Some machines machine cause a severe CORB/RIRB stall in certain
weird conditions, such as PA access at the start up together with
fglrx driver. This seems unable to be recovered without the controller
reset.
This patch allows the bus controller reset at critical errors so
that the communication gets recovered again.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
ICH6_GCTL_RESET was wrongly set to another bit by the commit
b21fadb9c1. This caused a problem when
the codec needs really a reset (e.g. recovering from the communication
error at probe).
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Added some missing register bits definitions to reduce magic numbers.
Also renamed some to follow the names on the datasheet.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In the single_cmd mode, the hardware cannot store the multiple replies
like on RIRB, thus each verb has to sync and wait for the response no
matter whether the return value is needed or not. Otherwise it may
result in a wrong return value from the previous verb.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In the single_cmd mode, the current driver code doesn't do any update
for RIRB just for any safety reason. But, actually the RIRB and
single_cmd mode don't conflict. Unsolicited events can be delivered
even while using the single_cmd mode.
This patch allows the handling of unsolicited events with single_cmd
mode, just always checking RIRB independent from single_cmd flag.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When a codec communication error occurs, the CORB/RIRB counters should
be reset first before re-issuing the verb. Simply call azx_free_cmd_io()
and azx_init_cmd_io() to achieve that.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
ASUS W5Fm needs the fixed codec-slots to probe to override the BIOS
problem like W5F.
Tested-by: Alp Kılıç <kilic.alp@gmail.com>
Signed-off-by: Ozan Çağlayan <ozan@pardus.org.tr>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Add proper prefix to each kernel message in hda_intel.c.
Also, avoid the unneeded prefix when CONFIG_SND_VERBOSE_PRINTK is used
together with snd_print*().
Reference: bko#13207
http://bugzilla.kernel.org/show_bug.cgi?id=13207
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The PCI entries of Creative with HD-audio class can be the devices
with emu20k1/emu20k2 chips. These are supported better by snd-ctxfi
driver. With that driver, the device will mutate from HD-audio to
its native class.
This patch adds a simple ifdef to avoid the conflict of device probe
between snd-hda-intel and snd-ctxfi drivers. 1102:0009 seems still
OK to be added as it has no emu20kx chip, and is a pure HD-audio
device.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The current error-recovery scheme for the codec communication errors
doesn't work always well. Especially falling back to the
single-command mode causes the fatal problem on many systems.
In this patch, the problematic verb is re-issued again after the error
(even with polling mode) instead of the single-cmd mode. The
single-cmd mode will be used only when specified via the command
option explicitly, mainly just for testing.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Check the length to copy via strlen() beforehand to avoid the stack
corruption, or use strlcpy() to be safe in HD-audio codes.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Added the support for Creative SB X-Fi boards with UAA (HD-audio) mode.
In the HD-audio mode, no multiple streams are supported by just it
behaves like a normal HD-audio device.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Commit fa00e046b4
added a new bitfield not adjacent to other
bitfields in the same struct. Moved the new one.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
* 'master' of git://git.alsa-project.org/alsa-kernel:
[ALSA] intel8x0: add one retry to the ac97_clock measurement routine
[ALSA] intel8x0: fix wrong conditions in ac97_clock measure routine
[ALSA] intel8x0: do not use zero value from PICB register
[ALSA] intel8x0: an attempt to make ac97_clock measurement more reliable
[ALSA] pcm-midlevel: Add more strict buffer position checks based on jiffies
[ALSA] hda_intel: fix unexpected ring buffer positions
This is the second go through of the old DMA_nBIT_MASK macro,and there're not
so many of them left,so I put them into one patch.I hope this is the last round.
After this the definition of the old DMA_nBIT_MASK macro could be removed.
Signed-off-by: Yang Hongyang <yanghy@cn.fujitsu.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Tony Lindgren <tony@atomide.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Greg KH <greg@kroah.com>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
I found two issues with ICH7-M (it should be related to other HDA chipsets
as well):
- the ring buffer position is not reset when stream restarts (after xrun) -
solved by moving azx_stream_reset() call from open() to prepare() callback
and reset posbuf to zero (it might be filled with hw later than position()
callback is called)
- irq_ignore flag should be set also when ring buffer memory area is not
changed in prepare() callback - this patch replaces irq_ignore with
more universal check based on jiffies clock
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
So far, the prepare callback is called multiple times, BDL entries
are reset and re-programmed at each time.
This patch adds the check to avoid the reset of BDL entries when the
same parameters are used.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The position-buffer on ATI controllers are unreliable as well as
on VIA chips, thus the same workaround for DMA position reading as
VIA is useful for ATI.
Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
ATI controllers (at least some SB0600 models) appear buggy to handle
64bit DMA. As a workaround, reset GCAP bit0 and let the driver to
use only 32bit DMA on these controllers.
Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Revert the Toshiba probe_mask quirk for 2.6.29 kernel
(commit 38f1df27e3).
In the current tree, the digital-only codec is handled properly so
no codec conflict should occur.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Some Toshiba laptops have another ALC268 codec on slot#3 that conflicts
with the primary codec. The codec#3 is for the digital I/O, and should
be fixed by the driver, but it'd need a bunch of changes.
So, let's fix the probe problem temporarily by setting the default
probe_mask value.
Reference: kernel bugzilla #12735http://bugzilla.kernel.org/show_bug.cgi?id=12735
Tested-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Added the Device IDs for MCP89 HD audio controller.
Removed the IDs of MCP7B cause this chipset had been cancelled.
Signed-off-by: Peer Chen <peerchen@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This reverts commit 32e176c14d.
That commit caused a regression with suspend on Thinkpad SL300.
Reference: kernel bug#12711
http://bugzilla.kernel.org/show_bug.cgi?id=12711
Tested-by: Alexandre Rostovtsev <tetromino@gmail.com>
Acked-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
ASUS W5F needs the fixed codec-slots to probe to override the BIOS
problem.
Tested-by: Giovanni Moser Frainer <giovanni@redix.com.br>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Some devices have broken BIOS and they don't set the codec probe-bit
properly after cleared by the driver. This makes the driver skipping
the necessary codec slots.
Since BIOS update isn't always easy, now the semantics of probe_mask
option is changed a bit. When it contains the bit 8 (0x100), the
lower bits are used to probe that slots regardless of codec-probe bits
returned by the hardware.
For example, probe_mask=0x103 will force to probe the codec slot #0
and #1.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
snd-hda-intel driver used schedule_work() fot the delayed DMA pointer
updates, but this has several potential problems:
- it may block other eventsd works longer
- it may deadlock when probing fails and flush_scheduled_work() is
called during probe callback (as probe callback itself could be
invoked from eventd)
This patch adds an own workq for each driver instance to solve these
problems.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Added probe_only module option to hd-audio driver.
This option specifies whether the driver creates and initializes the
codec-parser after probing. When this option is set, the driver skips
the codec parsing and initialization but gives you proc and other
accesses. It's useful to see the initial codec state for debugging.
The default of this value is off, so the default behavior is as same
as before.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Fixed the compile warning regarding the unused function when built
with CONFIG_PM=n:
sound/pci/hda/hda_intel.c:1905: warning: ‘snd_hda_codecs_inuse’ defined but not used
snd_hda_codecs_inuse() is used only in the resume callback.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Move the restoration of the standard PCI configuration registers
in the snd_hda_intel driver to a ->resume_early() callback executed
with interrupts disabled, since doing that with interrupts enabled
may lead to problems in some cases.
This patch addresses the regression from 2.6.26 tracked as
http://bugzilla.kernel.org/show_bug.cgi?id=12121 .
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Moved power_save field initialization inside a proper ifdef
to fix a build error without CONFIG_SND_HDA_POWER_SAVE.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Move power_save option into hda_intel.c, and make a field in hda_bus,
instead of keeping module parameters in separate files.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Minor clean ups: move snd_hda_codecs_inuse() into hda_intel.c and
make static. Also, make snd_hda_query_supported_pcm() static
as it's used only in hda_codec.c.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Medion MD96630 has ALC268 codec on slot#2 although it's not used
for any purpose. This codec conflicts with the primiary codec ALC888
on slot#0, and gives mixer errors.
This patch adds a corresponding entry to probe_mask blacklist.
Reference: Novell bnc#412528
https://bugzilla.novell.com/show_bug.cgi?id=412528
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Replace 5 free-and-return-err blocks with goto-out-free ones.
This makes the main logic more outstanding.
Signed-off-by: Wu Fengguang <wfg@linux.intel.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The patch uses HD Audio PCI class code to detect AMD HD Audio cards.
Signed-off-by: Libin Yang <libin.yang@amd.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
BIOS on Dell Studio Desktop tells wrong codec probe masks.
This patch gives the preset mask value to avoid invalid access.
Reference: Novell bug#440907
https://bugzilla.novell.com/show_bug.cgi?id=440907
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When an error occurs during the codec probing, typically accessing to an
non-existing codec slot, the controller chip gets often screwed up and
can no longer communicate with the codecs.
This patch adds a preparation phase just to probe codec addresses before
actually creating codec instances. If any error occurs during this
probing phase, the driver resets the controller to recover.
This will (hopefully) fix the famous "single_cmd" errors.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The hda_bus ops callback take struct hda_bus pointer.
Also, the command callback takes the composed command word, instead of
each small bits in arguments.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The current snd-hda-intel driver seems blocking the power-off on some
devices like eeepc. Although this is likely a BIOS problem, we can add
a workaround by disabling IRQ lines before power-off operation.
This patch adds the reboot notifier to achieve it.
The detailed problem description is found in bug#11889:
http://bugme.linux-foundation.org/show_bug.cgi?id=11889
Tested-by: Luiz Fernando N. Capitulino <lcapitulino@mandriva.com.br>
Cc: stable@kernel.org
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Removed the old workaround to avoid the non-existing codec slot.
The current code should work without that workaround. If any,
we can add a quirk table.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Use the newly introduced pci_ioremap_bar() function in sound/.
pci_ioremap_bar() just takes a pci device and a bar number, with the goal
of making it really hard to get wrong, while also having a central place
to stick sanity checks.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When read RIRB buffer immediately after RIRB interrupt received,
sometimes the data will be "0x0". If we wait for some time, the data
in buffer will be correct. This issue only occurred with AMD Phenom cpu.
So we set this "needs_damn_long_delay" flag.
Signed-off-by: Wei Ni <wni@nvidia.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Allow probing of 4 codecs on known good situations.
On some known bad situations, it should be avoided.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Add a workaround for bad DMA-position reporting on VIA chipset.
Signed-off-by: Joseph Chan <josephchan@via.com.tw>
[modified and cleaned up by tiwai]
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Check the continuity of allocated pages to reduce the BDL size as much
as possible so that it can use more than 1MB buffers.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Clean up SG-buffer helper functions and macros. Helpers take substream
as arguments now.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
When bdl_pos_adj=0 is given, disable the position-check and the delayed
period update mechanism. Usually bdl_pos_adj=0 is set only for the
debugging purpose on really broken hardwares. It's better to disable
the extra complexity in such a case.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Kill snd_assert() in sound/pci/*, either removed or replaced with
if () with snd_BUG_ON().
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Enable the snoop for nvidia hda controller to avoid data coherence issue.
Signed-off-by: Peer Chen <peerchen@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
It seems NVidia and other hardwares require the alignment for period
update timing. For satisfying this condition, align the position
adjustment for delayed wake-up to the initial bdl_pos_adj value.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Use bdl_pos_adj=32 as default except for Intel hardwares confirmed
to work with bdl_pos_adj=1. Looks like ATI and NVidia require this
higher value.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
The pending IRQ handling is a very hackish workaround and should be
avoided as much as possible via a larger bdl_pos_adj option value.
Put a warning message if this situation occurs so that the user may have
a chance to notice that something is wrong.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
ATI SB controllers seem to report the DMA ahead in the amount of FIFO.
Thus bdl_pos_adj should be 32 for them as default.
Also, the default value is set to -1, which means to make the driver
to choose the appropriate value.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
The option bdl_pos_adj should be provided for each card instance instead of
a global one because the value depends rather on each controller-chip.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
position_fix=3 is the option to correct the DMA position with the
FIFO size. But, it never worked correctly, and we have now more other
workarounds for the DMA position fixes. Thus better to remove it.
Also, change POS_FIX_NONE to POS_FIX_LPIB to represent its real role
better.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Added a new option, bdl_pos_adj, to adjust the delay of IRQ-wakeup
timing.
Most HD-audio hardwares have a problem that a BDL IRQ is issued before
actually the data and the DMA pointer are updated.
We have already a mechanism to force to delay snd_pcm_period_elapsed()
calls via workq, but this costs much CPU, and typically the delay is
within one sample. Thus, it's more clever to adjust the BDL entries
instead.
The new option adds the size of the delay in frames. As default,
it's set to 1 -- that is, one sample delay. Even the hardware is
really correct, one sample delay is relatively harmless in comparison
with reporting wrong positions.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
It turned out that some ICH9-based boards use SD3 for the audio codec
where the current driver code doesn't probe. Since we have a better
codec slot check now, it must be safe to increase this to 4.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Added the missing PCI ID for ICH9 controller (8086:2911)
Signed-off-by: Kailang Yang <kailang@realtek.com.tw>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Many HD-audio controllers seem inaccurate about the IRQ timing of
PCM period updates. This has caused problems on audio quality; e.g.
JACK doesn't work with two periods.
This patch fixes the problem by checking the current DMA position
at IRQ handler and delays the period-update via a workq if it's
inaccurate.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
free_irq() calls synchronize_irq() for you, so there is no need for
drivers to manually do the same thing (again). Thus, calls where
sync-irq immediately precedes free-irq can be simplified.
However, during this audit several bugs were noticed, where free-irq is
preceded by a "irq >= 0" check... but the sync-irq call is not covered
by the same check.
So, where sync-irq could not be eliminated completely, the missing check
was added.
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The hda-intel driver has a problem at power-off on ASUS P5AD2.
It's caused when the position-buffer is enabled -- most likely a
hardware-specific problem.
This patch adds a quirk to avoid the unnecessary enablement of
position-buffer.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Add the Device IDs of nvidia MCP79 HD audio controller to hda_intel.c
Signed-off-by: Peer Chen <peerchen@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
sparc32:
sound/pci/hda/hda_intel.c: In function 'azx_create':
sound/pci/hda/hda_intel.c:1838: error: 'DMA_64BIT_MASK' undeclared (first use in this function)
sound/pci/hda/hda_intel.c:1838: error: (Each undeclared identifier is reported only once
sound/pci/hda/hda_intel.c:1838: error: for each function it appears in.)
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The HD-audio hardware usually supports 64bit address for DMA and other
buffers. The patch enables the feature if supported.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In the current scheme, PCM device numbers are assigned incrementally
in the order of codecs. This causes problems when the codec number
is irregular, e.g. codec #0 for HDMI and codec #1 for analog. Then
the HDMI becomes the first PCM, which is picked up as the default
output device. Unfortuantely this doesn't work well with normal
setups.
This patch introduced the fixed device numbers for the PCM types,
namely, analog, SPDIF, HDMI and modem. The PCM devices are assigned
according to the corresponding PCM type. After this patch, HDMI will
be always assigned to PCM #3, SPDIF to PCM #1, and the first analog
to PCM #0, etc.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The driver gets Oops with ATI HDMI devices due to the wrong calculation
of index for playback streams. This patch fixes it. Reference:
https://bugtrack.alsa-project.org/alsa-bug/view.php?id=3746
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch adds support for Intel's SCH mobile chipset.
Signed-off-by: Tobin Davis <tdavis@dsl-only.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
This patch adds the Intel ICH10 HD Audio Controller DeviceID's.
Signed-off-by: Jason Gaston <jason.d.gaston@intel.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
In azx_[rirb_]get_response(), the timeout is checked at the end of the loop.
It's better to be checked just after the check of the RIRB index to avoid
a bogus error with a too long msleep().
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Some machines with STAC codecs seem to have problems (e.g. no audible
playback) when the delay in codec-read routine is too short.
I still don't figure out which command sequence causes this problem
(due to lack of test hardware), but it's known that increasing the
delay fixes. So, added a stupid workaround here temporarily...
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
This patch removes hardcoded values for the number of streams supported
by the southbridge in most chipsets, and reads these values from the
chipset directly. Most systems are hardwired for 4 streams in each
direction, but newer chipsets change that capability.
Signed-off-by: Tobin Davis <tdavis@dsl-only.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Fixes multiple device support by incrementing the static device counter
at the end of the azx_probe() call. Without this, subsequent probes would
always use the index specified for the first card.
Signed-off-by: Andrew Paprocki <andrew@ishiboo.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
This header file exists only for some hacks to adapt alsa-driver
tree. It's useless for building in the kernel. Let's move a few
lines in it to sound/core.h and remove it.
With this patch, sound/driver.h isn't removed but has just a single
compile warning to include it. This should be really killed in
future.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
