linux-stable.git/drivers/gpu/drm/amd/display/amdgpu_dm, branch linux-4.18.y Linux kernel stable tree drm/amd/display: Signal hw_done() after waiting for flip_done() 2018-11-04T13:51:47+00:00 Shirish S shirish.s@amd.com 2018-09-24T13:31:47+00:00 43cb734246c0d7fdea02aab272ae550f76a6949f [ Upstream commit 987bf116445db5d63a5c2ed94c4479687d9c9973 ] 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 <shirish.s@amd.com> Signed-off-by: Leo Li <sunpeng.li@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org> 
[ Upstream commit 987bf116445db5d63a5c2ed94c4479687d9c9973 ]

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 <shirish.s@amd.com>
Signed-off-by: Leo Li <sunpeng.li@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
drm/amd/display: Use requested HDMI aspect ratio 2018-09-15T07:46:58+00:00 Leo (Sunpeng) Li sunpeng.li@amd.com 2018-07-19T12:22:16+00:00 823057fc0a2c088bcbecfaac933a131e93385b75 commit e11d41472a50742c16d53c968e143fb498fa482f upstream. [Why] The DRM mode's HDMI picture aspect ratio field was never saved in dc_stream's timing struct. This causes us to mistake a new stream to have the same timings as the old, even though the user has requested a different aspect ratio. [How] Save DRM's aspect ratio field within dc_stream's timing struct. Bug: https://bugs.freedesktop.org/show_bug.cgi?id=107153 Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com> Reviewed-by: Mikita Lipski <Mikita.Lipski@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> 
commit e11d41472a50742c16d53c968e143fb498fa482f upstream.

[Why]
The DRM mode's HDMI picture aspect ratio field was never saved in
dc_stream's timing struct. This causes us to mistake a new stream to
have the same timings as the old, even though the user has requested a
different aspect ratio.

[How]
Save DRM's aspect ratio field within dc_stream's timing struct.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=107153
Signed-off-by: Leo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: Mikita Lipski <Mikita.Lipski@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drm/amd/display: Read back max backlight value at boot 2018-09-15T07:46:54+00:00 David Francis David.Francis@amd.com 2018-07-18T20:03:30+00:00 ac5395827d489b1a9484b2474cb1dba72fb9bd98 [ Upstream commit 53a53f8687faf492df2644d8c18ff0217fc18730 ] [Why] If there is no program explicitly setting the backlight brightness (for example, during a minimal install of linux), the hardware defaults to maximum brightness but the backlight_device defaults to 0 value. Thus, settings displays the wrong brightness value. [How] When creating the backlight device, set brightness to max Signed-off-by: David Francis <David.Francis@amd.com> Reviewed-by: Harry Wentland <Harry.Wentland@amd.com> Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <alexander.levin@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> 
[ Upstream commit 53a53f8687faf492df2644d8c18ff0217fc18730 ]

[Why]
If there is no program explicitly setting the backlight
brightness (for example, during a minimal install of linux), the
hardware defaults to maximum brightness but the backlight_device
defaults to 0 value.  Thus, settings displays the wrong brightness
value.

[How]
When creating the backlight device, set brightness to max

Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drm/amd/display: Guard against null crtc in CRC IRQ 2018-09-15T07:46:50+00:00 Nicholas Kazlauskas nicholas.kazlauskas@amd.com 2018-08-03T14:40:20+00:00 243b6c6a90cc7332a41d889a52b18297d193ea26 [ Upstream commit dddc0557e3a02499ce336b1e2e67f5afaecccc80 ] [Why] A null pointer deference can occur if crtc is null in amdgpu_dm_crtc_handle_crc_irq. This can happen if get_crtc_by_otg_inst returns NULL during dm_crtc_high_irq, leading to a hang in some IGT test cases. [How] Check that CRTC is non-null before accessing its fields. Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Reviewed-by: Sun peng Li <Sunpeng.Li@amd.com> Acked-by: Leo Li <sunpeng.li@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <alexander.levin@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> 
[ Upstream commit dddc0557e3a02499ce336b1e2e67f5afaecccc80 ]

[Why]

A null pointer deference can occur if crtc is null in
amdgpu_dm_crtc_handle_crc_irq. This can happen if get_crtc_by_otg_inst
returns NULL during dm_crtc_high_irq, leading to a hang in some IGT
test cases.

[How]

Check that CRTC is non-null before accessing its fields.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Sun peng Li <Sunpeng.Li@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drm/amd/display: Convert 10kHz clks from PPLib into kHz for Vega 2018-07-12T14:48:57+00:00 Harry Wentland harry.wentland@amd.com 2018-07-12T14:23:21+00:00 d89d01f2232b90e8024ae07ff5ab213521875448 The driver is expecting clock frequency in kHz, while SMU returns the values in 10kHz, which causes the bandwidth validation to fail 4.18 has the faulty clock assignment in pp_to_dc_clock_levels_with_latency only, which is only used by Vega. Make sure we multiply these values by 10 here, as we do for other ASICs as powerplay assigned them wrong. 4.19 has the proper fix in powerplay. v2: Add Fixes tag v3: Fixes -> Bugzilla, with simplified link Bugzilla: https://bugs.freedesktop.org/107082 Signed-off-by: Mikita Lipski <mikita.lipski@amd.com> Signed-off-by: Harry Wentland <harry.wentland@amd.com> Acked-by: Michel Dänzer <michel.daenzer@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 
The driver is expecting clock frequency in kHz, while SMU returns
the values in 10kHz, which causes the bandwidth validation to fail

4.18 has the faulty clock assignment in pp_to_dc_clock_levels_with_latency
only, which is only used by Vega. Make sure we multiply these values
by 10 here, as we do for other ASICs as powerplay assigned them
wrong. 4.19 has the proper fix in powerplay.

v2: Add Fixes tag
v3: Fixes -> Bugzilla, with simplified link

Bugzilla: https://bugs.freedesktop.org/107082

Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Acked-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Revert "drm/amd/display: Don't return ddc result and read_bytes in same return value" 2018-07-10T19:35:46+00:00 Alex Deucher alexander.deucher@amd.com 2018-07-10T17:56:45+00:00 5292221d6ddfed75e5b46cd42237a677094b99f3 This reverts commit 018d82e5f02ef3583411bcaa4e00c69786f46f19. This breaks DDC in certain cases. Revert for 4.18 and previous kernels. For 4.19, this is fixed with the following more extensive patches: drm/amd/display: Serialize is_dp_sink_present drm/amd/display: Break out function to simply read aux reply drm/amd/display: Return aux replies directly to DRM drm/amd/display: Right shift AUX reply value sooner than later drm/amd/display: Read AUX channel even if only status byte is returned Link: https://lists.freedesktop.org/archives/amd-gfx/2018-July/023788.html Acked-by: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org 
This reverts commit 018d82e5f02ef3583411bcaa4e00c69786f46f19.

This breaks DDC in certain cases.  Revert for 4.18 and previous kernels.
For 4.19, this is fixed with the following more extensive patches:
drm/amd/display: Serialize is_dp_sink_present
drm/amd/display: Break out function to simply read aux reply
drm/amd/display: Return aux replies directly to DRM
drm/amd/display: Right shift AUX reply value sooner than later
drm/amd/display: Read AUX channel even if only status byte is returned

Link: https://lists.freedesktop.org/archives/amd-gfx/2018-July/023788.html
Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
drm/amd/display: add a check for display depth validity 2018-07-05T15:19:53+00:00 Mikita Lipski mikita.lipski@amd.com 2018-07-04T21:27:56+00:00 413ff0b942481c7ac2e800abbbac5af318a65e61 [why] HDMI 2.0 fails to validate 4K@60 timing with 10 bpc [how] Adding a helper function that would verify if the display depth assigned would pass a bandwidth validation. Drop the display depth by one level till calculated pixel clk is lower than maximum TMDS clk. Bugzilla: https://bugs.freedesktop.org/106959 Tested-by: Mike Lothian <mike@fireburn.co.uk> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Mikita Lipski <mikita.lipski@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 
[why]
HDMI 2.0 fails to validate 4K@60 timing with 10 bpc
[how]
Adding a helper function that would verify if the display depth
assigned would pass a bandwidth validation.
Drop the display depth by one level till calculated pixel clk
is lower than maximum TMDS clk.

Bugzilla: https://bugs.freedesktop.org/106959

Tested-by: Mike Lothian <mike@fireburn.co.uk>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: adding ycbcr420 pixel encoding for hdmi 2018-07-05T15:19:18+00:00 Mikita Lipski mikita.lipski@amd.com 2018-07-04T21:26:26+00:00 a6311be8f062db5fe93a08a3722fae53a58f2499 [why] HDMI EDID's VSDB contains spectial timings for specifically YCbCr 4:2:0 colour space. In those cases we need to verify if the mode provided is one of the special ones has to use YCbCr 4:2:0 pixel encoding for display info. [how] Verify if the mode is using specific ycbcr420 colour space with the help of DRM helper function and assign the mode to use ycbcr420 pixel encoding. Tested-by: Mike Lothian <mike@fireburn.co.uk> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Mikita Lipski <mikita.lipski@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 
[why]
HDMI EDID's VSDB contains spectial timings for specifically
YCbCr 4:2:0 colour space. In those cases we need to verify
if the mode provided is one of the special ones has to use
YCbCr 4:2:0 pixel encoding for display info.
[how]
Verify if the mode is using specific ycbcr420 colour space with
the help of DRM helper function and assign the mode to use
ycbcr420 pixel encoding.

Tested-by: Mike Lothian <mike@fireburn.co.uk>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drm/amd/display: release spinlock before committing updates to stream 2018-06-27T19:35:53+00:00 Shirish S shirish.s@amd.com 2018-06-26T04:02:39+00:00 4de9f38bb2cce3a4821ffb8a83d6b08f6e37d905 Currently, amdgpu_do_flip() spinlocks crtc->dev->event_lock and releases it only after committing updates to the stream. dc_commit_updates_for_stream() should be moved out of spinlock for the below reasons: 1. event_lock is supposed to protect access to acrct->pflip_status _only_ 2. dc_commit_updates_for_stream() has potential sleep's and also its not appropriate to be in an atomic state for such long sequences of code. Signed-off-by: Shirish S <shirish.s@amd.com> Suggested-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com> Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org 
Currently, amdgpu_do_flip() spinlocks crtc->dev->event_lock and
releases it only after committing updates to the stream.

dc_commit_updates_for_stream() should be moved out of
spinlock for the below reasons:

1. event_lock is supposed to protect access to acrct->pflip_status _only_
2. dc_commit_updates_for_stream() has potential sleep's
   and also its not appropriate to be  in an atomic state
   for such long sequences of code.

Signed-off-by: Shirish S <shirish.s@amd.com>
Suggested-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
drm/amdgpu: Count disabled CRTCs in commit tail earlier 2018-06-22T19:55:25+00:00 Lyude Paul lyude@redhat.com 2018-06-21T20:48:26+00:00 fe2a19652918a5247418aed48a247414a5e45fe2 This fixes a regression I accidentally reduced that was picked up by kasan, where we were checking the CRTC atomic states after DRM's helpers had already freed them. Example: ================================================================== BUG: KASAN: use-after-free in amdgpu_dm_atomic_commit_tail.cold.50+0x13d/0x15a [amdgpu] Read of size 1 at addr ffff8803a697b071 by task kworker/u16:0/7 CPU: 7 PID: 7 Comm: kworker/u16:0 Tainted: G O 4.18.0-rc1Lyude-Upstream+ #1 Hardware name: HP HP ZBook 15 G4/8275, BIOS P70 Ver. 01.21 05/02/2018 Workqueue: events_unbound commit_work [drm_kms_helper] Call Trace: dump_stack+0xc1/0x169 ? dump_stack_print_info.cold.1+0x42/0x42 ? kmsg_dump_rewind_nolock+0xd9/0xd9 ? printk+0x9f/0xc5 ? amdgpu_dm_atomic_commit_tail.cold.50+0x13d/0x15a [amdgpu] print_address_description+0x6c/0x23c ? amdgpu_dm_atomic_commit_tail.cold.50+0x13d/0x15a [amdgpu] kasan_report.cold.6+0x241/0x2fd amdgpu_dm_atomic_commit_tail.cold.50+0x13d/0x15a [amdgpu] ? commit_planes_to_stream.constprop.45+0x13b0/0x13b0 [amdgpu] ? cpu_load_update_active+0x290/0x290 ? finish_task_switch+0x2bd/0x840 ? __switch_to_asm+0x34/0x70 ? read_word_at_a_time+0xe/0x20 ? strscpy+0x14b/0x460 ? drm_atomic_helper_wait_for_dependencies+0x47d/0x7e0 [drm_kms_helper] commit_tail+0x96/0xe0 [drm_kms_helper] process_one_work+0x88a/0x1360 ? create_worker+0x540/0x540 ? __sched_text_start+0x8/0x8 ? move_queued_task+0x760/0x760 ? call_rcu_sched+0x20/0x20 ? vsnprintf+0xcda/0x1350 ? wait_woken+0x1c0/0x1c0 ? mutex_unlock+0x1d/0x40 ? init_timer_key+0x190/0x230 ? schedule+0xea/0x390 ? __schedule+0x1ea0/0x1ea0 ? need_to_create_worker+0xe4/0x210 ? init_worker_pool+0x700/0x700 ? try_to_del_timer_sync+0xbf/0x110 ? del_timer+0x120/0x120 ? __mutex_lock_slowpath+0x10/0x10 worker_thread+0x196/0x11f0 ? flush_rcu_work+0x50/0x50 ? __switch_to_asm+0x34/0x70 ? __switch_to_asm+0x34/0x70 ? __switch_to_asm+0x40/0x70 ? __switch_to_asm+0x34/0x70 ? __switch_to_asm+0x40/0x70 ? __switch_to_asm+0x34/0x70 ? __switch_to_asm+0x40/0x70 ? __schedule+0x7d6/0x1ea0 ? migrate_swap_stop+0x850/0x880 ? __sched_text_start+0x8/0x8 ? save_stack+0x8c/0xb0 ? kasan_kmalloc+0xbf/0xe0 ? kmem_cache_alloc_trace+0xe4/0x190 ? kthread+0x98/0x390 ? ret_from_fork+0x35/0x40 ? ret_from_fork+0x35/0x40 ? deactivate_slab.isra.67+0x3c4/0x5c0 ? kthread+0x98/0x390 ? kthread+0x98/0x390 ? set_track+0x76/0x120 ? schedule+0xea/0x390 ? __schedule+0x1ea0/0x1ea0 ? wait_woken+0x1c0/0x1c0 ? kasan_unpoison_shadow+0x30/0x40 ? parse_args.cold.15+0x17a/0x17a ? flush_rcu_work+0x50/0x50 kthread+0x2d4/0x390 ? kthread_create_worker_on_cpu+0xc0/0xc0 ret_from_fork+0x35/0x40 Allocated by task 1124: kasan_kmalloc+0xbf/0xe0 kmem_cache_alloc_trace+0xe4/0x190 dm_crtc_duplicate_state+0x78/0x130 [amdgpu] drm_atomic_get_crtc_state+0x147/0x410 [drm] page_flip_common+0x57/0x230 [drm_kms_helper] drm_atomic_helper_page_flip+0xa6/0x110 [drm_kms_helper] drm_mode_page_flip_ioctl+0xc4b/0x10a0 [drm] drm_ioctl_kernel+0x1d4/0x260 [drm] drm_ioctl+0x433/0x920 [drm] amdgpu_drm_ioctl+0x11d/0x290 [amdgpu] do_vfs_ioctl+0x1a1/0x13d0 ksys_ioctl+0x60/0x90 __x64_sys_ioctl+0x6f/0xb0 do_syscall_64+0x147/0x440 entry_SYSCALL_64_after_hwframe+0x44/0xa9 Freed by task 1124: __kasan_slab_free+0x12e/0x180 kfree+0x92/0x1a0 drm_atomic_state_default_clear+0x315/0xc40 [drm] __drm_atomic_state_free+0x35/0xd0 [drm] drm_atomic_helper_update_plane+0xac/0x350 [drm_kms_helper] __setplane_internal+0x2d6/0x840 [drm] drm_mode_cursor_universal+0x41e/0xbe0 [drm] drm_mode_cursor_common+0x49f/0x880 [drm] drm_mode_cursor_ioctl+0xd8/0x130 [drm] drm_ioctl_kernel+0x1d4/0x260 [drm] drm_ioctl+0x433/0x920 [drm] amdgpu_drm_ioctl+0x11d/0x290 [amdgpu] do_vfs_ioctl+0x1a1/0x13d0 ksys_ioctl+0x60/0x90 __x64_sys_ioctl+0x6f/0xb0 do_syscall_64+0x147/0x440 entry_SYSCALL_64_after_hwframe+0x44/0xa9 The buggy address belongs to the object at ffff8803a697b068 which belongs to the cache kmalloc-1024 of size 1024 The buggy address is located 9 bytes inside of 1024-byte region [ffff8803a697b068, ffff8803a697b468) The buggy address belongs to the page: page:ffffea000e9a5e00 count:1 mapcount:0 mapping:ffff88041e00efc0 index:0x0 compound_mapcount: 0 flags: 0x8000000000008100(slab|head) raw: 8000000000008100 ffffea000ecbc208 ffff88041e000c70 ffff88041e00efc0 raw: 0000000000000000 0000000000170017 00000001ffffffff 0000000000000000 page dumped because: kasan: bad access detected Memory state around the buggy address: ffff8803a697af00: fb fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc ffff8803a697af80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc >ffff8803a697b000: fc fc fc fc fc fc fc fc fc fc fc fc fc fb fb fb ^ ffff8803a697b080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ffff8803a697b100: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ================================================================== So, we fix this by counting the number of CRTCs this atomic commit disabled early on in the function before their atomic states have been freed, then use that count later to do the appropriate number of RPM puts at the end of the function. Acked-by: Michel Dänzer <michel.daenzer@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Cc: stable@vger.kernel.org Fixes: 97028037a38ae ("drm/amdgpu: Grab/put runtime PM references in atomic_commit_tail()") Signed-off-by: Lyude Paul <lyude@redhat.com> Cc: Michel Dänzer <michel@daenzer.net> Reported-by: Michel Dänzer <michel@daenzer.net> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 
This fixes a regression I accidentally reduced that was picked up by
kasan, where we were checking the CRTC atomic states after DRM's helpers
had already freed them. Example:

==================================================================
BUG: KASAN: use-after-free in amdgpu_dm_atomic_commit_tail.cold.50+0x13d/0x15a [amdgpu]
Read of size 1 at addr ffff8803a697b071 by task kworker/u16:0/7

CPU: 7 PID: 7 Comm: kworker/u16:0 Tainted: G           O      4.18.0-rc1Lyude-Upstream+ #1
Hardware name: HP HP ZBook 15 G4/8275, BIOS P70 Ver. 01.21 05/02/2018
Workqueue: events_unbound commit_work [drm_kms_helper]
Call Trace:
 dump_stack+0xc1/0x169
 ? dump_stack_print_info.cold.1+0x42/0x42
 ? kmsg_dump_rewind_nolock+0xd9/0xd9
 ? printk+0x9f/0xc5
 ? amdgpu_dm_atomic_commit_tail.cold.50+0x13d/0x15a [amdgpu]
 print_address_description+0x6c/0x23c
 ? amdgpu_dm_atomic_commit_tail.cold.50+0x13d/0x15a [amdgpu]
 kasan_report.cold.6+0x241/0x2fd
 amdgpu_dm_atomic_commit_tail.cold.50+0x13d/0x15a [amdgpu]
 ? commit_planes_to_stream.constprop.45+0x13b0/0x13b0 [amdgpu]
 ? cpu_load_update_active+0x290/0x290
 ? finish_task_switch+0x2bd/0x840
 ? __switch_to_asm+0x34/0x70
 ? read_word_at_a_time+0xe/0x20
 ? strscpy+0x14b/0x460
 ? drm_atomic_helper_wait_for_dependencies+0x47d/0x7e0 [drm_kms_helper]
 commit_tail+0x96/0xe0 [drm_kms_helper]
 process_one_work+0x88a/0x1360
 ? create_worker+0x540/0x540
 ? __sched_text_start+0x8/0x8
 ? move_queued_task+0x760/0x760
 ? call_rcu_sched+0x20/0x20
 ? vsnprintf+0xcda/0x1350
 ? wait_woken+0x1c0/0x1c0
 ? mutex_unlock+0x1d/0x40
 ? init_timer_key+0x190/0x230
 ? schedule+0xea/0x390
 ? __schedule+0x1ea0/0x1ea0
 ? need_to_create_worker+0xe4/0x210
 ? init_worker_pool+0x700/0x700
 ? try_to_del_timer_sync+0xbf/0x110
 ? del_timer+0x120/0x120
 ? __mutex_lock_slowpath+0x10/0x10
 worker_thread+0x196/0x11f0
 ? flush_rcu_work+0x50/0x50
 ? __switch_to_asm+0x34/0x70
 ? __switch_to_asm+0x34/0x70
 ? __switch_to_asm+0x40/0x70
 ? __switch_to_asm+0x34/0x70
 ? __switch_to_asm+0x40/0x70
 ? __switch_to_asm+0x34/0x70
 ? __switch_to_asm+0x40/0x70
 ? __schedule+0x7d6/0x1ea0
 ? migrate_swap_stop+0x850/0x880
 ? __sched_text_start+0x8/0x8
 ? save_stack+0x8c/0xb0
 ? kasan_kmalloc+0xbf/0xe0
 ? kmem_cache_alloc_trace+0xe4/0x190
 ? kthread+0x98/0x390
 ? ret_from_fork+0x35/0x40
 ? ret_from_fork+0x35/0x40
 ? deactivate_slab.isra.67+0x3c4/0x5c0
 ? kthread+0x98/0x390
 ? kthread+0x98/0x390
 ? set_track+0x76/0x120
 ? schedule+0xea/0x390
 ? __schedule+0x1ea0/0x1ea0
 ? wait_woken+0x1c0/0x1c0
 ? kasan_unpoison_shadow+0x30/0x40
 ? parse_args.cold.15+0x17a/0x17a
 ? flush_rcu_work+0x50/0x50
 kthread+0x2d4/0x390
 ? kthread_create_worker_on_cpu+0xc0/0xc0
 ret_from_fork+0x35/0x40

Allocated by task 1124:
 kasan_kmalloc+0xbf/0xe0
 kmem_cache_alloc_trace+0xe4/0x190
 dm_crtc_duplicate_state+0x78/0x130 [amdgpu]
 drm_atomic_get_crtc_state+0x147/0x410 [drm]
 page_flip_common+0x57/0x230 [drm_kms_helper]
 drm_atomic_helper_page_flip+0xa6/0x110 [drm_kms_helper]
 drm_mode_page_flip_ioctl+0xc4b/0x10a0 [drm]
 drm_ioctl_kernel+0x1d4/0x260 [drm]
 drm_ioctl+0x433/0x920 [drm]
 amdgpu_drm_ioctl+0x11d/0x290 [amdgpu]
 do_vfs_ioctl+0x1a1/0x13d0
 ksys_ioctl+0x60/0x90
 __x64_sys_ioctl+0x6f/0xb0
 do_syscall_64+0x147/0x440
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

Freed by task 1124:
 __kasan_slab_free+0x12e/0x180
 kfree+0x92/0x1a0
 drm_atomic_state_default_clear+0x315/0xc40 [drm]
 __drm_atomic_state_free+0x35/0xd0 [drm]
 drm_atomic_helper_update_plane+0xac/0x350 [drm_kms_helper]
 __setplane_internal+0x2d6/0x840 [drm]
 drm_mode_cursor_universal+0x41e/0xbe0 [drm]
 drm_mode_cursor_common+0x49f/0x880 [drm]
 drm_mode_cursor_ioctl+0xd8/0x130 [drm]
 drm_ioctl_kernel+0x1d4/0x260 [drm]
 drm_ioctl+0x433/0x920 [drm]
 amdgpu_drm_ioctl+0x11d/0x290 [amdgpu]
 do_vfs_ioctl+0x1a1/0x13d0
 ksys_ioctl+0x60/0x90
 __x64_sys_ioctl+0x6f/0xb0
 do_syscall_64+0x147/0x440
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

The buggy address belongs to the object at ffff8803a697b068
 which belongs to the cache kmalloc-1024 of size 1024
The buggy address is located 9 bytes inside of
 1024-byte region [ffff8803a697b068, ffff8803a697b468)
The buggy address belongs to the page:
page:ffffea000e9a5e00 count:1 mapcount:0 mapping:ffff88041e00efc0 index:0x0 compound_mapcount: 0
flags: 0x8000000000008100(slab|head)
raw: 8000000000008100 ffffea000ecbc208 ffff88041e000c70 ffff88041e00efc0
raw: 0000000000000000 0000000000170017 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected

Memory state around the buggy address:
 ffff8803a697af00: fb fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
 ffff8803a697af80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
>ffff8803a697b000: fc fc fc fc fc fc fc fc fc fc fc fc fc fb fb fb
                                                             ^
 ffff8803a697b080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
 ffff8803a697b100: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================

So, we fix this by counting the number of CRTCs this atomic commit disabled
early on in the function before their atomic states have been freed, then use
that count later to do the appropriate number of RPM puts at the end of the
function.

Acked-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Cc: stable@vger.kernel.org
Fixes: 97028037a38ae ("drm/amdgpu: Grab/put runtime PM references in atomic_commit_tail()")
Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: Michel Dänzer <michel@daenzer.net>
Reported-by: Michel Dänzer <michel@daenzer.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>