IH tracks pending retry faults in a hash table for fast lookup in
interrupt context. Each VM has a short FIFO of pending VM faults for
processing in a bottom half.
The IH prescreening stage adds retry faults and filters out repeated
retry interrupts to minimize the impact of interrupt storms.
It's the VM's responsibility remove pending faults once they are
handled. For now this is only done when the VM is destroyed.
v2:
- Made the hash table smaller and the FIFO longer. I never want the
FIFO to fill up, because that would make prescreen take longer.
128 pending page faults should be enough to keep migrations busy.
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Christian König <christian.koenig@amd.com> (v1)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This adds a statically sized closed hash table implementation with
low memory and CPU overhead. The API is inspired by kfifo.
Storing, retrieving and deleting data does not involve any dynamic
memory management, which makes it ideal for use in interrupt context.
Static memory usage per entry comprises a 32 or 64 bit hash key, two
bits for occupancy tracking and the value size stored in the table.
No list heads or pointers are needed. Therefore this data structure
should be quite cache-friendly, too.
It uses linear probing and lazy deletion. During lookups free space
is reclaimed and entries relocated to speed up future lookups.
v2: squash in do_div and _BITOPS_LONG_SHIFT fixes
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Allows assigning a PASID to a VM for identifying VMs involved in page
faults. The global PASID manager is also exported in the KFD
interface so that AMDGPU and KFD can share the PASID space.
PASIDs of different sizes can be requested. On APUs, the PASID size
is deterined by the capabilities of the IOMMU. So KFD must be able
to allocate PASIDs in a smaller range.
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
currently, for CI asics,
use dpm by default, amdgpu.dpm=-1.
when set amdgpu.dpm=1, enable powplay.
when set amdgpu.dpm=0, disable both dpm and powerplay.
when powerplay is stable on CI asics, ci_dpm will
be removed.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This ports support for CI asics (Bonaire, Hawaii)
to the powerplay smumgr
v2: warning fix (Alex)
v3: squash in fix for thermal (Tom)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Avoid intermediate negative numbers when doing calculations with a mix
of signed and unsigned variables where implicit conversions can lead
to unexpected results.
When kernel queue buffer wraps around to 0, we need to check that rptr
won't be overwritten by the new packet.
Signed-off-by: Yong Zhao <yong.zhao@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
There are already CHIP_* definitions under amd_shared.h file on amdgpu
side, so KFD should reuse them rather than defining new ones.
Using enum for asic type requires default cases on switch statements
to prevent compiler warnings. WARN on unsupported ASICs. It should never
get there because KFD should not be initialized on unsupported devices.
v2: Replace BUG() with WARN and error return
Signed-off-by: Yong Zhao <Yong.Zhao@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
The hard-coded values related to VMID were removed in KFD, as those
values can be calculated in the KFD initialization function.
v2: remove unnecessary local variable
Signed-off-by: Yong Zhao <yong.zhao@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Adjust latencies and timeouts for dequeueing with HWS and consolidate
them in one place. Make them longer to allow long running waves to
complete without causing a timeout. The timeout is twice as long as the
latency plus some buffer to make sure we don't detect a timeout
prematurely.
Change timeouts for dequeueing HQDs without HWS. KFD_UNMAP_LATENCY is
more consistent with what the HWS does for user queues.
Signed-off-by: Yong Zhao <yong.zhao@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
The timeout in milliseconds should not be regarded as jiffies. This
commit fixed that.
v2:
- use msecs_to_jiffies
- change timeout_ms parameter to unsigned int to match msecs_to_jiffies
Signed-off-by: Yong Zhao <yong.zhao@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
When we do suspend/resume through "sudo pm-suspend" while there is
HSA activity running, upon resume we will encounter HWS hanging, which
is caused by memory read/write failures. The root cause is that when
suspend, we neglected to unbind pasid from kfd device.
Another major change is that the bind/unbinding is changed to be
performed on a per process basis, instead of whether there are queues
in dqm.
v2:
- free IOMMU device if kfd_bind_processes_to_device fails in kfd_resume
- add comments to kfd_bind/unbind_processes_to/from_device
- minor cleanups
Signed-off-by: Yong Zhao <yong.zhao@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
The idea is to let kfd init and resume function share the same code path
as much as possible, rather than to have two copies of almost identical
code. That way improves the code readability and maintainability.
Signed-off-by: Yong Zhao <yong.zhao@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
some trivial amdkfd cleanups
* tag 'drm-amdkfd-next-2017-09-02' of git://people.freedesktop.org/~gabbayo/linux:
drm/amdkfd: pass queue's mqd when destroying mqd
drm/amdkfd: remove memset before memcpy
uapi linux/kfd_ioctl.h: only use __u32 and __u64
Pull drm AMD fixes from Dave Airlie:
"Just had a single AMD fixes pull from Alex for rc1"
* tag 'drm-fixes-for-v4.14-rc1' of git://people.freedesktop.org/~airlied/linux:
drm/amdgpu: revert "fix deadlock of reservation between cs and gpu reset v2"
drm/amdgpu: remove duplicate return statement
drm/amdgpu: check memory allocation failure
drm/amd/amdgpu: fix BANK_SELECT on Vega10 (v2)
drm/amdgpu: inline amdgpu_ttm_do_bind again
drm/amdgpu: fix amdgpu_ttm_bind
drm/amdgpu: remove the GART copy hack
drm/ttm:fix wrong decoding of bo_count
drm/ttm: fix missing inc bo_count
drm/amdgpu: set sched_hw_submission higher for KIQ (v3)
drm/amdgpu: move default gart size setting into gmc modules
drm/amdgpu: refine default gart size
drm/amd/powerplay: ACG frequency added in PPTable
drm/amdgpu: discard commands of killed processes
drm/amdgpu: fix and cleanup shadow handling
drm/amdgpu: add automatic per asic settings for gart_size
drm/amdgpu/gfx8: fix spelling typo in mqd allocation
drm/amd/powerplay: unhalt mec after loading
drm/amdgpu/virtual_dce: Virtual display doesn't support disable vblank immediately
drm/amdgpu: Fix huge page updates with CPU
Several users have complained that the tile table update broke Oland
support. Despite several attempts to fix it, the root cause is still
unknown at this point and no solution is available. As it is not
acceptable to leave a known regression breaking a major functionality
in the kernel for several releases, let's just reverse this
optimization for now. It can be implemented again later if and only
if the breakage is understood and fixed.
As there were no complaints for Hainan so far, only the Oland part of
the offending commit is reverted. Optimization is preserved on
Hainan, so this commit isn't an actual revert of the original.
This fixes bug #194761:
https://bugzilla.kernel.org/show_bug.cgi?id=194761
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Jean Delvare <jdelvare@suse.de>
Fixes: f8d9422ef8 ("drm/amdgpu: update tile table for oland/hainan")
Cc: Flora Cui <Flora.Cui@amd.com>
Cc: Junwei Zhang <Jerry.Zhang@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Marek Olšák <maraeo@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
We adjusted the BO flags for USWC handling, but those never took effect
because the placement was passed in instead of generated inside this
function.
v2: better commit message
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fixes crash when trying to unload the amdgpu module before the fbdev
framebuffer was initialized, which can happen since the DRM fbdev helper
code supports deferred setup.
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
