drm/i915: Disallow plane x+w>stride on ilk+ with X-tiling

ilk+ planes get notably unhappy when the plane x+w exceeds
the stride. This wasn't a problem previously because we
always aligned SURF to the closest tile boundary so the
x offset never got particularly large. But now with async
flips we have to align to 256KiB instead and thus this
becomes a real issue.

On ilk/snb/ivb it looks like the accesses just wrap
early to the next tile row when scanout goes past the
SURF+n*stride boundary, hsw/bdw suffer more heavily and
start to underrun constantly. i965/g4x appear to be immune.
vlv/chv I've not yet checked.

Let's borrow another trick from the skl+ code and search
backwards for a better SURF offset in the hopes of getting the
x offset below the limit. IIRC when I ran into a similar issue
on skl years ago it was causing the hardware to fall over
pretty hard as well.

And let's be consistent and include i965/g4x in the check
as well, just in case I just got super lucky somehow when
I wasn't able to reproduce the issue. Not that it really
matters since we still use 4k SURF alignment for i965/g4x
anyway.

Fixes: 6ede6b0616b2 ("drm/i915: Implement async flips for vlv/chv")
Fixes: 4bb18054adc4 ("drm/i915: Implement async flip for ilk/snb")
Fixes: 2a636e240c77 ("drm/i915: Implement async flip for ivb/hsw")
Fixes: cda195f13abd ("drm/i915: Implement async flips for bdw")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210209021918.16234-1-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
(cherry picked from commit 59fb8218c8e5001f854e7d5fdb5fb135cba58102)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
[Rodrigo also exported some functions from intel_display.c during backport]

drm/i915: Reject 446-480MHz HDMI clock on GLK

The BXT/GLK DPLL can't generate certain frequencies. We already
reject the 233-240MHz range on both. But on GLK the DPLL max
frequency was bumped from 300MHz to 594MHz, so now we get to
also worry about the 446-480MHz range (double the original
problem range). Reject any frequency within the higher
problematic range as well.

Cc: stable@vger.kernel.org
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/3000
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210203093044.30532-1-ville.syrjala@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
(cherry picked from commit 41751b3e5c1ac656a86f8d45a8891115281b729e)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915/gt: Flush before changing register state

Flush; invalidate; change registers; invalidate; flush.

Will this finally work on every device? Or will Baytrail complain again?

On the positive side, we immediately see the benefit of having hsw-gt1 in
CI.

Fixes: ace44e13e577 ("drm/i915/gt: Clear CACHE_MODE prior to clearing residuals")
Testcase: igt/gem_render_tiled_blits # hsw-gt1
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Akeem G Abodunrin <akeem.g.abodunrin@intel.com>
Acked-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210125220247.31701-1-chris@chris-wilson.co.uk
(cherry picked from commit d30bbd62b1bfd9e0a33c3583c5a9e5d66f60cbd7)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915: Disable atomics in L3 for gen9

Enabling atomic operations in L3 leads to unrecoverable GPU hangs, as
the machine stops responding milliseconds after receipt of the reset
request [GDRT]. By disabling the cached atomics, the hang do not occur
and we presume the GPU would reset normally for similar hangs.

Sadly this is a shotgun approach, but since the impact is critical it is
better to err on the safe side and work back from there.

Reported-by: Jason Ekstrand <jason@jlekstrand.net>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=110998
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Jason Ekstrand <jason@jlekstrand.net>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Jason Ekstrand <jason@jlesktrand.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20210125220152.24070-1-chris@chris-wilson.co.uk
Cc: stable@vger.kernel.org
(cherry picked from commit b267c7ae0ad5b437b068f46919b17f85000154b4)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915/gem: Move freeze/freeze_late next to suspend/suspend_late

Push the hibernate pm routines next to the suspend pm routines in
gem/i915_gem_pm.c. This has the side-effect of putting the wbinvd()
abusers next to each other.

Reported-by: Guenter Roeck <linux@roeck-us.net>
Fixes: 30d2bfd09383 ("drm/i915/gem: Almagamate clflushes on freeze")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210123145543.10533-1-chris@chris-wilson.co.uk
(cherry picked from commit 6d8f02207420e76db693a00ccb44792474e297fc)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915/gem: Fix oops in error handling code

This code will Oops when it tries to i915_gem_object_free(obj) because
"obj" is an error pointer.

Fixes: 97d553963250 ("drm/i915/region: convert object_create into object_init")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/YA6FkPn5S4ZDUGxq@mwanda
(cherry picked from commit ad8db423a30f0ac39a5483dfd726058135ff2bd2)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915/gvt: fix uninitialized return in intel_gvt_update_reg_whitelist()

Smatch found an uninitialized variable bug in this code:

    drivers/gpu/drm/i915/gvt/cmd_parser.c:3191 intel_gvt_update_reg_whitelist()
    error: uninitialized symbol 'ret'.

The first thing that Smatch complains about is that "ret" isn't set if
we don't enter the "for_each_engine(engine, &dev_priv->gt, id) {" loop.
Presumably we always have at least one engine so that's a false
positive.

But it's definitely a bug to not set "ret" if i915_gem_object_pin_map()
fails.

Let's fix the bug and silence the false positive.

Fixes: 493f30cd086e ("drm/i915/gvt: parse init context to update cmd accessible reg whitelist")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/YA6F3oF8mRaNQWjb@mwanda
(cherry picked from commit 784f70e17e6bc423a04fb6524634a76f68ab1192)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/i915: Restrict DRM_I915_DEBUG to developer builds

Let's not encourage everybody to build i915's debug code, and certainly
not the build robots who need to scrutinise the production build. Since
CI will complain if the debug build is broken, having the other build
bots focus on the builds we don't cover ourselves should improve the
build coverage.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Fixes: 4f86975f539d ("drm/i915: Add DEBUG_GEM to the recommended CI config")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210122091058.5145-1-chris@chris-wilson.co.uk
(cherry picked from commit c442f658299d59b327a4bf21457ec8ece936f133)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

drm/amd/display: fix unused variable warning

After all users of the 'dm' warnings got hidden in an #ifdef,
the compiler started warning about it being unused:

drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm.c:5380:33: error: unused variable 'dm' [-Werror,-Wunused-variable]

Add another such #ifdef.

Fixes: 98ab5f3513f9 ("drm/amd/display: Fix deadlock during gpu reset v3")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210125124849.102037-1-arnd@kernel.org

Merge tag 'amd-drm-next-5.12-2021-02-03' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-5.12-2021-02-03:

amdgpu:
- Display fixes and cleanups
- Vangogh fixes
- Fix possible race when there are timeouts on two rings
- SR-IOV fixes
- Add missing license
- DCE 10/12 bpc fixes
- Display MALL fixes
- Fix SMU user preference settings persistence
- Fix retry in gem allocate
- Add new PCI DID
- Fix for manual fan speed control on cards where it was problematic
- Fix regression in pinning GTT
- Misc display fixes
- Misc code cleanups

amdkfd:
- Fix config handling
- Fix regression in buffer free

From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210204045717.3823-1-alexander.deucher@amd.com
Signed-off-by: Dave Airlie <airlied@redhat.com>

Merge tag 'mediatek-drm-next-5.12' of https://git./linux/kernel/git/chunkuang.hu/linux into drm-next

Mediatek DRM Next for Linux 5.12

1. Decouple Mediatek DRM sub driver
2. Share mtk mutex driver for both DRM and MDP
3. Add support for SoC MT8183

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210204151750.7550-1-chunkuang.hu@kernel.org

drm/mediatek: Add matrix_bits private data for ccorr

Add matrix_bits and coeffs_precision to ccorr private data:
- matrix bits of mt8183 is 10
- matrix bits of mt8192 is 11

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Fix ccorr size config

Fix setting to follow hardware datasheet. The original error setting
affects mt8192 display.

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Separate ccorr module

ccorr ctm matrix bits will be different in mt8192.

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Enable dither function

Enable dither function to improve the display quality.

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Enable OVL_LAYER_SMI_ID_EN for multi-layer usecase

Enable OVL_LAYER_SMI_ID_EN for multi-layer usecase, without this patch,
ovl will hang up when more than 1 layer enabled.

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Add support for SoC MT8183

1. Add ovl private data
2. Add rdma private data
3. Add gamma privte data
4. Add main and external path module for crtc create

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Add has_dither private data for gamma

Not all SoC has dither function in gamma module.
Add private data to control this function setting.

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Separate gamma module

mt8183 gamma module will different with mt8173,
so separate gamma for adding private data.

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Add mtk_dither_set_common() function

Current implementation of mtk_dither_set() cast dev data to
struct mtk_ddp_comp_dev. But other devices with different dev data
would also call this function.

Separate necessary parameters out so other device components (dither,
gamma) can call this function.

Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

soc / drm: mediatek: Move mtk mutex driver to soc folder

mtk mutex is used by DRM and MDP driver, and its function is SoC-specific,
so move it to soc folder.

Signed-off-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Acked-by: Matthias Brugger <matthias.bgg@gmail.com>

drm/mediatek: Automatically search unclaimed mtk mutex in mtk_mutex_get()

Moving mutex resource management from client driver to  mutex driver
could prevent client drivers negotiating for resource management.

Signed-off-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Change disp/ddp term to mutex in mtk mutex driver

mtk mutex is used by both drm and mdp driver, so change disp/ddp term to
mutex to show that it's a common driver for drm and mdp.

Signed-off-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Rename file mtk_drm_ddp to mtk_mutex

After mmsys routing function is moved out of mtk_drm_ddp.c, mtk_drm_ddp.c
has only mtk mutex function, so rename it to match the function in it.

Signed-off-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Remove redundant file including

Those file includings are useless, so remove them.

Signed-off-by: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Add fifo_size into rdma private data

Get the fifo size from device tree
because each rdma in the same SoC may have different fifo size

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Fix aal size config

The orginal setting is not correct, fix it to follow hardware data sheet.
If keep this error setting, mt8173/mt8183 display ok
but mt8192 display abnormal.

Fixes: 0664d1392c26 ("drm/mediatek: Add AAL engine basic function")

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

dt-bindings: mediatek: add description for postmask

add description for postmask
postmask is used control round corner for display frame

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: mtk_dpi: Create connector for bridges

Similar to commit a9d9fea74be2
("drm/mediatek: mtk_dsi: Create connector for bridges"):

Use the drm_bridge_connector helper to create a connector for pipelines
that use drm_bridge. This allows splitting connector operations across
multiple bridges when necessary, instead of having the last bridge in
the chain creating the connector and handling all connector operations
internally.

Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

dt-bindings: mediatek: add rdma-fifo-size description for mt8183 display

rdma fifo size may be different even in same SOC, add this
property to the corresponding rdma

Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

drm/mediatek: Check if fb is null

It's possible that state->base.fb is null. Add a check before access its
format.

Fixes: b6b1bb980ec4 ("drm/mediatek: Turn off Alpha bit when plane format has no alpha")
Signed-off-by: Yongqiang Niu <yongqiang.niu@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

Merge tag 'drm-intel-next-2021-01-29' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

- WARN if plane src coords are too big (Ville)
- Prevent double YUV range correction on HDR planes (Andres)
- DP MST related Fixes (Sean, Imre)
- More clean-up around DRAM detection code (Jose)
- Actually async flips enable for all ilk+ platforms (Ville)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210129225328.GA1041349@intel.com

Merge tag 'drm/tegra/for-5.12-rc1' of ssh://git.freedesktop.org/git/tegra/linux into drm-next

drm/tegra: Changes for v5.12-rc1

Adds support for newer firmware image versions of the Video Image
Composer (VIC) and adds a comment clarifying the use of the STREAMID
registers. Fixes a couple of issues with display and gr2d on older
Tegra SoCs such as Tegra114, as well as a runtime PM reference leak.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thierry Reding <thierry.reding@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210129193807.3653456-1-thierry.reding@gmail.com

drm/amdgpu: Prevent shift wrapping in amdgpu_read_mask()

If the user passes a "level" value which is higher than 31 then that
leads to shift wrapping.  The undefined behavior will lead to a
syzkaller stack dump.

Fixes: 5632708f4452 ("drm/amd/powerplay: add dpm force multiple levels on cz/tonga/fiji/polaris (v2)")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amd/display: remove unneeded variable: "pattern"

Remove unneeded variable: "pattern".

Signed-off-by: Bernard Zhao <bernard@vivo.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Disable GFXOFF when GFX DPM or PG disabled

Check GFX DPM and PG bit before enable GFXOFF on Vangogh
smu post init.

Signed-off-by: Jinzhou Su <Jinzhou.Su@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.121

This version brings along following fixes:
- Better handling of dummy p-state table
- Workaround for some legacy DP-VGA dongles
- Add Freesync HDMI support to DMCU
- Enable "trigger_hotplug" debugfs on all outputs
- fix initial bounding box values for dcn3.02
- implement support for DID2.0 dsc passthrough
- fix calculation for the pwl backlight curve
- Fix multiple memory leaks

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.50

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Reject too small viewport size when validating plane

[why]
Overlay won't move to a new positon if viewport size is smaller than
what can be handled. It'd either disappear or stay at the old
position. This condition is for example hit if overlay is moved too
much outside of left or top edge of the screen, but it applies to
any non-cursor plane type.

[how]
Reject this contidion at validation time. This gives the calling
level a chance to handle this gracefully and avoid inconsistent
behaivor.

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Better handling of dummy p-state table

[Why]
Some scenarios where we use a UCLK frequency in between dummy p-state table
entries result in a p-state hang, due to the table not having a close
enough match, so the default DPM0 latency is used, which can be too long to
support dummy p-state switching in these scenarios.

[How]
 - old: match if current freq is within +- margin of table entry
 - new: find largest table entry that is lower than current freq + margin
   - lower than DPM0 will still use DPM0

Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Workaround for some legacy DP-VGA dongles

[Why]
Maximum resolution is 1440*900 when connecting to FHD monitor via some DP-VGA
dongles. The display EDID reading fails over AUX/I2C via DP->VGA dongle, and
this leads to the maximum resolution 1920*1080 cannot be obtained from EDID.

[How]
Provide a workaround for some legacy DP-VGA dongles with a longer aux delay.

Signed-off-by: Taimur Hassan <syed.hassan@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Decrement refcount of dc_sink before reassignment

[why]
An old dc_sink state is causing a memory leak because it is missing a
dc_sink_release before a new dc_sink is assigned back to
aconnector->dc_sink.

[how]
Decrement the dc_sink refcount before reassigning it to a new dc_sink.

Signed-off-by: Victor Lu <victorchengchi.lu@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Free atomic state after drm_atomic_commit

[why]
drm_atomic_commit was changed so that the caller must free their
drm_atomic_state reference on successes.

[how]
Add drm_atomic_commit_put after drm_atomic_commit call in
dm_force_atomic_commit.

Signed-off-by: Victor Lu <victorchengchi.lu@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove unused force_ignore_link_settings debug option

[why]
Remove force_ignore_link_settings debug option as it is no longer used.

Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Reviewed-by: Eric Bernstein <Eric.Bernstein@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add Freesync HDMI support to DMCU

[Why]
Adding support for Freesync HDMI to DC and DMCU

[How]
Create DC interface and implementation on top of DMCU to support
parsing CEA blocks in DMCU.

Signed-off-by: Stylon Wang <stylon.wang@amd.com>
Reviewed-by: Hersen Wu <hersenxs.wu@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Enable "trigger_hotplug" debugfs on all outputs

[Why]
Per-connector debugfs entry "trigger_hotplug" is available on DP/eDP only.
New IGT tests need this entry to test other outputs.

[How]
Enable this debugfs entry on all types of connectors

Signed-off-by: Stylon Wang <stylon.wang@amd.com>
Reviewed-by: Mikita Lipski <Mikita.Lipski@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix CW4 programming for dmub30 cached inbox

[Why]
The conditions for whether we used cached vs non-cached inbox1 depend
on a version check that mismatches what the shared helpers in dmub20
implement.

[How]
Use the dmub_dcn20_use_cached_inbox check for dmub_dcn30 as well.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix initial bounding box values for dcn3.02

[Why]
Initial bounding box values are updated in dcn30_update_bw_bounding_box
 but they use dcn3_0_soc and dcn3_0_ip instead of dcn3_02_soc and
 dcn3_02_ip

[How]
Add dcn302_update_bw_bounding_box and
 dcn302_get_optimal_dcfclk_fclk_for_uclk so it uses
 dcn3_02_soc and dcn3_02_ip.
Use sr_exit_time_us, sr_enter_plus_exit_time_us,
 from dcn30 on dcn302 to fix flicker on eDP.
 Also use dram_clock_change_latency_us from dcn30.

Signed-off-by: Samson Tam <Samson.Tam@amd.com>
Reviewed-by: Joshua Aberback <Joshua.Aberback@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: implement support for DID2.0 dsc passthrough

[Why]
Some panels contain active converters (e.g. DP to MIPI) which only support
restricted DSC configurations.  DID2.0 adds support for such displays to
explicitly define per timing BPP restrictions on DSC.  Ignoring these
restrictions leads to blackscreen.

[How]
Add parsing in DID2.0 parser to get this bpp info.
Add support in DSC module to constraint target bpp based
on this info.

Signed-off-by: Jun Lei <jun.lei@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix dc_sink kref count in emulated_link_detect

[why]
prev_sink is not used anywhere else in the function and the reference to
it from dc_link is replaced with a new dc_sink.

[how]
Change dc_sink_retain(prev_sink) to dc_sink_release(prev_sink).

Signed-off-by: Victor Lu <victorchengchi.lu@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Release DSC before acquiring

[why]
Need to unassign DSC from pipes that are not using it
so other pipes can acquire it. That is needed for
asic's that have unmatching number of DSC engines from
the number of pipes.

[how]
Before acquiring dsc to stream resources, first remove it.

Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Reviewed-by: Eryk Brol <Eryk.Brol@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Revert "Fix EDID parsing after resume from suspend"

This reverts commit b24bdc37d03a0478189e20a50286092840f414fa.
It caused memory leak after S3 on 4K HDMI displays.

Signed-off-by: Stylon Wang <stylon.wang@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix calculation for the pwl backlight curve

[Why]
The PWL backlight curve is used by the firmware to convert between
brightness and linear PWM value.
Driver has a backlight LUT, but the firmware holds a PWL curve and
interpolates between points.

The calculations are incorrect leading to slightly off backlight values
being programmed.

[How]
Fix the PWL backlight curve threshold/offset calculations

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Josip Pavic <Josip.Pavic@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.120

This DC update brings improvements in multiple areas. In summary, we highlight:
 - Fix display detection on HDMI ComboPHY
 - Drop SOC bounding box hookup
 - Fix DPCD values

Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: [FW Promotion] Release 0.0.49

- Add field for passing line time for a frame

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add more Clock Sources to DCN2.1

[WHY]
When enabling HDMI on ComboPHY, there are not
enough clock sources to complete display detection.

[HOW]
Initialize more clock sources.

Signed-off-by: Sung Lee <sung.lee@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: correct some hdcp variable naming

[why]
In HDCP update stream config interface, some variables are named as
xxx_supported, but in fact the variable indicates whether or not xxx_enabled.
Correct the naming so it is less confusing to read the code.

Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Reviewed-by: George Shen <George.Shen@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Set power_gated to true for seamless boot pipe init

[Why]
In seamless boot without a flip case, the flag power_gated didn't
get cleared when resetting path mode because the plane_state is null.
The following sequence will cause this issue:
    1. OS call set mode to clone/extended
    2. Reset path mode to remove edp

[How]
Set power gated default to true in seamless boot pipe

Signed-off-by: Lewis Huang <Lewis.Huang@amd.com>
Reviewed-by: Martin Leung <Martin.Leung@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: reuse current context instead of recreating one

[Why]
Currently we discard the current context and recreate it. The current
context is what is applied to the HW so we should be re-using this
rather than creating a new context.

Recreating the context can lead to mismatch between new context and the
current context

For example: gsl groups get changed when we create a new context this
can cause issues in a multi display config (with flip immediate) because
we don't align the existing gsl groups in the new and current context.
If we reuse the current context the gsl group assignment stays the same.

[How]
Instead of discarding the current context, we instead just copy the
current state and add/remove planes and streams.

Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add null pointer check to is_dig_enabled func

[WHY]
Safeguarding as pointer may be null in diagnostic environment

Signed-off-by: Michael Strauss <michael.strauss@amd.com>
Reviewed-by: Sung Lee <Sung.Lee@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: initialize i2c speed if not initialized in dcnxxx__resource.c

Some dcnxxx__resource.c do not initialize the i2c speed; this patch adds
the required initialization at dc_construct().

Signed-off-by: Brendan Steve Leder <brendanSteve.Leder@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix DPCD translation for LTTPR AUX_RD_INTERVAL

[Why]
The translation between the DPCD value and the specified AUX_RD_INTERVAL
in the DP spec do not match.

[How]
Update values to match the spec.

Signed-off-by: George Shen <george.shen@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Drop SOC bounding box hookup in DM/DC

[Why]
There aren't any ASIC where we use these binaries and they aren't
useful for future use since it's inconvenient to extend and maintain
these structures.

[How]
Drop the support from DM and DC for now.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Acked-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Hersen Wu <hersenxs.wu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: enable freesync for A+A configs

Some newer APUs can scanout directly from GTT, that saves us from
allocating another bounce buffer in VRAM and enables freesync in such
configurations.

Without this patch creating a framebuffer from the imported BO will
fail and userspace will fall back to a copy.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Shashank Sharma <shashank.sharma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: remove some useless code for vangogh

This patch is to remove some useless code for vangogh.
In the earlier code, vangogh can't finish all the sequence of
smu late init. But now vangogh has one stable work state,so
remove the useless code.

Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Acked-by: Xiaomeng Hou <Xiaomeng.Hou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: fill in the data member of v2 gpu metrics table for vangogh

This patch is to fill in the data member of v2 gpu metrics
table for vangogh.

Signed-off-by: Xiaojian Du <Xiaojian.Du@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/gfx10: update CGTS_TCC_DISABLE and CGTS_USER_TCC_DISABLE register offsets for VGH

For Vangogh:
The offset of the CGTS_TCC_DISABLE is 0x5006 by calculation.
The offset of the CGTS_USER_TCC_DISABLE is 0x5007 by calculation.

Signed-off-by: chen gong <curry.gong@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Fix a false positive when pin non-VRAM memory

Flag TTM_PL_FLAG_CONTIGUOUS is only valid for VRAM domain. So fix the
false positive by checking memory type too.

Suggested-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Limit the maximum size of contiguous VRAM that can be encapsulated by an instance of DRM memory node

[Why]
Enable 1:1 mapping between VRAM of a DRM node and a scatterlist node

[How]
Ensure construction of DRM node to not exceed specified limit

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Ramesh Errabolu <Ramesh.Errabolu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: fix null pointer panic while free buffer in kfd

In drm_gem_object_free, it will call funcs of drm buffer obj. So
kfd_alloc should use amdgpu_gem_object_create instead of
amdgpu_bo_create to initialize the funcs as amdgpu_gem_object_funcs.

[  396.231390] amdgpu: Release VA 0x7f76b4ada000 - 0x7f76b4add000
[  396.231394] amdgpu:   remove VA 0x7f76b4ada000 - 0x7f76b4add000 in entry 0000000085c24a47
[  396.231408] BUG: kernel NULL pointer dereference, address: 0000000000000000
[  396.231445] #PF: supervisor read access in kernel mode
[  396.231466] #PF: error_code(0x0000) - not-present page
[  396.231484] PGD 0 P4D 0
[  396.231495] Oops: 0000 [#1] SMP NOPTI
[  396.231509] CPU: 7 PID: 1352 Comm: clinfo Tainted: G           OE     5.11.0-rc2-custom #1
[  396.231537] Hardware name: AMD Celadon-RN/Celadon-RN, BIOS WCD0401N_Weekly_20_04_0 04/01/2020
[  396.231563] RIP: 0010:drm_gem_object_free+0xc/0x22 [drm]
[  396.231606] Code: eb ec 48 89 c3 eb e7 0f 1f 44 00 00 55 48 89 e5 48 8b bf 00 06 00 00 e8 72 0d 01 00 5d c3 0f 1f 44 00 00 48 8b 87 40 01 00 00 <48> 8b 00 48 85 c0 74 0b 55 48 89 e5 e8 54 37 7c db 5d c3 0f 0b c3
[  396.231666] RSP: 0018:ffffb4704177fcf8 EFLAGS: 00010246
[  396.231686] RAX: 0000000000000000 RBX: ffff993a0d0cc400 RCX: 0000000000003113
[  396.231711] RDX: 0000000000000001 RSI: e9cda7a5d0791c6d RDI: ffff993a333a9058
[  396.231736] RBP: ffffb4704177fdd0 R08: ffff993a03855858 R09: 0000000000000000
[  396.231761] R10: ffff993a0d1f7158 R11: 0000000000000001 R12: 0000000000000000
[  396.231785] R13: ffff993a0d0cc428 R14: 0000000000003000 R15: ffffb4704177fde0
[  396.231811] FS:  00007f76b5730740(0000) GS:ffff993b275c0000(0000) knlGS:0000000000000000
[  396.231840] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  396.231860] CR2: 0000000000000000 CR3: 000000016d2e2000 CR4: 0000000000350ee0
[  396.231885] Call Trace:
[  396.231897]  ? amdgpu_amdkfd_gpuvm_free_memory_of_gpu+0x24c/0x25f [amdgpu]
[  396.232056]  ? __dynamic_dev_dbg+0xcd/0x100
[  396.232076]  kfd_ioctl_free_memory_of_gpu+0x91/0x102 [amdgpu]
[  396.232214]  kfd_ioctl+0x211/0x35b [amdgpu]
[  396.232341]  ? kfd_ioctl_get_queue_wave_state+0x52/0x52 [amdgpu]

Fixes: 246cb7e49a70 ("drm/amdgpu: Introduce GEM object functions")
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Tested-by: Changfeng <changzhu@amd.com>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix the issue that retry constantly once the buffer is oversize

We cannot modify initial_domain every time while the retry starts. That
will cause the busy waiting that unable to switch to GTT while the vram
is not enough.

Fixes: f8aab60422c3 ("drm/amdgpu: Initialise drm_gem_object_funcs for imported BOs")

Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/swsmu: use percent rather than rpm for manual fan control (v2)

On some boards the rpm interface apparently does not work at all
leading to the fan not spinning or spinning at strange speeds.
Both interfaces work properly on the boards I have.
Let's try and use the percent interface instead.

v2: rebase on revert

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1408
Reviewed-by: Evan Quan <evan.quan@amd.com> (v1)
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "drm/amdgpu/swsmu: drop set_fan_speed_percent (v2)"

On some boards the rpm interface apparently does not work at all
leading to the fan not spinning or spinning at strange speeds.
Revert this for now to fix 5.10, 5.11.  The follow on patch
fixes this properly for 5.12.

This reverts commit 8d6e65adc25e23fabbc5293b6cd320195c708dca.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1408
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Simplify bool conversion

Fix the following coccicheck warning:
./drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c:3137:35-40:
WARNING: conversion to bool not needed here

Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Suggested-by: Yang Li <oswb@linux.alibaba.com>
Signed-off-by: Abaci Team <abaci-bugfix@linux.alibaba.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amdgpu: Add Missing Sienna Cichlid DID

The purpose of this patch is to add a missing device ID for Sienna Cichlid.
The missing ID "0x73A1" is now added to the "amdgpu_drv.c" file.

Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Reviewed-by: Kent Russell <kent.russell@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: correct string parsing for extra separator characters case

Supplement of previous fix.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/pm: Account for extra separator characters in sysfs interface

Currently, the the input to the pp_od_clk_voltage sysfs interface has
two adjacent separator characters, i.e. "\n\0", then we try to parse an
argument out of the empty string, whereas we really should just ignore
this case, and treat any number of adjacent separators as one separator
for arguments.

We do this here by simply skipping the argument parsing for all
empy-string arguments.

An example test case would be `echo 's  1 900' > pp_od_clk_voltage`
(note the two spaces after the 's'.

This also solves the issue where the written string ends with both a
newline, and a nul-terminator, '\n\0', as is the case for `echo 's 1
900' > pp_od_clk_voltage` since the recent rebase.

Got lost during code rebase/merge. No need to port this to other
branches.

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Matt Coffin <mcoffin13@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Revert "Revert "drm/amdgpu: Fix NULL dereference in dpm sysfs handlers""

This reverts commit 2456c290a7889be492cb96092b62d16c11176f72. Got lost
during code rebase/merge. No need to port this to other branches.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge branch 'linux-5.12' of git://github.com/skeggsb/linux into drm-next

Nothing too major here, I actually thought I'd sent most of these
right before the new year, but that apparently got lost in the bustle:
- Turing MMU fault recovery fixes
- Fix mDP connectors being reported as eDP to userspace
- Fixes for audio locking, and other bit-rot from DRM changes since
atomic support was written
- Misc other minor fixes.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Ben Skeggs <skeggsb@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CACAvsv7yLfhuVbYa-4g0vxVt93OaC7Sodiz2R-TDHu-MoofEdw@mail.gmail.com

drm/i915/dp: Prevent setting the LTTPR LT mode if no LTTPRs are detected

Atm, the driver programs explicitly the default transparent link
training mode (0x55) to DP_PHY_REPEATER_MODE even if no LTTPRs are
detected.

This conforms to the spec (3.6.6.1):
"DP upstream devices that do not enable the Non-transparent mode of
 LTTPRs shall program the PHY_REPEATER_MODE register (DPCD Address
 F0003h) to 55h (default) prior to link training"

however writing the default value to this DPCD register seems to cause
occasional link training errors at least for a DELL WD19TB TBT dock, when
no LTTPRs are detected.

Writing to DP_PHY_REPEATER_MODE will also cause an unnecessary timeout
on systems without any LTTPR.

To fix the above two issues let's assume that setting the default mode
is redundant when no LTTPRs are detected. Keep the existing behavior and
program the default mode if more than 8 LTTPRs are detected or in case
the read from DP_PHY_REPEATER_CNT returns an invalid value.

References: https://gitlab.freedesktop.org/drm/intel/-/issues/2801
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Khaled Almahallawy <khaled.almahallawy@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210118183143.1145707-1-imre.deak@intel.com

drm/i915: Implement async flips for vlv/chv

Add support for async flips on vlv/chv. Unlike all the other
platforms vlv/chv do not use the async flip bit in DSPCNTR and
instead we select between async vs. sync flips based on the
surface address register. The normal DSPSURF generates sync
flips DSPADDR_VLV generates async flips. And as usual the
interrupt bits are different from the other platforms.

Cc: Karthik B S <karthik.b.s@intel.com>
Cc: Vandita Kulkarni <vandita.kulkarni@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210111163711.12913-12-ville.syrjala@linux.intel.com
Reviewed-by: Karthik B S <karthik.b.s@intel.com>

drm/i915: Implement async flip for ilk/snb

Add support for async flips on ivb/hsw. Again no need for any
workarounds and just have to deal with the interrupt bits being
shuffled around a bit.

Cc: Karthik B S <karthik.b.s@intel.com>
Cc: Vandita Kulkarni <vandita.kulkarni@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210111163711.12913-11-ville.syrjala@linux.intel.com
Reviewed-by: Karthik B S <karthik.b.s@intel.com>

drm/i915: Implement async flip for ivb/hsw

Add support for async flips on ivb/hsw. Unlike bdw+ we don't need
any workarounds to disable async flips. Apart from that the only
real difference from the bdw implementation is the location of the
flip_done interrupt bits.

Cc: Karthik B S <karthik.b.s@intel.com>
Cc: Vandita Kulkarni <vandita.kulkarni@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210111163711.12913-10-ville.syrjala@linux.intel.com
Reviewed-by: Karthik B S <karthik.b.s@intel.com>

drm/i915: Implement async flips for bdw

Implement async flip support for BDW. The implementation is
similar to the skl+ code. And just like skl/bxt/glk bdw also
needs the disable w/a, thus we need to plumb the desired state
of the async flip all the way down to i9xx_plane_ctl_crtc().

According to the spec we do need to bump the surface alignment
to 256KiB for this. Async flips require an X-tiled buffer so
we don't have to worry about linear.

Cc: Karthik B S <karthik.b.s@intel.com>
Cc: Vandita Kulkarni <vandita.kulkarni@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210111163711.12913-9-ville.syrjala@linux.intel.com
Reviewed-by: Karthik B S <karthik.b.s@intel.com>

drm/i915: Limit plane stride to below TILEOFF.x limit

Limit pre-skl plane stride to below 4k or 8k pixels (depending on
the platform). We do this in order guarantee that TILEOFF/OFFSET.x
does not get too big.

Currently this is not a problem as we align SURF to 4k, and so
TILEOFF/OFFSET only have to deal with a single tile's worth of
pixels. But for async flips we're going to have to bump SURF
alignment to 256k, and thus we can no longer guarantee
TILEOFF/OFFSET.x will stay within acceptable bounds. We can avoid
this by borrowing a trick from the skl+ code and limit the max
plane stride to whatever value we can fit into TILEOFF/OFFSET.x.

The slight downside is that we may end up doing GTT remapping in
a few more cases where previously we did not have to. But since
that will only happen with huge buffers I'm not really concerned
about it.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210111163711.12913-3-ville.syrjala@linux.intel.com
Reviewed-by: Karthik B S <karthik.b.s@intel.com>

drm/i915: Rename is_16gb_dimm to wm_lv_0_adjust_needed

As it now it is always required for GEN12+ the is_16gb_dimm name
do not make sense for GEN12+.

v2:
- Updated comment on top of "dram_info->wm_lv_0_adjust_needed =
!IS_GEN9_LP(i915);"

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210128164312.91160-3-jose.souza@intel.com

drm/i915/gen11+: Only load DRAM information from pcode

Up to now we were reading some DRAM information from MCHBAR register
and from pcode what is already not good but some GEN12(TGL-H and ADL-S)
platforms have MCHBAR DRAM information in different offsets.

This was notified to HW team that decided that the best alternative is
always apply the 16gb_dimm watermark adjustment for GEN12+ platforms
and read the remaning DRAM information needed to other display
programming from pcode.

So here moving the DRAM pcode function to intel_dram.c, removing
the duplicated fields from intel_qgv_info, setting and using
information from dram_info.

v2:
- bring back num_points to intel_qgv_info as num_qgv_point can be
overwritten in icl_get_qgv_points()
- add gen12_get_dram_info() and simplify gen11_get_dram_info()

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210128164312.91160-2-jose.souza@intel.com

drm/i915: Nuke not needed members of dram_info

Valid, ranks and bandwidth_kbps are set into dram_info but are not
used anywhere else so nuking it.

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210128164312.91160-1-jose.souza@intel.com

Merge tag 'drm-intel-next-2021-01-27' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

- HDCP 2.2 and HDCP 1.4 Gen12 DP MST support (Anshuman)
- Fix DP vswing settings and handling (Imre, Ville)
- Various display code clean-up (Jani, Ville)
- Various display refactoring, including split out of pps, aux, and fdi (Ja\
ni, Dave)
- Add DG1 missing workarounds (Jose)
- Fix display color conversion (Chris, Ville)
- Try to guess PCH type even without ISA bridge (Zhenyu)
- More backlight refactor (Lyude)
- Support two CSC module on gen11 and later (Lee)
- Async flips for all ilk+ platforms (Ville)
- Clear color support for TGL (RK)
- Add a helper to read data from a GEM object page (Imre)
- VRR/Adaptive Sync Enabling on DP/eDP for TGL+ (Manasi, Ville Aditya)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210127140822.GA711686@intel.com

drm/nouveau/kms: handle mDP connectors

In some cases we have the handle those explicitly as the fallback
connector type detection fails and marks those as eDP connectors.

Attempting to use such a connector with mutter leads to a crash of mutter
as it ends up with two eDP displays.

Information is taken from the official DCB documentation.

Cc: stable@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org
Cc: Ben Skeggs <bskeggs@redhat.com>
Reported-by: Mark Pearson <markpearson@lenovo.com>
Tested-by: Mark Pearson <markpearson@lenovo.com>
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: bail out of nouveau_channel_new if channel init fails

Unprivileged user can crash kernel by using DRM_IOCTL_NOUVEAU_CHANNEL_ALLOC
ioctl. This was reported by trinity[1] fuzzer.

[   71.073906] nouveau 0000:01:00.0: crashme[1329]: channel failed to initialise, -17
[   71.081730] BUG: kernel NULL pointer dereference, address: 00000000000000a0
[   71.088928] #PF: supervisor read access in kernel mode
[   71.094059] #PF: error_code(0x0000) - not-present page
[   71.099189] PGD 119590067 P4D 119590067 PUD 1054f5067 PMD 0
[   71.104842] Oops: 0000 [#1] SMP NOPTI
[   71.108498] CPU: 2 PID: 1329 Comm: crashme Not tainted 5.8.0-rc6+ #2
[   71.114993] Hardware name: AMD Pike/Pike, BIOS RPK1506A 09/03/2014
[   71.121213] RIP: 0010:nouveau_abi16_ioctl_channel_alloc+0x108/0x380 [nouveau]
[   71.128339] Code: 48 89 9d f0 00 00 00 41 8b 4c 24 04 41 8b 14 24 45 31 c0 4c 8d 4b 10 48 89 ee 4c 89 f7 e8 10 11 00 00 85 c0 75 78 48 8b 43 10 <8b> 90 a0 00 00 00 41 89 54 24 08 80 7d 3d 05 0f 86 bb 01 00 00 41
[   71.147074] RSP: 0018:ffffb4a1809cfd38 EFLAGS: 00010246
[   71.152526] RAX: 0000000000000000 RBX: ffff98cedbaa1d20 RCX: 00000000000003bf
[   71.159651] RDX: 00000000000003be RSI: 0000000000000000 RDI: 0000000000030160
[   71.166774] RBP: ffff98cee776de00 R08: ffffdc0144198a08 R09: ffff98ceeefd4000
[   71.173901] R10: ffff98cee7e81780 R11: 0000000000000001 R12: ffffb4a1809cfe08
[   71.181214] R13: ffff98cee776d000 R14: ffff98cec519e000 R15: ffff98cee776def0
[   71.188339] FS:  00007fd926250500(0000) GS:ffff98ceeac80000(0000) knlGS:0000000000000000
[   71.196418] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   71.202155] CR2: 00000000000000a0 CR3: 0000000106622000 CR4: 00000000000406e0
[   71.209297] Call Trace:
[   71.211777]  ? nouveau_abi16_ioctl_getparam+0x1f0/0x1f0 [nouveau]
[   71.218053]  drm_ioctl_kernel+0xac/0xf0 [drm]
[   71.222421]  drm_ioctl+0x211/0x3c0 [drm]
[   71.226379]  ? nouveau_abi16_ioctl_getparam+0x1f0/0x1f0 [nouveau]
[   71.232500]  nouveau_drm_ioctl+0x57/0xb0 [nouveau]
[   71.237285]  ksys_ioctl+0x86/0xc0
[   71.240595]  __x64_sys_ioctl+0x16/0x20
[   71.244340]  do_syscall_64+0x4c/0x90
[   71.248110]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[   71.253162] RIP: 0033:0x7fd925d4b88b
[   71.256731] Code: Bad RIP value.
[   71.259955] RSP: 002b:00007ffc743592d8 EFLAGS: 00000206 ORIG_RAX: 0000000000000010
[   71.267514] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fd925d4b88b
[   71.274637] RDX: 0000000000601080 RSI: 00000000c0586442 RDI: 0000000000000003
[   71.281986] RBP: 00007ffc74359340 R08: 00007fd926016ce0 R09: 00007fd926016ce0
[   71.289111] R10: 0000000000000003 R11: 0000000000000206 R12: 0000000000400620
[   71.296235] R13: 00007ffc74359420 R14: 0000000000000000 R15: 0000000000000000
[   71.303361] Modules linked in: rfkill sunrpc snd_hda_codec_realtek snd_hda_codec_generic ledtrig_audio snd_hda_intel snd_intel_dspcfg snd_hda_codec snd_hda_core edac_mce_amd snd_hwdep kvm_amd snd_seq ccp snd_seq_device snd_pcm kvm snd_timer snd irqbypass soundcore sp5100_tco pcspkr crct10dif_pclmul crc32_pclmul ghash_clmulni_intel wmi_bmof joydev i2c_piix4 fam15h_power k10temp acpi_cpufreq ip_tables xfs libcrc32c sd_mod t10_pi sg nouveau video mxm_wmi i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops ttm broadcom bcm_phy_lib ata_generic ahci drm e1000 crc32c_intel libahci serio_raw tg3 libata firewire_ohci firewire_core wmi crc_itu_t dm_mirror dm_region_hash dm_log dm_mod
[   71.365269] CR2: 00000000000000a0

simplified reproducer
---------------------------------8<----------------------------------------
/*
 * gcc -o crashme crashme.c
 * ./crashme /dev/dri/renderD128
 */

struct drm_nouveau_channel_alloc {
uint32_t     fb_ctxdma_handle;
uint32_t     tt_ctxdma_handle;

int          channel;
uint32_t     pushbuf_domains;

/* Notifier memory */
uint32_t     notifier_handle;

/* DRM-enforced subchannel assignments */
struct {
uint32_t handle;
uint32_t grclass;
} subchan[8];
uint32_t nr_subchan;
};

static struct drm_nouveau_channel_alloc channel;

int main(int argc, char *argv[]) {
int fd;
int rv;

if (argc != 2)
die("usage: %s <dev>", 0, argv[0]);

if ((fd = open(argv[1], O_RDONLY)) == -1)
die("open %s", errno, argv[1]);

if (ioctl(fd, DRM_IOCTL_NOUVEAU_CHANNEL_ALLOC, &channel) == -1 &&
errno == EACCES)
die("ioctl %s", errno, argv[1]);

close(fd);

printf("PASS\n");

return 0;
}
---------------------------------8<----------------------------------------

[1] https://github.com/kernelslacker/trinity

Fixes: eeaf06ac1a55 ("drm/nouveau/svm: initial support for shared virtual memory")
Signed-off-by: Frantisek Hrbata <frantisek@hrbata.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: Fix locking for audio callbacks

Noticed that I wasn't paying close enough attention the last time I looked
at our audio callbacks, as I completely missed the fact that we were
figuring out which audio-enabled connector goes to each encoder by checking
it's state, but without grabbing any of the appropriate modesetting locks
to do so.

That being said however: trying to grab modesetting locks in our audio
callbacks would be very painful due to the potential for locking inversion
between HDA and DRM. So, let's instead just copy what i915 does again - add
our own audio lock to protect audio related state, and store each audio
enabled connector in each nouveau_encoder struct so that we don't need to
check any atomic states.

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

drm/nouveau/kms/nv50-: Use nouveau_encoder->crtc in get_eld callback

drm_encoder->crtc is deprecated for atomic drivers, but
nouveau_encoder->crtc is safe.

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

drm/nouveau/kms/nv50-: Lookup current encoder/crtc from atomic state

Despite being an atomic driver, nouveau has a lot of leftover code that
relies on retrieving information regarding the new atomic state from
members of drm_encoder and drm_crtc. The first field being used,
drm_encoder.crtc, is deprecated for atomic drivers. The second field being
used is drm_crtc.state, which is only really sensible to use outside of an
atomic modeset.

So, add some helpers to lookup the current crtc for a given outp from the
atomic state. Then, convert most of the code in dispnv50/disp.c to use said
new helper, along with the relevant DRM atomic helpers for retrieving the
new encoder/crtc combinations for a new atomic state.

Note that we don't get rid of the nouveau_encoder.crtc field entirely for
three reasons:

- Legacy modesetting for pre-nv50 still uses it
- It doesn't cause any locking issues
- We need it for the HDA callbacks, as grabbing atomic modesetting locks in
  those would be a mess.

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

drm/nouveau/kms/nv50-: Reverse args for nv50_outp_get_(old|new)_connector()

Just to be more consistent with the order of args that DRM helpers like
drm_atomic_get_new_crtc_state() use.

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

drm/nouveau/kms/nv50-: s/armh/asyh/ in nv50_msto_atomic_enable()

I have a strange dejavu feeling that I tried to submit a patch for this in
the past, but that it was rejected. I can't remember though, but I'm
further convinced this patch is the right thing to do anyway.

We label the to-be-committed head state in nv50_msto_atomic_enable() as
armh. Normally armh implies a state which is currently armed in hardware.
nv50_msto_atomic_enable() is called _after_ drm_atomic_swap_state()
however, but before the commit tail ends, which means that said state is
not actually armed on hardware.

As well - take note that this is the same convention followed in all of the
other atomic_enable() callbacks.

So, let's correct this to asyh.

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

drm/nouveau/kms/nv50-: Rename encoder->atomic_(enable|disable) callbacks

No functional changes, just change the function names to match the
callbacks they provide.

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

drm/nouveau/kms/nv50-: Remove (nv_encoder->crtc) checks in ->disable callbacks

Noticed these in both the disable (which we'll be getting rid of in a
moment) and the atomic disable callbacks: both callback types check whether
or not there's actually a CRTC assigned to the given encoder. However, as
->atomic_disable and ->disable will never be called without a CRTC assigned
to the given encoder there's no point in this check. So just remove it.

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

drm/nouveau/fifo/tu102: Turing channel preemption fix

Previous hardware allowed a MMU fault to be generated by software to
trigger a context switch for engine recovery. Turing has the capability
to preempt all work from a specific runlist processor and removed the
registers currently used for triggering MMU faults. Attempting to access
these non-existent registers results in further errors, so use the
runlist preemption register instead.

Signed-off-by: Alistair Popple <apopple@nvidia.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/tu102: FIFO interrupt fixes for Turing

Some of the low level FIFO interrupt status bits have changed for
Turing. Update the handling of these to match the hardware.

Signed-off-by: Alistair Popple <apopple@nvidia.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>