This patch implements support for VRAM page table entry compression.
PTE construction is enhanced to identify physically contiguous page
ranges and mark them in the PTE fragment field. L1/L2 TLB support is
enabled for 64KB (SI/CIK) and 256KB (NI) PTE fragments, significantly
improving TLB utilization for VRAM allocations.
Linear store bandwidth is improved from 60GB/s to 125GB/s on Pitcairn.
Unigine Heaven 3.0 sees an average improvement from 24.7 to 27.7 FPS
on default settings at 1920x1200 resolution with vsync disabled.
See main comment in radeon_vm.c for a technical description.
v2 (chk): rebased and simplified.
v3 (chk): add missing hw setup
v4 (chk): rebased on current drm-fixes-3.15
v5 (chk): fix comments and commit text
Signed-off-by: Jay Cornwall <jay@jcornwall.me>
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Also add golden registers, update firmware loading functions.
Signed-off-by: Samuel Li <samuel.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
It would appear this bug has been copy/pasted many times without being noticed.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Testing the update pending bit directly after issuing an
update is nonsense cause depending on the pixel clock the
CRTC needs a bit of time to execute the flip even when we
are in the VBLANK period.
This is just a non invasive patch to solve the problem at
hand, a more complete and cleaner solution should follow
in the next merge window.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=76564
v2: fix source IDs for CRTC2-6
Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org
Avoid a possible segfault.
Noticed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Need to swap on BE.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org
This fixes fast color clear with 1D-tiled single-sample surfaces
and Hyper-Z corruption with 1D-tiled depth surfaces.
Even though it seems it is not needed for 1D tiling, CMASK and HTILE are
always 2D-tiled, thus the hw needs to know the actual pipe configuration
for CMASK and HTILE addressing no matter what the tiling mode of the surface
is.
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iQEcBAABAgAGBQJTJlUvAAoJEHm+PkMAQRiGOhYH/1I+Bc7N7Rjr6QQAtBIy0GPC
XMqSE/gpgxlvRneQbQsvTUlPnWRhgzLGendT9HFKawkaQ0UNuZdRVyBHGFmpuED8
RlbicVVuuEZabrxEnCd7UPvYvEyK5pLIFpCRs5B+ManB1qLki2Ar03ymH1NRxOde
edmPbSUFo2aONITrEBm7tqT3cShTmBaDGP/zU0TNDMNrpVVDbHZolSNu2z4xOTa5
GqAOEbluLQ6jP3yxWur/V3Lk3W7pB6TabfX4o6UZu0F3iFnJxRMIzHXrI3o4yLTj
HEwmB3npfc8DIUk4oik7RkN+aqxDcdg/rBLQD63+xxt6zCkP+0q16brC0R67qWE=
=n9Ml
-----END PGP SIGNATURE-----
Merge tag 'v3.14-rc7' into drm-next
Linux 3.14-rc7
Backmerge to help out Intel guys.
When we disable the rings, set the status properly. If
not other code pathes may try and use the rings which are
not functional at this point.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Moving the pm resume up in the init order to fix
dpm seems to have regressed somes cases with the old
pm code. Move it back to late resume.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The reason for the false positives was fixed quite some time ago and since
most engines can still execute NOPs while being locked up it leads to false
negatives.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
In all cases where it really matters we are using the read functions anyway.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Some of the vce clocks are automatic, others need to
be manually enabled. For ease, just disable cg when
vce is active.
v2: rebased
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Only VCE 2.0 support so far.
v2: squashing multiple patches into this one
v3: add IRQ support for CIK, major cleanups,
basic code documentation
v4: remove HAINAN from chipset list
Signed-off-by: Christian König <christian.koenig@amd.com>
If we are not able to properly initialize one of the gpu
engines for buffer paging, we limit vram to the size of
the cpu visible aperture. We generally either use the gfx
or dma engine to do this. Clean up the size limiting code
to only adjust the size based on what ring is selected
for buffer paging rather than making assumptions about which
engine is selected for paging.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
This is the preferred flushing method on CIK.
Note, this only works on the PFP so the engine bit must be
set.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm-intel-next-2014-01-10:
- final bits for runtime D3 on Haswell from Paul (now enabled fully)
- parse the backlight modulation freq information in the VBT from Jani
(but not yet used)
- more watermark improvements from Ville for ilk-ivb and bdw
- bugfixes for fastboot from Jesse
- watermark fix for i830M (but not yet everything)
- vlv vga hotplug w/a (Imre)
- piles of other small improvements, cleanups and fixes all over
Note that the pull request includes a backmerge of the last drm-fixes
pulled into Linus' tree - things where getting a bit too messy. So the
shortlog also contains a bunch of patches from Linus tree. Please yell if
you want me to frob it for you a bit.
* 'drm-intel-next' of git://people.freedesktop.org/~danvet/drm-intel: (609 commits)
drm/i915/bdw: make sure south port interrupts are enabled properly v2
drm/i915: Include more information in disabled hotplug interrupt warning
drm/i915: Only complain about a rogue hotplug IRQ after disabling
drm/i915: Only WARN about a stuck hotplug irq ONCE
drm/i915: s/hotplugt_status_gen4/hotplug_status_g4x/
To support HSA on KV, we need to limit the number of vmids and pipes
that are available for radeon's use with KV.
This patch reserves VMIDs 8-15 for amdkfd (so radeon can only use VMIDs
0-7) and also makes radeon thinks that KV has only a single MEC with a single
pipe in it
v3: Use define for static vmid allocation in radeon
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@amd.com>
pci config reset is a low level reset that resets
the entire chip from the bus interface. It can
be more reliable if soft reset fails.
v2: fix rebase
v3: hide behind module parameter
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fill in asic family specific versions rather than
using the generic version. This lets us handle asic
specific differences more easily. In this case, we
disable sw swapping of the rtpr writeback value on
r6xx+ since the hw does it for us. Fixes bogus
rptr readback on BE systems.
v2: remove missed cpu_to_le32(), add comments
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We need more control over the ordering of dpm init with
respect to the rest of the asic. Specifically, the SMC
has to be initialized before the rlc and cg/pg. The pm
code currently initializes late in the driver, but we need
it to happen much earlier so move pm handling into the asic
specific callbacks.
This makes dpm more reliable and makes clockgating work
properly on CIK parts and should help on SI parts as well.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We need to reorder the driver init sequence to better accomodate
dpm which needs to be loaded earlier in the init sequence. Move
fw init up so that it's available for dpm init.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This will allow userspace to correctly program the PA_SC_RASTER_CONFIG
register, so it can be considered a fix.
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Only the render backends of the first shader engine were enabled. The others
were erroneously disabled. Enabling the other render backends improves
performance a lot.
Unigine Sanctuary on Bonaire:
Before: 15 fps
After: 90 fps
Judging from the fan noise, the GPU was also underclocked when the other
render backends were disabled, resulting in horrible performance. The fan is
a lot noisy under load now.
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
This is required to properly calculate the tiling parameters
in userspace.
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
A single doorbell page is plenty for cik kms compute.
Use a single page and manage doorbell allocation by
individual doorbells rather than pages. Identify
doorbells by their index rather than byte offset.
Signed-off-by: Andrew Lewycky <Andrew.Lewycky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
To workaround bugs and/or certain limits it's sometimes
useful to fall back to waiting on fences.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
This just updates the firmware loading functions
to look for the appropriate firmware files for
hawaii.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The formula needs to be adjusted since there are 4 RBs
per SH rather than 2 as on previous asics.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The golden register settings are optimal settings for
certain registers from the hardware team.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Bit a bit -fixes pull request in the merge window than usual dua to two
feauture-y things:
- Display CRCs are now enabled on all platforms, including the odd DP case
on gm45/vlv. Since this is a testing-only feature it should ever hurt,
but I figured it'll help with regression-testing -fixes. So I left it
in and didn't postpone it to 3.14.
- Display power well refactoring from Imre. Would have caused major pain
conflict with the bdw stage 1 patches if I'd postpone this to -next.
It's only an relatively small interface rework, so shouldn't cause pain.
It's also been in my tree since almost 3 weeks already.
That accounts for about two thirds of the pull, otherwise just bugfixes:
- vlv backlight fix from Jesse/Jani
- vlv vblank timestamp fix from Jesse
- improved edp detection through vbt from Ville (fixes a vlv issue)
- eDP vdd fix from Paulo
- fixes for dvo lvds on i830M
- a few smaller things all over
Note: This contains a backmerge of v3.12. Since the -internal branch
always applied on top of -nightly I need that unified base to merge bdw
patches. So you'll get a conflict with radeon connector props when pulling
this (and nouveau/master will also conflict a bit when Ben doesn't
rebase). The backmerge itself only had conflicts in drm/i915.
There's also a tiny conflict between Jani's backlight fix and your sysfs
lifetime fix in drm-next.
* tag 'drm-intel-fixes-2013-11-07' of git://people.freedesktop.org/~danvet/drm-intel: (940 commits)
drm/i915/vlv: use per-pipe backlight controls v2
drm/i915: make backlight functions take a connector
drm/i915: move opregion asle request handling to a work queue
drm/i915/vlv: use PIPE_START_VBLANK interrupts on VLV
drm/i915: Make intel_dp_is_edp() less specific
drm/i915: Give names to the VBT child device type bits
drm/i915/vlv: enable HDA display audio for Valleyview2
drm/i915/dvo: call ->mode_set callback only when the port is running
drm/i915: avoid unclaimed registers when capturing the error state
drm/i915: Enable DP port CRC for the "auto" source on g4x/vlv
drm/i915: scramble reset support for DP port CRC on vlv
drm/i915: scramble reset support for DP port CRC on g4x
drm/i916: add "auto" pipe CRC source
...
Conflicts:
MAINTAINERS
drivers/gpu/drm/i915/intel_panel.c
drivers/gpu/drm/nouveau/core/subdev/mc/base.c
drivers/gpu/drm/radeon/atombios_encoders.c
drivers/gpu/drm/radeon/radeon_connectors.c
The DMA ring seems to be stable now.
v2: remove pt_ring_index as well
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
It is said to cause hangs.
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The buffers needs to be in little endian format.
Noticed-by: Sylvain BERTRAND <sylware@legeek.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Allows you to enable dither in the display hardware
when the monitor supports lower a lower bpc than the
current framebuffer format.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The FMT blocks control how data is sent from the backend
of the display pipe to to monitor. Proper set up of the
FMT blocks are required for 30bpp formats. Additionally,
dithering can be enabled on for better display with 18 and
24bpp displays. The exception is LVDS/eDP which atom
takes care of in the SelectCRTC_Source table. For now
just enable truncation until we test dithering more.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stop leaking IB memory and scratch register space when the test fails.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Missing ULL when calculating the amount of vram
leads to an overflow when the amount of vram is >= 4G.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
There are multiple valid values, not just 0 or 1. Required
to properly support 2D tiling in the userspace drivers.
Cc: stable@vger.kernel.org
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The string is encoded from the MSB to the LSB of the register.
Cc: stable@vger.kernel.org
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This adds spinlocks to protect access to other
indirect register apertures. These indirect spaces are
used pretty infrequently and we haven't had an reported
problems, but better safe than sorry.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The internal cp interrupts need to be enabled and
disabled at specific times in order clockgating to
work properly. This patch changes the handling
of the CP_INT_CNTL register to respect the current
state of the internal CP interrupts when making
changes to the other interrupts in CP_INT_CNTL.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Alex writes:
This is the radeon drm-next request. Big changes include:
- support for dpm on CIK parts
- support for ASPM on CIK parts
- support for berlin GPUs
- major ring handling cleanup
- remove the old 3D blit code for bo moves in favor of CP DMA or sDMA
- lots of bug fixes
[airlied: fix up a bunch of conflicts from drm_order removal]
* 'drm-next-3.12' of git://people.freedesktop.org/~agd5f/linux: (898 commits)
drm/radeon/dpm: make sure dc performance level limits are valid (CI)
drm/radeon/dpm: make sure dc performance level limits are valid (BTC-SI) (v2)
drm/radeon: gcc fixes for extended dpm tables
drm/radeon: gcc fixes for kb/kv dpm
drm/radeon: gcc fixes for ci dpm
drm/radeon: gcc fixes for si dpm
drm/radeon: gcc fixes for ni dpm
drm/radeon: gcc fixes for trinity dpm
drm/radeon: gcc fixes for sumo dpm
drm/radeonn: gcc fixes for rv7xx/eg/btc dpm
drm/radeon: gcc fixes for rv6xx dpm
drm/radeon: gcc fixes for radeon_atombios.c
drm/radeon: enable UVD interrupts on CIK
drm/radeon: fix init ordering for r600+
drm/radeon/dpm: only need to reprogram uvd if uvd pg is enabled
drm/radeon: check the return value of uvd_v1_0_start in uvd_v1_0_init
drm/radeon: split out radeon_uvd_resume from uvd_v4_2_resume
radeon kms: fix uninitialised hotplug work usage in r100_irq_process()
drm/radeon/audio: set up the sads on DCE3.2 asics
drm/radeon: fix handling of variable sized arrays for router objects
...
Conflicts:
drivers/gpu/drm/i915/i915_dma.c
drivers/gpu/drm/i915/i915_gem_dmabuf.c
drivers/gpu/drm/i915/intel_pm.c
drivers/gpu/drm/radeon/cik.c
drivers/gpu/drm/radeon/ni.c
drivers/gpu/drm/radeon/r600.c
The same as on evergreen.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reported-by: FrankR Huang <FrankR.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
The vram scratch buffer needs to be initialized
before the mc is programmed otherwise we program
0 as the GPU address of the default GPU fault
page. In most cases we put vram at zero anyway and
reserve a page for the legacy vga buffer so in practice
this shouldn't cause any problems, but better to make
it correct.
Was changed in:
6fab3febf6
Reported-by: FrankR Huang <FrankR.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
For powergating, we just need to re-init the registers, there
is no need to restore the uvd BOs. This just adds needless
work when powergating uvd for playback while the system is
on. We only need to restore the uvd BOs on an actual resume
from suspend or when the driver loads.
This fixes multi-stream UVD playback on KB systems.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
This fills in the GPU specific details for berlin
GPU cores so that the driver will work with them.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
We need to allocate line buffer to each display when
setting up the watermarks. Failure to do so can lead
to a blank screen. This fixes blank screen problems
on dce8 asics.
Based on an initial fix from:
Jay Cornwall <jay.cornwall@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
The format of the clearstate buffer used for pg (powergating)
changed between NI and SI. This formats it properly for what
the hardware expects on SI+.
v2: fix addresses
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Clockgating requires signalling between the CP and the
RLC to work properly. Resetting the CP block in the
CP resume code messed up the internal coordination
between the blocks. Removing the reset allows gfx
clockgating to work properly. However, when gfx clock
gating is enabled, there is a strange interaction with
dpm which causes the chip to stay in the high performance
level all the time, so leave gfx clockgating disabled
for now.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
- use new cg/pg flags for finer grained clock and
powergating control
- restructure the cg/pg code so it can be called from
other components such as dpm
v2: fix build breakage from rebase
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Similar to DCE4/5, but supports multiple audio pins
which can be assigned per afmt block.
v2: rework the driver to handle more than one audio
pin.
v3: try different dto reg
v4: properly program dto
v5 (ck): change dto programming order
v6: program speaker allocation block
v7: rebase
v8: rebase on Rafał's changes
v9: integrated Rafał's comments, update to latest
drm_edid_to_speaker_allocation API
v10: add missing line break in error message
v11: add back audio enabled messages
v12: fix copy paste typo in r600_audio_enable
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Rafał Miłecki <zajec5@gmail.com>
Similar to separating the UVD code, just put the DMA
functions into separate files.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Our different hardware blocks are actually completely
separated, so it doesn't make much sense any more to
structure the code by pure chipset generations.
Start restructuring the code by separating our the UVD block.
v2: updated commit message
v3: rebased and restructurized start/stop functions for kv dpm.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Now that we have callbacks for [rw]ptr handling we can
remove the special handling for the DMA rings and use
the callbacks instead.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The hardware just doesn't support this correctly.
Disable it before we accidentally write anywhere we shouldn't.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
When we PG (powergate) UVD, we need to re-initialize it
before we can use it again.
v2: rebase on UVD stop fixes
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This adds dpm support for btc asics. This includes:
- dynamic engine clock scaling
- dynamic memory clock scaling
- dynamic voltage scaling
- dynamic pcie gen switching
Set radeon.dpm=1 to enable.
v2: remove unused radeon_atombios.c changes,
make missing smc ucode non-fatal
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This adds dpm support for KB/KV asics. This includes:
- dynamic engine clock scaling
- dynamic voltage scaling
- power containment
- shader power scaling
Set radeon.dpm=1 to enable.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Calculate the low and high watermarks based on the low and high
clocks for the current power state. The dynamic pm hw will select
the appropriate watermark based on the internal dpm state.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Otherwise just reinitialize from scratch on resume,
and so make it more likely to succeed.
Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
For r6xx+ asics. This mirrors the behavior of pre-r6xx
asics. We need to program the MC even if something
else in startup() fails. Failure to do so results in
an unusable GPU.
Based on a fix from: Mark Kettenis <kettenis@openbsd.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Removing the clock/power or resetting the VCPU can cause
hangs if that happens in the middle of a register write.
Stall the memory and register bus before putting the VCPU
into reset. Keep it in reset when unloading the module or
suspending.
Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Last driver and pretty obviously a major user of this little function.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@gmail.com>
Changing the UVD BOs offset on suspend/resume doesn't work because the VCPU
internally keeps pointers to it. Just keep it always pinned and save the
content manually.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=66425
v2: fix compiler warning
v3: fix CIK support
Note: a version of this patch needs to go to stable.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Helpful for debugging GPUVM errors as we can see what
hw block and page generated the fault in the log.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Avoid creating temporary platform device that will lead to issue
when several radeon gpu are in same computer. Instead directly use
the radeon device for requesting firmware.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The compute rings use RELEASE_MEM rather then EOP
packets for writing fences and there is no SYNC_PFP_ME
packet on the compute rings.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Type 2 packets are deprecated on CIK MEC and we should use
type 3 nop packets. Setting the count field to the max value
(0x3fff) indicates that only one dword should be skipped
like a type 2 packet.
v2: add comment to code
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
On CIK, the compute rings work slightly differently than
on previous asics, however the basic concepts are the same.
The main differences:
- New MEC engines for compute queues
- Multiple queues per MEC:
- CI/KB: 1 MEC, 4 pipes per MEC, 8 queues per pipe = 32 queues
- KV: 2 MEC, 4 pipes per MEC, 8 queues per pipe = 64 queues
- Queues can be allocated and scheduled by another queue
- New doorbell aperture allows you to assign space in the aperture
for the wptr which allows for userspace access to queues
v2: add wptr shadow, fix eop setup
v3: fix comment
v4: switch to new callback method
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
The doorbell aperture is a PCI BAR whose pages can be
mapped to compute resources for things like wptrs
for userspace queues.
This patch maps the BAR and sets up a simple allocator
to allocate pages from the BAR.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Allows us to select instanced registers based on:
- ME (micro engine
- Pipe
- Queue
- VMID
Switch MC setup to use this new function.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
v2: agd5f: fix clock dividers setup for bonaire
v3: agd5f: rebase
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
v2: update to latest driver changes
v3: properly tear down vm on suspend
v4: fix up irq init ordering
v5: remove outdated comment
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Async page table updates using the sDMA engine. sDMA has a
special packet for updating entries for contiguous pages
that reduces overhead.
v2: add support for and use the CP for now.
v3: update for 2 level PTs
v4: rebase, fix DMA packet
v5: switch to using an IB
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Update the page table base address and flush the
VM TLB using the sDMA.
V2: update for 2 level PTs
V3: update vm flush
V4: update SH_MEM* regs
V5: switch back to old style VM TLB invalidate
V6: fix packet formatting
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
CIK has new asynchronous DMA engines called sDMA
(system DMA). Each engine supports 1 ring buffer
for kernel and gfx and 2 userspace queues for compute.
TODO: fill in the compute setup.
v2: update to the latest reset code
v3: remove ib_parse
v4: fix copy_dma()
v5: drop WIP compute sDMA queues
v6: rebase
v7: endian fixes for IB
v8: cleanup for release
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
RLC handles the interrupt controller and other tasks
on the GPU.
v2: add documentation
v3: update programming sequence
v4: additional setup
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Update the page table base address and flush the
VM TLB using the CP.
v2: update for 2 level PTs
v3: use new packet for invalidate
v4: update SH_MEM* regs when flushing the VM
v5: add pfp sync, go back to old style vm TLB invalidate
v6: fix hdp flush packet count
v7: use old style HDP flush
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
For gfx ring only. Compute is still todo.
v2: add documentation
v3: update to latest reset changes, integrate emit update patch.
v4: fix count on wait_reg_mem for HDP flush
v5: use old hdp flush method for fence
v6: set valid bit for IB
v7: cleanup for release
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Sets up the GFX ring and loads ucode for GFX and Compute.
Todo:
- handle compute queue setup.
v2: add documentation
v3: integrate with latest reset changes
v4: additional init fixes
v5: scratch reg write back no longer supported on CIK
v6: properly set CP_RB0_BASE_HI
v7: rebase
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Currently the driver required 6 sets of ucode:
1. pfp - pre-fetch parser, part of the GFX CP
2. me - micro engine, part of the GFX CP
3. ce - constant engine, part of the GFX CP
4. rlc - interrupt, etc. controller
5. mc - memory controller (discrete cards only)
6. mec - compute engines, part of Compute CP
V2: add documentation
V3: update MC ucode
V4: rebase
V5: update mc ucode
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Redirect invalid memory accesses to the default page
instead of locking up the memory controller.
v2: rebase on top of 2 level PTs
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The vm callbacks are the same as the SI ones right now
(same regs and bits). We could share the SI variants, and
I may yet do that, but I figured I would add CIK specific
ones for now in case we need to change anything.
V2: add documentation, minor fixes.
V3: integrate vram offset fixes for APUs
V4: enable 2 level VM PTs
V5: index SH_MEM_* regs properly
V6: add ib_parse()
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
v2: split soft reset into compute and gfx. Still need
to make reset more fine grained, but this should be a
start.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
v2: tiling fixes
v3: more tiling fixes
v4: more tiling fixes
v5: additional register init
v6: rebase
v7: fix gb_addr_config for KV/KB
v8: drop wip KV bits for now, add missing config reg
v9: fix cu count on Bonaire
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
