child process clone vm objects from svm->apertures if parent process
doesn't free memory before fork. fmm_clear_all_mem suppose to clear the
apertures in forked child process but this only works if gpu_vm is not
NULL. parent process call hsaKmtCloseKFD reset gpu_vm to NULL and then
fork, then child process will not clear svm->apertures.
As a result, the child process will allocate vm object with same address
and add to aperture, there are duplicate vm objects with same address
in aperture. Then mapping to GPU will find the wrong vm object and
create incorrect GPU mapping cause rocrtst IPC test VM fault. The issue
happened with HSA_USERPTR_FOR_PAGED_MEM=1.
The fix is to clear vm objects in all apertures in clear_after_fork.
Change-Id: I92e42a967075a634a3f475b915c8242d82077ecb
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
This reverts commit 632ad3a749.
This change causes KFDTest failed on gfx803. The first hsaKmtCreateEvent
call allocate system memory for events_page because global variable
events_page is NULL. And this events page vm address should not be freed
until the process exit.
The change to destrory objects in hsaKmtCloseKFD removes events page. As
a result, KFDTest call hsaKmtOpenKFD again and then allocate memory will
get same events_page vm address on gfx803, and map this vm failed because
the vm conflict with events_page mapping.
KFDTest passed on VG10, gfx906 because allocate memory get different vm
address. hsaKmtCreateEvent still works fine as the driver keeps the
events page mapping of the process.
We should only destroy objects in fork cloned child process regardless
if gpu_vm is NULL or not.
Change-Id: I174ef65321cbd6074c855c2021318fe961c8c72c
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
/proc/cpuinfo are opened, read, and closed multiple times. Once for vendor
name and multiple times for model name -- each node opens once. For example
in a 2 CPUs + 4 GPUs system, it'll be opened 7 times. This patch reads it
one time and stores it in a cpuinfo buffer. This cpuinfo buffer is freed
when the snapshot is done.
Also replace returns with gotos inside the snapshot to avoid possible
memory leak.
Change-Id: Iaf26a6c7e7323a8651d137c3706179449b9e3c80
Signed-off-by: Amber Lin <Amber.Lin@amd.com>
adding test cases to query debugger pending events
Change-Id: I089754c508e476ce7b19e1cbd84235e4474b30c4
Signed-off-by: Jonathan Kim <Jonathan.Kim@amd.com>
Reviewed-by: Kent Russell <Kent.Russell@amd.com>
Reserve half of dma32 zone for non-NUMA system.
Change-Id: Id7aea7b6ff6cc1cc7983ecd95f8078b7f1be630c
Signed-off-by: Eric Huang <JinhuiEric.Huang@amd.com>
Add data out for enable trap to return poll fd to user space.
Add query debug events interface.
Change-Id: Ia4afde1cf167e6aa61d502380a8b329ee89d5f44
Signed-off-by: Jonathan Kim <Jonathan.Kim@amd.com>
Otherwise the parent call hsaKmtCloseKFD and then fork child process,
child process will duplicate the vm_objects from the parent.
Change-Id: Ia6ffc51cbae983b6a7cdc58ccf3b11ebe4087d97
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
The NodeId parameter is redundant and can be retrieved
from QueueId parameter.
Change-Id: I12853849b868b304bd27633fa7653ba644d69026
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
Only run the P2P test over the specified source and destination nodes if user already specify them
Change-Id: Ia3c0195cead7f46e3e28507f3255d8c59a287ab8
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
On GFX10, the wave size is determined by the COMPUTE_DISPATCH_INITIATOR value passed to
DISPATCH_DIRECT.CS_W32_EN, default 0 value was giving 64 lane waves
Change-Id: Ie8c407a24bd2825757ec481be62247b35047e5ca
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
We should use the SE number reflected by NumShaderBanks of the node
rather than hardcoding it.
Change-Id: I945fb001f81ce506249cf485a7ce25aee8219bc7
Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Ignore requests to bind to invalid NUMA nodes. This affects only
legacy applications (such as KFDTest) that allocate system memory
as paged memory with a GPU node ID.
Change-Id: I81e514af6d0c1ab2ed5229adeeca1fa0ab2a0685
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Addressed by:
4066dcd kfdtest: increase BigBufStressTest timeout and avoid VM fault
36776e9 kfdtest: avoid BigBufStressTest run on NUMA node 0
Change-Id: If21c6e42b4cf6aada1f74e77f0d8d1a2fdebcdb8
Signed-off-by: Cole Nelson <cole.nelson@amd.com>
Remove the CHIP name from the shader ISA and add wave_size(32) to make the same
shader can be used for both GFX9 and GFX10
Change-Id: I16ea72f87980c3d9c11298e20c06a0a073fe9a28
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
KFDGraphicsInterop.RegisterForeignDeviceMem looks like it is running
now. Re-enable it for kfdtest for all platforms.
Change-Id: I6f6ee9cd11da793c5d525d8676bfc6d5bd8007bb
Signed-off-by: Philip Cox <Philip.Cox@amd.com>
v_add_u32 was removed from gfx10, use carry-out explicit instruction
v_add_co_u32 instead on both gfx9 and gfx10
Change-Id: I1fcd5956844457a676757ad13bdce7f5304bb34b
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Modified shaders in KFD memory test to support gfx10.
There is no gprs register for flat_scratch on gfx10.
Use s_setreg_b32 instruction to set flat scratch base
address register
Change-Id: I505156a046056b61ce2d873343feb50ce635274a
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
This functionality doesn't work on GFX9+, and was disabled for gfx802.
Remove the test altogether for now, especially since some kernel changes
broke it on gfx803, and the functionality is deprecated now anyways. Leave
the code for reference, but "#if 0" it to prevent it from compiling or
being in the kfdtest binary
Change-Id: I848b4f23201f18612cbdc122a5b46e4010c4af2a
Some SDMA packet format might be different among asic versions
Change-Id: Ic7eda7554c23e3972e168480874ca67a92677346
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Because not all ASICs (like gfx908) have GFX rings, we should use SDMA
rings instead of GFX rings.
Change-Id: Ibcc9f9e555302ba4ce25ac76c2ca73b8c3962a58
Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
use family ID as parameter when construct the packets
Change-Id: I6c1706954ab7b8cbb8bef2aab16edf21f5e1abf0
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Add release_mem and acquire_mem pm4 packet format for nv
Change-Id: I172407c3418005922c17937e1e43f57d153ea732
Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Currently the test only covers relatively small buffers sizes. It's
useful to test buffer sizes up to 1GB to see the impact of features
that target the efficiency of large buffer allocations and mappings.
Change-Id: I2e8d5afd482894dbe2166f32d38091199b9c15e6
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
This test is designed to reproduce soft-hangs cause by HWS running
with oversubscription.
Change-Id: I49861522b3ff5ba50df5ddc968545c35ccb25353
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Create KFDMultiProcessTest base class for tests forking multiple
child processes. Derive KFDEvictTest from that class.
Change-Id: Ie5f3362c45be2b807bf7a83839ab3820352a67f9
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
mbind walks through pages to setup vma memory policy. So we need do mmap
to create vma mappings first, then call mbind. mbind will do nothing if
vma does not exist.
And add numa available check before executing mbind, and return NULL to
hsaKmtAllocMemory if mbind failed.
Change-Id: I28ab661885d807ca51ef90e87230669dc80f10ec
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
This will faciliate ASIC bringup, including under simulation environment.
Change-Id: Ie027a77a2498cba739fea51f404d9843ce8dbeae
Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
BasicAddressWatch causes issues where KFDEvictTest and
KFDQMTest.OverSubscribeCpQueues fails, and results in a GPU hang/reset.
PM4EventInterrupt just hangs indefinitely. Remove them for now to allow
the kernel merges to resume, and figure out what happened in the nv10
merge to cause it
Change-Id: I418f9561ecb3e71bc52ac48ea363fcbde82a8e2b
The SDMA blacklist should contain all tests that use SDMA. It will
be applied to all ASICs that are know to have SDMA stability issues.
Change-Id: I53e723382c12f99bddf9c535000e27737a7ea1f6
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
1. Use s_mov_b32 to move 0xcafe to s18. s_movk_i32 is a sign extention move
instruction. Oxcafe will be extended to 0xffffcafe which is not desired
2. Add wait to s_load_dword instruction to make sure memory read finish before
the next store instruction.
Change-Id: I665d1d471019edfaba5693e07cdc567d4103573f
Signed-off-by: Oak Zeng <Oak.Zeng@amd.com>
If TTM eviction and restore happens, it may takes very long time if
retry, the longest time is 5 minutes during my test. There is chance
packet is submited to queue while eviction, we have to increase the
Wait4PacketConsumption timeout.
The queue will continue to execute after eviction and restore. If we
upmap the memory from GPU while queue is evicted, this will cause VM
fault. Change to unmap memory after queue is destroyed.
Change-Id: I1b44e2274ea7b83398b2e3293578dad6947cb5af
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Because dma32 zone is on node 0, use all system memory on node 0 will
cause TTM eviction to free dma32 zone for other devices which only
work with 32bit physical address. The TTM eviction and restore may take
too long and cause queue timeout.
Running on other NUMA nodes, the NUMA default memory policy is
MPOL_PREFERRED, means TTM will get pages from local node first, and then
get remaining pages from other nodes. Check /proc/buddyinfo can confirm
this.
Reset NUMA bind to all after the test.
Change-Id: I39b373c07a2d5aa396f5c7602bffabab0481930f
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Philip Yang
Yong Zhao
Amber Lin
Jonathan Kim
Eric Huang
Oak Zeng
shaoyunl
Konstantin Zhuravlyov
Felix Kuehling
Cole Nelson
Philip Cox
Kent Russell