With runtime power save feature enabled, Headphone hotplug
event will not be detected while controller/codec in D3. HDA has
feature WAKEEN to let codec wake up system if controller is in D3 or
system in S3.(HDA Spec 4.5.9.2/3). Codec can send out INT or wake up
controller depending on whether CIE or GIE enabled.(Figure 4, Interupt
structure).

The controller must be in RESET mode after enter runtime-suspend, otherwise
it will not be waken up even if codec send out wake-up event. And STATESTS
will be cleared after controller brought out of RESET mode.

This patch only enable WAKEEN for runtime-suspend(Controller D3) mode,
not for system S3 mode. with tool "evtest", Headphone hotplug events
could be cought and reported successfully.

[fixed an unused variable warning by tiwai]
Signed-off-by: NWang Xingchao <xingchao.wang@linux.intel.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Register STATESTS is 16-bit length, use correct API for read/write.
Signed-off-by: NWang Xingchao <xingchao.wang@linux.intel.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

This patch is a cleanup to the previous patch "reset hda link during system/
runtime suspend".

In this patch
- azx_enter_link_reset() and azx_exit_link_reset() are defined for entering and
  exiting the link reset respectively. azx_link_reset() is no longer used and
  replaced by azx_enter_link_reset().
- azx_reset() reuses the above two new functions for a link reset cycle
Signed-off-by: NMengdong Lin <mengdong.lin@intel.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

If all the codecs report ClkStopOK (OK to stop bus clock) after being put to
D3, this patch will reset the HDA link before the controller is put to D3.

So the link will be in reset during system or runtime suspend, the bus clock
stops and the codecs are in D3(ClkStop) state.

This may help to reduce power consumption by dozens of mW on some peripheral
hda codecs.
Signed-off-by: NMengdong Lin <mengdong.lin@intel.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

For Intel Haswell chip, HDA controller and codec have
power well dependency from GPU side. This patch added support
to request/release power well in audio driver. Power save
feature should be enabled to get runtime power saving.

There's deadlock when request_module(i915) in azx_probe.
It looks like:
device_lock(audio pci device) -> azx_probe -> module_request
(or symbol_request) -> modprobe (userspace) -> i915 init ->
drm_pci_init -> pci_register_driver -> bus_add_driver -> driver_attach ->
which in turn tries all locks on pci bus, and when it tries the one on the
audio device, it will deadlock.

This patch introduce a work to store remaining probe stuff, and let
request_module run in safe work context.
Signed-off-by: NWang Xingchao <xingchao.wang@linux.intel.com>
Reviewed-by: NTakashi Iwai <tiwai@suse.de>
Reviewed-by: NLiam Girdwood <liam.r.girdwood@intel.com>
Reviewed-by: NDavid Henningsson <david.henningsson@canonical.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

This is a preliminary work for the upcoming Haswell HDMI audio fixes.

azx_first_init() function can be safely called after the f/w loader,
since the f/w loader doesn't require the sound hardware initialization
beforehand.  Moving it into azx_probe_continue() cleans up the code
flow a bit.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The device can support runtime PM no matter whether it support
signal wakeup or not. For some chips like Haswell which doesnot
support PME by default, this patch let haswell Display HD-A controller
enter runtime suspend, and bring more power saving whith power-well
feature enabled.
Signed-off-by: NWang Xingchao <xingchao.wang@linux.intel.com>
Reviewed-by: NTakashi Iwai <tiwai@suse.de>
Reviewed-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

When a codec is powered off, some systems don't respond properly after
D3 FG transition, while the driver still expects the response and
tries to fall back to different modes (polling and single-cmd).  When
the fallback happens, the driver stays in that mode, and falling back
to the single-cmd mode means it'll loose the unsol event handling,
too.

The unresponsiveness at D3 isn't too serious, thus this fallback is
mostly superfluous.  We can gracefully ignore the error there so that
the driver keeps the normal operation mode.

This patch adds a new bit flag for codec read/write, set in the power
transition stage, which is notified to the controller driver via a new
bus->no_response_fallback flag.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

As drvdata is cleared to NULL at probe failure or at removal by the
driver core, we don't have to call pci_set_drvdata(pci, NULL) any
longer in each driver.

The only remaining pci_set_drvdata(NULL) is in azx_firmware_cb() in
hda_intel.c.  Since this function itself releases the card instance,
we need to clear drvdata here as well, so that it won't be released
doubly in the remove callback.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Add HD Audio Device PCI ID for the Intel BayTrail platform.
Signed-off-by: NChew, Chiau Ee <chiau.ee.chew@intel.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

The audio driver mistakenly allows 64 bit addresses to be created for
the audio driver on Nvidia GPUs.  Unfortunately, the hardware normally
only supports up to 40 bits of DMA.  This can cause system panics as
well as misdirected data when the address is > 40 bits as the upper
part the address is truncated.
Signed-off-by: NMike Travis <travis@sgi.com>
Reviewed-by: NMike Habeck <habeck@sgi.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

For capture, the delay through the codec contributes to the time stamp
of the sample recorded at the A to D.  Rename the codec time stamp
function appropriately.
Signed-off-by: NDylan Reid <dgreid@chromium.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

For playback add the codec-side delay to the timestamp, for capture
subtract it.  This brings the timestamps in line with the time that
was recently added to the delay reporting.
Signed-off-by: NDylan Reid <dgreid@chromium.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Add a new codec PCM ops, get_delay(), to obtain the codec/stream-
specific PCM delay count.  When it's NULL, nothing changes.

This new feature was requested for CA0132, which has significant
delays in the path depending on the running DSP code.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

This reverts commit 6ab31741.

The commit [6ab31741: ALSA: hda - Allow power_save_controller option
override DCAPS] changed the behavior of power_save_controller so that
it can override the driver capability.  This assumed that this option
is rarely changed dynamically unlike power_save option.  Too naive.

It turned out that the user-space power-management tool tries to set
power_save_controller option to 1 together with power_save option
without knowing what's actually doing.  This enabled forcibly the
runtime PM of the controller,  which is known to be broken om many
chips thus disabled as default.

So, the only sane fix is to revert this commit again.  It was intended
to ease debugging/testing for runtime PM enablement, but obviously we
need another way for it.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=56171Reported-and-tested-by: NNikita Tsukanov <keks9n@gmail.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

The current DSP loader code abuses snd_hda_lock_devices() for ensuring
the DSP loader not conflicting with the other normal operations.  But
this trick obviously doesn't work for the PM resume since the streams
are kept opened there where snd_hda_lock_devices() returns -EBUSY.
That means we need another lock mechanism instead of abuse.

This patch provides the new lock state to azx_dev.  Theoretically it's
possible that the DSP loader conflicts with the stream that has been
already assigned for another PCM.  If it's running, the DSP loader
should simply fail.  If not -- it's the case for PM resume --, we
should assign this stream temporarily to the DSP loader, and take it
back to the PCM after finishing DSP loading.  If the PCM is operated
during the DSP loading, it should get an error, too.
Reported-and-tested-by: NDylan Reid <dgreid@chromium.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

We've got a regression report wrt the IRQ issue related with the
power-save on a Dell machine, and disabling runtime PM works around.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=53441

Cc: <stable@vger.kernel.org> [v3.7+]
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

This patch fixes a few obvious bugs in DSP loader stuff:
- Fix possible memory leaks in the error path
- Avoid double-free calls in dma_reset()
- Properly set/unset WC bits for DMA buffers
- Add missing error status checks
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

This patch adds the HD Audio Device IDs for the Intel Wellsburg PCH
Signed-off-by: NJames Ralston <james.d.ralston@intel.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

... looks like we need this for stable operations.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Add new PCI ID 0x0a0c for Haswell ULT platform.
Signed-off-by: NWang Xingchao <xingchao.wang@linux.intel.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

For non-snoop mode, we fiddle with the page attributes of CORB/RIRB
and the position buffer, but also the ring buffers.  The problem is
that the current code blindly assumes that the buffer is contiguous.
However, the ring buffers may be SG-buffers, thus a wrong vmapped
address is passed there, leading to Oops.

This patch fixes the handling for SG-buffers.

Bugzilla: https://bugzilla.novell.com/show_bug.cgi?id=800701

Cc: <stable@vger.kernel.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Currently we use LPIB forcibly for both playback and capture for
Poulsbo and Oaktrail devices, and this seems rather problematic.
The recent fix for LPIB delay count seems working well with these
devices, so let's enable it instead.
Reported-by: NMartin Weishart <martin.weishart@telosalliance.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Add an overflow check of CORB in HD-audio controller and codec drivers
so that flood of sequential writes would work properly.
In the controller side, add a check of CORB read-pointer to make
returning -EAGAIN when it's full.  Meanwhile in the codec side, when
-EAGAIN error is received, it retries the write after flushing the
pending verbs (calling get_response() essentially does it).
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Change the power_save_controller option to bint from bool so that user
can override the runtime PM capability bit and force to enable or
disable the runtime PM.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

We've got a few bug reports that the runtime D3 results in the dead
HD-audio controller.  It seems that the problem is in a deeper level
than the sound driver itself, so as a temporal solution, disable the
feature for these controllers again.
Reported-and-tested-by: NVincent Blut <vincent.debian@free.fr>
Reported-and-tested-by: NMaurizio Avogadro <mavoga@gmail.com>
Cc: <stable@vger.kernel.org> [v3.7]
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Resuming a switcheroo'd HDA controller hangs since the completion
is one-shot (thus works the first time). Fix by using completions
that explictly need rearming, so remain fired before.
Signed-off-by: NDaniel J Blueman <daniel@quora.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

The runtime_idle callback is the right place to check the suspend
capability, but currently we do it wrongly in the runtime_suspend
callback.  This leads to a kernel error message like:
   pci_pm_runtime_suspend(): azx_runtime_suspend+0x0/0x50 [snd_hda_intel] returns -11
and the runtime PM core would even repeat the attempts.
Reported-and-tested-by: NBorislav Petkov <bp@alien8.de>
Cc: <stable@vger.kernel.org> [v3.7]
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

The HD-audio driver artificially calls the suspend and the resume code
path in the VGA switcheroo state changes.  When a machine goes to
suspend, it tries to suspend the device again, and it stalls at
snd_power_wait().

This patch adds checks whether the devices were already in (forced)
suspend in PM callbacks for avoiding the doubly suspend.
Reported-by: NDaniel J Blueman <daniel@quora.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

When the HD-audio controller is disabled (e.g. via vga switcheroo) but
the driver is still accessing it, it spews floods of "spurious
response" kernel messages.  It's because CORB/RIRB WP reads 0xff, and
the driver tries to fill up until this number.

This patch changes the CORB/RIRB WP reads to word instead of byte, and
add the check of the read value.  If it's 0xffff, the controller is
supposed to be disabled, so the further action will be skipped.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Reducing the time on HDA link reset can help to reduce the driver loading
time. So we replace msleep with usleep_range to get more accurate time
control and change the value to a smaller one. And a 100ms timeout is set
for both entering and exiting the link reset.
Signed-off-by: NXingchao Wang <xingchao.wang@intel.com>
Signed-off-by: NMengdong Lin <mengdong.lin@intel.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Since __devinit* have been removed completely, DELAYED_INIT*_MARK in
hda_intel.c became NOP.  Let's rip them off.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

CONFIG_HOTPLUG is going away as an option.  As result the __dev*
markings will be going away.

Remove use of __devinit, __devexit_p, __devinitdata, __devinitconst,
and __devexit.
Signed-off-by: NBill Pemberton <wfp5p@virginia.edu>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

When printing, use a prefix of the PCI domain, bus, device and function
as in other drivers, to differentiate multiple devices.

Important for reporting and debugging. A future step is to tidy this up with
dev_printk et al.

v2: Move conversion specifier into call site, preventing build issues
v3: Refactor for Takashi's for-next branch
Signed-off-by: NDaniel J Blueman <daniel@quora.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

The recent fix for vga switcheroo race in commit 128960a9 opened yet
another race.  At the time the audio driver starts probing, user may
turn off D-GPU off.  But at this moment, the audio driver still
doesn't register the vga switcheroo client, thus the switching isn't
notified.  Then the hardware gets off out of sudden, resulting in
invalid reads and lots of "spurious response" error messages.

For solving this situation, the following changes have been done in
this patch:
- Move again vga switcheroo registration to the very early stage of
  the probing; this also requires to set pci drvdata properly before
  registration
- Introduce the completion to synchronize the driver probe at vga
  switcheroo callbacks; this assures that the whole probing finished
  before executing the callbacks
Reported-by: NDaniel J Blueman <daniel@quora.org>
Tested-by: NDaniel J Blueman <daniel@quora.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Pass DMA buffer pointers in calls to setup_bdle().
Add DSP loader callback routines to controller.
Add new DSP loader switch to Kconfig to turn off DSP firmware.
Signed-off-by: NIan Minett <ian_minett@creativelabs.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

The new firmware code tries to re-read the formerly read firmware
files before suspend.  Thus it's wiser to keep the "patch" firmware in
the driver for avoiding this unnecessary re-reading.

Of course, this will consume a bit of memory for unused stuff, but
the patch fw is supposed to be fairly small, so it's more benefit in
the end.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

We've got a report that the runtime PM may make the codec the
unresponsive on AMD platforms.  Since the feature has been tested only
on the recent Intel platforms, it's safer to limit the support to such
devices for now.

This patch adds a new DCAPS bit flag indicating the runtime PM
support, and mark it for Intel controllers.
Reported-and-tested-by: NJulian Wollrath <jwollrath@web.de>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

The audio chipset used in Teradici's Tera2 host cards is the same as that in
the 1200 host cards. This patch allows ALSA to recognize the Tera2 cards.
Signed-off-by: NLars R. Damerow <lars@pixar.com>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Add a couple of tracepoints to snd-hda-intel for tracing the position
and the trigger timings.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>