Drop the "kernel" and sg parameter and give the BO type to create
explicit to amdgpu_bo_create instead of figuring it out from the
parameters.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Just set the GTT domain as mandatory, so that the BO is validated into
it on first use. This allows us to setup the sg table later on.
Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Roger He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Instead of moving this to the SYSTEM domain just drop the backing store
and let the resulting allocation be freed.
Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Roger He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Instead of calculating the size in bytes just to recalculate the number
of pages from it pass the BO directly to the function.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
this patch actually refactor mailbox implmentations, and
all below changes are needed together to fix all those mailbox
handshake issues exposured by heavey TDR test.
1)refactor all mailbox functions based on byte accessing for mb_control
reason is to avoid touching non-related bits when writing trn/rcv part of
mailbox_control, this way some incorrect INTR sent to hypervisor
side could be avoided, and it fixes couple handshake bug.
2)trans_msg function re-impled: put a invalid
logic before transmitting message to make sure the ACK bit is in
a clear status, otherwise there is chance that ACK asserted already
before transmitting message and lead to fake ACK polling.
(hypervisor side have some tricks to workaround ACK bit being corrupted
by VF FLR which hase an side effects that may make guest side ACK bit
asserted wrongly), and clear TRANS_MSG words after message transferred.
3)for mailbox_flr_work, it is also re-worked: it takes the mutex lock
first if invoked, to block gpu recover's participate too early while
hypervisor side is doing VF FLR. (hypervisor sends FLR_NOTIFY to guest
before doing VF FLR and sentds FLR_COMPLETE after VF FLR done, and
the FLR_NOTIFY will trigger interrupt to guest which lead to
mailbox_flr_work being invoked)
This can avoid the issue that mailbox trans msg being cleared by its VF FLR.
4)for mailbox_rcv_irq IRQ routine, it should only peek msg and schedule
mailbox_flr_work, instead of ACK to hypervisor itself, because FLR_NOTIFY
msg sent from hypervisor side doesn't need VF's ACK (this is because
VF's ACK would lead to hypervisor clear its trans_valid/msg, and this
would cause handshake bug if trans_valid/msg is cleared not due to
correct VF ACK but from a wrong VF ACK like this "FLR_NOTIFY" one)
This fixed handshake bug that sometimes GUEST always couldn't receive
"READY_TO_ACCESS_GPU" msg from hypervisor.
5)seperate polling time limite accordingly:
POLL ACK cost no more than 500ms
POLL MSG cost no more than 12000ms
POLL FLR finish cost no more than 500ms
6) we still need to set adev into in_gpu_reset mode after we received
FLR_NOTIFY from host side, this can prevent innocent app wrongly succesed
to open amdgpu dri device.
FLR_NOFITY is received due to an IDLE hang detected from hypervisor side
which indicating GPU is already die in this VF.
v2:
use MACRO as the offset of mailbox_control register
don't test if NOTIFY_CMPL event in rcv_msg since it won't
recieve that message anymore
Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Pixel Ding <Pixel.Ding@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
mailbox registers can be accessed with a byte boundry according
to BIF team, so this patch prepares register byte access
and will be used by following patches.
Actually, for mailbox registers once the byte field is touched even not changed,
the mailbox behaves, so we need the byte width accessing to those sort of regs.
Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Pixel Ding <Pixel.Ding@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The commit d296278fd372003fc69588acfd0c0c5edbdf4874 added support for
detecting DDR4 but omitted the label that is printed out in
amdgpu_bo_init() resulting in a KASAN error.
Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The amdgpu_ucode_fini_bo should be called after gfx_v8_0_hw_fini,
or it will have KCQ disable failed issue.
For Tonga, as it firstly finishes SMC block, and the SMC hw fini
will call amdgpu_ucode_fini, which will lead the amdgpu_ucode_fini_bo
called before gfx_v8_0_hw_fini, this is incorrect.
Signed-off-by: Emily Deng <Emily.Deng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The amdgpu_pm_sysfs_fini should call before amdgpu_device_ip_fini,
or the adev->pm.dpm_enabled would be set to 0, then the device files
related to pp won't be removed by amdgpu_pm_sysfs_fini when unload
driver.
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>
Major points for this pull request:
- Add dGPU support for amdkfd initialization code and queue handling. It's
not complete support since the GPUVM part is missing (the under debate stuff).
- Enable PCIe atomics for dGPU if present
- Various adjustments to the amdgpu<-->amdkfd interface for dGPUs
- Refactor IOMMUv2 code to allow loading amdkfd without IOMMUv2 in the system
- Add HSA process eviction code in case of system memory pressure
- Various fixes and small changes
* tag 'drm-amdkfd-next-2018-03-11' of git://people.freedesktop.org/~gabbayo/linux: (24 commits)
uapi: Fix type used in ioctl parameter structures
drm/amdkfd: Implement KFD process eviction/restore
drm/amdkfd: Add GPUVM virtual address space to PDD
drm/amdkfd: Remove unaligned memory access
drm/amdkfd: Centralize IOMMUv2 code and make it conditional
drm/amdgpu: Add submit IB function for KFD
drm/amdgpu: Add GPUVM memory management functions for KFD
drm/amdgpu: add amdgpu_sync_clone
drm/amdgpu: Update kgd2kfd_shared_resources for dGPU support
drm/amdgpu: Add KFD eviction fence
drm/amdgpu: Remove unused kfd2kgd interface
drm/amdgpu: Fix wrong mask in get_atc_vmid_pasid_mapping_pasid
drm/amdgpu: Fix header file dependencies
drm/amdgpu: Replace kgd_mem with amdgpu_bo for kernel pinned gtt mem
drm/amdgpu: remove useless BUG_ONs
drm/amdgpu: Enable KFD initialization on dGPUs
drm/amdkfd: Add dGPU device IDs and device info
drm/amdkfd: Add dGPU support to kernel_queue_init
drm/amdkfd: Add dGPU support to the MQD manager
drm/amdkfd: Add dGPU support to the device queue manager
...
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae3 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a07 ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
More stuff for 4.17. Highlights:
- More fixes for "wattman" like functionality (fine grained clk/voltage control)
- Add more power profile infrastucture (context based dpm)
- SR-IOV fixes
- Add iomem debugging interface for use with umr
- Powerplay and cgs cleanups
- DC fixes and cleanups
- ttm improvements
- Misc cleanups all over
* 'drm-next-4.17' of git://people.freedesktop.org/~agd5f/linux: (143 commits)
drm/amdgpu:Always save uvd vcpu_bo in VM Mode
drm/amdgpu:Correct max uvd handles
drm/amdgpu: replace iova debugfs file with iomem (v3)
drm/amd/display: validate plane format on primary plane
drm/amdgpu: Clean sdma wptr register when only enable wptr polling
drm/amd/amdgpu: re-add missing GC 9.1 and SDMA0 4.1 sh_mask header files
drm/amdgpu: give warning before sleep in kiq_r/wreg
drm/amdgpu: further mitigate workaround for i915
drm/amdgpu: drop gtt->adev
drm/amdgpu: add amdgpu_evict_gtt debugfs entry
drm/amd/pp: Add #ifdef checks for CONFIG_ACPI
drm/amd/pp: fix "Delete the wrapper layer of smu_allocate/free_memory"
drm/amd/pp: Drop wrapper functions for upper/lower_32_bits
drm/amdgpu: Delete cgs wrapper functions for gpu memory manager
drm/amd/pp: Delete the wrapper layer of smu_allocate/free_memory
drm/amd/pp: Remove cgs wrapper function for temperature update
Revert "drm/amd/pp: Add a pp feature mask bit for AutoWattman feature"
drm/amd/pp: Add auto power profilng switch based on workloads (v2)
drm/amd/pp: Revert gfx/compute profile switch sysfs
drm/amd/pp: Fix sclk in highest two levels when compute on smu7
...
In dce110, the plane configuration is such that plane 0
or the primary plane should be rendered with only RGB data.
This patch adds the validation to ensure that no video data
is rendered on plane 0.
Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
HW Engineer's Notes:
During switch from vga->extended, if we set the VGA_TEST_ENABLE and then
hit the VGA_TEST_RENDER_START, then the DCHUBP timing gets updated correctly.
Then vBIOS will have it poll for the VGA_TEST_RENDER_DONE and unset
VGA_TEST_ENABLE, to leave it in the same state as before.
Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
While checking plane states for updates during atomic check, we create
dc_plane_states in preparation. These dc states should be freed if
something errors.
Although the input transfer function is also freed by
dc_plane_state_release(), we should free it (on error) under the same
scope as where it is created.
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Before dig fe is enabled, infoframe can't be programmed. So in
suspend resume case our infoframe programmming was not going through.
This change changes the sequence so that infoframe is programmed
after.
Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fixing null-deref on Vega10 due to regression after
'fix cursor related Pstate hang' change.
Added null checks in setting cursor position.
Signed-off-by: Roman Li <Roman.Li@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Move cursor programming to inside the OTG_MASTER_UPDATE_LOCK
If graphics plane go from 1 pipe to hsplit, the cursor updates
after mpc programming and unlock. Which means there is a window
of time where cursor is enabled on the wrong pipe if it's on
the right side of the screen (i.e. case where cursor need to
move from pipe 0 to pipe 3 post split). This will cause pstate hang.
Solution is to program the cursor while still locked.
Signed-off-by: Eric Yang <Eric.Yang2@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Because AMDGPU_CRTC_IRQ_VLINE1 = 6, it expected 6 more crtcs to be
programed with disabled irq state in amdgpu_irq_disable_all. That caused errors and accessed
the wrong memory location.
Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
There is no need to call drm_mode_set_crtcinfo() again once
crtc timing is decided. Otherwise non-native/unsupported timing
might get overwritten.
Signed-off-by: Jerry (Fangzhi) Zuo <Jerry.Zuo@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
There's no good place in DC to cover all place where stream signal should
be updated. update_stream_signal depends on timing which comes from DM.
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We've got a helper function to call dc_create_stream_for_sink and one
other place that calls it directly. Make sure we call the helper
functions always since we need to update a bunch of things in stream and
don't want to miss that.
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
When topology changed and rehook up MST display to the same DP
connector, need to take care of drm_dp_mst_port object.
Due to the topology is changed, drm_dp_mst_port and corresponding
i2c_algorithm object could be NULL in such situation.
Signed-off-by: Jerry (Fangzhi) Zuo <Jerry.Zuo@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The below commit
"drm/atomic: Try to preserve the crtc enabled state in drm_atomic_remove_fb, v2"
introduces a slight behavioral change to rmfb. Instead of disabling a crtc
when the primary plane is disabled, it now preserves it.
This change leads to BUG hit while performing atomic commit on amd driver.
As a fix this patch ensures that we disable the CRTC's with NULL FB by returning
-EINVAL and hence triggering fall back to the old behavior and turning off the
crtc in atomic_remove_fb().
V2: Added error check for plane_state and removed sanity check for crtc.
Signed-off-by: Shirish S <shirish.s@amd.com>
Signed-off-by: Pratik Vishwakarma <Pratik.Vishwakarma@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch updates the dc's plane state with the parameters set by the
user side.
This is needed to validate the plane capabilities with the parameters
user space wants to set.
Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
CZ & ST support uptil a limit 2:1 downscaling, this patch
adds validate_plane hook, that shall be used to validate
the plane attributes sent by the user space based
on dce110 capabilities.
Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
amdgpu_dm_atomic_check() is used to validate the entire configuration of
planes and crtc's that the user space wants to commit.
However amdgpu_dm_atomic_check() depends upon DRM_MODE_ATOMIC_ALLOW_MODESET
flag else its mostly dummy.
Its not mandatory for the user space to set DRM_MODE_ATOMIC_ALLOW_MODESET,
and in general its not set either along with DRM_MODE_ATOMIC_TEST_ONLY.
Considering its importantance, this patch defers the allow_modeset check
in dm_update_planes_state(), so that there shall be scope to validate
the configuration sent from user space, without impacting the population
of dc/dm related data structures.
Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
it is required if a platform supports PCIe root complex
core voltage reduction. After receiving this notification,
SBIOS can apply default PCIe root complex power policy.
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>
Cc: stable@vger.kernel.org
This allows access to pages allocated through the driver with optional
IOMMU mapping.
v2: Fix number of bytes copied and add write method
v3: drop check for kmap return
Original-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
In dce110, the plane configuration is such that plane 0
or the primary plane should be rendered with only RGB data.
This patch adds the validation to ensure that no video data
is rendered on plane 0.
Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
