drm/nouveau/kms/nv50-: initial overlay support

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/gk104-: add support for [XA]2R10G10B10 formats

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/gk104-: support additional cursor sizes

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: separate blocklinear vs linear pitch

Will be required to support Volta.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: handle degamma LUT from window channels

Required to eventually support DRM colour management APIs, and to
support Volta.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: plane updates don't always require image_set()

When only the position of a window changes, there's no need to submit
an image update as well.

Will be required to support the overlays, and Volta windows.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: store window visibility in state

Window visibility is going to become a little more complicated with the
upcoming LUT changes, so store the calculated value to avoid needing to
recalculate the armed state again.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: simplify swap interval handling

This is just cleaning up some left-overs from when we needed a custom
legacy page flip implementation.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: decouple window state changes, and update method submisssion

This will be required to support Volta.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: simplify tracking of channel interlocks

Instead of windows returning their core channel interlock mask if they
know core has been modified, it's recorded unconditionally and used if
required when update methods are emitted.

This will be required to support Volta.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: determine MST support from DP Info Table

GV100 doesn't support MST, use the information provided in VBIOS tables to
detect its presence instead.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: extend window image data for stereo/planar formats

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: move drm format->hw conversion into common code

This will be required to support additional HW features.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: unify set/clr masks

This is a simplification that'll be used to improve interlock handling.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: allow specification of valid heads for a window

This will be required to support Volta, where window ID != head.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: split base implementation by hardware class

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: split core implementation by hardware class

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: split each resource type into their own source files

There should be no code changes here, just shuffling stuff around.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50: abstract OR interfaces so the code can be split

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50: handle SetControlOutputResource from head

Removes duplicated code from OR-specific functions.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: abstract head interfaces so the code can be split

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50: modify core allocation so the code can be split

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: modify base allocation so the code can be split

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: modify cursor allocation so the code can be split

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: modify overlay allocation so the code can be split

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: move fb ctxdma tracking into windows

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: fix i2c-over-aux on anx9805

We don't support address-only transactions there.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: move code underneath dispnv50/

The code is about to be split up, and this matches dispnv04.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms: move display class instantiation to library

This function is useful outside of DRM code.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/drm/nv50-: remove allocation of sw class

Hasn't been required for a long time.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: no need to create ctxdma for push buffers on fermi and up

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: remove fence wait code from deferred client work handler

Fences attached to deferred client work items now originate from channels
belonging to the client, meaning we can be certain they've been signalled
before we destroy a client.

This closes a race that could happen if the dma_fence_wait_timeout() call
didn't succeed.  When the fence was later signalled, a use-after-free was
possible.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gem: tie deferred unmapping of buffers to VMA fence completion

As VMAs are per-client, unlike buffers, this allows us to avoid referencing
foreign fences (those that belong to another client/driver) from the client
deferred work handler, and prevent some not-fun race conditions that can be
triggered when a fence stalls.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gem: attach fences to VMAs to track GPU usage

An upcoming patch will use these to fix issues related to the deferred
unmapping of GEM objects.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gem: lookup VMAs for buffers referenced by pushbuf ioctl

We previously only did this for push buffers, but an upcoming patch will
need to attach fences to all VMAs to resolve another issue.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gp102-: setup stencil zbc

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gp100-: use correct registers for zbc colour/depth setup

These were missed the first time around due to the driver version I traced
using the older registers still.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gp100-: fix attrib cb setup

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gp100-: fix pagepool setup

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-gm10x: update register lists

There are differences on GM200 and newer too, but we can't fix them there
as they come from firmware packages.

A request has been made to NVIDIA to release updated firmware.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: swap bundle and pagepool

Makes it easier to diff against RM traces.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: calculate and use sm mapping table

There's a number of places that require this data, so let's separate out
the calculations to ensure they remain consistent.

This is incorrect for GM200 and newer, but will produce the same results
as we did before.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: port zcull tile mapping calculations from NVGPU

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: port tile mapping calculations from NVGPU

There's also a couple of hardcoded tables for a couple of very specific
configurations that NVGPU's algorithm didn't work for.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise trap_mp

Required to support Volta.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: add missing reset sequence before golden context init

RM and NVGPU both have a variant of this, we probably should too.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: delete duplicated grctx init code

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: update r408840 where required

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: update 419a3c where required

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise r418e94

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise r419e00

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: update 419eb0 where required

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: note missing 418800 modifications

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-gf119: update 419cb8 where required

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: support firmware-provided bundle/method everywhere

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise tpc_mask + apply fixes from traces

We weren't placing higher TPC IDs in the right place on some configurations.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise r419f78 + apply fixes from traces

Removed from GK110[B]/GK208 as RM traces show it not being touched.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise gpc_tpc_nr

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise r406500

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise dist_skip_table + improve algorithm

The algorithm for GM200 and newer matches RM for all the boards I have, but
I don't have enough data to try and figure something out for earlier boards,
so these will still write zeroes to the table as we did before.

The code in NVGPU isn't helpful here, it appears to handle specific cases.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-gf119: modify max_ways_evict where required

I don't think this is done after Fermi, NVGPU used to do it but removed
the code, and I've not seen RM traces touching it either.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise alpha_beta_tables + improve algorithms

I haven't yet been able to find a fully programatic way of calculating the
same mapping as NVIDIA for GF100-GF119, so the algorithm partially depends
on data tables for specific configurations.

I couldn't find traces for every possibility, so the algorithm will switch
to a mapping similar to what GK104-GM10x use if it encounters one.  We did
the wrong thing before anyway, so shouldn't matter too much.

The algorithm used in the GK104 implementation was ported from NVGPU.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise rop_mapping

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise r4060a8 + apply fixes from traces

Also fixes some GPUs where we write too many registers.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise tpc_per_gpc

GM20B now also shares the same code, as NVGPU shows it doesn't need
special treatment.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise sm_id/tpc_nr

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise patch_ltc, noting missing init

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: support firmware-provided sw_ctx everywhere

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: delete duplicated init code

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_400054

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: apply be exception fixes from traces

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_shader_exceptions

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_504430

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_tex_hww_esr

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_ppc_exceptions

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_419c9c + apply fixes from traces

Deliberately removed from non-GP100, as RM doesn't touch it.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_419eb4 + apply fixes from traces

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_419cc0 + apply fixes from traces

Pulled some init out of main per-GPC/TPC loops to match RM.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_sked_hww_esr

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_40601c

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_ds_hww_esr_2

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_fecs_exceptions + apply fixes from traces

The value for GF100 has changed here, but it matches RM now.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100: write 0x400124 during init

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_swdx_pes_mask

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: implement another chunk of bios-provided init

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_rop_active_fbps

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_num_active_ltcs

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_zcull

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_vsc_stream_master

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_bios

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: support clkgate_pack everywhere

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise r405a14

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: support firmware-provided sw_nonctx everywhere

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: virtualise init_gpc_mmu + apply fixes from traces

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gp100-: force individual channels into a channel group

RM does this for some reason, and is enforced in HW on Volta.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gm107-: write instance address in channel runlist entry

RM does this for some reason.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk208-: write pbdma timeout regs during initialisation

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk110-: support writing channel group runlist entries

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: poll for runlist update completion

Newer HW doesn't appear to send this event, which will cause long delays
in runlist updates if they don't complete immediately.

RM doesn't use these events anywhere, and an NVGPU commit message notes
that polling is the preferred method even on HW that supports the event.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: add interfaces to support different runlist layouts

This will be required to support features on newer hardware.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>