From d77ef138ff572409ab93d492e5e6c826ee6fb21d Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Wed, 15 Aug 2018 15:00:11 -0400
Subject: drm/nouveau/drm/nouveau: Fix bogus drm_kms_helper_poll_enable()
 placement

Turns out this part is my fault for not noticing when reviewing
9a2eba337cace ("drm/nouveau: Fix drm poll_helper handling"). Currently
we call drm_kms_helper_poll_enable() from nouveau_display_hpd_work().
This makes basically no sense however, because that means we're calling
drm_kms_helper_poll_enable() every time we schedule the hotplug
detection work. This is also against the advice mentioned in
drm_kms_helper_poll_enable()'s documentation:

 Note that calls to enable and disable polling must be strictly ordered,
 which is automatically the case when they're only call from
 suspend/resume callbacks.

Of course, hotplugs can't really be ordered. They could even happen
immediately after we called drm_kms_helper_poll_disable() in
nouveau_display_fini(), which can lead to all sorts of issues.

Additionally; enabling polling /after/ we call
drm_helper_hpd_irq_event() could also mean that we'd miss a hotplug
event anyway, since drm_helper_hpd_irq_event() wouldn't bother trying to
probe connectors so long as polling is disabled.

So; simply move this back into nouveau_display_init() again. The race
condition that both of these patches attempted to work around has
already been fixed properly in

  d61a5c106351 ("drm/nouveau: Fix deadlock on runtime suspend")

Fixes: 9a2eba337cace ("drm/nouveau: Fix drm poll_helper handling")
Signed-off-by: Lyude Paul <lyude@redhat.com>
Acked-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Cc: Lukas Wunner <lukas@wunner.de>
Cc: Peter Ujfalusi <peter.ujfalusi@ti.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_display.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c b/drivers/gpu/drm/nouveau/nouveau_display.c
index 139368b31916..7db01ea7fd41 100644
--- a/drivers/gpu/drm/nouveau/nouveau_display.c
+++ b/drivers/gpu/drm/nouveau/nouveau_display.c
@@ -355,8 +355,6 @@ nouveau_display_hpd_work(struct work_struct *work)
 	pm_runtime_get_sync(drm->dev->dev);
 
 	drm_helper_hpd_irq_event(drm->dev);
-	/* enable polling for external displays */
-	drm_kms_helper_poll_enable(drm->dev);
 
 	pm_runtime_mark_last_busy(drm->dev->dev);
 	pm_runtime_put_sync(drm->dev->dev);
@@ -411,6 +409,11 @@ nouveau_display_init(struct drm_device *dev)
 	if (ret)
 		return ret;
 
+	/* enable connector detection and polling for connectors without HPD
+	 * support
+	 */
+	drm_kms_helper_poll_enable(dev);
+
 	/* enable hotplug interrupts */
 	drm_connector_list_iter_begin(dev, &conn_iter);
 	nouveau_for_each_non_mst_connector_iter(connector, &conn_iter) {
-- 
cgit v1.2.3


From 611ce855420a6e8b9ff47af5f47431d52c7709f8 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Wed, 15 Aug 2018 15:00:12 -0400
Subject: drm/nouveau: Remove duplicate poll_enable() in
 pmops_runtime_suspend()

Since actual hotplug notifications don't get disabled until
nouveau_display_fini() is called, all this will do is cause any hotplugs
that happen between this drm_kms_helper_poll_disable() call and the
actual hotplug disablement to potentially be dropped if ACPI isn't
around to help us.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Acked-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Cc: stable@vger.kernel.org
Cc: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_drm.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c
index c7ec86d6c3c9..5fdc1fbe2ee5 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_drm.c
@@ -835,7 +835,6 @@ nouveau_pmops_runtime_suspend(struct device *dev)
 		return -EBUSY;
 	}
 
-	drm_kms_helper_poll_disable(drm_dev);
 	nouveau_switcheroo_optimus_dsm();
 	ret = nouveau_do_suspend(drm_dev, true);
 	pci_save_state(pdev);
-- 
cgit v1.2.3


From 7fec8f5379fb6eddabc0aaef6d2304c366808f97 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Wed, 15 Aug 2018 15:00:13 -0400
Subject: drm/nouveau/drm/nouveau: Fix deadlock with fb_helper with async RPM
 requests

Currently, nouveau uses the generic drm_fb_helper_output_poll_changed()
function provided by DRM as it's output_poll_changed callback.
Unfortunately however, this function doesn't grab runtime PM references
early enough and even if it did-we can't block waiting for the device to
resume in output_poll_changed() since it's very likely that we'll need
to grab the fb_helper lock at some point during the runtime resume
process. This currently results in deadlocking like so:

[  246.669625] INFO: task kworker/4:0:37 blocked for more than 120 seconds.
[  246.673398]       Not tainted 4.18.0-rc5Lyude-Test+ #2
[  246.675271] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[  246.676527] kworker/4:0     D    0    37      2 0x80000000
[  246.677580] Workqueue: events output_poll_execute [drm_kms_helper]
[  246.678704] Call Trace:
[  246.679753]  __schedule+0x322/0xaf0
[  246.680916]  schedule+0x33/0x90
[  246.681924]  schedule_preempt_disabled+0x15/0x20
[  246.683023]  __mutex_lock+0x569/0x9a0
[  246.684035]  ? kobject_uevent_env+0x117/0x7b0
[  246.685132]  ? drm_fb_helper_hotplug_event.part.28+0x20/0xb0 [drm_kms_helper]
[  246.686179]  mutex_lock_nested+0x1b/0x20
[  246.687278]  ? mutex_lock_nested+0x1b/0x20
[  246.688307]  drm_fb_helper_hotplug_event.part.28+0x20/0xb0 [drm_kms_helper]
[  246.689420]  drm_fb_helper_output_poll_changed+0x23/0x30 [drm_kms_helper]
[  246.690462]  drm_kms_helper_hotplug_event+0x2a/0x30 [drm_kms_helper]
[  246.691570]  output_poll_execute+0x198/0x1c0 [drm_kms_helper]
[  246.692611]  process_one_work+0x231/0x620
[  246.693725]  worker_thread+0x214/0x3a0
[  246.694756]  kthread+0x12b/0x150
[  246.695856]  ? wq_pool_ids_show+0x140/0x140
[  246.696888]  ? kthread_create_worker_on_cpu+0x70/0x70
[  246.697998]  ret_from_fork+0x3a/0x50
[  246.699034] INFO: task kworker/0:1:60 blocked for more than 120 seconds.
[  246.700153]       Not tainted 4.18.0-rc5Lyude-Test+ #2
[  246.701182] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[  246.702278] kworker/0:1     D    0    60      2 0x80000000
[  246.703293] Workqueue: pm pm_runtime_work
[  246.704393] Call Trace:
[  246.705403]  __schedule+0x322/0xaf0
[  246.706439]  ? wait_for_completion+0x104/0x190
[  246.707393]  schedule+0x33/0x90
[  246.708375]  schedule_timeout+0x3a5/0x590
[  246.709289]  ? mark_held_locks+0x58/0x80
[  246.710208]  ? _raw_spin_unlock_irq+0x2c/0x40
[  246.711222]  ? wait_for_completion+0x104/0x190
[  246.712134]  ? trace_hardirqs_on_caller+0xf4/0x190
[  246.713094]  ? wait_for_completion+0x104/0x190
[  246.713964]  wait_for_completion+0x12c/0x190
[  246.714895]  ? wake_up_q+0x80/0x80
[  246.715727]  ? get_work_pool+0x90/0x90
[  246.716649]  flush_work+0x1c9/0x280
[  246.717483]  ? flush_workqueue_prep_pwqs+0x1b0/0x1b0
[  246.718442]  __cancel_work_timer+0x146/0x1d0
[  246.719247]  cancel_delayed_work_sync+0x13/0x20
[  246.720043]  drm_kms_helper_poll_disable+0x1f/0x30 [drm_kms_helper]
[  246.721123]  nouveau_pmops_runtime_suspend+0x3d/0xb0 [nouveau]
[  246.721897]  pci_pm_runtime_suspend+0x6b/0x190
[  246.722825]  ? pci_has_legacy_pm_support+0x70/0x70
[  246.723737]  __rpm_callback+0x7a/0x1d0
[  246.724721]  ? pci_has_legacy_pm_support+0x70/0x70
[  246.725607]  rpm_callback+0x24/0x80
[  246.726553]  ? pci_has_legacy_pm_support+0x70/0x70
[  246.727376]  rpm_suspend+0x142/0x6b0
[  246.728185]  pm_runtime_work+0x97/0xc0
[  246.728938]  process_one_work+0x231/0x620
[  246.729796]  worker_thread+0x44/0x3a0
[  246.730614]  kthread+0x12b/0x150
[  246.731395]  ? wq_pool_ids_show+0x140/0x140
[  246.732202]  ? kthread_create_worker_on_cpu+0x70/0x70
[  246.732878]  ret_from_fork+0x3a/0x50
[  246.733768] INFO: task kworker/4:2:422 blocked for more than 120 seconds.
[  246.734587]       Not tainted 4.18.0-rc5Lyude-Test+ #2
[  246.735393] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[  246.736113] kworker/4:2     D    0   422      2 0x80000080
[  246.736789] Workqueue: events_long drm_dp_mst_link_probe_work [drm_kms_helper]
[  246.737665] Call Trace:
[  246.738490]  __schedule+0x322/0xaf0
[  246.739250]  schedule+0x33/0x90
[  246.739908]  rpm_resume+0x19c/0x850
[  246.740750]  ? finish_wait+0x90/0x90
[  246.741541]  __pm_runtime_resume+0x4e/0x90
[  246.742370]  nv50_disp_atomic_commit+0x31/0x210 [nouveau]
[  246.743124]  drm_atomic_commit+0x4a/0x50 [drm]
[  246.743775]  restore_fbdev_mode_atomic+0x1c8/0x240 [drm_kms_helper]
[  246.744603]  restore_fbdev_mode+0x31/0x140 [drm_kms_helper]
[  246.745373]  drm_fb_helper_restore_fbdev_mode_unlocked+0x54/0xb0 [drm_kms_helper]
[  246.746220]  drm_fb_helper_set_par+0x2d/0x50 [drm_kms_helper]
[  246.746884]  drm_fb_helper_hotplug_event.part.28+0x96/0xb0 [drm_kms_helper]
[  246.747675]  drm_fb_helper_output_poll_changed+0x23/0x30 [drm_kms_helper]
[  246.748544]  drm_kms_helper_hotplug_event+0x2a/0x30 [drm_kms_helper]
[  246.749439]  nv50_mstm_hotplug+0x15/0x20 [nouveau]
[  246.750111]  drm_dp_send_link_address+0x177/0x1c0 [drm_kms_helper]
[  246.750764]  drm_dp_check_and_send_link_address+0xa8/0xd0 [drm_kms_helper]
[  246.751602]  drm_dp_mst_link_probe_work+0x51/0x90 [drm_kms_helper]
[  246.752314]  process_one_work+0x231/0x620
[  246.752979]  worker_thread+0x44/0x3a0
[  246.753838]  kthread+0x12b/0x150
[  246.754619]  ? wq_pool_ids_show+0x140/0x140
[  246.755386]  ? kthread_create_worker_on_cpu+0x70/0x70
[  246.756162]  ret_from_fork+0x3a/0x50
[  246.756847]
           Showing all locks held in the system:
[  246.758261] 3 locks held by kworker/4:0/37:
[  246.759016]  #0: 00000000f8df4d2d ((wq_completion)"events"){+.+.}, at: process_one_work+0x1b3/0x620
[  246.759856]  #1: 00000000e6065461 ((work_completion)(&(&dev->mode_config.output_poll_work)->work)){+.+.}, at: process_one_work+0x1b3/0x620
[  246.760670]  #2: 00000000cb66735f (&helper->lock){+.+.}, at: drm_fb_helper_hotplug_event.part.28+0x20/0xb0 [drm_kms_helper]
[  246.761516] 2 locks held by kworker/0:1/60:
[  246.762274]  #0: 00000000fff6be0f ((wq_completion)"pm"){+.+.}, at: process_one_work+0x1b3/0x620
[  246.762982]  #1: 000000005ab44fb4 ((work_completion)(&dev->power.work)){+.+.}, at: process_one_work+0x1b3/0x620
[  246.763890] 1 lock held by khungtaskd/64:
[  246.764664]  #0: 000000008cb8b5c3 (rcu_read_lock){....}, at: debug_show_all_locks+0x23/0x185
[  246.765588] 5 locks held by kworker/4:2/422:
[  246.766440]  #0: 00000000232f0959 ((wq_completion)"events_long"){+.+.}, at: process_one_work+0x1b3/0x620
[  246.767390]  #1: 00000000bb59b134 ((work_completion)(&mgr->work)){+.+.}, at: process_one_work+0x1b3/0x620
[  246.768154]  #2: 00000000cb66735f (&helper->lock){+.+.}, at: drm_fb_helper_restore_fbdev_mode_unlocked+0x4c/0xb0 [drm_kms_helper]
[  246.768966]  #3: 000000004c8f0b6b (crtc_ww_class_acquire){+.+.}, at: restore_fbdev_mode_atomic+0x4b/0x240 [drm_kms_helper]
[  246.769921]  #4: 000000004c34a296 (crtc_ww_class_mutex){+.+.}, at: drm_modeset_backoff+0x8a/0x1b0 [drm]
[  246.770839] 1 lock held by dmesg/1038:
[  246.771739] 2 locks held by zsh/1172:
[  246.772650]  #0: 00000000836d0438 (&tty->ldisc_sem){++++}, at: ldsem_down_read+0x37/0x40
[  246.773680]  #1: 000000001f4f4d48 (&ldata->atomic_read_lock){+.+.}, at: n_tty_read+0xc1/0x870

[  246.775522] =============================================

After trying dozens of different solutions, I found one very simple one
that should also have the benefit of preventing us from having to fight
locking for the rest of our lives. So, we work around these deadlocks by
deferring all fbcon hotplug events that happen after the runtime suspend
process starts until after the device is resumed again.

Changes since v7:
 - Fixup commit message - Daniel Vetter

Changes since v6:
 - Remove unused nouveau_fbcon_hotplugged_in_suspend() - Ilia

Changes since v5:
 - Come up with the (hopefully final) solution for solving this dumb
   problem, one that is a lot less likely to cause issues with locking in
   the future. This should work around all deadlock conditions with fbcon
   brought up thus far.

Changes since v4:
 - Add nouveau_fbcon_hotplugged_in_suspend() to workaround deadlock
   condition that Lukas described
 - Just move all of this out of drm_fb_helper. It seems that other DRM
   drivers have already figured out other workarounds for this. If other
   drivers do end up needing this in the future, we can just move this
   back into drm_fb_helper again.

Changes since v3:
- Actually check if fb_helper is NULL in both new helpers
- Actually check drm_fbdev_emulation in both new helpers
- Don't fire off a fb_helper hotplug unconditionally; only do it if
  the following conditions are true (as otherwise, calling this in the
  wrong spot will cause Bad Things to happen):
  - fb_helper hotplug handling was actually inhibited previously
  - fb_helper actually has a delayed hotplug pending
  - fb_helper is actually bound
  - fb_helper is actually initialized
- Add __must_check to drm_fb_helper_suspend_hotplug(). There's no
  situation where a driver would actually want to use this without
  checking the return value, so enforce that
- Rewrite and clarify the documentation for both helpers.
- Make sure to return true in the drm_fb_helper_suspend_hotplug() stub
  that's provided in drm_fb_helper.h when CONFIG_DRM_FBDEV_EMULATION
  isn't enabled
- Actually grab the toplevel fb_helper lock in
  drm_fb_helper_resume_hotplug(), since it's possible other activity
  (such as a hotplug) could be going on at the same time the driver
  calls drm_fb_helper_resume_hotplug(). We need this to check whether or
  not drm_fb_helper_hotplug_event() needs to be called anyway

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Cc: stable@vger.kernel.org
Cc: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/dispnv50/disp.c   |  2 +-
 drivers/gpu/drm/nouveau/nouveau_display.c |  2 +-
 drivers/gpu/drm/nouveau/nouveau_fbcon.c   | 57 +++++++++++++++++++++++++++++++
 drivers/gpu/drm/nouveau/nouveau_fbcon.h   |  5 +++
 4 files changed, 64 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c b/drivers/gpu/drm/nouveau/dispnv50/disp.c
index 8412119bd940..aec6ee1ff4e0 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@@ -2074,7 +2074,7 @@ nv50_disp_atomic_state_alloc(struct drm_device *dev)
 static const struct drm_mode_config_funcs
 nv50_disp_func = {
 	.fb_create = nouveau_user_framebuffer_create,
-	.output_poll_changed = drm_fb_helper_output_poll_changed,
+	.output_poll_changed = nouveau_fbcon_output_poll_changed,
 	.atomic_check = nv50_disp_atomic_check,
 	.atomic_commit = nv50_disp_atomic_commit,
 	.atomic_state_alloc = nv50_disp_atomic_state_alloc,
diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c b/drivers/gpu/drm/nouveau/nouveau_display.c
index 7db01ea7fd41..42e7c35e3fba 100644
--- a/drivers/gpu/drm/nouveau/nouveau_display.c
+++ b/drivers/gpu/drm/nouveau/nouveau_display.c
@@ -293,7 +293,7 @@ nouveau_user_framebuffer_create(struct drm_device *dev,
 
 static const struct drm_mode_config_funcs nouveau_mode_config_funcs = {
 	.fb_create = nouveau_user_framebuffer_create,
-	.output_poll_changed = drm_fb_helper_output_poll_changed,
+	.output_poll_changed = nouveau_fbcon_output_poll_changed,
 };
 
 
diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.c b/drivers/gpu/drm/nouveau/nouveau_fbcon.c
index 844498c4267c..0f64c0a1d4b3 100644
--- a/drivers/gpu/drm/nouveau/nouveau_fbcon.c
+++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.c
@@ -466,6 +466,7 @@ nouveau_fbcon_set_suspend_work(struct work_struct *work)
 	console_unlock();
 
 	if (state == FBINFO_STATE_RUNNING) {
+		nouveau_fbcon_hotplug_resume(drm->fbcon);
 		pm_runtime_mark_last_busy(drm->dev->dev);
 		pm_runtime_put_sync(drm->dev->dev);
 	}
@@ -487,6 +488,61 @@ nouveau_fbcon_set_suspend(struct drm_device *dev, int state)
 	schedule_work(&drm->fbcon_work);
 }
 
+void
+nouveau_fbcon_output_poll_changed(struct drm_device *dev)
+{
+	struct nouveau_drm *drm = nouveau_drm(dev);
+	struct nouveau_fbdev *fbcon = drm->fbcon;
+	int ret;
+
+	if (!fbcon)
+		return;
+
+	mutex_lock(&fbcon->hotplug_lock);
+
+	ret = pm_runtime_get(dev->dev);
+	if (ret == 1 || ret == -EACCES) {
+		drm_fb_helper_hotplug_event(&fbcon->helper);
+
+		pm_runtime_mark_last_busy(dev->dev);
+		pm_runtime_put_autosuspend(dev->dev);
+	} else if (ret == 0) {
+		/* If the GPU was already in the process of suspending before
+		 * this event happened, then we can't block here as we'll
+		 * deadlock the runtime pmops since they wait for us to
+		 * finish. So, just defer this event for when we runtime
+		 * resume again. It will be handled by fbcon_work.
+		 */
+		NV_DEBUG(drm, "fbcon HPD event deferred until runtime resume\n");
+		fbcon->hotplug_waiting = true;
+		pm_runtime_put_noidle(drm->dev->dev);
+	} else {
+		DRM_WARN("fbcon HPD event lost due to RPM failure: %d\n",
+			 ret);
+	}
+
+	mutex_unlock(&fbcon->hotplug_lock);
+}
+
+void
+nouveau_fbcon_hotplug_resume(struct nouveau_fbdev *fbcon)
+{
+	struct nouveau_drm *drm;
+
+	if (!fbcon)
+		return;
+	drm = nouveau_drm(fbcon->helper.dev);
+
+	mutex_lock(&fbcon->hotplug_lock);
+	if (fbcon->hotplug_waiting) {
+		fbcon->hotplug_waiting = false;
+
+		NV_DEBUG(drm, "Handling deferred fbcon HPD events\n");
+		drm_fb_helper_hotplug_event(&fbcon->helper);
+	}
+	mutex_unlock(&fbcon->hotplug_lock);
+}
+
 int
 nouveau_fbcon_init(struct drm_device *dev)
 {
@@ -505,6 +561,7 @@ nouveau_fbcon_init(struct drm_device *dev)
 
 	drm->fbcon = fbcon;
 	INIT_WORK(&drm->fbcon_work, nouveau_fbcon_set_suspend_work);
+	mutex_init(&fbcon->hotplug_lock);
 
 	drm_fb_helper_prepare(dev, &fbcon->helper, &nouveau_fbcon_helper_funcs);
 
diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.h b/drivers/gpu/drm/nouveau/nouveau_fbcon.h
index a6f192ea3fa6..db9d52047ef8 100644
--- a/drivers/gpu/drm/nouveau/nouveau_fbcon.h
+++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.h
@@ -41,6 +41,9 @@ struct nouveau_fbdev {
 	struct nvif_object gdi;
 	struct nvif_object blit;
 	struct nvif_object twod;
+
+	struct mutex hotplug_lock;
+	bool hotplug_waiting;
 };
 
 void nouveau_fbcon_restore(void);
@@ -68,6 +71,8 @@ void nouveau_fbcon_set_suspend(struct drm_device *dev, int state);
 void nouveau_fbcon_accel_save_disable(struct drm_device *dev);
 void nouveau_fbcon_accel_restore(struct drm_device *dev);
 
+void nouveau_fbcon_output_poll_changed(struct drm_device *dev);
+void nouveau_fbcon_hotplug_resume(struct nouveau_fbdev *fbcon);
 extern int nouveau_nofbaccel;
 
 #endif /* __NV50_FBCON_H__ */
-- 
cgit v1.2.3


From 6833fb1ec120bf078e1a527c573a09d4de286224 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Wed, 15 Aug 2018 15:00:14 -0400
Subject: drm/nouveau/drm/nouveau: Use pm_runtime_get_noresume() in
 connector_detect()

It's true we can't resume the device from poll workers in
nouveau_connector_detect(). We can however, prevent the autosuspend
timer from elapsing immediately if it hasn't already without risking any
sort of deadlock with the runtime suspend/resume operations. So do that
instead of entirely avoiding grabbing a power reference.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Cc: stable@vger.kernel.org
Cc: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_connector.c | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c
index 51932c72334e..31b31a35c8fe 100644
--- a/drivers/gpu/drm/nouveau/nouveau_connector.c
+++ b/drivers/gpu/drm/nouveau/nouveau_connector.c
@@ -555,12 +555,16 @@ nouveau_connector_detect(struct drm_connector *connector, bool force)
 		nv_connector->edid = NULL;
 	}
 
-	/* Outputs are only polled while runtime active, so acquiring a
-	 * runtime PM ref here is unnecessary (and would deadlock upon
-	 * runtime suspend because it waits for polling to finish).
+	/* Outputs are only polled while runtime active, so resuming the
+	 * device here is unnecessary (and would deadlock upon runtime suspend
+	 * because it waits for polling to finish). We do however, want to
+	 * prevent the autosuspend timer from elapsing during this operation
+	 * if possible.
 	 */
-	if (!drm_kms_helper_is_poll_worker()) {
-		ret = pm_runtime_get_sync(connector->dev->dev);
+	if (drm_kms_helper_is_poll_worker()) {
+		pm_runtime_get_noresume(dev->dev);
+	} else {
+		ret = pm_runtime_get_sync(dev->dev);
 		if (ret < 0 && ret != -EACCES)
 			return conn_status;
 	}
@@ -638,10 +642,8 @@ detect_analog:
 
  out:
 
-	if (!drm_kms_helper_is_poll_worker()) {
-		pm_runtime_mark_last_busy(connector->dev->dev);
-		pm_runtime_put_autosuspend(connector->dev->dev);
-	}
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
 
 	return conn_status;
 }
-- 
cgit v1.2.3


From 3e1a12754d4df5804bfca5dedf09d2ba291bdc2a Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Wed, 15 Aug 2018 15:00:15 -0400
Subject: drm/nouveau: Fix deadlocks in nouveau_connector_detect()

When we disable hotplugging on the GPU, we need to be able to
synchronize with each connector's hotplug interrupt handler before the
interrupt is finally disabled. This can be a problem however, since
nouveau_connector_detect() currently grabs a runtime power reference
when handling connector probing. This will deadlock the runtime suspend
handler like so:

[  861.480896] INFO: task kworker/0:2:61 blocked for more than 120 seconds.
[  861.483290]       Tainted: G           O      4.18.0-rc6Lyude-Test+ #1
[  861.485158] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[  861.486332] kworker/0:2     D    0    61      2 0x80000000
[  861.487044] Workqueue: events nouveau_display_hpd_work [nouveau]
[  861.487737] Call Trace:
[  861.488394]  __schedule+0x322/0xaf0
[  861.489070]  schedule+0x33/0x90
[  861.489744]  rpm_resume+0x19c/0x850
[  861.490392]  ? finish_wait+0x90/0x90
[  861.491068]  __pm_runtime_resume+0x4e/0x90
[  861.491753]  nouveau_display_hpd_work+0x22/0x60 [nouveau]
[  861.492416]  process_one_work+0x231/0x620
[  861.493068]  worker_thread+0x44/0x3a0
[  861.493722]  kthread+0x12b/0x150
[  861.494342]  ? wq_pool_ids_show+0x140/0x140
[  861.494991]  ? kthread_create_worker_on_cpu+0x70/0x70
[  861.495648]  ret_from_fork+0x3a/0x50
[  861.496304] INFO: task kworker/6:2:320 blocked for more than 120 seconds.
[  861.496968]       Tainted: G           O      4.18.0-rc6Lyude-Test+ #1
[  861.497654] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[  861.498341] kworker/6:2     D    0   320      2 0x80000080
[  861.499045] Workqueue: pm pm_runtime_work
[  861.499739] Call Trace:
[  861.500428]  __schedule+0x322/0xaf0
[  861.501134]  ? wait_for_completion+0x104/0x190
[  861.501851]  schedule+0x33/0x90
[  861.502564]  schedule_timeout+0x3a5/0x590
[  861.503284]  ? mark_held_locks+0x58/0x80
[  861.503988]  ? _raw_spin_unlock_irq+0x2c/0x40
[  861.504710]  ? wait_for_completion+0x104/0x190
[  861.505417]  ? trace_hardirqs_on_caller+0xf4/0x190
[  861.506136]  ? wait_for_completion+0x104/0x190
[  861.506845]  wait_for_completion+0x12c/0x190
[  861.507555]  ? wake_up_q+0x80/0x80
[  861.508268]  flush_work+0x1c9/0x280
[  861.508990]  ? flush_workqueue_prep_pwqs+0x1b0/0x1b0
[  861.509735]  nvif_notify_put+0xb1/0xc0 [nouveau]
[  861.510482]  nouveau_display_fini+0xbd/0x170 [nouveau]
[  861.511241]  nouveau_display_suspend+0x67/0x120 [nouveau]
[  861.511969]  nouveau_do_suspend+0x5e/0x2d0 [nouveau]
[  861.512715]  nouveau_pmops_runtime_suspend+0x47/0xb0 [nouveau]
[  861.513435]  pci_pm_runtime_suspend+0x6b/0x180
[  861.514165]  ? pci_has_legacy_pm_support+0x70/0x70
[  861.514897]  __rpm_callback+0x7a/0x1d0
[  861.515618]  ? pci_has_legacy_pm_support+0x70/0x70
[  861.516313]  rpm_callback+0x24/0x80
[  861.517027]  ? pci_has_legacy_pm_support+0x70/0x70
[  861.517741]  rpm_suspend+0x142/0x6b0
[  861.518449]  pm_runtime_work+0x97/0xc0
[  861.519144]  process_one_work+0x231/0x620
[  861.519831]  worker_thread+0x44/0x3a0
[  861.520522]  kthread+0x12b/0x150
[  861.521220]  ? wq_pool_ids_show+0x140/0x140
[  861.521925]  ? kthread_create_worker_on_cpu+0x70/0x70
[  861.522622]  ret_from_fork+0x3a/0x50
[  861.523299] INFO: task kworker/6:0:1329 blocked for more than 120 seconds.
[  861.523977]       Tainted: G           O      4.18.0-rc6Lyude-Test+ #1
[  861.524644] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[  861.525349] kworker/6:0     D    0  1329      2 0x80000000
[  861.526073] Workqueue: events nvif_notify_work [nouveau]
[  861.526751] Call Trace:
[  861.527411]  __schedule+0x322/0xaf0
[  861.528089]  schedule+0x33/0x90
[  861.528758]  rpm_resume+0x19c/0x850
[  861.529399]  ? finish_wait+0x90/0x90
[  861.530073]  __pm_runtime_resume+0x4e/0x90
[  861.530798]  nouveau_connector_detect+0x7e/0x510 [nouveau]
[  861.531459]  ? ww_mutex_lock+0x47/0x80
[  861.532097]  ? ww_mutex_lock+0x47/0x80
[  861.532819]  ? drm_modeset_lock+0x88/0x130 [drm]
[  861.533481]  drm_helper_probe_detect_ctx+0xa0/0x100 [drm_kms_helper]
[  861.534127]  drm_helper_hpd_irq_event+0xa4/0x120 [drm_kms_helper]
[  861.534940]  nouveau_connector_hotplug+0x98/0x120 [nouveau]
[  861.535556]  nvif_notify_work+0x2d/0xb0 [nouveau]
[  861.536221]  process_one_work+0x231/0x620
[  861.536994]  worker_thread+0x44/0x3a0
[  861.537757]  kthread+0x12b/0x150
[  861.538463]  ? wq_pool_ids_show+0x140/0x140
[  861.539102]  ? kthread_create_worker_on_cpu+0x70/0x70
[  861.539815]  ret_from_fork+0x3a/0x50
[  861.540521]
               Showing all locks held in the system:
[  861.541696] 2 locks held by kworker/0:2/61:
[  861.542406]  #0: 000000002dbf8af5 ((wq_completion)"events"){+.+.}, at: process_one_work+0x1b3/0x620
[  861.543071]  #1: 0000000076868126 ((work_completion)(&drm->hpd_work)){+.+.}, at: process_one_work+0x1b3/0x620
[  861.543814] 1 lock held by khungtaskd/64:
[  861.544535]  #0: 0000000059db4b53 (rcu_read_lock){....}, at: debug_show_all_locks+0x23/0x185
[  861.545160] 3 locks held by kworker/6:2/320:
[  861.545896]  #0: 00000000d9e1bc59 ((wq_completion)"pm"){+.+.}, at: process_one_work+0x1b3/0x620
[  861.546702]  #1: 00000000c9f92d84 ((work_completion)(&dev->power.work)){+.+.}, at: process_one_work+0x1b3/0x620
[  861.547443]  #2: 000000004afc5de1 (drm_connector_list_iter){.+.+}, at: nouveau_display_fini+0x96/0x170 [nouveau]
[  861.548146] 1 lock held by dmesg/983:
[  861.548889] 2 locks held by zsh/1250:
[  861.549605]  #0: 00000000348e3cf6 (&tty->ldisc_sem){++++}, at: ldsem_down_read+0x37/0x40
[  861.550393]  #1: 000000007009a7a8 (&ldata->atomic_read_lock){+.+.}, at: n_tty_read+0xc1/0x870
[  861.551122] 6 locks held by kworker/6:0/1329:
[  861.551957]  #0: 000000002dbf8af5 ((wq_completion)"events"){+.+.}, at: process_one_work+0x1b3/0x620
[  861.552765]  #1: 00000000ddb499ad ((work_completion)(&notify->work)#2){+.+.}, at: process_one_work+0x1b3/0x620
[  861.553582]  #2: 000000006e013cbe (&dev->mode_config.mutex){+.+.}, at: drm_helper_hpd_irq_event+0x6c/0x120 [drm_kms_helper]
[  861.554357]  #3: 000000004afc5de1 (drm_connector_list_iter){.+.+}, at: drm_helper_hpd_irq_event+0x78/0x120 [drm_kms_helper]
[  861.555227]  #4: 0000000044f294d9 (crtc_ww_class_acquire){+.+.}, at: drm_helper_probe_detect_ctx+0x3d/0x100 [drm_kms_helper]
[  861.556133]  #5: 00000000db193642 (crtc_ww_class_mutex){+.+.}, at: drm_modeset_lock+0x4b/0x130 [drm]

[  861.557864] =============================================

[  861.559507] NMI backtrace for cpu 2
[  861.560363] CPU: 2 PID: 64 Comm: khungtaskd Tainted: G           O      4.18.0-rc6Lyude-Test+ #1
[  861.561197] Hardware name: LENOVO 20EQS64N0B/20EQS64N0B, BIOS N1EET78W (1.51 ) 05/18/2018
[  861.561948] Call Trace:
[  861.562757]  dump_stack+0x8e/0xd3
[  861.563516]  nmi_cpu_backtrace.cold.3+0x14/0x5a
[  861.564269]  ? lapic_can_unplug_cpu.cold.27+0x42/0x42
[  861.565029]  nmi_trigger_cpumask_backtrace+0xa1/0xae
[  861.565789]  arch_trigger_cpumask_backtrace+0x19/0x20
[  861.566558]  watchdog+0x316/0x580
[  861.567355]  kthread+0x12b/0x150
[  861.568114]  ? reset_hung_task_detector+0x20/0x20
[  861.568863]  ? kthread_create_worker_on_cpu+0x70/0x70
[  861.569598]  ret_from_fork+0x3a/0x50
[  861.570370] Sending NMI from CPU 2 to CPUs 0-1,3-7:
[  861.571426] NMI backtrace for cpu 6 skipped: idling at intel_idle+0x7f/0x120
[  861.571429] NMI backtrace for cpu 7 skipped: idling at intel_idle+0x7f/0x120
[  861.571432] NMI backtrace for cpu 3 skipped: idling at intel_idle+0x7f/0x120
[  861.571464] NMI backtrace for cpu 5 skipped: idling at intel_idle+0x7f/0x120
[  861.571467] NMI backtrace for cpu 0 skipped: idling at intel_idle+0x7f/0x120
[  861.571469] NMI backtrace for cpu 4 skipped: idling at intel_idle+0x7f/0x120
[  861.571472] NMI backtrace for cpu 1 skipped: idling at intel_idle+0x7f/0x120
[  861.572428] Kernel panic - not syncing: hung_task: blocked tasks

So: fix this by making it so that normal hotplug handling /only/ happens
so long as the GPU is currently awake without any pending runtime PM
requests. In the event that a hotplug occurs while the device is
suspending or resuming, we can simply defer our response until the GPU
is fully runtime resumed again.

Changes since v4:
- Use a new trick I came up with using pm_runtime_get() instead of the
  hackish junk we had before

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Cc: stable@vger.kernel.org
Cc: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_connector.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c
index 31b31a35c8fe..76660bc1ccfb 100644
--- a/drivers/gpu/drm/nouveau/nouveau_connector.c
+++ b/drivers/gpu/drm/nouveau/nouveau_connector.c
@@ -1107,6 +1107,26 @@ nouveau_connector_hotplug(struct nvif_notify *notify)
 	const struct nvif_notify_conn_rep_v0 *rep = notify->data;
 	const char *name = connector->name;
 	struct nouveau_encoder *nv_encoder;
+	int ret;
+
+	ret = pm_runtime_get(drm->dev->dev);
+	if (ret == 0) {
+		/* We can't block here if there's a pending PM request
+		 * running, as we'll deadlock nouveau_display_fini() when it
+		 * calls nvif_put() on our nvif_notify struct. So, simply
+		 * defer the hotplug event until the device finishes resuming
+		 */
+		NV_DEBUG(drm, "Deferring HPD on %s until runtime resume\n",
+			 name);
+		schedule_work(&drm->hpd_work);
+
+		pm_runtime_put_noidle(drm->dev->dev);
+		return NVIF_NOTIFY_KEEP;
+	} else if (ret != 1 && ret != -EACCES) {
+		NV_WARN(drm, "HPD on %s dropped due to RPM failure: %d\n",
+			name, ret);
+		return NVIF_NOTIFY_DROP;
+	}
 
 	if (rep->mask & NVIF_NOTIFY_CONN_V0_IRQ) {
 		NV_DEBUG(drm, "service %s\n", name);
@@ -1124,6 +1144,8 @@ nouveau_connector_hotplug(struct nvif_notify *notify)
 		drm_helper_hpd_irq_event(connector->dev);
 	}
 
+	pm_runtime_mark_last_busy(drm->dev->dev);
+	pm_runtime_put_autosuspend(drm->dev->dev);
 	return NVIF_NOTIFY_KEEP;
 }
 
-- 
cgit v1.2.3


From 0445f7537d0742e4f8bcf594a8d81fb901fd131e Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Wed, 15 Aug 2018 15:15:11 -0400
Subject: drm/nouveau: Remove useless poll_enable() call in
 switcheroo_set_state()

This doesn't do anything, drm_kms_helper_poll_enable() gets called in
nouveau_pmops_resume()->nouveau_display_resume()->nouveau_display_init()
already.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Cc: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_vga.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_vga.c b/drivers/gpu/drm/nouveau/nouveau_vga.c
index 3da5a4305aa4..09b1d8151881 100644
--- a/drivers/gpu/drm/nouveau/nouveau_vga.c
+++ b/drivers/gpu/drm/nouveau/nouveau_vga.c
@@ -46,7 +46,6 @@ nouveau_switcheroo_set_state(struct pci_dev *pdev,
 		pr_err("VGA switcheroo: switched nouveau on\n");
 		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
 		nouveau_pmops_resume(&pdev->dev);
-		drm_kms_helper_poll_enable(dev);
 		dev->switch_power_state = DRM_SWITCH_POWER_ON;
 	} else {
 		pr_err("VGA switcheroo: switched nouveau off\n");
-- 
cgit v1.2.3


From 0d7b2d4def679cae3bf2728fc31be7f8a48ceab3 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Wed, 15 Aug 2018 15:15:12 -0400
Subject: drm/nouveau: Remove useless poll_disable() call in
 switcheroo_set_state()

This won't do anything but potentially make us miss hotplugs. We already
call drm_kms_helper_poll_disable() in
nouveau_pmops_suspend()->nouveau_display_suspend()->nouveau_display_fini()

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Cc: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_vga.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_vga.c b/drivers/gpu/drm/nouveau/nouveau_vga.c
index 09b1d8151881..8f1ce4833230 100644
--- a/drivers/gpu/drm/nouveau/nouveau_vga.c
+++ b/drivers/gpu/drm/nouveau/nouveau_vga.c
@@ -50,7 +50,6 @@ nouveau_switcheroo_set_state(struct pci_dev *pdev,
 	} else {
 		pr_err("VGA switcheroo: switched nouveau off\n");
 		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
-		drm_kms_helper_poll_disable(dev);
 		nouveau_switcheroo_optimus_dsm();
 		nouveau_pmops_suspend(&pdev->dev);
 		dev->switch_power_state = DRM_SWITCH_POWER_OFF;
-- 
cgit v1.2.3


From 7326ead9828e5eb5c6030d80310241c404e919f9 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Wed, 15 Aug 2018 15:15:13 -0400
Subject: drm/nouveau: Remove useless poll_enable() call in drm_load()

Again, this doesn't do anything. drm_kms_helper_poll_enable() will have
already been called in nouveau_display_init()

Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Cc: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_drm.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c
index 5fdc1fbe2ee5..04f704b77a3c 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_drm.c
@@ -592,10 +592,8 @@ nouveau_drm_load(struct drm_device *dev, unsigned long flags)
 		pm_runtime_allow(dev->dev);
 		pm_runtime_mark_last_busy(dev->dev);
 		pm_runtime_put(dev->dev);
-	} else {
-		/* enable polling for external displays */
-		drm_kms_helper_poll_enable(dev);
 	}
+
 	return 0;
 
 fail_dispinit:
-- 
cgit v1.2.3


From b26b4590dd53e012526342e749c423e6c0e73437 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Thu, 9 Aug 2018 18:22:05 -0400
Subject: drm/nouveau: Only write DP_MSTM_CTRL when needed

Currently, nouveau will re-write the DP_MSTM_CTRL register for an MST
hub every time it receives a long HPD pulse on DP. This isn't actually
necessary and additionally, has some unintended side effects.

With the P50 I've got here, rewriting DP_MSTM_CTRL constantly seems to
make it rather likely (1 out of 5 times usually) that bringing up MST
with it's ThinkPad dock will fail and result in sideband messages timing
out in the middle. Afterwards, successive probes don't manage to get the
dock to communicate properly over MST sideband properly.

Many times sideband message timeouts from MST hubs are indicative of
either the source or the sink dropping an ESI event, which can cause
DRM's perspective of the topology's current state to go out of sync with
reality. While it's tough to really know for sure what's happening to
the dock, using userspace tools to write to DP_MSTM_CTRL in the middle
of the MST link probing process does appear to make things flaky. It's
possible that when we write to DP_MSTM_CTRL, the function that gets
triggered to respond in the dock's firmware temporarily puts it in a
state where it might end up not reporting an ESI to the source, or ends
up dropping a sideband message we sent it.

So, to fix this we make it so that when probing an MST topology, we
respect it's current state. If the dock's already enabled, we simply
read DP_MSTM_CTRL and disable the topology if it's value is not what we
expected. Otherwise, we perform the normal MST probing dance. We avoid
taking any action except if the state of the MST topology actually
changes.

This fixes MST sideband message timeouts and detection failures on my
P50 with its ThinkPad dock.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Cc: Karol Herbst <karolherbst@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/dispnv50/disp.c | 45 ++++++++++++++++++++++++++-------
 1 file changed, 36 insertions(+), 9 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c b/drivers/gpu/drm/nouveau/dispnv50/disp.c
index aec6ee1ff4e0..1a06c165a8df 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@@ -1142,31 +1142,58 @@ nv50_mstm_enable(struct nv50_mstm *mstm, u8 dpcd, int state)
 int
 nv50_mstm_detect(struct nv50_mstm *mstm, u8 dpcd[8], int allow)
 {
-	int ret, state = 0;
+	struct drm_dp_aux *aux;
+	int ret;
+	bool old_state, new_state;
+	u8 mstm_ctrl;
 
 	if (!mstm)
 		return 0;
 
-	if (dpcd[0] >= 0x12) {
-		ret = drm_dp_dpcd_readb(mstm->mgr.aux, DP_MSTM_CAP, &dpcd[1]);
+	mutex_lock(&mstm->mgr.lock);
+
+	old_state = mstm->mgr.mst_state;
+	new_state = old_state;
+	aux = mstm->mgr.aux;
+
+	if (old_state) {
+		/* Just check that the MST hub is still as we expect it */
+		ret = drm_dp_dpcd_readb(aux, DP_MSTM_CTRL, &mstm_ctrl);
+		if (ret < 0 || !(mstm_ctrl & DP_MST_EN)) {
+			DRM_DEBUG_KMS("Hub gone, disabling MST topology\n");
+			new_state = false;
+		}
+	} else if (dpcd[0] >= 0x12) {
+		ret = drm_dp_dpcd_readb(aux, DP_MSTM_CAP, &dpcd[1]);
 		if (ret < 0)
-			return ret;
+			goto probe_error;
 
 		if (!(dpcd[1] & DP_MST_CAP))
 			dpcd[0] = 0x11;
 		else
-			state = allow;
+			new_state = allow;
 	}
 
-	ret = nv50_mstm_enable(mstm, dpcd[0], state);
+	if (new_state == old_state) {
+		mutex_unlock(&mstm->mgr.lock);
+		return new_state;
+	}
+
+	ret = nv50_mstm_enable(mstm, dpcd[0], new_state);
 	if (ret)
-		return ret;
+		goto probe_error;
+
+	mutex_unlock(&mstm->mgr.lock);
 
-	ret = drm_dp_mst_topology_mgr_set_mst(&mstm->mgr, state);
+	ret = drm_dp_mst_topology_mgr_set_mst(&mstm->mgr, new_state);
 	if (ret)
 		return nv50_mstm_enable(mstm, dpcd[0], 0);
 
-	return mstm->mgr.mst_state;
+	return new_state;
+
+probe_error:
+	mutex_unlock(&mstm->mgr.lock);
+	return ret;
 }
 
 static void
-- 
cgit v1.2.3


From fa3cdf8d0b092c4561f9f017dfac409eb7644737 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Thu, 9 Aug 2018 18:22:06 -0400
Subject: drm/nouveau: Reset MST branching unit before enabling

When probing a new MST device, it's not safe to make any assumptions
about it's current state. While most well mannered MST hubs will just
disable the branching unit on hotplug disconnects, this isn't enough to
save us from various other scenarios that might have resulted in
something writing to the MST branching unit before we got control of it.
This could happen if a previous probe we tried failed, if we're booting
in kexec context and the hub is still in the state the last kernel put
it in, etc.

Luckily; there is no reason we can't just reset the branching unit
every time we enable a new topology. So, fix this by resetting it on
enabling new topologies to ensure that we always start off with a clean,
unmodified topology state on MST sinks.

This fixes occasional hard-lockups on my P50's laptop dock (e.g. AUX
times out all DPCD trasactions) observed after multiple docks, undocks,
and module reloads.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Cc: Karol Herbst <karolherbst@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/dispnv50/disp.c | 20 ++++++++++++--------
 1 file changed, 12 insertions(+), 8 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c b/drivers/gpu/drm/nouveau/dispnv50/disp.c
index 1a06c165a8df..5691dfa1db6f 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@@ -1123,17 +1123,21 @@ nv50_mstm_enable(struct nv50_mstm *mstm, u8 dpcd, int state)
 	int ret;
 
 	if (dpcd >= 0x12) {
-		ret = drm_dp_dpcd_readb(mstm->mgr.aux, DP_MSTM_CTRL, &dpcd);
+		/* Even if we're enabling MST, start with disabling the
+		 * branching unit to clear any sink-side MST topology state
+		 * that wasn't set by us
+		 */
+		ret = drm_dp_dpcd_writeb(mstm->mgr.aux, DP_MSTM_CTRL, 0);
 		if (ret < 0)
 			return ret;
 
-		dpcd &= ~DP_MST_EN;
-		if (state)
-			dpcd |= DP_MST_EN;
-
-		ret = drm_dp_dpcd_writeb(mstm->mgr.aux, DP_MSTM_CTRL, dpcd);
-		if (ret < 0)
-			return ret;
+		if (state) {
+			/* Now, start initializing */
+			ret = drm_dp_dpcd_writeb(mstm->mgr.aux, DP_MSTM_CTRL,
+						 DP_MST_EN);
+			if (ret < 0)
+				return ret;
+		}
 	}
 
 	return nvif_mthd(disp, 0, &args, sizeof(args));
-- 
cgit v1.2.3


From 79e765ad665da4b8aa7e9c878bd2fef837f6fea5 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Thu, 16 Aug 2018 16:13:13 -0400
Subject: drm/nouveau/drm/nouveau: Prevent handling ACPI HPD events too early

On most systems with ACPI hotplugging support, it seems that we always
receive a hotplug event once we re-enable EC interrupts even if the GPU
hasn't even been resumed yet.

This can cause problems since even though we schedule hpd_work to handle
connector reprobing for us, hpd_work synchronizes on
pm_runtime_get_sync() to wait until the device is ready to perform
reprobing. Since runtime suspend/resume callbacks are disabled before
the PM core calls ->suspend(), any calls to pm_runtime_get_sync() during
this period will grab a runtime PM ref and return immediately with
-EACCES. Because we schedule hpd_work from our ACPI HPD handler, and
hpd_work synchronizes on pm_runtime_get_sync(), this causes us to launch
a connector reprobe immediately even if the GPU isn't actually resumed
just yet. This causes various warnings in dmesg and occasionally, also
prevents some displays connected to the dedicated GPU from coming back
up after suspend. Example:

usb 1-4: USB disconnect, device number 14
usb 1-4.1: USB disconnect, device number 15
WARNING: CPU: 0 PID: 838 at drivers/gpu/drm/nouveau/include/nvkm/subdev/i2c.h:170 nouveau_dp_detect+0x17e/0x370 [nouveau]
CPU: 0 PID: 838 Comm: kworker/0:6 Not tainted 4.17.14-201.Lyude.bz1477182.V3.fc28.x86_64 #1
Hardware name: LENOVO 20EQS64N00/20EQS64N00, BIOS N1EET77W (1.50 ) 03/28/2018
Workqueue: events nouveau_display_hpd_work [nouveau]
RIP: 0010:nouveau_dp_detect+0x17e/0x370 [nouveau]
RSP: 0018:ffffa15143933cf0 EFLAGS: 00010293
RAX: 0000000000000000 RBX: ffff8cb4f656c400 RCX: 0000000000000000
RDX: ffffa1514500e4e4 RSI: ffffa1514500e4e4 RDI: 0000000001009002
RBP: ffff8cb4f4a8a800 R08: ffffa15143933cfd R09: ffffa15143933cfc
R10: 0000000000000000 R11: 0000000000000000 R12: ffff8cb4fb57a000
R13: ffff8cb4fb57a000 R14: ffff8cb4f4a8f800 R15: ffff8cb4f656c418
FS:  0000000000000000(0000) GS:ffff8cb51f400000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f78ec938000 CR3: 000000073720a003 CR4: 00000000003606f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 ? _cond_resched+0x15/0x30
 nouveau_connector_detect+0x2ce/0x520 [nouveau]
 ? _cond_resched+0x15/0x30
 ? ww_mutex_lock+0x12/0x40
 drm_helper_probe_detect_ctx+0x8b/0xe0 [drm_kms_helper]
 drm_helper_hpd_irq_event+0xa8/0x120 [drm_kms_helper]
 nouveau_display_hpd_work+0x2a/0x60 [nouveau]
 process_one_work+0x187/0x340
 worker_thread+0x2e/0x380
 ? pwq_unbound_release_workfn+0xd0/0xd0
 kthread+0x112/0x130
 ? kthread_create_worker_on_cpu+0x70/0x70
 ret_from_fork+0x35/0x40
Code: 4c 8d 44 24 0d b9 00 05 00 00 48 89 ef ba 09 00 00 00 be 01 00 00 00 e8 e1 09 f8 ff 85 c0 0f 85 b2 01 00 00 80 7c 24 0c 03 74 02 <0f> 0b 48 89 ef e8 b8 07 f8 ff f6 05 51 1b c8 ff 02 0f 84 72 ff
---[ end trace 55d811b38fc8e71a ]---

So, to fix this we attempt to grab a runtime PM reference in the ACPI
handler itself asynchronously. If the GPU is already awake (it will have
normal hotplugging at this point) or runtime PM callbacks are currently
disabled on the device, we drop our reference without updating the
autosuspend delay. We only schedule connector reprobes when we
successfully managed to queue up a resume request with our asynchronous
PM ref.

This also has the added benefit of preventing redundant connector
reprobes from ACPI while the GPU is runtime resumed!

Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Cc: Karol Herbst <kherbst@redhat.com>
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1477182#c41
Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_display.c | 26 ++++++++++++++++++++------
 1 file changed, 20 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c b/drivers/gpu/drm/nouveau/nouveau_display.c
index 42e7c35e3fba..e4024af5a46f 100644
--- a/drivers/gpu/drm/nouveau/nouveau_display.c
+++ b/drivers/gpu/drm/nouveau/nouveau_display.c
@@ -377,15 +377,29 @@ nouveau_display_acpi_ntfy(struct notifier_block *nb, unsigned long val,
 {
 	struct nouveau_drm *drm = container_of(nb, typeof(*drm), acpi_nb);
 	struct acpi_bus_event *info = data;
+	int ret;
 
 	if (!strcmp(info->device_class, ACPI_VIDEO_CLASS)) {
 		if (info->type == ACPI_VIDEO_NOTIFY_PROBE) {
-			/*
-			 * This may be the only indication we receive of a
-			 * connector hotplug on a runtime suspended GPU,
-			 * schedule hpd_work to check.
-			 */
-			schedule_work(&drm->hpd_work);
+			ret = pm_runtime_get(drm->dev->dev);
+			if (ret == 1 || ret == -EACCES) {
+				/* If the GPU is already awake, or in a state
+				 * where we can't wake it up, it can handle
+				 * it's own hotplug events.
+				 */
+				pm_runtime_put_autosuspend(drm->dev->dev);
+			} else if (ret == 0) {
+				/* This may be the only indication we receive
+				 * of a connector hotplug on a runtime
+				 * suspended GPU, schedule hpd_work to check.
+				 */
+				NV_DEBUG(drm, "ACPI requested connector reprobe\n");
+				schedule_work(&drm->hpd_work);
+				pm_runtime_put_noidle(drm->dev->dev);
+			} else {
+				NV_WARN(drm, "Dropped ACPI reprobe event due to RPM error: %d\n",
+					ret);
+			}
 
 			/* acpi-video should not generate keypresses for this */
 			return NOTIFY_BAD;
-- 
cgit v1.2.3


From 2f7ca781fd382cf8dde73ed36dfdd93fd05b3332 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Tue, 7 Aug 2018 17:32:48 -0400
Subject: drm/nouveau/drm/nouveau: Don't forget to cancel hpd_work on
 suspend/unload

Currently, there's nothing in nouveau that actually cancels this work
struct. So, cancel it on suspend/unload. Otherwise, if we're unlucky
enough hpd_work might try to keep running up until the system is
suspended.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_display.c | 9 ++++++---
 drivers/gpu/drm/nouveau/nouveau_display.h | 2 +-
 drivers/gpu/drm/nouveau/nouveau_drm.c     | 2 +-
 3 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c b/drivers/gpu/drm/nouveau/nouveau_display.c
index e4024af5a46f..540c0cbbfcee 100644
--- a/drivers/gpu/drm/nouveau/nouveau_display.c
+++ b/drivers/gpu/drm/nouveau/nouveau_display.c
@@ -442,7 +442,7 @@ nouveau_display_init(struct drm_device *dev)
 }
 
 void
-nouveau_display_fini(struct drm_device *dev, bool suspend)
+nouveau_display_fini(struct drm_device *dev, bool suspend, bool runtime)
 {
 	struct nouveau_display *disp = nouveau_display(dev);
 	struct nouveau_drm *drm = nouveau_drm(dev);
@@ -467,6 +467,9 @@ nouveau_display_fini(struct drm_device *dev, bool suspend)
 	}
 	drm_connector_list_iter_end(&conn_iter);
 
+	if (!runtime)
+		cancel_work_sync(&drm->hpd_work);
+
 	drm_kms_helper_poll_disable(dev);
 	disp->fini(dev);
 }
@@ -635,11 +638,11 @@ nouveau_display_suspend(struct drm_device *dev, bool runtime)
 			}
 		}
 
-		nouveau_display_fini(dev, true);
+		nouveau_display_fini(dev, true, runtime);
 		return 0;
 	}
 
-	nouveau_display_fini(dev, true);
+	nouveau_display_fini(dev, true, runtime);
 
 	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
 		struct nouveau_framebuffer *nouveau_fb;
diff --git a/drivers/gpu/drm/nouveau/nouveau_display.h b/drivers/gpu/drm/nouveau/nouveau_display.h
index 54aa7c3fa42d..ff92b54ce448 100644
--- a/drivers/gpu/drm/nouveau/nouveau_display.h
+++ b/drivers/gpu/drm/nouveau/nouveau_display.h
@@ -62,7 +62,7 @@ nouveau_display(struct drm_device *dev)
 int  nouveau_display_create(struct drm_device *dev);
 void nouveau_display_destroy(struct drm_device *dev);
 int  nouveau_display_init(struct drm_device *dev);
-void nouveau_display_fini(struct drm_device *dev, bool suspend);
+void nouveau_display_fini(struct drm_device *dev, bool suspend, bool runtime);
 int  nouveau_display_suspend(struct drm_device *dev, bool runtime);
 void nouveau_display_resume(struct drm_device *dev, bool runtime);
 int  nouveau_display_vblank_enable(struct drm_device *, unsigned int);
diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c
index 04f704b77a3c..f1a119113d04 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_drm.c
@@ -627,7 +627,7 @@ nouveau_drm_unload(struct drm_device *dev)
 	nouveau_debugfs_fini(drm);
 
 	if (dev->mode_config.num_crtc)
-		nouveau_display_fini(dev, false);
+		nouveau_display_fini(dev, false, false);
 	nouveau_display_destroy(dev);
 
 	nouveau_bios_takedown(dev);
-- 
cgit v1.2.3


From d5986a1c4dcd00cb8b9eee4a56ee93868222a9a2 Mon Sep 17 00:00:00 2001
From: Lyude Paul <lyude@redhat.com>
Date: Thu, 30 Aug 2018 13:16:28 -0400
Subject: drm/nouveau: Fix nouveau_connector_ddc_detect()
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

It looks like that when we moved over to using
drm_connector_for_each_possible_encoder() in nouveau, that one rather
important part of this function got dropped by accident:

	/*          Right   v   here */
	for (i = 0; nv_encoder = NULL, i < DRM_CONNECTOR_MAX_ENCODER; i++) {
		int id = connector->encoder_ids[i];
		if (id == 0)
			break;

Since it's rather difficult to notice: the conditional in this loop is
actually:

	nv_encoder = NULL, i < DRM_CONNECTOR_MAX_ENCODER

Meaning that all early breaks result in nv_encoder keeping it's value,
otherwise nv_encoder = NULL. Ugh.

Since this got dropped, nouveau_connector_ddc_detect() now returns an
encoder for every single connector, regardless of whether or not it's
detected:

    [ 1780.056185] nouveau 0000:01:00.0: DRM: DDC responded, but no EDID for DP-2

So: fix this to ensure we only return an encoder if we actually found
one, and clean up the rest of the function while we're at it since it's
nearly impossible to read properly.

Changes since v1:
- Don't skip ddc probing for LVDS if we can't switch DDC through
  vga-switcheroo, just do the DDC probing without calling
  vga_switcheroo_lock_ddc() - skeggsb

Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Fixes: ddba766dd07e ("drm/nouveau: Use drm_connector_for_each_possible_encoder()")
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_connector.c | 49 ++++++++++++++++-------------
 1 file changed, 28 insertions(+), 21 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c
index 76660bc1ccfb..259ee5039125 100644
--- a/drivers/gpu/drm/nouveau/nouveau_connector.c
+++ b/drivers/gpu/drm/nouveau/nouveau_connector.c
@@ -412,9 +412,10 @@ nouveau_connector_ddc_detect(struct drm_connector *connector)
 	struct nouveau_connector *nv_connector = nouveau_connector(connector);
 	struct nouveau_drm *drm = nouveau_drm(dev);
 	struct nvkm_gpio *gpio = nvxx_gpio(&drm->client.device);
-	struct nouveau_encoder *nv_encoder = NULL;
+	struct nouveau_encoder *nv_encoder = NULL, *found = NULL;
 	struct drm_encoder *encoder;
-	int i, panel = -ENODEV;
+	int i, ret, panel = -ENODEV;
+	bool switcheroo_ddc = false;
 
 	/* eDP panels need powering on by us (if the VBIOS doesn't default it
 	 * to on) before doing any AUX channel transactions.  LVDS panel power
@@ -431,37 +432,43 @@ nouveau_connector_ddc_detect(struct drm_connector *connector)
 	drm_connector_for_each_possible_encoder(connector, encoder, i) {
 		nv_encoder = nouveau_encoder(encoder);
 
-		if (nv_encoder->dcb->type == DCB_OUTPUT_DP) {
-			int ret = nouveau_dp_detect(nv_encoder);
+		switch (nv_encoder->dcb->type) {
+		case DCB_OUTPUT_DP:
+			ret = nouveau_dp_detect(nv_encoder);
 			if (ret == NOUVEAU_DP_MST)
 				return NULL;
-			if (ret == NOUVEAU_DP_SST)
-				break;
-		} else
-		if ((vga_switcheroo_handler_flags() &
-		     VGA_SWITCHEROO_CAN_SWITCH_DDC) &&
-		    nv_encoder->dcb->type == DCB_OUTPUT_LVDS &&
-		    nv_encoder->i2c) {
-			int ret;
-			vga_switcheroo_lock_ddc(dev->pdev);
-			ret = nvkm_probe_i2c(nv_encoder->i2c, 0x50);
-			vga_switcheroo_unlock_ddc(dev->pdev);
-			if (ret)
+			else if (ret == NOUVEAU_DP_SST)
+				found = nv_encoder;
+
+			break;
+		case DCB_OUTPUT_LVDS:
+			switcheroo_ddc = !!(vga_switcheroo_handler_flags() &
+					    VGA_SWITCHEROO_CAN_SWITCH_DDC);
+		/* fall-through */
+		default:
+			if (!nv_encoder->i2c)
 				break;
-		} else
-		if (nv_encoder->i2c) {
+
+			if (switcheroo_ddc)
+				vga_switcheroo_lock_ddc(dev->pdev);
 			if (nvkm_probe_i2c(nv_encoder->i2c, 0x50))
-				break;
+				found = nv_encoder;
+			if (switcheroo_ddc)
+				vga_switcheroo_unlock_ddc(dev->pdev);
+
+			break;
 		}
+		if (found)
+			break;
 	}
 
 	/* eDP panel not detected, restore panel power GPIO to previous
 	 * state to avoid confusing the SOR for other output types.
 	 */
-	if (!nv_encoder && panel == 0)
+	if (!found && panel == 0)
 		nvkm_gpio_set(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff, panel);
 
-	return nv_encoder;
+	return found;
 }
 
 static struct nouveau_encoder *
-- 
cgit v1.2.3


From a43b16dda2d7485f5c5aed075c1dc9785e339515 Mon Sep 17 00:00:00 2001
From: Ben Skeggs <bskeggs@redhat.com>
Date: Tue, 28 Aug 2018 14:10:34 +1000
Subject: drm/nouveau: fix oops in client init failure path

The NV_ERROR macro requires drm->client to be initialised, which it may not
be at this stage of the init process.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_drm.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c
index f1a119113d04..74d2283f2c28 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_drm.c
@@ -230,7 +230,7 @@ nouveau_cli_init(struct nouveau_drm *drm, const char *sname,
 		mutex_unlock(&drm->master.lock);
 	}
 	if (ret) {
-		NV_ERROR(drm, "Client allocation failed: %d\n", ret);
+		NV_PRINTK(err, cli, "Client allocation failed: %d\n", ret);
 		goto done;
 	}
 
@@ -240,37 +240,37 @@ nouveau_cli_init(struct nouveau_drm *drm, const char *sname,
 			       }, sizeof(struct nv_device_v0),
 			       &cli->device);
 	if (ret) {
-		NV_ERROR(drm, "Device allocation failed: %d\n", ret);
+		NV_PRINTK(err, cli, "Device allocation failed: %d\n", ret);
 		goto done;
 	}
 
 	ret = nvif_mclass(&cli->device.object, mmus);
 	if (ret < 0) {
-		NV_ERROR(drm, "No supported MMU class\n");
+		NV_PRINTK(err, cli, "No supported MMU class\n");
 		goto done;
 	}
 
 	ret = nvif_mmu_init(&cli->device.object, mmus[ret].oclass, &cli->mmu);
 	if (ret) {
-		NV_ERROR(drm, "MMU allocation failed: %d\n", ret);
+		NV_PRINTK(err, cli, "MMU allocation failed: %d\n", ret);
 		goto done;
 	}
 
 	ret = nvif_mclass(&cli->mmu.object, vmms);
 	if (ret < 0) {
-		NV_ERROR(drm, "No supported VMM class\n");
+		NV_PRINTK(err, cli, "No supported VMM class\n");
 		goto done;
 	}
 
 	ret = nouveau_vmm_init(cli, vmms[ret].oclass, &cli->vmm);
 	if (ret) {
-		NV_ERROR(drm, "VMM allocation failed: %d\n", ret);
+		NV_PRINTK(err, cli, "VMM allocation failed: %d\n", ret);
 		goto done;
 	}
 
 	ret = nvif_mclass(&cli->mmu.object, mems);
 	if (ret < 0) {
-		NV_ERROR(drm, "No supported MEM class\n");
+		NV_PRINTK(err, cli, "No supported MEM class\n");
 		goto done;
 	}
 
-- 
cgit v1.2.3


From 51ed833c881b9d96557c773f6a37018d79e29a46 Mon Sep 17 00:00:00 2001
From: Ben Skeggs <bskeggs@redhat.com>
Date: Tue, 28 Aug 2018 14:10:42 +1000
Subject: drm/nouveau/mmu: don't attempt to dereference vmm without valid
 instance pointer

Fixes oopses in certain failure paths.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c
index de269eb482dd..7459def78d50 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c
@@ -1423,7 +1423,7 @@ nvkm_vmm_get(struct nvkm_vmm *vmm, u8 page, u64 size, struct nvkm_vma **pvma)
 void
 nvkm_vmm_part(struct nvkm_vmm *vmm, struct nvkm_memory *inst)
 {
-	if (vmm->func->part && inst) {
+	if (inst && vmm->func->part) {
 		mutex_lock(&vmm->mutex);
 		vmm->func->part(vmm, inst);
 		mutex_unlock(&vmm->mutex);
-- 
cgit v1.2.3


From 0a6986c6595e9afd20ff7280dab36431c1e467f8 Mon Sep 17 00:00:00 2001
From: Ben Skeggs <bskeggs@redhat.com>
Date: Tue, 4 Sep 2018 15:56:57 +1000
Subject: drm/nouveau/TBDdevinit: don't fail when PMU/PRE_OS is missing from
 VBIOS

This Falcon application doesn't appear to be present on some newer
systems, so let's not fail init if we can't find it.

TBD: is there a way to determine whether it *should* be there?

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm200.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm200.c b/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm200.c
index b80618e35491..d65959ef0564 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm200.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm200.c
@@ -158,7 +158,8 @@ gm200_devinit_post(struct nvkm_devinit *base, bool post)
 	}
 
 	/* load and execute some other ucode image (bios therm?) */
-	return pmu_load(init, 0x01, post, NULL, NULL);
+	pmu_load(init, 0x01, post, NULL, NULL);
+	return 0;
 }
 
 static const struct nvkm_devinit_func
-- 
cgit v1.2.3


From 606557708fa06ebf21372d8fabf6f97529ab2349 Mon Sep 17 00:00:00 2001
From: Ben Skeggs <bskeggs@redhat.com>
Date: Tue, 4 Sep 2018 15:57:04 +1000
Subject: drm/nouveau/disp: remove unused struct member

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c | 1 -
 drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h | 1 -
 2 files changed, 2 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c
index be9e7f8c3b23..4b6973f90309 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c
@@ -245,7 +245,6 @@ nvkm_outp_ctor(const struct nvkm_outp_func *func, struct nvkm_disp *disp,
 	outp->index = index;
 	outp->info = *dcbE;
 	outp->i2c = nvkm_i2c_bus_find(i2c, dcbE->i2c_index);
-	outp->or = ffs(outp->info.or) - 1;
 
 	OUTP_DBG(outp, "type %02x loc %d or %d link %d con %x "
 		       "edid %x bus %d head %x",
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h
index ea84d7d5741a..776e36972daa 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h
@@ -13,7 +13,6 @@ struct nvkm_outp {
 	struct dcb_output info;
 
 	struct nvkm_i2c_bus *i2c;
-	int or;
 
 	struct list_head head;
 	struct nvkm_conn *conn;
-- 
cgit v1.2.3


From f6d52b2172b1adfde010df34730290c282ee641b Mon Sep 17 00:00:00 2001
From: Ben Skeggs <bskeggs@redhat.com>
Date: Tue, 4 Sep 2018 15:57:07 +1000
Subject: drm/nouveau/disp: move eDP panel power handling

We need to do this earlier to prevent aux channel timeouts in resume
paths on certain systems.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nouveau_connector.c   | 23 +----------------
 drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c | 37 ++++++++++++++++++++++++---
 2 files changed, 35 insertions(+), 25 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c
index 259ee5039125..247f72cc4d10 100644
--- a/drivers/gpu/drm/nouveau/nouveau_connector.c
+++ b/drivers/gpu/drm/nouveau/nouveau_connector.c
@@ -409,26 +409,11 @@ static struct nouveau_encoder *
 nouveau_connector_ddc_detect(struct drm_connector *connector)
 {
 	struct drm_device *dev = connector->dev;
-	struct nouveau_connector *nv_connector = nouveau_connector(connector);
-	struct nouveau_drm *drm = nouveau_drm(dev);
-	struct nvkm_gpio *gpio = nvxx_gpio(&drm->client.device);
 	struct nouveau_encoder *nv_encoder = NULL, *found = NULL;
 	struct drm_encoder *encoder;
-	int i, ret, panel = -ENODEV;
+	int i, ret;
 	bool switcheroo_ddc = false;
 
-	/* eDP panels need powering on by us (if the VBIOS doesn't default it
-	 * to on) before doing any AUX channel transactions.  LVDS panel power
-	 * is handled by the SOR itself, and not required for LVDS DDC.
-	 */
-	if (nv_connector->type == DCB_CONNECTOR_eDP) {
-		panel = nvkm_gpio_get(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff);
-		if (panel == 0) {
-			nvkm_gpio_set(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff, 1);
-			msleep(300);
-		}
-	}
-
 	drm_connector_for_each_possible_encoder(connector, encoder, i) {
 		nv_encoder = nouveau_encoder(encoder);
 
@@ -462,12 +447,6 @@ nouveau_connector_ddc_detect(struct drm_connector *connector)
 			break;
 	}
 
-	/* eDP panel not detected, restore panel power GPIO to previous
-	 * state to avoid confusing the SOR for other output types.
-	 */
-	if (!found && panel == 0)
-		nvkm_gpio_set(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff, panel);
-
 	return found;
 }
 
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c
index 7c5bed29ffef..bb34ee77458e 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c
@@ -28,6 +28,7 @@
 
 #include <subdev/bios.h>
 #include <subdev/bios/init.h>
+#include <subdev/gpio.h>
 #include <subdev/i2c.h>
 
 #include <nvif/event.h>
@@ -491,7 +492,7 @@ done:
 	return ret;
 }
 
-static void
+static bool
 nvkm_dp_enable(struct nvkm_dp *dp, bool enable)
 {
 	struct nvkm_i2c_aux *aux = dp->aux;
@@ -505,7 +506,7 @@ nvkm_dp_enable(struct nvkm_dp *dp, bool enable)
 
 		if (!nvkm_rdaux(aux, DPCD_RC00_DPCD_REV, dp->dpcd,
 				sizeof(dp->dpcd)))
-			return;
+			return true;
 	}
 
 	if (dp->present) {
@@ -515,6 +516,7 @@ nvkm_dp_enable(struct nvkm_dp *dp, bool enable)
 	}
 
 	atomic_set(&dp->lt.done, 0);
+	return false;
 }
 
 static int
@@ -555,9 +557,38 @@ nvkm_dp_fini(struct nvkm_outp *outp)
 static void
 nvkm_dp_init(struct nvkm_outp *outp)
 {
+	struct nvkm_gpio *gpio = outp->disp->engine.subdev.device->gpio;
 	struct nvkm_dp *dp = nvkm_dp(outp);
+
 	nvkm_notify_put(&dp->outp.conn->hpd);
-	nvkm_dp_enable(dp, true);
+
+	/* eDP panels need powering on by us (if the VBIOS doesn't default it
+	 * to on) before doing any AUX channel transactions.  LVDS panel power
+	 * is handled by the SOR itself, and not required for LVDS DDC.
+	 */
+	if (dp->outp.conn->info.type == DCB_CONNECTOR_eDP) {
+		int power = nvkm_gpio_get(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff);
+		if (power == 0)
+			nvkm_gpio_set(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff, 1);
+
+		/* We delay here unconditionally, even if already powered,
+		 * because some laptop panels having a significant resume
+		 * delay before the panel begins responding.
+		 *
+		 * This is likely a bit of a hack, but no better idea for
+		 * handling this at the moment.
+		 */
+		msleep(300);
+
+		/* If the eDP panel can't be detected, we need to restore
+		 * the panel power GPIO to avoid breaking another output.
+		 */
+		if (!nvkm_dp_enable(dp, true) && power == 0)
+			nvkm_gpio_set(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff, 0);
+	} else {
+		nvkm_dp_enable(dp, true);
+	}
+
 	nvkm_notify_get(&dp->hpd);
 }
 
-- 
cgit v1.2.3


From e04cfdc9b7398c60dbc70212415ea63b6c6a93ae Mon Sep 17 00:00:00 2001
From: Ben Skeggs <bskeggs@redhat.com>
Date: Tue, 4 Sep 2018 15:57:09 +1000
Subject: drm/nouveau/disp: fix DP disable race

If a HPD pulse signalling the need to retrain the link occurs between
the KMS driver releasing the output and the supervisor interrupt that
finishes the teardown, it was possible get a NULL-ptr deref.

Avoid this by marking the link as inactive earlier.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c   | 17 ++++++++++++-----
 drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c |  6 +++---
 drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c |  2 ++
 drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h |  3 ++-
 4 files changed, 19 insertions(+), 9 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c
index bb34ee77458e..5f301e632599 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c
@@ -413,14 +413,10 @@ nvkm_dp_train(struct nvkm_dp *dp, u32 dataKBps)
 }
 
 static void
-nvkm_dp_release(struct nvkm_outp *outp, struct nvkm_ior *ior)
+nvkm_dp_disable(struct nvkm_outp *outp, struct nvkm_ior *ior)
 {
 	struct nvkm_dp *dp = nvkm_dp(outp);
 
-	/* Prevent link from being retrained if sink sends an IRQ. */
-	atomic_set(&dp->lt.done, 0);
-	ior->dp.nr = 0;
-
 	/* Execute DisableLT script from DP Info Table. */
 	nvbios_init(&ior->disp->engine.subdev, dp->info.script[4],
 		init.outp = &dp->outp.info;
@@ -429,6 +425,16 @@ nvkm_dp_release(struct nvkm_outp *outp, struct nvkm_ior *ior)
 	);
 }
 
+static void
+nvkm_dp_release(struct nvkm_outp *outp)
+{
+	struct nvkm_dp *dp = nvkm_dp(outp);
+
+	/* Prevent link from being retrained if sink sends an IRQ. */
+	atomic_set(&dp->lt.done, 0);
+	dp->outp.ior->dp.nr = 0;
+}
+
 static int
 nvkm_dp_acquire(struct nvkm_outp *outp)
 {
@@ -607,6 +613,7 @@ nvkm_dp_func = {
 	.fini = nvkm_dp_fini,
 	.acquire = nvkm_dp_acquire,
 	.release = nvkm_dp_release,
+	.disable = nvkm_dp_disable,
 };
 
 static int
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c
index f89c7b977aa5..def005dd5fda 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c
@@ -501,11 +501,11 @@ nv50_disp_super_2_0(struct nv50_disp *disp, struct nvkm_head *head)
 	nv50_disp_super_ied_off(head, ior, 2);
 
 	/* If we're shutting down the OR's only active head, execute
-	 * the output path's release function.
+	 * the output path's disable function.
 	 */
 	if (ior->arm.head == (1 << head->id)) {
-		if ((outp = ior->arm.outp) && outp->func->release)
-			outp->func->release(outp, ior);
+		if ((outp = ior->arm.outp) && outp->func->disable)
+			outp->func->disable(outp, ior);
 	}
 }
 
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c
index 4b6973f90309..9fcaf3147eb8 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c
@@ -93,6 +93,8 @@ nvkm_outp_release(struct nvkm_outp *outp, u8 user)
 	if (ior) {
 		outp->acquired &= ~user;
 		if (!outp->acquired) {
+			if (outp->func->release && outp->ior)
+				outp->func->release(outp);
 			outp->ior->asy.outp = NULL;
 			outp->ior = NULL;
 		}
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h
index 776e36972daa..96272ecccb59 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h
@@ -40,7 +40,8 @@ struct nvkm_outp_func {
 	void (*init)(struct nvkm_outp *);
 	void (*fini)(struct nvkm_outp *);
 	int (*acquire)(struct nvkm_outp *);
-	void (*release)(struct nvkm_outp *, struct nvkm_ior *);
+	void (*release)(struct nvkm_outp *);
+	void (*disable)(struct nvkm_outp *, struct nvkm_ior *);
 };
 
 #define OUTP_MSG(o,l,f,a...) do {                                              \
-- 
cgit v1.2.3


From 53b0cc46f27cfc2cadca609b503a7d92b5185a47 Mon Sep 17 00:00:00 2001
From: Ben Skeggs <bskeggs@redhat.com>
Date: Tue, 4 Sep 2018 15:57:11 +1000
Subject: drm/nouveau/disp/gm200-: enforce identity-mapped SOR assignment for
 LVDS/eDP panels

Fixes eDP backlight issues on more recent laptops.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
---
 drivers/gpu/drm/nouveau/nvkm/engine/disp/base.c | 14 ++++++++++++++
 drivers/gpu/drm/nouveau/nvkm/engine/disp/ior.h  |  1 +
 drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c | 15 ++++++++++++---
 drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h |  1 +
 4 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/base.c
index 32fa94a9773f..cbd33e87b799 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/base.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/base.c
@@ -275,6 +275,7 @@ nvkm_disp_oneinit(struct nvkm_engine *engine)
 	struct nvkm_outp *outp, *outt, *pair;
 	struct nvkm_conn *conn;
 	struct nvkm_head *head;
+	struct nvkm_ior *ior;
 	struct nvbios_connE connE;
 	struct dcb_output dcbE;
 	u8  hpd = 0, ver, hdr;
@@ -399,6 +400,19 @@ nvkm_disp_oneinit(struct nvkm_engine *engine)
 			return ret;
 	}
 
+	/* Enforce identity-mapped SOR assignment for panels, which have
+	 * certain bits (ie. backlight controls) wired to a specific SOR.
+	 */
+	list_for_each_entry(outp, &disp->outp, head) {
+		if (outp->conn->info.type == DCB_CONNECTOR_LVDS ||
+		    outp->conn->info.type == DCB_CONNECTOR_eDP) {
+			ior = nvkm_ior_find(disp, SOR, ffs(outp->info.or) - 1);
+			if (!WARN_ON(!ior))
+				ior->identity = true;
+			outp->identity = true;
+		}
+	}
+
 	i = 0;
 	list_for_each_entry(head, &disp->head, head)
 		i = max(i, head->id + 1);
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/ior.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/ior.h
index e0b4e0c5704e..19911211a12a 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/ior.h
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/ior.h
@@ -16,6 +16,7 @@ struct nvkm_ior {
 	char name[8];
 
 	struct list_head head;
+	bool identity;
 
 	struct nvkm_ior_state {
 		struct nvkm_outp *outp;
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c
index 9fcaf3147eb8..c62030c96fba 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.c
@@ -129,17 +129,26 @@ nvkm_outp_acquire(struct nvkm_outp *outp, u8 user)
 	if (proto == UNKNOWN)
 		return -ENOSYS;
 
+	/* Deal with panels requiring identity-mapped SOR assignment. */
+	if (outp->identity) {
+		ior = nvkm_ior_find(outp->disp, SOR, ffs(outp->info.or) - 1);
+		if (WARN_ON(!ior))
+			return -ENOSPC;
+		return nvkm_outp_acquire_ior(outp, user, ior);
+	}
+
 	/* First preference is to reuse the OR that is currently armed
 	 * on HW, if any, in order to prevent unnecessary switching.
 	 */
 	list_for_each_entry(ior, &outp->disp->ior, head) {
-		if (!ior->asy.outp && ior->arm.outp == outp)
+		if (!ior->identity && !ior->asy.outp && ior->arm.outp == outp)
 			return nvkm_outp_acquire_ior(outp, user, ior);
 	}
 
 	/* Failing that, a completely unused OR is the next best thing. */
 	list_for_each_entry(ior, &outp->disp->ior, head) {
-		if (!ior->asy.outp && ior->type == type && !ior->arm.outp &&
+		if (!ior->identity &&
+		    !ior->asy.outp && ior->type == type && !ior->arm.outp &&
 		    (ior->func->route.set || ior->id == __ffs(outp->info.or)))
 			return nvkm_outp_acquire_ior(outp, user, ior);
 	}
@@ -148,7 +157,7 @@ nvkm_outp_acquire(struct nvkm_outp *outp, u8 user)
 	 * but will be released during the next modeset.
 	 */
 	list_for_each_entry(ior, &outp->disp->ior, head) {
-		if (!ior->asy.outp && ior->type == type &&
+		if (!ior->identity && !ior->asy.outp && ior->type == type &&
 		    (ior->func->route.set || ior->id == __ffs(outp->info.or)))
 			return nvkm_outp_acquire_ior(outp, user, ior);
 	}
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h
index 96272ecccb59..6c8aa5cfed9d 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h
@@ -16,6 +16,7 @@ struct nvkm_outp {
 
 	struct list_head head;
 	struct nvkm_conn *conn;
+	bool identity;
 
 	/* Assembly state. */
 #define NVKM_OUTP_PRIV 1
-- 
cgit v1.2.3