Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
1,Changes on virtualization detections
2,Don't load smu & mc firmware if using sr-iov bios
3,skip vPost for sriov & force vPost if dev pass-through
v2: agd: squash in Rays's fix for the missed SI case
v3: agd: squash in additional fixes for CIK, SI, cleanup
Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
v1:
for gfx8, use CONTEXT_CONTROL package to dynamically
skip preamble CEIB and other load_xxx command in sequence.
v2:
support GFX7 as well.
remove cntxcntl in compute ring funcs because CPC doesn't
support this packet.
v3: fix reduntant judgement in cntxcntl.
v4: some cleanups, don't change cs_submit()
v5: keep old MESA supported & bump up KMS version.
Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Ack-by: Chunming Zhou <David1.Zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
reset the asic if adapter is not powerdown when doing freeze()
thaw() and restore(), in order to get a valid state of adapter.
v2: squash in warning fix from Rex
Signed-off-by: JimQu <Jim.Qu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Tested By: Shawn Starr <shawn.starr@rogers.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Normally on shutdown or reboot we don't care about necessarily
making sure the hw is in a good state because the system is about
to be powered down or reset. However, after a shutdown or reboot
in a VM, it's best to tear down the hw properly otherwise there
can be problems with the next VM use.
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The sbios always powers up the dGPU on resume.
Acked-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Ken Wang <Qingqing.Wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The old mechanism used a per-submission limit that didn't take previous
submissions within the same time frame into account. It also filled VRAM
slowly when VRAM usage dropped due to a big eviction or buffer deallocation.
This new method establishes a configurable MBps limit that is obeyed when
VRAM usage is very high. When VRAM usage is not very high, it gives
the driver the freedom to fill it quickly. The result is more consistent
performance.
It can't keep the BO move rate low if lots of evictions are happening due
to VRAM fragmentation, or if a big buffer is being migrated.
The amdgpu.moverate parameter can be used to set a non-default limit.
Measurements can be done to find out which amdgpu.moverate setting gives
the best results.
Mainly APUs and cards with small VRAM will benefit from this. For F1 2015,
anything with 2 GB VRAM or less will benefit.
Some benchmark results - F1 2015 (Tonga 2GB):
Limit MinFPS AvgFPS
Old code: 14 32.6
128 MB/s: 28 41
64 MB/s: 15.5 43
32 MB/s: 28.7 43.4
8 MB/s: 27.8 44.4
8 MB/s: 21.9 42.8 (different run)
Random drops in Min FPS can still occur (due to fragmented VRAM?), but
the average FPS is much better. 8 MB/s is probably a good limit for this
game & the current VRAM management. The random FPS drops are still to be
tackled.
v2: use a spinlock
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The old names were dragged over from radeon. The new ones
better match the naming conventions used in the driver.
No functional change.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Writes to this register are the preferred way to do NOPs.
Bump the driver version as well.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
For profiling.
v2: really bump the minor version
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
For virtual display feature, as there may be multiple GPUs,
for user could choose whiche GPU need to enable this feature, change
the type of virtual_display from int to char*. The variable will be set
like this virtual_display="xxxx:xx:xx.x;xxxx:xx:xx.x;".
Signed-off-by: Emily Deng <Emily.Deng@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
For virtual display feature, define on variable in amdgpu.ko. When want to
enable virtual display feature, need set the option "amdgpu.virtual_display=1".
And then disable vga render and crtc if have DCE engine.
Signed-off-by: Emily Deng <Emily.Deng@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We need to check on Polaris if UVD session context is allowed or not.
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>
This flag was being set unconditionally at runtime so just set it at
compile time instead.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Frank Binns <frank.binns@imgtec.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This parameter will allow disabling individual CUs on module load, e.g.
amdgpu.disable_cu=2.0.3,2.0.4 to disable CUs 3 and 4 of SE2.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
They allow disabling clock and power gating from the kernel command line,
which hopefully helps with diagnosing problems in the field.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Nicolai Hähnle <Nicolai.Haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The ATPX power control method does this for you.
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The platform d3 cold is used to power down the dGPU.
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
On PX systems without dGPU power control, use PCI_D3hot.
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch makes powercontainment feature configurable. Currently, the
powercontainment is not very stable, so add a module parameter to
enable/disable it via user mode.
Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Flora Cui <Flora.Cui@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Flora Cui <Flora.Cui@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This avoids problems with multiple GPUs. For example,
if the first GPU failed before amdgpu_fence_init() was
called, amdgpu_fence_slab_ref is still 0 and it will
get decremented in amdgpu_fence_driver_fini(). This
will lead to a crash during init of the second GPU since
amdgpu_fence_slab_ref is not 0.
v2: add functions for init/exit instead of
moving the variables into the driver.
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
TC_WB_ACTION must be set according to the docs
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This is the first big radeon/amdgpu pull request for 4.7. Highlights:
- Polaris support in amdgpu
Current display stack on par with other asics, for advanced features DAL is required
Power management support
Support for GFX, Compute, SDMA, UVD, VCE
- VCE and UVD init/fini cleanup in radeon
- GPUVM improvements
- Scheduler improvements
- Clockgating improvements
- Powerplay improvements
- TTM changes to support driver specific LRU update mechanism
- Radeon support for new Mesa features
- ASYNC pageflip support for radeon
- Lots of bug fixes and code cleanups
* 'drm-next-4.7' of git://people.freedesktop.org/~agd5f/linux: (180 commits)
drm/amdgpu: Replace rcu_assign_pointer() with RCU_INIT_POINTER()
drm/amdgpu: use drm_mode_vrefresh() rather than mode->vrefresh
drm/amdgpu/uvd6: add bypass support for fiji (v3)
drm/amdgpu/fiji: set UVD CG state when enabling UVD DPM (v2)
drm/powerplay: add missing clockgating callback for tonga
drm/amdgpu: Constify some tables
drm/amd/powerplay: Delete dead struct declaration
drm/amd/powerplay/hwmgr: don't add invalid voltage
drm/amd/powerplay/hwmgr: prevent VDDC from exceeding 2V
MAINTAINERS: Remove unneded wildcard for the Radeon/AMDGPU drivers
drm/radeon: add cayman VM support for append packet.
drm/amd/amdgpu: Add debugfs entries for smc/didt/pcie
drm/amd/amdgpu: Drop print_status callbacks.
drm/amd/powerplay: revise reading/writing pptable on Polaris10
drm/amd/powerplay: revise reading/writing pptable on Tonga
drm/amd/powerplay: revise reading/writing pptable on Fiji
drm/amd/powerplay: revise caching the soft pptable and add it's size
drm/amd/powerplay: add dpm force multiple levels on cz/tonga/fiji/polaris (v2)
drm/amd/powerplay: fix fan speed percent setting error on Polaris10
drm/amd/powerplay: fix bug dpm can't work when resume back on Polaris
...
Adjust to preferred code names.
Signed-off-by: Flora Cui <Flora.Cui@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch marks some compile-time constant tables 'const'.
The tables marked in this patch are the low hanging fruit
where little other changes were necesary to avoid casting
away constness etc. Also mark some tables that are private
to a file as static.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Nils Wallménius <nils.wallmenius@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This allows us to ditch a ton of ugly #ifdefs from a bunch of drm modeset
drivers.
v2: Make the dummy function actually return a sane value, spotted by
Ville.
v3: Because the patch is still in limbo there's no more drivers to
convert, noticed by Emil.
v4: Rebase once more, because hooray. I'll just go ahead an apply this
one later on to drm-misc.
Cc: Emil Velikov <emil.l.velikov@gmail.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
A few more fixes and cleanups for 4.6:
- DCE code cleanups
- HDP flush/invalidation fixes
- GPUVM fixes
- switch to drm_vblank_[on|off]
- PX fixes
- misc bug fixes
* 'drm-next-4.6' of git://people.freedesktop.org/~agd5f/linux: (50 commits)
drm/amdgpu: split pipeline sync out of SDMA vm_flush() as well
drm/amdgpu: Revert "add mutex for ba_va->valids/invalids"
drm/amdgpu: Revert "add lock for interval tree in vm"
drm/amdgpu: Revert "add spin lock to protect freed list in vm (v3)"
drm/amdgpu: reserve the PD during unmap and remove
drm/amdgpu: Fix two bugs in amdgpu_vm_bo_split_mapping
drm/radeon: Don't drop DP 2.7 Ghz link setup on some cards.
MAINTAINERS: update radeon entry to include amdgpu as well
drm/amdgpu: disable runtime pm on PX laptops without dGPU power control
drm/radeon: disable runtime pm on PX laptops without dGPU power control
drm/amd/amdgpu: Fix indentation in do_set_base() (DCEv8)
drm/amd/amdgpu: make afmt_init cleanup if alloc fails (DCEv8)
drm/amd/amdgpu: Move config init flag to bottom of sw_init (DCEv8)
drm/amd/amdgpu: Don't proceed into audio_fini if audio is disabled (DCEv8)
drm/amd/amdgpu: Fix identation in do_set_base() (DCEv10)
drm/amd/amdgpu: Make afmt_init cleanup if alloc fails (DCEv10)
drm/amd/amdgpu: Move initialized flag to bottom of sw_init (DCEv10)
drm/amd/amdgpu: Don't proceed in audio_fini if disabled (DCEv10)
drm/amd/amdgpu: Fix indentation in dce_v11_0_crtc_do_set_base()
drm/amd/amdgpu: Make afmt_init() cleanup if alloc fails (DCEv11)
...
We need them all the time.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
amdgpu must load only after amdkfd's loading has been completed. If that
is not enforced, then amdgpu's call into amdkfd's functions will cause a
kernel BUG.
When amdgpu and amdkfd are built as kernel modules, that rule is enforced
by the kernel's modules loading mechanism. When amdgpu and amdkfd are
built inside the kernel image, that rule is enforced by ordering in the
drm Makefile (amdkfd before amdgpu).
Instead of using drm Makefile ordering, we can now use deferred loading
as amdkfd now returns -EPROBE_DEFER in kgd2kfd_init() when it is not yet
loaded.
This patch defers amdgpu loading by propagating -EPROBE_DEFER to the
kernel's drivers loading infrastructure. That will put amdgpu into the
pending drivers list (see description in dd.c). Once amdkfd is loaded,
a call to kgd2kfd_init() will return successfully and amdgpu will be able
to load.
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
It doesn't currently do anything and there's no need for it
going forward since pci config reset will be required as a
fallback even when we have fine grained reset implemented.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Non-scheduler code is longer supported.
v2: agd: rebased on upstream
Signed-off-by: Chunming Zhou <David1.Zhou@amd.com>
Reviewed-by: Ken Wang <Qingqing.Wang@amd.com>
Reviewed-by: Monk Liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
No longer used.
Signed-off-by: Chunming Zhou <David1.Zhou@amd.com>
Reviewed-by: Ken Wang <Qingqing.Wang@amd.com>
Reviewed-by: Monk Liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Allows the user to force the supported pcie gen and lane
config on both the asic and the chipset.
Useful for debugging pcie problems and for virtualization
where we may not be able to query the pcie bridge caps.
Default to:
gen: chipset 1/2, asic 1/2/3
lanes: 1/2/4/8/16
v2: fix bare metal case
Reviewed-by: monk liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The amdgu_powerplay variable is global for multiple GPU instances.
v2: fold in Flora's module option change, protect adev reference in
macros
Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
This option can be used to enable the new powerplay implementation,
and it is disabled by default.
Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
amdgpu_pp_ip_funcs is introduced to handle the two code paths,
the legacy one and the new powerplay implementation.
CONFIG_DRM_AMD_POWERPLAY kernel configuration option is
introduced for the powerplay component.
v4: squash in fixes
v3: register debugfs file when powerplay module enable
v2: add amdgpu_ucode_init_bo in hw init when amdgpu_powerplay enable.
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Change the default scheduler queue size from 16 to 32.
Signed-off-by: Chunming Zhou <David1.Zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
More amdgpu and radeon stuff for drm-next. Stoney support is the big change.
The rest is just bug fixes and code cleanups. The Stoney stuff is pretty
low impact with respect to existing chips.
* 'drm-next-4.4' of git://people.freedesktop.org/~agd5f/linux:
drm/amdgpu: change VM size default to 64GB
drm/amdgpu: add Stoney pci ids
drm/amdgpu: update the core VI support for Stoney
drm/amdgpu: add VCE support for Stoney (v2)
drm/amdgpu: add UVD support for Stoney
drm/amdgpu: add GFX support for Stoney (v2)
drm/amdgpu: add SDMA support for Stoney (v2)
drm/amdgpu: add DCE support for Stoney
drm/amdgpu: Update SMC/DPM for Stoney
drm/amdgpu: add GMC support for Stoney
drm/amdgpu: add Stoney chip family
drm/amdgpu: fix the broken vm->mutex V2
drm/amdgpu: remove the unnecessary parameter adev for amdgpu_fence_wait_any()
drm/amdgpu: remove the exclusive lock
drm/amdgpu: remove old lockup detection infrastructure
drm: fix trivial typos
drm/amdgpu/dce: simplify suspend/resume
drm/amdgpu/gfx8: set TC_WB_ACTION_EN in RELEASE_MEM packet
drm/radeon: Use rdev->gem.mutex to protect hyperz/cmask owners
That's still small enough to not waste to much memory on PD/PTs.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
