precludes using aperture checks to discover buffer ownership.
Switches to ptrinfo to decide which device a buffer owned by.
This corrects faults in the legacy hsa_memory_copy api.
Change-Id: I5c7ce0216e1cdc96f836fc6fec9c3defdf4b9d90
On update, the removal will occur AFTER the new package is installed,
due to some stupidity with how yum/rpm does things. Only remove it if
we're doing a pure uninstall
Change-Id: I4982610828d8bc1f2d8691b1e4ee1718c89413cc
GPU Resource management can disable some of the GPU nodes.
The Kernel driver could be not aware of this.
Get from Kernel driver information of all the nodes and then filter it.
Change-Id: I4eeb126a5efce2192c35f5d2b72be1811e9ded32
Signed-off-by: Mike Li <Tianxinmike.Li@amd.com>
Currently the FindDRMRenderNode function will access the sysfs
directly to find the render node. It doesn't work with the
GPU management changes. Have changed code to call hsaKmtGetNodeProperties
instead.
Change-Id: I3bb537a323bc1e8c49f38d8aabc60c13e268aecd
Signed-off-by: Mike Li <Tianxinmike.Li@amd.com>
The existing call sysconf (_SC_NPROCESSORS_ONLN) provides the number of
processors available to the scheduler. When a KFD process is run under a
container environment, only a subset (cpuset) of processors are
available to the current process.
For getting CPU cache information use sched_getaffinity() to get the
number of processors available to the current process.
Change-Id: Ieac02f1f61c17e24ac34ba502968c69d3bc631cb
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Some infrastructures below,
Implement SdmaTimePacket which records the global GPU timestamp.
Introduce class AsyncMPSQ and AsyncMPMQ.
AsyncMPSQ is aka async multiple packet single queue. It takes a set of
packet when create and submits them to a GPU to run. While AsyncMPMQ is
aka async multiple packet multiple queue. It manages a set of AsyncMPSQ,
and use a forloop to do operations of AsyncMPSQ.
Implement sdma_multicopy helper functions.
Change-Id: I47e1d2ca9630113b2a1d85a0055f3f8ee629fb5f
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
For following test cases:
- KFDQMTest.QueueLatency
- KFDQMTest.BasicCuMaskingLinear
- KFDQMTest.BasicCuMaskingEven
- KFDMemoryTest.MMBandWidth
- KFDMemoryTest.MMapLarge
- KFDMemoryTest.MMBench
v2: xml element cannot start with a number, so change the key name of
MMBandWidth and MMBench accordingly
xml element cannot contain whitespaces, so trim whitespaces in "VRAM "
v3: introduce KFDLog-like way to use KFDRecord
Change-Id: Ifc3ed5657621252a7b39dccf1ef4f50a92593f77
Signed-off-by: Xiaojie Yuan <xiaojie.yuan@amd.com>
changzhu
Oak Zeng
shaoyunl
Sean Keely
Konstantin Zhuravlyov
Oak Zeng
Felix Kuehling
Mike Li
Gang Ba
Harish Kasiviswanathan
Kent Russell
Philip Cox
Evgeny
xinhui pan
Yong Zhao
Jay Cornwall
Scott Linder
Ramesh Errabolu
Xiaojie Yuan