MSIs were only problematic on some old, broken chipsets. But now that we
already see systems where PCI legacy interrupts are somewhat flaky, it's
really time to move to MSIs.

v2 (Ben Skeggs): blacklist BR02 boards
Signed-off-by: NLucas Stach <dev@lynxeye.de>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Commit ea9197cc effectively enabled the
use of an improved DAC detection code, but introduced a regression on
the original nv50 chipset, causing a ghost monitor to be detected.

v2 (Ben Skeggs): the offending line was likely a thinko, removed it for
all chipsets (tested nv50 and nve6 to cover entire range) and added
some additional debugging.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=67382Tested-by: NMartin Peres <martin.peres@labri.fr>
Signed-off-by: NEmil Velikov <emil.l.velikov@gmail.com>
Cc: <stable@vger.kernel.org> # 3.9+
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

i2c_bit_add_bus can call the pre_xfer function, which expects the func
pointer to be set. Pass in func to the port creation logic so that it is
set before i2c_bit_add_bus.

See https://bugs.freedesktop.org/show_bug.cgi?id=68456Reported-by: NHans-Peter Deifel <hpdeifel@gmx.de>
Tested-by: NHans-Peter Deifel <hpdeifel@gmx.de>
Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Therm uses 3 ptimer alarms. Two to drive the fan and one for polling the
temperature. When suspending/resuming, alarms will never be fired.
As we are checking if there isn't an alarm pending before rescheduling
another one, we end up never checking temperature or updating the
fan speed.

This commit also adds debug messages to be able to spot more easily
if this case happens again in the future. Sorry for the spam if you
activate the debug level though.
Tested-by: NDash Four <mr.dash.four@googlemail.com>

v2:
- fix temperature polling too
Signed-off-by: NMartin Peres <martin.peres@labri.fr>
Tested-by: NMartin Peres <martin.peres@labri.fr>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Since alarms don't play well with suspend, it is important every alarm
user cancels his tasks before suspending.

The task should be rescheduled on resume.
Signed-off-by: NMartin Peres <martin.peres@labri.fr>
Tested-by: NMartin Peres <martin.peres@labri.fr>
Tested-by: NDash Four <mr.dash.four@googlemail.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

This can be useful if some parts of Nouveau try to calculate the time
between two events.  Without this patch, the time difference would be
negative in the case where the computer is suspended/resumed between
two events.

This patch should fix fan speed probing when done while suspending/resuming.

Solve this by saving the current time before suspending and by restoring it
on resume.
Signed-off-by: NMartin Peres <martin.peres@labri.fr>
Tested-by: NMartin Peres <martin.peres@labri.fr>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

If the fan was in manual or auto mode, we should restore the fan speed
that was previously set when resuming.

The initial pwm value is saved when loading the module.
Signed-off-by: NMartin Peres <martin.peres@labri.fr>
Tested-by: NMartin Peres <martin.peres@labri.fr>
Tested-by: NDash Four <mr.dash.four@googlemail.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NEmil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: NMartin Peres <martin.peres@labri.fr>
Tested-by: NMartin Peres <martin.peres@labri.fr>
Tested-by: NDash Four <mr.dash.four@googlemail.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

The values are already stored on chipset specific basis in the ctor.
Make the most of them and simplify the code further by using a temporary
variable to avoid code duplication.
Signed-off-by: NEmil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

For NV98+, BSP/VP/PPP are all FUC-based engines. Hook them all up in the
same way as NVC0, but with a couple of different values. Also make sure
that the PPP engine is handled in the fifo/mc/vm.
Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

This hooks nouveau up to the runtime PM system to enable
dynamic power management for secondary GPUs in switchable
and optimus laptops.

a) rewrite suspend/resume printks to hide them during dynamic s/r
to avoid cluttering logs
b) add runtime pm suspend to irq handler, crtc display, ioctl handler,
connector status,
c) handle hdmi audio dynamic power on/off using magic register.

v0.5:
make sure we hit D3 properly
fix fbdev_set_suspend locking interaction, we only will poweroff if we have no
active crtcs/fbcon anyways.
add reference for active crtcs.
sprinkle mark last busy for autosuspend timeout

v0.6:
allow more flexible debugging - to avoid log spam
add option to enable/disable dynpm
got to D3Cold

v0.7:
add hdmi audio support.

v0.8:
call autosuspend from idle, so pci config space access doesn't go straight
back to sleep, this makes starting X faster.
only signal usage if we actually handle the irq, otherwise usb keeps us awake.
fix nv50 display active powerdown

v0.9:
use masking function to enable hdmi audio
set busy when we fail to suspend
Signed-off-by: NDave Airlie <airlied@redhat.com>

Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Allocating type=0 marks the memory as free. This allows the ltcg memory
to be allocated twice.

Add a BUG_ON in core/mm.c to prevent this ever happening again.
Signed-off-by: NMaarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Some registers were not initialized in init, this causes them to be
uninitialized after suspend.
Signed-off-by: NMaarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Commit dceef5d8 (drm/nouveau/fb: initialise vram controller as pfb
sub-object) moved some code around and introduced these null derefs.
pfb->ram is set to the new ram object outside of this ctor.
Reported-by: NRonald Uitermark <ronald645@gmail.com>
Tested-by: NRonald Uitermark <ronald645@gmail.com>
Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

There is no need to pass constants via stack. The width may be explicitly
specified in the format.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: NDavid Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Never used to be required, but a recent change made it necessary.
Reported-by: NMaarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Looks like the rewrite in commit ebb945a9 ("drm/nouveau: port all
engines to new engine module format") missed that one little detail.
Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Unused and irrelavant since the code move of DP training/linkcontrol interrupt
Signed-off-by: NEmil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Commits 0a9e2b95 (drm/nvd0/disp: move HDA codec setup to core) and
a4feaf4e (drm/nva3/disp: move hda codec handling to core) moved code
around but neglected to fill data up to 0x60 as before. This caused
/proc/asound/cardN/eld#3.0 to show eld_valid as 0. With this patch, that
file is again populated with the correct data.

See https://bugs.freedesktop.org/show_bug.cgi?id=67051Reported-and-tested-by: NAlex <alupu01@gmail.com>
Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

object->engine is null, which leads to a null deref down the line
Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Since the original merge of nouveau to upstream kernel, we were assuming
that nv90 (and later) cards have 32 lines.

Based on mmio traces of the binary driver, as well as PBUS error messages
during read/write of the e070/e074 registers, we can conclude that nv92
has only 16 lines whereas nv94 (and later) cards have 32.
Reported-and-tested-by: NDavid M. Lloyd <david.lloyd@redhat.com>
Signed-off-by: NEmil Velikov <emil.l.velikov@gmail.com>
Cc: dri-devel@lists.freedesktop.org
Cc: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Kernel panic caused by list corruption in ltcg seems to indicate a
concurrency issue.

Take mutex of pfb like nv50_ram_put() to eliminate concurrency.

V2: Separate critical section into separate function, avoid taking the
    lock twice on NVC0
Signed-off-by: NRoy Spliet <r.spliet@student.tudelft.nl>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

The current logic is wrong since we send fw->size >> 8 to the
card. Rounding the size up by 0x100 and 0x1000 didn't seem to help,
the card still hung, so go back to what the blob does -- 0x40000.
Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

This prevents 100% cpu usage on fermi cards when the exit interrupt
from the secret scrubber is not acked.
Signed-off-by: NMaarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

"drm/nve0-/gr: some new gpc registers can have multiple copies"
5ee86c41 caused a regression for nvc0, because the bit indicating last
transfer has occured was no longer set, resulting in random system lockups.
Reported-by: NRonald Uitermark <ronald645@gmail.com>
Tested-by: NRonald Uitermark <ronald645@gmail.com>
Signed-off-by: NMaarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

This is the nva3 counterpart to commit beba44b1 (drm/nv84/disp: Fix
HDMI audio regression). The regression happened as a result of
refactoring in commit 8e9e3d2d (drm/nv84/disp: move hdmi control into
core).
Reported-and-tested-by: NMax Baldwin <archerseven@gmail.com>
Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>
Cc: stable@vger.kernel.org

The commit

   commit 476e84e1
   Author: Ben Skeggs <bskeggs@redhat.com>
   Date:   Mon Feb 11 09:24:23 2013 +1000

       drm/nv50-/disp: initial supervisor support for off-chip encoders

changed the write mask in one of the interrupt functions for on-chip encoders,
causing a regression in certain VGA dual-head setups. This commit reintroduces
the mask thus resolving the regression

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66129Reported-and-Tested-by: NYves-Alexis <corsac@debian.org>
Cc: stable@vger.kernel.org [3.9+]
CC: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: NEmil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Page tables on nv50 take 48kB, which can be hard to allocate in one piece.
Let's use vmalloc.
Signed-off-by: NMarcin Slusarz <marcin.slusarz@gmail.com>
Cc: stable@vger.kernel.org [3.7+]
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

NVC1/NVD9 are the only chipsets that should have anything different
happen on them after this.  We previously weren't doing these
register modifications, and NVIDIA do.
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

Signed-off-by: NBen Skeggs <bskeggs@redhat.com>