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>
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>
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: ddba766dd0 ("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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
Turns out this part is my fault for not noticing when reviewing
9a2eba337c ("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
d61a5c1063 ("drm/nouveau: Fix deadlock on runtime suspend")
Fixes: 9a2eba337c ("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>
Core Changes:
- add support for DisplayPort CEC-Tunneling-over-AUX (Hans Verkuil)
- more doc updates (Daniel Vetter)
- fourcc: Add is_yuv field to drm_format_info (Ayan Kumar Halder)
- dma-buf: correctly place BUG_ON (Michel Dänzer)
Driver Changes:
- more vkms support(Rodrigo Siqueira)
- many fixes and small improments to all drivers
-----BEGIN PGP SIGNATURE-----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=HyBh
-----END PGP SIGNATURE-----
Merge tag 'drm-misc-next-2018-07-18' of git://anongit.freedesktop.org/drm/drm-misc into drm-next
drm-misc-next for 4.19:
Core Changes:
- add support for DisplayPort CEC-Tunneling-over-AUX (Hans Verkuil)
- more doc updates (Daniel Vetter)
- fourcc: Add is_yuv field to drm_format_info (Ayan Kumar Halder)
- dma-buf: correctly place BUG_ON (Michel Dänzer)
Driver Changes:
- more vkms support(Rodrigo Siqueira)
- many fixes and small improments to all drivers
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180718200826.GA20165@juma
Workaround for issues seen on systems with large amounts of RAM, caused
by display not supporting the same physical address limits as the other
parts of the GPU.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This was causing problems on a system with a large amount of RAM, where
display push buffers were being fetched incorrectly when placed in high
system memory addresses.
While this commit will resolve the issue on that particular system, the
issue will be avoided completely with another patch to more fully solve
problems with display and large amounts of system memory on Pascal.
It's still probably a good idea to disable this to prevent weird issues
in the future.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Depending on the kernel configuration, early ARM architecture setup code
may have attached the GPU to a DMA/IOMMU mapping that transparently uses
the IOMMU to back the DMA API. Tegra requires special handling for IOMMU
backed buffers (a special bit in the GPU's MMU page tables indicates the
memory path to take: via the SMMU or directly to the memory controller).
Transparently backing DMA memory with an IOMMU prevents Nouveau from
properly handling such memory accesses and causes memory access faults.
As a side-note: buffers other than those allocated in instance memory
don't need to be physically contiguous from the GPU's perspective since
the GPU can map them into contiguous buffers using its own MMU. Mapping
these buffers through the IOMMU is unnecessary and will even lead to
performance degradation because of the additional translation. One
exception to this are compressible buffers which need large pages. In
order to enable these large pages, multiple small pages will have to be
combined into one large (I/O virtually contiguous) mapping via the
IOMMU. However, that is a topic outside the scope of this fix and isn't
currently supported. An implementation will want to explicitly create
these large pages in the Nouveau driver, so detaching from a DMA/IOMMU
mapping would still be required.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Tested-by: Nicolas Chauvet <kwizart@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
In the quest to remove all stack VLA usage from the kernel[1], this
allocates the working buffers before starting the writing so it won't
abort in the middle. This needs an initial walk of the lists to figure
out how large the buffer should be.
[1] https://lkml.kernel.org/r/CA+55aFzCG-zNmZwX4A2FQpadafLfEzK6CC=qPXydAacU1RqZWA@mail.gmail.com
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This patch unifies the naming of DRM functions for reference counting
of struct drm_device. The resulting code is more aligned with the rest
of the Linux kernel interfaces.
Signed-off-by: Thomas Zimmermann <tdz@users.sourceforge.net>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This patch unifies the naming of DRM functions for reference counting
of struct drm_gem_object. The resulting code is more aligned with the
rest of the Linux kernel interfaces.
Signed-off-by: Thomas Zimmermann <tdz@users.sourceforge.net>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This patch unifies the naming of DRM functions for reference counting
of struct drm_framebuffer. The resulting code is more aligned with the
rest of the Linux kernel interfaces.
Signed-off-by: Thomas Zimmermann <tdz@users.sourceforge.net>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Similar to commit 0bf8bf50ed ("module: Remove
const attribute from alias for MODULE_DEVICE_TABLE")
Fixes many -Wduplicate-decl-specifier warnings due to the combination of
const typeof() of already const variables.
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Smatch complains that "value" can be uninitialized when kstrtol()
returns -ERANGE.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Noticed this as I was skimming through, if we fail to allocate memory
for cli we'll end up returning without dropping the runtime PM ref we
got. Additionally, we'll even return the wrong return code! (ret most
likely will == 0 here, we want -ENOMEM).
Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Fixes various reclocking related issues on prime systems.
Signed-off-by: Karol Herbst <karolherbst@gmail.com>
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
If the entry size is too small, default to invalid values for both
boost_id and tdp_id, so as to default to the base clock in both cases.
Signed-off-by: Karol Herbst <karolherbst@gmail.com>
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
With instantaneous high precision vblank timestamping
that updates at leading edge of vblank, the emulated
"hw vblank counter" from vblank timestamping, which
increments at leading edge of vblank, and reliable
page flip execution and completion at leading edge of
vblank, we should meet the requirements for fast/
immediate vblank irq disable/enable.
This is only allowed on nv50+ gpu's, ie. the ones with
atomic modesetting. One requirement for immediate vblank
disable is that high precision vblank timestamping works
reliably all the time on all connectors. This is not the
case on all pre-nv50 parts for analog VGA outputs, where we
currently don't always have support for scanout position
queries and therefore fall back to vblank interrupt
timestamping. The implementation in nv04_head_state() does
not return valid values for vblanks, vtotal, hblanks, htotal
for VGA outputs on all cards, but those are needed for scanout
position queries.
Testing on Linux-4.12-rc5 + drm-next on a GeForce 9500 GT
(NV G96) with timing measurement equipment indicates this
works fine, so allow immediate vblank disable for power
saving.
For debugging in case of unexpected trouble, booting
with kernel cmdline option drm.vblankoffdelay=0
(or echo 0 > /sys/module/drm/parameters/vblankoffdelay)
would keep vblank irqs permanently on to approximate old
behavior.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Use drm_*_get() and drm_*_put() helpers instead of drm_*_reference() and
drm_*_unreference() helpers.
Generated by: scripts/coccinelle/api/drm-get-put.cocci
Fixes: 30ed49b55b ("drm/nouveau/kms/nv50-: move code underneath dispnv50/")
Signed-off-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Julia Lawall <julia.lawall@lip6.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Change values to u32, there's no need for them to be 64-bit.
Reported-by: Colin Ian King <colin.king@canonical.com>
Suggested-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This allows to have the related MODULE_FIRMWARE tag only
on relevant arch (arm64).
This will saves about 400k on initramfs when not relevant
Signed-off-by: Nicolas Chauvet <kwizart@gmail.com>
Acked-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Currently nouveau doesn't actually expose the state debugfs file that's
usually provided for any modesetting driver that supports atomic, even
if nouveau is loaded with atomic=1. This is due to the fact that the
standard debugfs files that DRM creates for atomic drivers is called
when drm_get_pci_dev() is called from nouveau_drm.c. This happens well
before we've initialized the display core, which is currently
responsible for setting the DRIVER_ATOMIC cap.
So, move the atomic option into nouveau_drm.c and just add the
DRIVER_ATOMIC cap whenever it's enabled on the kernel commandline. This
shouldn't cause any actual issues, as the atomic ioctl will still fail
as expected even if the display core doesn't disable it until later in
the init sequence. This also provides the added benefit of being able to
use the state debugfs file to check the current display state even if
clients aren't allowed to modify it through anything other than the
legacy ioctls.
Additionally, disable the DRIVER_ATOMIC cap in nv04's display core, as
this was already disabled there previously.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This both uses the legacy modesetting structures in a racy manner, and
additionally also doesn't even check the right variable (enabled != the
CRTC is actually turned on for atomic).
This fixes issues on my P50 regarding the dedicated GPU not entering
runtime suspend.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
A CRTC being enabled doesn't mean it's on! It doesn't even necessarily
mean it's being used. This fixes runtime PM leaks on the P50 I've got
next to me.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
When MST and atomic were introduced to nouveau, another structure that
could contain a drm_connector embedded within it was introduced; struct
nv50_mstc. This meant that we no longer would be able to simply loop
through our connector list and assume that nouveau_connector() would
return a proper pointer for each connector, since the assertion that
all connectors coming from nouveau have a full nouveau_connector struct
became invalid.
Unfortunately, none of the actual code that looped through connectors
ever got updated, which means that we've been causing invalid memory
accesses for quite a while now.
An example that was caught by KASAN:
[ 201.038698] ==================================================================
[ 201.038792] BUG: KASAN: slab-out-of-bounds in nvif_notify_get+0x190/0x1a0 [nouveau]
[ 201.038797] Read of size 4 at addr ffff88076738c650 by task kworker/0:3/718
[ 201.038800]
[ 201.038822] CPU: 0 PID: 718 Comm: kworker/0:3 Tainted: G O 4.18.0-rc4Lyude-Test+ #1
[ 201.038825] Hardware name: LENOVO 20EQS64N0B/20EQS64N0B, BIOS N1EET78W (1.51 ) 05/18/2018
[ 201.038882] Workqueue: events nouveau_display_hpd_work [nouveau]
[ 201.038887] Call Trace:
[ 201.038894] dump_stack+0xa4/0xfd
[ 201.038900] print_address_description+0x71/0x239
[ 201.038929] ? nvif_notify_get+0x190/0x1a0 [nouveau]
[ 201.038935] kasan_report.cold.6+0x242/0x2fe
[ 201.038942] __asan_report_load4_noabort+0x19/0x20
[ 201.038970] nvif_notify_get+0x190/0x1a0 [nouveau]
[ 201.038998] ? nvif_notify_put+0x1f0/0x1f0 [nouveau]
[ 201.039003] ? kmsg_dump_rewind_nolock+0xe4/0xe4
[ 201.039049] nouveau_display_init.cold.12+0x34/0x39 [nouveau]
[ 201.039089] ? nouveau_user_framebuffer_create+0x120/0x120 [nouveau]
[ 201.039133] nouveau_display_resume+0x5c0/0x810 [nouveau]
[ 201.039173] ? nvkm_client_ioctl+0x20/0x20 [nouveau]
[ 201.039215] nouveau_do_resume+0x19f/0x570 [nouveau]
[ 201.039256] nouveau_pmops_runtime_resume+0xd8/0x2a0 [nouveau]
[ 201.039264] pci_pm_runtime_resume+0x130/0x250
[ 201.039269] ? pci_restore_standard_config+0x70/0x70
[ 201.039275] __rpm_callback+0x1f2/0x5d0
[ 201.039279] ? rpm_resume+0x560/0x18a0
[ 201.039283] ? pci_restore_standard_config+0x70/0x70
[ 201.039287] ? pci_restore_standard_config+0x70/0x70
[ 201.039291] ? pci_restore_standard_config+0x70/0x70
[ 201.039296] rpm_callback+0x175/0x210
[ 201.039300] ? pci_restore_standard_config+0x70/0x70
[ 201.039305] rpm_resume+0xcc3/0x18a0
[ 201.039312] ? rpm_callback+0x210/0x210
[ 201.039317] ? __pm_runtime_resume+0x9e/0x100
[ 201.039322] ? kasan_check_write+0x14/0x20
[ 201.039326] ? do_raw_spin_lock+0xc2/0x1c0
[ 201.039333] __pm_runtime_resume+0xac/0x100
[ 201.039374] nouveau_display_hpd_work+0x67/0x1f0 [nouveau]
[ 201.039380] process_one_work+0x7a0/0x14d0
[ 201.039388] ? cancel_delayed_work_sync+0x20/0x20
[ 201.039392] ? lock_acquire+0x113/0x310
[ 201.039398] ? kasan_check_write+0x14/0x20
[ 201.039402] ? do_raw_spin_lock+0xc2/0x1c0
[ 201.039409] worker_thread+0x86/0xb50
[ 201.039418] kthread+0x2e9/0x3a0
[ 201.039422] ? process_one_work+0x14d0/0x14d0
[ 201.039426] ? kthread_create_worker_on_cpu+0xc0/0xc0
[ 201.039431] ret_from_fork+0x3a/0x50
[ 201.039441]
[ 201.039444] Allocated by task 79:
[ 201.039449] save_stack+0x43/0xd0
[ 201.039452] kasan_kmalloc+0xc4/0xe0
[ 201.039456] kmem_cache_alloc_trace+0x10a/0x260
[ 201.039494] nv50_mstm_add_connector+0x9a/0x340 [nouveau]
[ 201.039504] drm_dp_add_port+0xff5/0x1fc0 [drm_kms_helper]
[ 201.039511] drm_dp_send_link_address+0x4a7/0x740 [drm_kms_helper]
[ 201.039518] drm_dp_check_and_send_link_address+0x1a7/0x210 [drm_kms_helper]
[ 201.039525] drm_dp_mst_link_probe_work+0x71/0xb0 [drm_kms_helper]
[ 201.039529] process_one_work+0x7a0/0x14d0
[ 201.039533] worker_thread+0x86/0xb50
[ 201.039537] kthread+0x2e9/0x3a0
[ 201.039541] ret_from_fork+0x3a/0x50
[ 201.039543]
[ 201.039546] Freed by task 0:
[ 201.039549] (stack is not available)
[ 201.039551]
[ 201.039555] The buggy address belongs to the object at ffff88076738c1a8
which belongs to the cache kmalloc-2048 of size 2048
[ 201.039559] The buggy address is located 1192 bytes inside of
2048-byte region [ffff88076738c1a8, ffff88076738c9a8)
[ 201.039563] The buggy address belongs to the page:
[ 201.039567] page:ffffea001d9ce200 count:1 mapcount:0 mapping:ffff88084000d0c0 index:0x0 compound_mapcount: 0
[ 201.039573] flags: 0x8000000000008100(slab|head)
[ 201.039578] raw: 8000000000008100 ffffea001da3be08 ffffea001da25a08 ffff88084000d0c0
[ 201.039582] raw: 0000000000000000 00000000000d000d 00000001ffffffff 0000000000000000
[ 201.039585] page dumped because: kasan: bad access detected
[ 201.039588]
[ 201.039591] Memory state around the buggy address:
[ 201.039594] ffff88076738c500: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[ 201.039598] ffff88076738c580: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[ 201.039601] >ffff88076738c600: 00 00 00 00 00 00 00 00 00 00 fc fc fc fc fc fc
[ 201.039604] ^
[ 201.039607] ffff88076738c680: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[ 201.039611] ffff88076738c700: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[ 201.039613] ==================================================================
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>
Every codepath in nouveau that loops through the connector list
currently does so using the old method, which is prone to race
conditions from MST connectors being created and destroyed. This has
been causing a multitude of problems, including memory corruption from
trying to access connectors that have already been freed!
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>
The bo array has req->nr_buffers elements so the > should be >= so we
don't read beyond the end of the array.
Fixes: a1606a9596 ("drm/nouveau: new gem pushbuf interface, bump to 0.0.16")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
It was possible for this to be skipped when shutting down MST streams, and
leaving the core channel interlocked with a wndw channel update that never
happens - leading to a hung display.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Tested-By: Lyude Paul <lyude@redhat.com>
Since there's very few callers of these I've decided to do them all in
one patch. With this the unecessarily long drm_mode_connector_ prefix
is gone from the codebase! The only exception being struct
drm_mode_connector_set_property, which is part of the uapi so can't be
renamed.
Again done with sed+some manual fixups for indent issues.
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180709084016.23750-8-daniel.vetter@ffwll.ch
Use drm_connector_for_each_possible_encoder() for iterating
connector->encoder_ids[]. A bit more convenient not having
to deal with the implementation details.
v2: Replace drm_for_each_connector_encoder_ids() with
drm_connector_for_each_possible_encoder() (Daniel)
v3: Initialize nv_encoder to NULL to shut up gcc/smatch
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: nouveau@lists.freedesktop.org
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180702152927.13351-1-ville.syrjala@linux.intel.com
Ctxdmas for cursors from all heads are setup in the core channel, and due
to us tracking allocated handles per-window, we were failing with -EEXIST
on multiple-head setups trying to allocate duplicate handles.
The cursor code is hardcoded to use the core channel vram ctxdma already,
so just skip ctxdma allocation for cursor fbs to fix the issue.
Fixes: 5bca1621c0 ("drm/nouveau/kms/nv50-: move fb ctxdma tracking into windows")
Reported-by: Adam Borowski <kilobyte@angband.pl>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
- Use overflow helpers in 2-factor allocators (Kees, Rasmus)
- Introduce overflow test module (Rasmus, Kees)
- Introduce saturating size helper functions (Matthew, Kees)
- Treewide use of struct_size() for allocators (Kees)
-----BEGIN PGP SIGNATURE-----
Comment: Kees Cook <kees@outflux.net>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=/3L8
-----END PGP SIGNATURE-----
Merge tag 'overflow-v4.18-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux
Pull overflow updates from Kees Cook:
"This adds the new overflow checking helpers and adds them to the
2-factor argument allocators. And this adds the saturating size
helpers and does a treewide replacement for the struct_size() usage.
Additionally this adds the overflow testing modules to make sure
everything works.
I'm still working on the treewide replacements for allocators with
"simple" multiplied arguments:
*alloc(a * b, ...) -> *alloc_array(a, b, ...)
and
*zalloc(a * b, ...) -> *calloc(a, b, ...)
as well as the more complex cases, but that's separable from this
portion of the series. I expect to have the rest sent before -rc1
closes; there are a lot of messy cases to clean up.
Summary:
- Introduce arithmetic overflow test helper functions (Rasmus)
- Use overflow helpers in 2-factor allocators (Kees, Rasmus)
- Introduce overflow test module (Rasmus, Kees)
- Introduce saturating size helper functions (Matthew, Kees)
- Treewide use of struct_size() for allocators (Kees)"
* tag 'overflow-v4.18-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
treewide: Use struct_size() for devm_kmalloc() and friends
treewide: Use struct_size() for vmalloc()-family
treewide: Use struct_size() for kmalloc()-family
device: Use overflow helpers for devm_kmalloc()
mm: Use overflow helpers in kvmalloc()
mm: Use overflow helpers in kmalloc_array*()
test_overflow: Add memory allocation overflow tests
overflow.h: Add allocation size calculation helpers
test_overflow: Report test failures
test_overflow: macrofy some more, do more tests for free
lib: add runtime test of check_*_overflow functions
compiler.h: enable builtin overflow checkers and add fallback code
One of the more common cases of allocation size calculations is finding
the size of a structure that has a zero-sized array at the end, along
with memory for some number of elements for that array. For example:
struct foo {
int stuff;
void *entry[];
};
instance = kmalloc(sizeof(struct foo) + sizeof(void *) * count, GFP_KERNEL);
Instead of leaving these open-coded and prone to type mistakes, we can
now use the new struct_size() helper:
instance = kmalloc(struct_size(instance, entry, count), GFP_KERNEL);
This patch makes the changes for kmalloc()-family (and kvmalloc()-family)
uses. It was done via automatic conversion with manual review for the
"CHECKME" non-standard cases noted below, using the following Coccinelle
script:
// pkey_cache = kmalloc(sizeof *pkey_cache + tprops->pkey_tbl_len *
// sizeof *pkey_cache->table, GFP_KERNEL);
@@
identifier alloc =~ "kmalloc|kzalloc|kvmalloc|kvzalloc";
expression GFP;
identifier VAR, ELEMENT;
expression COUNT;
@@
- alloc(sizeof(*VAR) + COUNT * sizeof(*VAR->ELEMENT), GFP)
+ alloc(struct_size(VAR, ELEMENT, COUNT), GFP)
// mr = kzalloc(sizeof(*mr) + m * sizeof(mr->map[0]), GFP_KERNEL);
@@
identifier alloc =~ "kmalloc|kzalloc|kvmalloc|kvzalloc";
expression GFP;
identifier VAR, ELEMENT;
expression COUNT;
@@
- alloc(sizeof(*VAR) + COUNT * sizeof(VAR->ELEMENT[0]), GFP)
+ alloc(struct_size(VAR, ELEMENT, COUNT), GFP)
// Same pattern, but can't trivially locate the trailing element name,
// or variable name.
@@
identifier alloc =~ "kmalloc|kzalloc|kvmalloc|kvzalloc";
expression GFP;
expression SOMETHING, COUNT, ELEMENT;
@@
- alloc(sizeof(SOMETHING) + COUNT * sizeof(ELEMENT), GFP)
+ alloc(CHECKME_struct_size(&SOMETHING, ELEMENT, COUNT), GFP)
Signed-off-by: Kees Cook <keescook@chromium.org>
Inserted wait-for-gr-idle in the places it seems that RM does it, seems
to prevent some random mmio timeouts on Quadro GV100.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
It's better to use "list_for_each_entry_from_reverse" for iterating list
than "for loop" as it makes the code more clear to read.
This patch replace "for loop" with "list_for_each_entry_from_reverse"
and "start" variable with "cstate" which helps in refactoring
the code and also "cstate" variable is more commonly used in the other
functions.
changes in v2:
"start" variable is removed, before "cstate" variable was removed
but "cstate" is more common so preferred "cstate" over "start".
Signed-off-by: Arushi Singhal <arushisinghal19971997@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
A NV34 GPU was seeing temp and pwm entries in hwmon, which would error
out when read. These should not have been visible, but also the whole
hwmon object should just not have been registered in the first place.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The method struct vga_switcheroo_handler::get_client_id() is defined
as returning an 'enum vga_switcheroo_client_id' but the implementation
in this driver, nouveau_dsm_get_client_id(), returns an 'int'.
Fix this by returning 'enum vga_switcheroo_client_id' in this driver too.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The method struct drm_connector_helper_funcs::mode_valid is defined
as returning an 'enum drm_mode_status' but the driver implementation
for this method uses an 'int' for it.
Fix this by using 'enum drm_mode_status' in the driver too.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
VEID support hacked in here, as it's the most convenient place for now.
Will be refined once it's better understood.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
When only the position of a window changes, there's no need to submit
an image update as well.
Will be required to support the overlays, and Volta windows.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Window visibility is going to become a little more complicated with the
upcoming LUT changes, so store the calculated value to avoid needing to
recalculate the armed state again.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Instead of windows returning their core channel interlock mask if they
know core has been modified, it's recorded unconditionally and used if
required when update methods are emitted.
This will be required to support Volta.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Fences attached to deferred client work items now originate from channels
belonging to the client, meaning we can be certain they've been signalled
before we destroy a client.
This closes a race that could happen if the dma_fence_wait_timeout() call
didn't succeed. When the fence was later signalled, a use-after-free was
possible.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
As VMAs are per-client, unlike buffers, this allows us to avoid referencing
foreign fences (those that belong to another client/driver) from the client
deferred work handler, and prevent some not-fun race conditions that can be
triggered when a fence stalls.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
We previously only did this for push buffers, but an upcoming patch will
need to attach fences to all VMAs to resolve another issue.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
These were missed the first time around due to the driver version I traced
using the older registers still.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
There are differences on GM200 and newer too, but we can't fix them there
as they come from firmware packages.
A request has been made to NVIDIA to release updated firmware.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
There's a number of places that require this data, so let's separate out
the calculations to ensure they remain consistent.
This is incorrect for GM200 and newer, but will produce the same results
as we did before.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
There's also a couple of hardcoded tables for a couple of very specific
configurations that NVGPU's algorithm didn't work for.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The algorithm for GM200 and newer matches RM for all the boards I have, but
I don't have enough data to try and figure something out for earlier boards,
so these will still write zeroes to the table as we did before.
The code in NVGPU isn't helpful here, it appears to handle specific cases.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
I don't think this is done after Fermi, NVGPU used to do it but removed
the code, and I've not seen RM traces touching it either.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
I haven't yet been able to find a fully programatic way of calculating the
same mapping as NVIDIA for GF100-GF119, so the algorithm partially depends
on data tables for specific configurations.
I couldn't find traces for every possibility, so the algorithm will switch
to a mapping similar to what GK104-GM10x use if it encounters one. We did
the wrong thing before anyway, so shouldn't matter too much.
The algorithm used in the GK104 implementation was ported from NVGPU.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Newer HW doesn't appear to send this event, which will cause long delays
in runlist updates if they don't complete immediately.
RM doesn't use these events anywhere, and an NVGPU commit message notes
that polling is the preferred method even on HW that supports the event.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
We didn't used to be aware that runlist/engine IDs weren't the same thing,
or that there was such variability in configuration between GPUs.
By exposing this information to a client, and giving it explicit control
of which runlist it's allocating a channel on, we're able to make better
choices.
The immediate effect of this is that on GPUs where CE0 is the "GRCE", we
will now be allocating a copy engine running asynchronously to GR for BO
migrations - as intended.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
We have a need to fetch data from GPU-specific sub-devices that is not
tied to any particular engine object.
This commit provides the framework to support such queries.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This will be required to support Volta, but also allows us to remove code
that's duplicated for each channel type already.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Introduces a new method of defining channels available from the display,
common to all channel types, allowing for more flexibility in available
channel types/counts, and reducing the amount of boiler-plate required.
This will be required to support Volta.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Engines are initialised on an as-needed basis, so this results in the
same behaviour, whilst allowing us to simplify things a bit.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
We should be reading registers to determine which subunits are really
present on a given board, and this needs to be done after DEVINIT.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Likely a rebase bug. Should have no impact in default configuration due
to using per-instance setting by default.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
In preparation to enabling -Wvla, remove VLA. In this particular
case directly use macro NVKM_MSGQUEUE_CMDLINE_SIZE instead of local
variable cmdline_size. Also, remove cmdline_size as it is not
actually useful anymore.
The use of stack Variable Length Arrays needs to be avoided, as they
can be a vector for stack exhaustion, which can be both a runtime bug
or a security flaw. Also, in general, as code evolves it is easy to
lose track of how big a VLA can get. Thus, we can end up having runtime
failures that are hard to debug.
Also, fixed as part of the directive to remove all VLAs from
the kernel: https://lkml.org/lkml/2018/3/7/621
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
gcc points out a buffer that is clearly too small to be used
in a meaningful way, as the 'sizeof(*args) + argc > sizeof(stack)'
will always fail:
In function 'memcpy',
inlined from 'nvif_vmm_map' at drivers/gpu/drm/nouveau/nvif/vmm.c:55:2:
include/linux/string.h:353:9: error: '__builtin_memcpy' offset 40 is out of the bounds [0, 16] of object 'stack' with type 'u8[16]' {aka 'unsigned char[16]'} [-Werror=array-bounds]
return __builtin_memcpy(p, q, size);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/gpu/drm/nouveau/nvif/vmm.c: In function 'nvif_vmm_map':
drivers/gpu/drm/nouveau/nvif/vmm.c:40:5: note: 'stack' declared here
This makes the buffer large enough so it should serve the purpose
that the author presumably had in mind. Alternatively we could
just get rid of it completely and simplify the code at the cost
of always doing the kmalloc (as we do in the current version).
Fixes: 920d2b5ef2 ("drm/nouveau/mmu: define user interfaces to mmu vmm opertaions")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
-----BEGIN PGP SIGNATURE-----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=4NJV
-----END PGP SIGNATURE-----
Merge drm-fixes-for-v4.17-rc6-urgent into drm-next
Need to backmerge some nouveau fixes to reduce
the nouveau -next conflicts a lot.
Signed-off-by: Dave Airlie <airlied@redhat.com>
-----BEGIN PGP SIGNATURE-----
iQEcBAABAgAGBQJauCZfAAoJEHm+PkMAQRiGWGUH/2rhdQDkoJpYWnjaQkolECG8
MxpGE7nmIIHxQcbSDdHTGJ8IhVm6Z5wZ7ym/PwCDTT043Y1y341sJrIwL2/nTG6d
HVidk8hFvgN6QzlzVAHT3ZZMII/V9Zt+VV5SUYLGnPAVuJNHo/6uzWlTU5g+NTFo
IquFDdQUaGBlkKqby+NoAFnkV1UAIkW0g22cfvPnlO5GMer0gusGyVNvVp7TNj3C
sqj4Hvt3RMDLMNe9RZ2pFTiOD096n8FWpYftZneUTxFImhRV3Jg5MaaYZm9SI3HW
tXrv/LChT/F1mi5Pkx6tkT5Hr8WvcrwDMJ4It1kom10RqWAgjxIR3CMm448ileY=
=YKUG
-----END PGP SIGNATURE-----
Backmerge tag 'v4.16-rc7' into drm-next
Linux 4.16-rc7
This was requested by Daniel, and things were getting
a bit hard to reconcile, most of the conflicts were
trivial though.
It's better to use list_entry instead of list_{next/prev}_entry
as it makes the code more clear to read.
This patch replace list_entry with list_{next/prev}_entry.
Signed-off-by: Arushi Singhal <arushisinghal19971997@gmail.com>
Acked-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1522000893-5331-3-git-send-email-arushisinghal19971997@gmail.com
Unbinding nouveau on a dual GPU MacBook Pro oopses because we iterate
over the bl_connectors list in nouveau_backlight_exit() but skipped
initializing it in nouveau_backlight_init(). Stacktrace for posterity:
BUG: unable to handle kernel NULL pointer dereference at 0000000000000010
IP: nouveau_backlight_exit+0x2b/0x70 [nouveau]
nouveau_display_destroy+0x29/0x80 [nouveau]
nouveau_drm_unload+0x65/0xe0 [nouveau]
drm_dev_unregister+0x3c/0xe0 [drm]
drm_put_dev+0x2e/0x60 [drm]
nouveau_drm_device_remove+0x47/0x70 [nouveau]
pci_device_remove+0x36/0xb0
device_release_driver_internal+0x157/0x220
driver_detach+0x39/0x70
bus_remove_driver+0x51/0xd0
pci_unregister_driver+0x2a/0xa0
nouveau_drm_exit+0x15/0xfb0 [nouveau]
SyS_delete_module+0x18c/0x290
system_call_fast_compare_end+0xc/0x6f
Fixes: b53ac1ee12 ("drm/nouveau/bl: Do not register interface if Apple GMUX detected")
Cc: stable@vger.kernel.org # v4.10+
Cc: Pierre Moreau <pierre.morrow@free.fr>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Commit 7110c89bb8852ff8b0f88ce05b332b3fe22bd11e ("mmu: swap out round
for ALIGN") replaced two calls to round/rounddown with ALIGN/ALIGN_DOWN,
but erroneously applied ALIGN_DOWN to a different variable (addr) and left
intended variable (tail) not rounded/ALIGNed.
As a result screen corruption, X lockups are observable. An example of kernel
log of affected system with NV98 card where it was bisected:
nouveau 0000:01:00.0: gr: TRAP_M2MF 00000002 [IN]
nouveau 0000:01:00.0: gr: TRAP_M2MF 00320951 400007c0 00000000 04000000
nouveau 0000:01:00.0: gr: 00200000 [] ch 1 [000fbbe000 DRM] subc 4 class 5039
mthd 0100 data 00000000
nouveau 0000:01:00.0: fb: trapped read at 0040000000 on channel 1
[0fbbe000 DRM]
engine 00 [PGRAPH] client 03 [DISPATCH] subclient 04 [M2M_IN] reason 00000006
[NULL_DMAOBJ]
Fixes bug 105173 ("[MCP79][Regression] Unhandled NULL pointer dereference in
nvkm_object_unmap since kernel 4.15")
https://bugs.freedesktop.org/show_bug.cgi?id=105173
Fixes: 7110c89bb885 ("mmu: swap out round for ALIGN ")
Tested-by: Pierre Moreau <pierre.morrow@free.fr>
Reviewed-by: Pierre Moreau <pierre.morrow@free.fr>
Signed-off-by: Maris Nartiss <maris.nartiss@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Cc: stable@vger.kernel.org # v4.15+
Instead of calculating the size in bytes just to recalculate the number
of pages from it pass the BO directly to the function.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Commit 5addcf0a5f ("nouveau: add runtime PM support (v0.9)") prevents
runtime suspend of the GPU if its integrated HDA controller is not bound
to a driver. The rationale appears to be that probing the HDA fails if
the GPU is in D3cold.
However we now use a device link to ensure that the GPU is runtime
resumed while the HDA controller is probed, rendering this safety
measure obsolete. Remove it.
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/77e0ab74f3377ea9b6acf8fab624acfb4f7dbeca.1520068884.git.lukas@wunner.de
Back in 2013, runtime PM for GPUs with integrated HDA controller was
introduced with commits 0d69704ae3 ("gpu/vga_switcheroo: add driver
control power feature. (v3)") and 246efa4a07 ("snd/hda: add runtime
suspend/resume on optimus support (v4)").
Briefly, the idea was that the HDA controller is forced on and off in
unison with the GPU.
The original code is mostly still in place even though it was never a
100% perfect solution: E.g. on access to the HDA controller, the GPU
is powered up via vga_switcheroo_runtime_resume_hdmi_audio() but there
are no provisions to keep it resumed until access to the HDA controller
has ceased: The GPU autosuspends after 5 seconds, rendering the HDA
controller inaccessible.
Additionally, a kludge is required when hda_intel.c probes: It has to
check whether the GPU is powered down (check_hdmi_disabled()) and defer
probing if so.
However in the meantime (in v4.10) the driver core has gained a feature
called device links which promises to solve such issues in a clean way:
It allows us to declare a dependency from the HDA controller (consumer)
to the GPU (supplier). The PM core then automagically ensures that the
GPU is runtime resumed as long as the HDA controller's ->probe hook is
executed and whenever the HDA controller is accessed.
By default, the HDA controller has a dependency on its parent, a PCIe
Root Port. Adding a device link creates another dependency on its
sibling:
PCIe Root Port
^ ^
| |
| |
HDA ===> GPU
The device link is not only used for runtime PM, it also guarantees that
on system sleep, the HDA controller suspends before the GPU and resumes
after the GPU, and on system shutdown the HDA controller's ->shutdown
hook is executed before the one of the GPU. It is a complete solution.
Using this functionality is as simple as calling device_link_add(),
which results in a dmesg entry like this:
pci 0000:01:00.1: Linked as a consumer to 0000:01:00.0
The code for the GPU-governed audio power management can thus be removed
(except where it's still needed for legacy manual power control).
The device link is added in a PCI quirk rather than in hda_intel.c.
It is therefore legal for the GPU to runtime suspend to D3cold even if
the HDA controller is not bound to a driver or if CONFIG_SND_HDA_INTEL
is not enabled, for accesses to the HDA controller will cause the GPU to
wake up regardless if they're occurring outside of hda_intel.c (think
config space readout via sysfs).
Contrary to the previous implementation, the HDA controller's power
state is now self-governed, rather than GPU-governed, whereas the GPU's
power state is no longer fully self-governed. (The HDA controller needs
to runtime suspend before the GPU can.)
It is thus crucial that runtime PM is always activated on the HDA
controller even if CONFIG_SND_HDA_POWER_SAVE_DEFAULT is set to 0 (which
is the default), lest the GPU stays awake. This is achieved by setting
the auto_runtime_pm flag on every codec and the AZX_DCAPS_PM_RUNTIME
flag on the HDA controller.
A side effect is that power consumption might be reduced if the GPU is
in use but the HDA controller is not, because the HDA controller is now
allowed to go to D3hot. Before, it was forced to stay in D0 as long as
the GPU was in use. (There is no reduction in power consumption on my
Nvidia GK107, but there might be on other chips.)
The code paths for legacy manual power control are adjusted such that
runtime PM is disabled during power off, thereby preventing the PM core
from resuming the HDA controller.
Note that the device link is not only added on vga_switcheroo capable
systems, but for *any* GPU with integrated HDA controller. The idea is
that the HDA controller streams audio via connectors located on the GPU,
so the GPU needs to be on for the HDA controller to do anything useful.
This commit implicitly fixes an unbalanced runtime PM ref upon unbind of
hda_intel.c: On ->probe, a runtime PM ref was previously released under
the condition "azx_has_pm_runtime(chip) || hda->use_vga_switcheroo", but
on ->remove a runtime PM ref was only acquired under the first of those
conditions. Thus, binding and unbinding the driver twice on a
vga_switcheroo capable system caused the runtime PM refcount to drop
below zero. The issue is resolved because the AZX_DCAPS_PM_RUNTIME flag
is now always set if use_vga_switcheroo is true.
For more information on device links please refer to:
https://www.kernel.org/doc/html/latest/driver-api/device_link.html
Documentation/driver-api/device_link.rst
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Reviewed-by: Peter Wu <peter@lekensteyn.nl>
Tested-by: Kai Heng Feng <kai.heng.feng@canonical.com> # AMD PowerXpress
Tested-by: Mike Lothian <mike@fireburn.co.uk> # AMD PowerXpress
Tested-by: Denis Lisov <dennis.lissov@gmail.com> # Nvidia Optimus
Tested-by: Peter Wu <peter@lekensteyn.nl> # Nvidia Optimus
Tested-by: Lukas Wunner <lukas@wunner.de> # MacBook Pro
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://patchwork.freedesktop.org/patch/msgid/51bd38360ff502a8c42b1ebf4405ee1d3f27118d.1520068884.git.lukas@wunner.de
