linux-stable.git/drivers/gpu/drm/amd/display, branch v4.19.7

drm/amd/amdgpu/dm: Fix dm_dp_create_fake_mst_encoder()

2018-11-27T15:13:05+00:00

[ Upstream commit 63237f8748bdf46dccf79ef8f98f05e9fe799162 ]

[why]
Removing connector reusage from DM to match the rest of the tree ended
up revealing an issue that was surprisingly subtle. The original amdgpu
code for DC that was submitted appears to have left a chunk in
dm_dp_create_fake_mst_encoder() that tries to find a "master encoder",
the likes of which isn't actually used or stored anywhere. It does so at
the wrong time as well by trying to access parts of the drm_connector
from the encoder init before it's actually been initialized. This
results in a NULL pointer deref on MST hotplugs:

[  160.696613] BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
[  160.697234] PGD 0 P4D 0
[  160.697814] Oops: 0010 [#1] SMP PTI
[  160.698430] CPU: 2 PID: 64 Comm: kworker/2:1 Kdump: loaded Tainted: G           O      4.19.0Lyude-Test+ #2
[  160.699020] Hardware name: HP HP ZBook 15 G4/8275, BIOS P70 Ver. 01.22 05/17/2018
[  160.699672] Workqueue: events_long drm_dp_mst_link_probe_work [drm_kms_helper]
[  160.700322] RIP: 0010:          (null)
[  160.700920] Code: Bad RIP value.
[  160.701541] RSP: 0018:ffffc9000029fc78 EFLAGS: 00010206
[  160.702183] RAX: 0000000000000000 RBX: ffff8804440ed468 RCX: ffff8804440e9158
[  160.702778] RDX: 0000000000000000 RSI: ffff8804556c5700 RDI: ffff8804440ed000
[  160.703408] RBP: ffff880458e21800 R08: 0000000000000002 R09: 000000005fca0a25
[  160.704002] R10: ffff88045a077a3d R11: ffff88045a077a3c R12: ffff8804440ed000
[  160.704614] R13: ffff880458e21800 R14: ffff8804440e9000 R15: ffff8804440e9000
[  160.705260] FS:  0000000000000000(0000) GS:ffff88045f280000(0000) knlGS:0000000000000000
[  160.705854] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  160.706478] CR2: ffffffffffffffd6 CR3: 000000000200a001 CR4: 00000000003606e0
[  160.707124] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  160.707724] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  160.708372] Call Trace:
[  160.708998]  ? dm_dp_add_mst_connector+0xed/0x1d0 [amdgpu]
[  160.709625]  ? drm_dp_add_port+0x2fa/0x470 [drm_kms_helper]
[  160.710284]  ? wake_up_q+0x54/0x70
[  160.710877]  ? __mutex_unlock_slowpath.isra.18+0xb3/0x110
[  160.711512]  ? drm_dp_dpcd_access+0xe7/0x110 [drm_kms_helper]
[  160.712161]  ? drm_dp_send_link_address+0x155/0x1e0 [drm_kms_helper]
[  160.712762]  ? drm_dp_check_and_send_link_address+0xa3/0xd0 [drm_kms_helper]
[  160.713408]  ? drm_dp_mst_link_probe_work+0x4b/0x80 [drm_kms_helper]
[  160.714013]  ? process_one_work+0x1a1/0x3a0
[  160.714667]  ? worker_thread+0x30/0x380
[  160.715326]  ? wq_update_unbound_numa+0x10/0x10
[  160.715939]  ? kthread+0x112/0x130
[  160.716591]  ? kthread_create_worker_on_cpu+0x70/0x70
[  160.717262]  ? ret_from_fork+0x35/0x40
[  160.717886] Modules linked in: amdgpu(O) vfat fat snd_hda_codec_generic joydev i915 chash gpu_sched ttm i2c_algo_bit drm_kms_helper snd_hda_codec_hdmi hp_wmi syscopyarea iTCO_wdt sysfillrect sparse_keymap sysimgblt fb_sys_fops snd_hda_intel usbhid wmi_bmof drm snd_hda_codec btusb snd_hda_core intel_rapl btrtl x86_pkg_temp_thermal btbcm btintel coretemp snd_pcm crc32_pclmul bluetooth psmouse snd_timer snd pcspkr i2c_i801 mei_me i2c_core soundcore mei tpm_tis wmi tpm_tis_core hp_accel ecdh_generic lis3lv02d tpm video rfkill acpi_pad input_polldev hp_wireless pcc_cpufreq crc32c_intel serio_raw tg3 xhci_pci xhci_hcd [last unloaded: amdgpu]
[  160.720141] CR2: 0000000000000000

Somehow the connector reusage DM was using for MST connectors managed to
paper over this issue entirely; hence why this was never caught until
now.

[how]
Since this code isn't used anywhere and seems useless anyway, we can
just drop it entirely. This appears to fix the issue on my HP ZBook with
an AMD WX4150.

Signed-off-by: Lyude Paul 
Reviewed-by: Harry Wentland 
Signed-off-by: Alex Deucher 
Signed-off-by: Sasha Levin

drm/amd/display: Drop reusing drm connector for MST

2018-11-27T15:13:05+00:00

[ Upstream commit 0e6613e46fed29316f33acf86e1d1568288638b5 ]

[why]
It is not safe to keep existing connector while entire topology
has been removed. Could lead potential impact to uapi.
Entirely unregister all the connectors on the topology,
and use a new set of connectors when the topology is plugged back
on.

[How]
Remove the drm connector entirely each time when the
corresponding MST topology is gone.
When hotunplug a connector (e.g., DP2)
1. Remove connector from userspace.
2. Drop it's reference.
When hotplug back on:
1. Detect new topology, and create new connectors.
2. Notify userspace with sysfs hotplug event.
3. Reprobe new connectors, and reassign CRTC from old (e.g., DP2)
to new (e.g., DP3) connector.

Signed-off-by: Jerry (Fangzhi) Zuo 
Reviewed-by: Harry Wentland 
Reviewed-by: Lyude Paul 
Signed-off-by: Alex Deucher 
Signed-off-by: Sasha Levin

drm/amd/display: Stop leaking planes

2018-11-27T15:13:05+00:00

[ Upstream commit 02680efbb10be0d2c867fe722ae23d588f6bebef ]

[Why]
drm_plane_cleanup does not free the plane.

[How]
Call drm_primary_helper_destroy which will also free the plane.

Signed-off-by: Harry Wentland 
Acked-by: Alex Deucher 
Signed-off-by: Alex Deucher 
Signed-off-by: Sasha Levin

drm/amd/display: fix gamma not being applied

2018-11-21T08:19:10+00:00

[ Upstream commit 30049754ab7c4b6148dd3cd64af7d54850604582 ]

[WHY]
Previously night light forced a full update by
applying a  transfer function update regardless of if it was changed.
This logic was removed,

Now gamma surface updates are only applied when there is also a plane
info update, this does not work in cases such as using the night light
slider.

[HOW]
When moving the night light slider we will perform a full update if
the gamma has changed and there is a surface, even when the surface
has not changed. Also get stream updates in setgamma prior to
update planes and stream.

Signed-off-by: SivapiriyanKumarasamy 
Reviewed-by: Anthony Koo 
Acked-by: Bhawanpreet Lakha 
Signed-off-by: Alex Deucher 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

drm/amd/display: Raise dispclk value for dce120 by 15%

2018-11-21T08:19:10+00:00

[ Upstream commit 481f576c6c21bf0446eaa23623ef0262e9a5387c ]

[Why]

The DISPCLK value was previously requested to be 15% higher for all
ASICs that went through the dce110 bandwidth code path. As part of a
refactoring of dce_clocks and the dce110 set bandwidth codepath this
was removed for power saving considerations.

That change caused display corruption under certain hardware
configurations with Vega10.

[How]

The 15% DISPCLK increase is brought back but only on dce110 for now.
This is should be a temporary workaround until the root cause is sorted
out for why this occurs on Vega (or other ASICs, if reported).

Tested-by: Nick Sarnie 
Signed-off-by: Nicholas Kazlauskas 
Reviewed-by: Harry Wentland 
Acked-by: Bhawanpreet Lakha 
Signed-off-by: Alex Deucher 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

drm/amd/display: fix bug of accessing invalid memory

2018-11-21T08:19:08+00:00

[ Upstream commit 43c3ff27a47d83d153c4adc088243ba594582bf5 ]

[Why]
A loop inside of build_evenly_distributed_points function that traverse through
the array of points become an infinite loop when m_GammaUpdates does not
get assigned to any value.

[How]
In DMColor, clear m_gammaIsValid bit just before writting all Zeromem for
m_GammaUpdates, to prevent calling build_evenly_distributed_points
before m_GammaUpdates gets assigned to some value.

Signed-off-by: Su Sung Chung 
Reviewed-by: Aric Cyr 
Acked-by: Bhawanpreet Lakha 
Signed-off-by: Alex Deucher 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman

drm/amd/display: Signal hw_done() after waiting for flip_done()

2018-10-04T15:21:03+00:00

In amdgpu_dm_commit_tail(), wait until flip_done() is signaled before
we signal hw_done().

[Why]

This is to temporarily address a paging error that occurs when a
nonblocking commit contends with another commit, particularly in a
mirrored display configuration where at least 2 CRTCs are updated.
The error occurs in drm_atomic_helper_wait_for_flip_done(), when we
attempt to access the contents of new_crtc_state->commit.

Here's the sequence for a mirrored 2 display setup (irrelevant steps
left out for clarity):

**THREAD 1**                        | **THREAD 2**
                                    |
Initialize atomic state for flip    |
                                    |
Queue worker                        |
                                   ...

                                    | Do work for flip
                                    |
                                    | Signal hw_done() on CRTC 1
                                    | Signal hw_done() on CRTC 2
                                    |
                                    | Wait for flip_done() on CRTC 1

                                <---- **PREEMPTED BY THREAD 1**

Initialize atomic state for cursor  |
update (1)                          |
                                    |
Do cursor update work on both CRTCs |
                                    |
Clear atomic state (2)              |
**DONE**                            |
                                   ...
                                    |
                                    | Wait for flip_done() on CRTC 2
                                    | *ERROR*
                                    |

The issue starts with (1). When the atomic state is initialized, the
current CRTC states are duplicated to be the new_crtc_states, and
referenced to be the old_crtc_states. (The new_crtc_states are to be
filled with update data.)

Some things to note:

* Due to the mirrored configuration, the cursor updates on both CRTCs.

* At this point, the pflip IRQ has already been handled, and flip_done
  signaled on all CRTCs. The cursor commit can therefore continue.

* The old_crtc_states used by the cursor update are the **same states**
  as the new_crtc_states used by the flip worker.

At (2), the old_crtc_state is freed (*), and the cursor commit
completes. We then context switch back to the flip worker, where we
attempt to access the new_crtc_state->commit object. This is
problematic, as this state has already been freed.

(*) Technically, 'state->crtcs[i].state' is freed, which was made to
    reference old_crtc_state in drm_atomic_helper_swap_state()

[How]

By moving hw_done() after wait_for_flip_done(), we're guaranteed that
the new_crtc_state (from the flip worker's perspective) still exists.
This is because any other commit will be blocked, waiting for the
hw_done() signal.

Note that both the i915 and imx drivers have this sequence flipped
already, masking this problem.

Signed-off-by: Shirish S 
Signed-off-by: Leo Li 
Reviewed-by: Harry Wentland 
Signed-off-by: Alex Deucher

drm/amd/display: Fix Edid emulation for linux

2018-09-27T15:05:21+00:00

[Why]
EDID emulation didn't work properly for linux, as we stop programming
if nothing is connected physically.

[How]
We get a flag from DRM when we want to do edid emulation. We check if
this flag is true and nothing is connected physically, if so we only
program the front end using VIRTUAL_SIGNAL.

Signed-off-by: Bhawanpreet Lakha 
Reviewed-by: Harry Wentland 
Acked-by: Leo Li 
Signed-off-by: Alex Deucher

drm/amd/display: Fix Vega10 lightup on S3 resume

2018-09-27T15:03:12+00:00

[Why]
There have been a few reports of Vega10 display remaining blank
after S3 resume. The regression is caused by workaround for mode
change on Vega10 - skip set_bandwidth if stream count is 0.
As a result we skipped dispclk reset on suspend, thus on resume
we may skip the clock update assuming it hasn't been changed.
On some systems it causes display blank or 'out of range'.

[How]
Revert "drm/amd/display: Fix Vega10 black screen after mode change"
Verified that it hadn't cause mode change regression.

Signed-off-by: Roman Li 
Reviewed-by: Sun peng Li 
Acked-by: Leo Li 
Signed-off-by: Alex Deucher

drm/amd/display: Fix memory leak caused by missed dc_sink_release

2018-08-27T19:53:37+00:00

[Why]
There is currently an intermittent hang from a memory leak in
DTN stress testing. It is caused by unfreed memory during driver
disable.

[How]
Do a dc_sink_release in the case that skips it incorrectly.

Signed-off-by: SivapiriyanKumarasamy 
Reviewed-by: Aric Cyr 
Acked-by: Leo Li 
Signed-off-by: Alex Deucher