Commit Graph

201 Commits

Author SHA1 Message Date
Sean Keely 1e0d690948 Use ptrinfo rather than apertures in hsa_memory_copy
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
2018-10-11 13:34:53 -04:00
Jay Cornwall f1ffbc3286 Revert "Extend SDMA disable list until firmware stability resolved"
This reverts commit 5e1ccdc4a9.

Change-Id: I17b379e4d0e49a79dc8d4a60f01ea424fda24f02
2018-10-05 15:17:27 -04:00
Ramesh Errabolu 01eea21d6c Capture number of Numa Nodes present on system
Change-Id: Ic789a6b9da8e316cb483e50b0fe9faa03798f97c
2018-09-18 16:27:30 -05:00
Ramesh Errabolu f007870792 ROCr changes to enable small BAR P2P over xGMI
Change-Id: I6aaa3fe2565cdf7e15d58a7484d6bd5916ffff64
2018-09-17 22:54:40 -04:00
Sean Keely 3357cadeec Check fill addresses for alignment.
Check was documented but missing.

Change-Id: I97951635d794fd22e20c25d20e9d0e35035254af
2018-09-05 16:34:19 -04:00
Sean Keely 2843988dd7 Remove redundant initialization.
LinkInfo is already initialized to zero in its default constructor.

Change-Id: Ifa4fb886cce9b474c6879c9c82744044ab394082
2018-08-29 19:36:07 -04:00
Sean Keely 56ed5c8904 Refactor blocking sdma commands.
Remove fence pool and use two signals.  Two signals allows overlapped
submission and copy while reducing thread busy polling.

Change-Id: Idb5f8e4c7f482a596ffce9e7799191fdd785a216
2018-08-29 19:13:23 -04:00
Sean Keely e0839ab27e Implement SDMA copy rect for gfx9.
Fix pitch overflow due to small element detection.
Add wide pitch 2D copy handling.
Cleanup code duplication.

Change-Id: I93b1584aba8e5964957eb7ab3544df806ca3e2f9
2018-08-29 19:13:07 -04:00
Sean Keely aca00b7238 Add debug checking of time stamps validity.
Can only check that the signal has some time stamp, can't check if
the translating agent matches the last used agent or not.

Change-Id: I62943a864318808059c617280bb65a269dfadd1b
2018-08-26 12:36:35 -04:00
Sean Keely cd8e5c1da8 Expose ROCr build ID.
Adds HSA_AMD_SYSTEM_INFO_BUILD_VERSION=0x200 to hsa_system_info_t.
This returns a const char* pointing at the build string (git describe).

Change-Id: I73e6612482bf6ffc4037fd365808eb9211a650ad
2018-08-20 20:44:32 -05:00
Sean Keely 6c47780620 Experimental flag to swap copy agent for async copy APIs.
Adds env flag HSA_REV_COPY_DIR.  If set to 1 async copy will
copy from dst device to src device rather than from src to dst.

Change-Id: I3095642066fa026dc112c2eac06db9393341cd7e
2018-08-09 10:58:14 -04:00
Jay Cornwall 5e1ccdc4a9 Extend SDMA disable list until firmware stability resolved
Change-Id: I5e21cb761ae970ba2b68edd97b1564b36ca1f0f4
2018-08-08 11:20:14 -05:00
Sean Keely 35a270ef7e Do not initialize runtime internal queues based on mapping memory to a GPU.
Conserves VMIDs when multiple processes are in use and memory operations
are not GPU specific.  For instance HIP API hipHostMalloc does not accept
a target GPU so when used with one process per GPU (ie GPU == MPI rank) we can
quickly exceed the available VMID slots if every process consumes a VMID on
every GPU.

Change-Id: Ib6fa051290089f71581029c09f9a44b9992237d1
2018-07-13 19:58:45 -04:00
Jay Cornwall e388a23344 Add hsa_amd_queue_set_priority extension function
Controls dispatch and wavefront scheduling arbitration across quees.

Change-Id: I498f4898b544f79b8fb8514bf7e789ca9da29462
2018-06-19 19:41:28 -05:00
Sean Keely 3e3aa37750 Enable SDMA use without platform atomic support.
SDMA will use atomic completion fences if KFD reports 64bit atomic support.
Otherwise it will fall back to store completion fences.

Change-Id: I12b76f8a74ec3ee96372c250f9824d846051536e
2018-06-12 15:38:44 -04:00
Sean Keely f09eb2e8c7 Move SDMA dependencies back to hardware.
SDMA poll packet preemption has been fixed.

Change-Id: I3da878c433d4594a169e3bc8f173d3651448fd2d
2018-05-29 23:32:47 -04:00
Sean Keely c593dfc6bf Enable SDMA conditionally based on link atomic suport.
Avoids using non-atomic SDMA fences by default since that path can duplicate fences.
If HSA_ENABLE_SDMA is set this will override copy path selection and may use
non-atomic fences.

Change-Id: I4747e9a766f7f649d21ddf6bfded047ac26fd60e
2018-05-29 23:32:34 -04:00
Jay Cornwall 536823482b Handle llvm.trap only in gfx9 trap handler
llvm.debugtrap and other trap IDs are reserved and should not place
the queue into an error state.

Change-Id: I98193a35ac7da94c4a42ee75d87754ee552ebea0
2018-05-04 13:15:23 -05:00
Evgeny 0e0be791ec Tool load failure report changing to unconditional print bcos it's already is controlled with the env var
Change-Id: I91b400ba94575a32005e825e6b41bda05c55b357
2018-05-03 22:31:17 -05:00
Sean Keely a20cceb425 Relax large scratch cutoff.
Optimize for few queues rather than max.

Change-Id: I6531427319d3b2109b70d98fdb51daee7ffe4403
2018-04-30 07:25:22 -04:00
Sean Keely 5f25619bb7 Enable large scratch on GFX8.
Ensure system release fence is set on GFX8 large scratch using packets.

Change-Id: I13cfdcd35969482ea6e95e0b352f5cb3a0454b86
2018-04-30 07:24:53 -04:00
Evgeny 0fcd2fa56e aqlprofile: get version API
Change-Id: I3a85c088bfff3f54d8829e17cdafd7dfcdfb0c1d
2018-04-27 23:11:16 -04:00
Evgeny b37027e347 aqlprofile: read API
Change-Id: I896b1fbf1c19608197ac0a99b9d467d8c1bee775
2018-04-27 11:00:08 -05:00
Sean Keely 7cd6e366ed Workaround SDMA poll packet preemption.
Use async. signal handler to satisfy dependencies for SDMA blits.

Change-Id: Ifa8d3ee6810509f400a568ca2387ac6ab3ab7c36
2018-04-26 02:00:33 -05:00
Qingchuan Shi 49d2175c74 debug suport for queue error.
1/ Revised debug event handler to handle different events.
2/ Added queue error handler using the callback in queue create, which will print out wave info when queue in error state.
3/ Preempt queue instead of destory queue when queue error state.

Change-Id: Ib727d208de9caf1c72c76d42268483b24aaebde8
2018-04-20 14:25:16 -04:00
Sean Keely b66764e4c6 Disable large scratch on GFX8.
Temporary pending firmware fix.

Change-Id: Id1b1ecef421bc97327fd0d2e6225549a6e81dba0
2018-04-19 20:26:12 -05:00
Hari Thangirala 3e0cd85d69 Allow HSA_ENABLE_SDMA to override runtime defaults.
Change-Id: I2305304228010157bfb589c365f4a998577231cd
2018-04-10 12:56:48 -04:00
Konstantin Zhuravlyov 7ef70f7eaa Bring naming on par with the spec (hsa-runtime)
Change-Id: Ie1903c90a195cf95b186eb5552131a20af408adf
2018-04-10 09:15:02 -04:00
Ramesh Errabolu f25d59cab2 Compute size of command buffer based on support for HDP Flush
Change-Id: I4987a262c191a91cd845fe18002c314a95a9ed8c
2018-04-07 13:36:09 -04:00
Sean Keely 7caf9633f6 Support large scratch allocations and reclaim.
Also improve small_heap used for scratch region allocation.

Change-Id: Ib7311b663b38968d88ebc355b81e12c0863dc541
2018-04-05 21:51:56 -04:00
Jay Cornwall df964343a3 Support new first-level trap handler ABI
- Ignore exceptions passed to the second-level handler
- Restore SQ_WAVE_IB_STS and SQ_WAVE_STATUS before exiting trap

Change-Id: I872c111c030d94eae644ae073df3c2e508f42f45
2018-04-04 11:01:14 -05:00
Sean Keely b6f0248f53 Respect new memory model requirements at queue destroy.
Spec requires GPU release fences and CPU acquire fences at queue destroy.
Also update the recognized status codes.

Change-Id: If9166f5149f65417c7057ff7c0f69f6ac094d6ab
2018-04-04 08:13:00 -04:00
Sean Keely 6df9ba97ce Sequence queue error callbacks with queue destroy.
HSA v1.2 update.

Change-Id: I13975e71b2c1ea5b7738236f5d02df84312ad00c
2018-04-04 08:12:58 -04:00
Ramesh Errabolu 987f3f97aa Enable sDMA packet HDP Flush on Gfx9 and later devices
Change-Id: I85922e5266883ef7e9eed3565e2c3b209009d294
2018-04-02 11:47:59 -04:00
Konstantin Zhuravlyov e4c6a3ae28 Convert old target names to old new target names.
Change-Id: I701651d74f353e234556e4bf5d50d63c598e5f15
2018-03-26 11:59:13 -04:00
Sean Keely cd46954cc4 Cleanup in blit kernel management code.
Remove unused function (FenceRelease), add comments to barrier packet settings,
correct profiling controls to work with queue wrappers.

Change-Id: I45bb26227bcc2b78edb8ad5dc497603c33234e18
2018-03-20 22:19:54 -04:00
Sean Keely f4521ce782 Revert "Reduce to only one internal compute queue."
This reverts commit 0eb534e3cf.

Change-Id: Ifcc5e148457243a6cf9ef277da7ab7c4e10f6fc9
2018-03-20 22:19:44 -04:00
Wilkin 8e3d26c617 ROCm Runtime Support for respecting target xnack setting
This includes the changes provided by Konstantin, "Add xnack from elf header" (Change 136389).

Change-Id: I95e51141caa0d7c21903b09212c02e4906ec54a3
2018-03-20 16:57:15 -04:00
Sean Keely ac5ccb45b7 Use atomic variable for Runtime ref_count_.
Change-Id: Ic4d0ad9ff93d0cc52cfe2df006ee3436d5960b07
2018-03-06 03:45:14 -06:00
Sean Keely 31c05d2fc7 Add exception safety to Runtime::Acquire.
Change-Id: Ia2a9baf08bb56971412f1ac3914592612de5f134
2018-02-28 05:21:07 -06:00
Sean Keely 95c926059d Improve fragment map reporting format.
Change-Id: I85d09d085b08de46271ec902c766a8609a4b921a
2018-02-09 14:03:03 -05:00
Sean Keely 9212e7a09f Emit fragment map and thunk ptr info with VM faults.
Change-Id: If1302f674df7a636529c64bf66dfdda755a70c32
2018-02-09 14:02:26 -05:00
Sean Keely 0eb534e3cf Reduce to only one internal compute queue.
Change-Id: Ie42ecb3b242077624d74caeabfcd418dbbd9ff3e
2018-02-09 14:02:15 -05:00
Sean Keely bd5dd47ca1 Defer creation of internal queues and blits until first needed.
Change-Id: I2e61d7e102f38389d806d9eb24beda910573157b
2018-02-09 14:02:07 -05:00
James Edwards b913795c31 : Fix compilation errors with gcc 7.2.1 for hsa runtime.
Change-Id: I3356388753ca78cc0f1e0c3188220d7f3f60283d
2018-02-07 09:22:39 -06:00
Evgeny 8a8d7ad814 ExecutePM4 queue full check fix
Change-Id: Id56ece6d3f5eab1ef3a2758922022f0996c1efe4
2018-02-05 19:35:39 -06:00
Chris Freehill 3449f7dea6 Don't support platform atomics for gfx9XX
Change-Id: I302c862494e221ae2b6b3e1a843f06586b0b28ba
2018-02-02 18:21:16 -05:00
Sean Keely f59b001c75 Guard against IPC signal use of when profiling copy APIs.
Also update IPC signal API text to allow single process profiling with IPC signals.

Change-Id: I90b246623129d57183acb4ba1789beec360547c3
2018-01-31 19:05:32 -05:00
Sean Keely 91f559802d Revert CRAT table workaround.
Change-Id: Ic2bf9e1fb1d00c5a31d52560e0eb37e0ae1ab08a
2018-01-30 18:26:53 -06:00
Chris Freehill 563581223c Report physical memory instead of virtual memory
Change-Id: I18105e3982a96aea40e05cd78521c0c3acf75de4
2017-12-20 22:11:50 -04:00