platform/kernel/linux-rpi.git
4 years agoALSA: ctl: remove unused macro for timestamping of elem_value
Takashi Sakamoto [Mon, 23 Dec 2019 02:39:18 +0000 (11:39 +0900)]
ALSA: ctl: remove unused macro for timestamping of elem_value

In a former commit, 'tstamp' member was removed from 'struct
snd_ctl_elem_value' in a middle way toward solution of Y2038 issue. In a
protocol of ALSA control interface, this member is designed to deliver
timestamp information in the value structure when the target element
supports SNDRV_CTL_ELEM_ACCESS_TIMESTAMP flag.

Actually, the feature is neither used by kernel space nor user space,
especiall alsa-lib has no API for the feature. Therefore it's reasonable
to remove both of them. Practically, the timestamp information
corresponds to no information about type of clock ID. It can bring
confusions to applications.

Reference: a4e7dd35b9da ("ALSA: Avoid using timespec for struct snd_ctl_elem_value")
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Link: https://lore.kernel.org/r/20191223023921.8151-2-o-takashi@sakamocchi.jp
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: uapi: Drop unneeded typedefs
Takashi Iwai [Fri, 20 Dec 2019 16:15:55 +0000 (17:15 +0100)]
ALSA: uapi: Drop unneeded typedefs

We kept some typedefs in uapi/sound/*.h so that the programs in
alsa-tools can be built.  Now that alsa-lib takes these and applies
the workarounds in its own, we don't need these typedefs any longer in
the kernel uapi side.  Let's drop them.

Link: https://lore.kernel.org/r/20191220161555.20232-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: uapi: Drop asound.h inclusion from asoc.h
Takashi Iwai [Fri, 20 Dec 2019 15:34:15 +0000 (16:34 +0100)]
ALSA: uapi: Drop asound.h inclusion from asoc.h

The asound.h isn't always available while asoc.h itself is distributed
in alsa-lib package.  So we need to avoid the unnecessary inclusion of
asound.h from there.

Link: https://lore.kernel.org/r/20191220153415.2740-6-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: uapi: Fix typos and header inclusion in asound.h
Takashi Iwai [Fri, 20 Dec 2019 15:34:14 +0000 (16:34 +0100)]
ALSA: uapi: Fix typos and header inclusion in asound.h

The recent changes in uapi/asoundlib.h caused some build errors in
alsa-lib side because of a typo and the new included files.
Basically asound.h is supposed to be usable also on non-Linux systems,
so we've tried to avoid the Linux-specific include files.

This patch is an attempt to recover from those changes.

Fixes: 3ddee7f88aaf ("ALSA: Avoid using timespec for struct snd_pcm_status")
Fixes: 80fe7430c708 ("ALSA: add new 32-bit layout for snd_pcm_mmap_status/control")
Link: https://lore.kernel.org/r/20191220153415.2740-5-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hdspm: Drop linux/types.h inclusion in uapi header
Takashi Iwai [Fri, 20 Dec 2019 15:34:13 +0000 (16:34 +0100)]
ALSA: hdspm: Drop linux/types.h inclusion in uapi header

The hdspm.h uapi header has been used also from non-Linux or platforms
that don't have linux/*.h.  It was OK in the past because alsa-lib
contained the modified version of this header file, but now it tries
to the verbatim copy, so it broke the build.  This fixes it again.

Link: https://lore.kernel.org/r/20191220153415.2740-4-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hdsp: Make uapi/hdsp.h compilable again
Takashi Iwai [Fri, 20 Dec 2019 15:34:12 +0000 (16:34 +0100)]
ALSA: hdsp: Make uapi/hdsp.h compilable again

Recently alsa-lib updated its content of sound/hdsp.h just by copying
the latest Linus kernel uapi/*.h, and this broke the build of
alsa-tools programs.  We used to modify the headers so that they can
be built without asoundlib.h and linux kernel headers, and the
verbatim copy doesn't work as is.

This patch removes again the linux/types.h inclusion and drop __user
prefix that broke the build and adjusts the corresponding code.

Link: https://lore.kernel.org/r/20191220153415.2740-3-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: emu10k1: Make uapi/emu10k1.h compilable again
Takashi Iwai [Fri, 20 Dec 2019 15:34:11 +0000 (16:34 +0100)]
ALSA: emu10k1: Make uapi/emu10k1.h compilable again

Recently we updated the content in alsa-lib uapi header files by just
copying from the latest Linus kernel uapi/*.h, and noticed that it
broke the build of some alsa-tools programs.  The reason is that we
used to have a modified version in the past, so that the program can
be built without referring to the unexported stuff like
snd_ctl_elem_id or __user prefix.

This patch attempts to restore that, i.e. dropping the stuff that
can't be referred in the user-space.  For adapting the changes in
uapi/emu10k1.h, the emu10k1 driver code is also slightly modified.
Most of changes are pointer cast.

Link: https://lore.kernel.org/r/20191220153415.2740-2-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoMerge branch 'for-linus' into for-next
Takashi Iwai [Wed, 18 Dec 2019 19:05:39 +0000 (20:05 +0100)]
Merge branch 'for-linus' into for-next

Taking the 5.5 devel branch back into the main devel branch.
A USB-audio fix needs to be adjusted to adapt the changes that have
been formerly applied for stop_sync.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: usb-audio: set the interface format after resume on Dell WD19
Hui Wang [Wed, 18 Dec 2019 13:26:50 +0000 (21:26 +0800)]
ALSA: usb-audio: set the interface format after resume on Dell WD19

Recently we found the headset-mic on the Dell Dock WD19 doesn't work
anymore after s3 (s2i or deep), this problem could be workarounded by
closing (pcm_close) the app and then reopening (pcm_open) the app, so
this bug is not easy to be detected by users.

When problem happens, retire_capture_urb() could still be called
periodically, but the size of captured data is always 0, it could be
a firmware bug on the dock. Anyway I found after resuming, the
snd_usb_pcm_prepare() will be called, and if we forcibly run
set_format() to set the interface and its endpoint, the capture
size will be normal again. This problem and workaound also apply to
playback.

To fix it in the kernel, add a quirk to let set_format() run
forcibly once after resume.

Signed-off-by: Hui Wang <hui.wang@canonical.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20191218132650.6303-1-hui.wang@canonical.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: usx2y: Adjust indentation in snd_usX2Y_hwdep_dsp_status
Nathan Chancellor [Wed, 18 Dec 2019 03:42:57 +0000 (20:42 -0700)]
ALSA: usx2y: Adjust indentation in snd_usX2Y_hwdep_dsp_status

Clang warns:

../sound/usb/usx2y/usX2Yhwdep.c:122:3: warning: misleading indentation;
statement is not part of the previous 'if' [-Wmisleading-indentation]
        info->version = USX2Y_DRIVER_VERSION;
        ^
../sound/usb/usx2y/usX2Yhwdep.c:120:2: note: previous statement is here
        if (us428->chip_status & USX2Y_STAT_CHIP_INIT)
        ^
1 warning generated.

This warning occurs because there is a space before the tab on this
line. Remove it so that the indentation is consistent with the Linux
kernel coding style and clang no longer warns.

This was introduced before the beginning of git history so no fixes tag.

Link: https://github.com/ClangBuiltLinux/linux/issues/831
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Link: https://lore.kernel.org/r/20191218034257.54535-1-natechancellor@gmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoMerge tag 'y2038-alsa-v8-signed' of git://git.kernel.org:/pub/scm/linux/kernel/git...
Takashi Iwai [Tue, 17 Dec 2019 22:12:39 +0000 (23:12 +0100)]
Merge tag 'y2038-alsa-v8-signed' of git://git./linux/kernel/git/arnd/playground into for-next

ALSA: Fix year 2038 issue for sound subsystem

This is a series I worked on with Baolin in 2017 and 2018, but we
never quite managed to finish up the last pieces. During the
ALSA developer meetup at ELC-E 2018 in Edinburgh, a decision was
made to go with this approach for keeping best compatibility
with existing source code, and then I failed to follow up by
resending the patches.

Now I have patches for all remaining time_t uses in the kernel,
so it's absolutely time to revisit them. I have done more
review of the patches myself and found a couple of minor issues
that I have fixed up, otherwise the series is still the same as
before.

Conceptually, the idea of these patches is:

- 64-bit applications should see no changes at all, neither
  compile-time nor run-time.

- 32-bit code compiled with a 64-bit time_t currently
  does not work with ALSA, and requires kernel changes and/or
  sound/asound.h changes

- Most 32-bit code using these interfaces will work correctly
  on a modified kernel, with or without the uapi header changes.

- 32-bit code using SNDRV_TIMER_IOCTL_TREAD requires the
  updated header file for 64-bit time_t support

- 32-bit i386 user space with 64-bit time_t is broken for
  SNDRV_PCM_IOCTL_STATUS, SNDRV_RAWMIDI_IOCTL_STATUS and
  SNDRV_PCM_IOCTL_SYNC_PTR because of i386 alignment. This is also
  addressed by the updated uapi header.

- PCM mmap is currently supported on native x86 kernels
  (both 32-bit and 64-bit) but not for compat mode. This series breaks
  the 32-bit native mmap support for 32-bit time_t, but instead allows
  it for 64-bit time_t on both native and compat kernels. This seems to
  be the best trade-off, as mmap support is optional already, and most
  32-bit code runs in compat mode anyway.

- I've tried to avoid breaking compilation of 32-bit code
  as much as possible. Anything that does break however is likely code
  that is already broken on 64-bit time_t and needs source changes to
  fix them.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/arnd/playground.git y2038-alsa-v8
[2] https://lore.kernel.org/lkml/CAK8P3a2Os66+iwQYf97qh05W2JP8rmWao8zmKoHiXqVHvyYAJA@mail.gmail.com/T/#m6519cb07cfda08adf1dedea6596bb98892b4d5dc

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Changes since v7: (Arnd):
 - Fix a typo found by Ben Hutchings

Changes since v6: (Arnd):
 - Add a patch to update the API versions
 - Hide a timespec reference in #ifndef __KERNEL__ to remove the
   last reference to time_t
 - Use a more readable way to do padding and describe it in the
   changelog
 - Rebase to linux-5.5-rc1, changing include/sound/soc-component.h
   and sound/drivers/aloop.c as needed.

Changes since v5 (Arnd):
 - Rebased to linux-5.4-rc4
 - Updated to completely remove timespec and time_t references from alsa
 - found and fixed a few bugs

Changes since v4 (Baolin):
 - Add patch 5 to change trigger_tstamp member of struct snd_pcm_runtime.
 - Add patch 8 to change internal timespec.
 - Add more explanation in commit message.
 - Use ktime_get_real_ts64() in patch 6.
 - Split common code out into a separate function in patch 6.
 - Fix tu->tread bug in patch 6 and remove #if __BITS_PER_LONG == 64 macro.

Changes since v3:
 - Move struct snd_pcm_status32 to pcm.h file.
 - Modify comments and commit message.
 - Add new patch2 ~ patch6.

Changes since v2:
 - Renamed all structures to make clear.
 - Remove CONFIG_X86_X32 macro and introduced new compat_snd_pcm_status64_x86_32.

Changes since v1:
 - Add one macro for struct snd_pcm_status_32 which only active in 32bits kernel.
 - Convert pcm_compat.c to use struct snd_pcm_status_64.
 - Convert pcm_native.c to use struct snd_pcm_status_64.

4 years agoMerge tag 'asoc-fix-v5.5-rc2' of https://git.kernel.org/pub/scm/linux/kernel/git...
Takashi Iwai [Tue, 17 Dec 2019 13:18:32 +0000 (14:18 +0100)]
Merge tag 'asoc-fix-v5.5-rc2' of https://git./linux/kernel/git/broonie/sound into for-linus

ASoC: Fixes for v5.5

A collection of fixes since the merge window, mostly driver specific but
there's a few in the core that clean up fallout from the refactorings
done in the last cycle.

4 years agoALSA: cmipci: Allow disabling MPU port via module option
Takashi Iwai [Tue, 17 Dec 2019 08:14:48 +0000 (09:14 +0100)]
ALSA: cmipci: Allow disabling MPU port via module option

Patrick May reported that his sound card with CMI8378 chip causes a
crash / reboot when accessing the MIDI port that isn't actually
present on the board.  Moreover, despite of the documentation,
passing mpu_port=0 doesn't disable the MIDI port on this board.

It implies that the chip is a newer revision and the MPU401 port is
integrated and mapped on the PCI register.  For this chip model, the
driver enables the MPU port unconditionally, so far.

Although fixing the unexpected reboot would be the best solution, it's
not so trivial to identify the cause.  So, as a plan B, this patch
extends the existing mpu_port option usage to allow disabling the port
by specifying the value 0, just like we applied for fm_port option in
commit 2f24d159d5ac ("[ALSA] cmipci - Allow to disable integrated FM
port").  As default, the MPU port is still enabled, but user can pass
mpu_port=0 to disable it.

Reported-and-tested-by: Patrick May <dusthillresident@gmail.com>
Link: https://lore.kernel.org/r/20191217081448.1144-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hda - Downgrade error message for single-cmd fallback
Takashi Iwai [Mon, 16 Dec 2019 15:12:24 +0000 (16:12 +0100)]
ALSA: hda - Downgrade error message for single-cmd fallback

We made the error message for the CORB/RIRB communication clearer by
upgrading to dev_WARN() so that user can notice better.  But this
struck us like a boomerang: now it caught syzbot and reported back as
a fatal issue although it's not really any too serious bug that worth
for stopping the whole system.

OK, OK, let's be softy, downgrade it to the standard dev_err() again.

Fixes: dd65f7e19c69 ("ALSA: hda - Show the fatal CORB/RIRB error more clearly")
Reported-by: syzbot+b3028ac3933f5c466389@syzkaller.appspotmail.com
Link: https://lore.kernel.org/r/20191216151224.30013-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoASoC: wm8962: fix lambda value
Shengjiu Wang [Wed, 11 Dec 2019 11:57:22 +0000 (19:57 +0800)]
ASoC: wm8962: fix lambda value

According to user manual, it is required that FLL_LAMBDA > 0
in all cases (Integer and Franctional modes).

Fixes: 9a76f1ff6e29 ("ASoC: Add initial WM8962 CODEC driver")
Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
Acked-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://lore.kernel.org/r/1576065442-19763-1-git-send-email-shengjiu.wang@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
4 years agoALSA: hda: Comment about snd_hdac_bus_update_rirb() and spinlock
Takashi Iwai [Fri, 13 Dec 2019 16:30:05 +0000 (17:30 +0100)]
ALSA: hda: Comment about snd_hdac_bus_update_rirb() and spinlock

The call of snd_hdac_bus_update_rirb() needs the bus->reg_lock
spinlock protection for concurrency.  Comment about it more
explicitly.

Link: https://lore.kernel.org/r/20191213163005.19116-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hda: Fix regression by strip mask fix
Takashi Iwai [Sat, 14 Dec 2019 17:52:17 +0000 (18:52 +0100)]
ALSA: hda: Fix regression by strip mask fix

The commit e38e486d66e2 ("ALSA: hda: Modify stream stripe mask only
when needed") tried to address the regression by the unconditional
application of the stripe mask, but this caused yet another
regression for the previously working devices.  Namely, the patch
clears the azx_dev->stripe flag at snd_hdac_stream_clear(), but this
may be called multiple times before restarting the stream, so this
ended up with clearance of the flag for the whole time.

This patch fixes the regression by moving the azx_dev->stripe flag
clearance at the counter-part, the close callback of HDMI codec
driver instead.

Fixes: e38e486d66e2 ("ALSA: hda: Modify stream stripe mask only when needed")
BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=205855
BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=204477
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20191214175217.31852-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hda/ca0132 - Fix work handling in delayed HP detection
Takashi Iwai [Fri, 13 Dec 2019 08:51:11 +0000 (09:51 +0100)]
ALSA: hda/ca0132 - Fix work handling in delayed HP detection

CA0132 has the delayed HP jack detection code that is invoked from the
unsol handler, but it does a few weird things: it contains the cancel
of a work inside the work handler, and yet it misses the cancel-sync
call at (runtime-)suspend.  This patch addresses those issues.

Fixes: 15c2b3cc09a3 ("ALSA: hda/ca0132 - Fix possible workqueue stall")
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20191213085111.22855-4-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hda/ca0132 - Avoid endless loop
Takashi Iwai [Fri, 13 Dec 2019 08:51:10 +0000 (09:51 +0100)]
ALSA: hda/ca0132 - Avoid endless loop

Introduce a timeout to dspio_clear_response_queue() so that it won't
be caught in an endless loop even if the hardware doesn't respond
properly.

Fixes: a73d511c4867 ("ALSA: hda/ca0132: Add unsol handler for DSP and jack detection")
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20191213085111.22855-3-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hda/ca0132 - Keep power on during processing DSP response
Takashi Iwai [Fri, 13 Dec 2019 08:51:09 +0000 (09:51 +0100)]
ALSA: hda/ca0132 - Keep power on during processing DSP response

We need to keep power on while processing the DSP response via unsol
event.  Each snd_hda_codec_read() call does the power management, so
it should work normally, but still it's safer to keep the power up for
the whole function.

Fixes: a73d511c4867 ("ALSA: hda/ca0132: Add unsol handler for DSP and jack detection")
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20191213085111.22855-2-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: pcm: Avoid possible info leaks from PCM stream buffers
Takashi Iwai [Wed, 11 Dec 2019 15:57:42 +0000 (16:57 +0100)]
ALSA: pcm: Avoid possible info leaks from PCM stream buffers

The current PCM code doesn't initialize explicitly the buffers
allocated for PCM streams, hence it might leak some uninitialized
kernel data or previous stream contents by mmapping or reading the
buffer before actually starting the stream.

Since this is a common problem, this patch simply adds the clearance
of the buffer data at hw_params callback.  Although this does only
zero-clear no matter which format is used, which doesn't mean the
silence for some formats, but it should be OK because the intention is
just to clear the previous data on the buffer.

Reported-by: Lionel Koenig <lionel.koenig@gmail.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20191211155742.3213-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: control: remove useless assignment in .info callback of PCM chmap element
Takashi Sakamoto [Sat, 14 Dec 2019 13:13:51 +0000 (22:13 +0900)]
ALSA: control: remove useless assignment in .info callback of PCM chmap element

Control elements for PCM chmap return information to userspace abount
the maximum number of available PCM channels as the number of values
in the element.

In current implementation the number is once initialized to zero, then
assigned to. This is useless and this commit fixes it.

Fixes: 2d3391ec0ecc ("ALSA: PCM: channel mapping API implementation")
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Link: https://lore.kernel.org/r/20191214131351.28950-1-o-takashi@sakamocchi.jp
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hda: Unify get_response handling
Takashi Iwai [Thu, 12 Dec 2019 19:11:01 +0000 (20:11 +0100)]
ALSA: hda: Unify get_response handling

Now most of the get_response handling became quite similar between
HDA-core and legacy drivers, and the only differences are:

- the handling of extra-long polling delay for some codecs
- the debug message for the stalled communication

and both are worth to share in the common code.

This patch unifies the code into snd_hdac_bus_get_response(), and use
this from the legacy get_response callback.  It results in a good
amount of code reduction in the end.

Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Link: https://lore.kernel.org/r/20191212191101.19517-3-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hda: Use waitqueue for RIRB in HDA-core helper, too
Takashi Iwai [Thu, 12 Dec 2019 19:11:00 +0000 (20:11 +0100)]
ALSA: hda: Use waitqueue for RIRB in HDA-core helper, too

This patch implements the same logic that was done for the legacy
HD-audio controller driver by the commit 88452da92ba2 ("ALSA: hda: Use
standard waitqueue for RIRB wakeup") to the HDA-core helper code,
too.  This makes snd_hdac_bus_get_response() waiting for the response
with bus->rirb_wq instead of polling when bus->polling is false.
It'll save both CPU time and response latency.

Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Link: https://lore.kernel.org/r/20191212191101.19517-2-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: bump uapi version numbers
Arnd Bergmann [Wed, 13 Nov 2019 16:49:14 +0000 (17:49 +0100)]
ALSA: bump uapi version numbers

Change SNDRV_PCM_VERSION, SNDRV_RAWMIDI_VERSION and SNDRV_TIMER_VERSION
to indicate the addition of the time64 version of the mmap interface and
these ioctl commands:

SNDRV_PCM_IOCTL_SYNC
SNDRV_RAWMIDI_IOCTL_STATUS
SNDRV_PCM_IOCTL_STATUS
SNDRV_PCM_IOCTL_STATUS_EXT
SNDRV_TIMER_IOCTL_TREAD
SNDRV_TIMER_IOCTL_STATUS

32-bit applications built with 64-bit time_t require both the headers
and the running kernel to support at least the new API version. When
built with earlier kernel headers, some of these may not work
correctly, so applications are encouraged to fail compilation like

 #if SNDRV_PCM_VERSION < SNDRV_PROTOCOL_VERSION(2, 0, 15)
 extern int __fail_build_for_time_64[sizeof(long) - sizeof(time_t)];
 #endif

or provide their own updated copy of the header file.
At runtime, the interface is unchanged for 32-bit time_t, but new
kernels are required to work with user compiled with 64-bit time_t.

A runtime check can be used to detect old kernel versions and
warn about those.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
4 years agoALSA: add new 32-bit layout for snd_pcm_mmap_status/control
Arnd Bergmann [Tue, 24 Apr 2018 12:06:15 +0000 (20:06 +0800)]
ALSA: add new 32-bit layout for snd_pcm_mmap_status/control

The snd_pcm_mmap_status and snd_pcm_mmap_control interfaces are one of the
trickiest areas to get right when moving to 64-bit time_t in user space.

The snd_pcm_mmap_status structure layout is incompatible with user space
that uses a 64-bit time_t, so we need a new layout for it. Since the
SNDRV_PCM_IOCTL_SYNC_PTR ioctl combines it with snd_pcm_mmap_control
into snd_pcm_sync_ptr, we need to change those two as well.

Both structures are also exported via an mmap() operation on certain
architectures, and this suffers from incompatibility between 32-bit
and 64-bit user space. As we have to change both structures anyway,
this is a good opportunity to fix the mmap() problem as well, so let's
standardize on the existing 64-bit layout of the structure where possible.

The downside is that we lose mmap() support for existing 32-bit x86 and
powerpc applications, adding that would introduce very noticeable runtime
overhead and complexity. My assumption here is that not too many people
will miss the removed feature, given that:

- Almost all x86 and powerpc users these days are on 64-bit kernels,
the majority of today's 32-bit users are on architectures that never
supported mmap (ARM, MIPS, ...).
- It never worked in compat mode (it was intentionally disabled there)
- The application already needs to work with a fallback to
SNDRV_PCM_IOCTL_SYNC_PTR, which will keep working with both the old
and new structure layout.

Both the ioctl() and mmap() based interfaces are changed at the same
time, as they are based on the same structures. Unlike other interfaces,
we change the uapi header to export both the traditional structure and
a version that is portable between 32-bit and 64-bit user space code
and that corresponds to the existing 64-bit layout. We further check the
__USE_TIME_BITS64 macro that will be defined by future C library versions
whenever we use the new time_t definition, so any existing user space
source code will not see any changes until it gets rebuilt against a new
C library. However, the new structures are all visible in addition to the
old ones, allowing applications to explicitly request the new structures.

In order to detect the difference between the old snd_pcm_mmap_status and
the new __snd_pcm_mmap_status64 structure from the ioctl command number,
we rely on one quirk in the structure definition: snd_pcm_mmap_status
must be aligned to alignof(time_t), which leads the compiler to insert
four bytes of padding in struct snd_pcm_sync_ptr after 'flags' and a
corresponding change in the size of snd_pcm_sync_ptr itself. On x86-32
(and only there), the compiler doesn't use 64-bit alignment in structure,
so I'm adding an explicit pad in the structure that has no effect on the
existing 64-bit architectures but ensures that the layout matches for x86.

The snd_pcm_uframes_t type compatibility requires another hack: we can't
easily make that 64 bit wide, so I leave the type as 'unsigned long',
but add padding before and after it, to ensure that the data is properly
aligned to the respective 64-bit field in the in-kernel structure.

For the SNDRV_PCM_MMAP_OFFSET_STATUS/CONTROL constants that are used
as the virtual file offset in the mmap() function, we also have to
introduce new constants that depend on hte __USE_TIME_BITS64 macro:
The existing macros are renamed to SNDRV_PCM_MMAP_OFFSET_STATUS_OLD
and SNDRV_PCM_MMAP_OFFSET_CONTROL_OLD, they continue to work fine on
64-bit architectures, but stop working on native 32-bit user space.
The replacement _NEW constants are now used by default for user space
built with __USE_TIME_BITS64, those now work on all new kernels for x86,
ppc and alpha (32 and 64 bit, native and compat). It might be a good idea
for a future alsa-lib to support both the _OLD and _NEW macros and use
the corresponding structures directly. Unmodified alsa-lib source code
will retain the current behavior, so it will no longer be able to use
mmap() for the status/control structures on 32-bit systems, until either
the C library gets updated to 64-bit time_t or alsa-lib gets updated to
support both mmap() layouts.

Co-developed-with: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
4 years agoALSA: move snd_pcm_ioctl_sync_ptr_compat into pcm_native.c
Arnd Bergmann [Tue, 24 Apr 2018 12:06:14 +0000 (20:06 +0800)]
ALSA: move snd_pcm_ioctl_sync_ptr_compat into pcm_native.c

This is a preparation patch, moving the compat handler for
snd_pcm_ioctl_sync_ptr_compat from pcm_compat.c to pcm_native.c.
No other changes are indented.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
4 years agoALSA: Avoid using timespec for struct snd_timer_tread
Baolin Wang [Tue, 24 Apr 2018 12:06:13 +0000 (20:06 +0800)]
ALSA: Avoid using timespec for struct snd_timer_tread

The struct snd_timer_tread will use 'timespec' type variables to record
timestamp, which is not year 2038 safe on 32bits system.

Since the struct snd_timer_tread is passed through read() rather than
ioctl(), and the read syscall has no command number that lets us pick
between the 32-bit or 64-bit version of this structure.

Thus we introduced one new command SNDRV_TIMER_IOCTL_TREAD64 and new
struct snd_timer_tread64 replacing timespec with s64 type to handle
64bit time_t. That means we will set tu->tread = TREAD_FORMAT_64BIT
when user space has a 64bit time_t, then we will copy to user with
struct snd_timer_tread64. Otherwise we will use 32bit time_t variables
when copying to user.

Moreover this patch replaces timespec type with timespec64 type and
related y2038 safe APIs.

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
4 years agoALSA: hda: tegra: Fix unused variable compile warning
Takashi Iwai [Thu, 12 Dec 2019 08:05:18 +0000 (09:05 +0100)]
ALSA: hda: tegra: Fix unused variable compile warning

Forgot to remove the variable declaration as well in the last commit.
  sound/pci/hda/hda_tegra.c: In function 'hda_tegra_runtime_suspend':
  sound/pci/hda/hda_tegra.c:169:19: warning: unused variable 'bus' [-Wunused-variable]

Fixes: f36da9406e66 ("ALSA: hda: Support PCM sync_stop")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Link: https://lore.kernel.org/r/20191212080518.6522-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: Avoid using timespec for struct snd_rawmidi_status
Baolin Wang [Tue, 24 Apr 2018 12:06:12 +0000 (20:06 +0800)]
ALSA: Avoid using timespec for struct snd_rawmidi_status

The struct snd_rawmidi_status will use 'timespec' type variables to record
timestamp, which is not year 2038 safe on 32bits system.

Thus we introduced 'struct snd_rawmidi_status32' and 'struct snd_rawmidi_status64'
to handle 32bit time_t and 64bit time_t in native mode, which replace
timespec with s64 type.

In compat mode, we renamed or introduced new structures to handle 32bit/64bit
time_t in compatible mode. The 'struct snd_rawmidi_status32' and
snd_rawmidi_ioctl_status32() are used to handle 32bit time_t in compat mode.
'struct compat_snd_rawmidi_status64' is used to handle 64bit time_t.

When glibc changes time_t to 64-bit, any recompiled program will issue ioctl
commands that the kernel does not understand without this patch.

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
4 years agoALSA: Avoid using timespec for struct snd_pcm_status
Baolin Wang [Tue, 24 Apr 2018 12:06:11 +0000 (20:06 +0800)]
ALSA: Avoid using timespec for struct snd_pcm_status

The struct snd_pcm_status will use 'timespec' type variables to record
timestamp, which is not year 2038 safe on 32bits system.

Userspace will use SNDRV_PCM_IOCTL_STATUS and SNDRV_PCM_IOCTL_STATUS_EXT
as commands to issue ioctl() to fill the 'snd_pcm_status' structure in
userspace. The command number is always defined through _IOR/_IOW/IORW,
so when userspace changes the definition of 'struct timespec' to use
64-bit types, the command number also changes.

Thus in the kernel, we now need to define two versions of each such ioctl
and corresponding ioctl commands to handle 32bit time_t and 64bit time_t
in native mode:
struct snd_pcm_status32 {
......

s32 trigger_tstamp_sec;
s32 trigger_tstamp_nsec;

......

s32 audio_tstamp_sec;
s32 audio_tstamp_nsec;

......
};

struct snd_pcm_status64 {
......

s32 trigger_tstamp_sec;
s32 trigger_tstamp_nsec;

......

s32 audio_tstamp_sec;
s32 audio_tstamp_nsec;

......
};

Moreover in compat file, we renamed or introduced new structures to handle
32bit/64bit time_t in compatible mode. The 'struct snd_pcm_status32' and
snd_pcm_status_user32() are used to handle 32bit time_t in compat mode.
'struct compat_snd_pcm_status64' and snd_pcm_status_user_compat64() are used
to handle 64bit time_t.

The implicit padding before timespec is made explicit to avoid incompatible
structure layout between 32-bit and 64-bit x86 due to the different
alignment requirements, and the snd_pcm_status structure is now hidden
from the kernel to avoid relying on the timespec definitio definitionn

Finally we can replace SNDRV_PCM_IOCTL_STATUS and SNDRV_PCM_IOCTL_STATUS_EXT
with new commands and introduce new functions to fill new 'struct snd_pcm_status64'
instead of using unsafe 'struct snd_pcm_status'. Then in future, the new
commands can be matched when userspace changes 'timespec' to 64bit type
to make a size change of 'struct snd_pcm_status'. When glibc changes time_t
to 64-bit, any recompiled program will issue ioctl commands that the kernel
does not understand without this patch.

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
4 years agoALSA: Avoid using timespec for struct snd_ctl_elem_value
Baolin Wang [Tue, 24 Apr 2018 12:06:10 +0000 (20:06 +0800)]
ALSA: Avoid using timespec for struct snd_ctl_elem_value

The struct snd_ctl_elem_value will use 'timespec' type variables to record
timestamp, which is not year 2038 safe on 32bits system.

Since there are no drivers will implemented the tstamp member of the
struct snd_ctl_elem_value, and also the stucture size will not be changed
if we change timespec to s64 for tstamp member of struct snd_ctl_elem_value.

From Takashi's comments, "In the library, applications are not expected
to access to this structure directly. The applications get opaque pointer
to the structure and must use any control APIs to operate it. Actually the
library produce no API to handle 'struct snd_ctl_elem_value.tstamp'. This
means that we can drop this member from alsa-lib without decline of
functionality." Thus we can simply remove the tstamp member to avoid using
the type which is not year 2038 safe on 32bits system.

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
4 years agoALSA: Avoid using timespec for struct snd_timer_status
Baolin Wang [Tue, 24 Apr 2018 12:06:09 +0000 (20:06 +0800)]
ALSA: Avoid using timespec for struct snd_timer_status

struct snd_timer_status uses 'timespec' type variables to record
timestamp, which will be changed to an incompatible layout with
updated user space using 64-bit time_t.

To handle both the old and the new layout on 32-bit architectures,
this patch introduces 'struct snd_timer_status32' and 'struct snd_timer_status64'
to handle 32bit time_t and 64bit time_t in native mode and compat mode,
which replaces timespec with s64 type.

When glibc changes time_t to 64-bit, any recompiled program will issue
ioctl commands that the kernel does not understand without this patch.

In the public uapi header, snd_timer_status is now guarded by
an #ifndef __KERNEL__ to avoid referencing 'struct timespec'.
The timespec definition will be removed from the kernel to prevent
new y2038 bugs and to avoid the conflict with an incompatible libc
type of the same name.

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
4 years agoALSA: Replace timespec with timespec64
Baolin Wang [Tue, 24 Apr 2018 12:06:08 +0000 (20:06 +0800)]
ALSA: Replace timespec with timespec64

Since timespec is not year 2038 safe on 32bit system, and we need to
convert all timespec variables to timespec64 type for sound subsystem.

This patch is used to do preparation for following patches, that will
convert all structures defined in uapi/sound/asound.h to use 64-bit
time_t.

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
4 years agoALSA: usb-audio: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:54 +0000 (07:34 +0100)]
ALSA: usb-audio: Support PCM sync_stop

USB-audio driver had some implementation of its own sync-stop
mechanism.  This patch moved a part of it to the common PCM sync_stop
ops.

Link: https://lore.kernel.org/r/20191210063454.31603-56-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: vxpocket: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:53 +0000 (07:34 +0100)]
ALSA: vxpocket: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the threaded
interrupt handler.  Set card->sync_irq for enabling the missing
sync_stop PCM operation.

Link: https://lore.kernel.org/r/20191210063454.31603-55-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: pdaudiocf: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:52 +0000 (07:34 +0100)]
ALSA: pdaudiocf: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the threaded
interrupt handler.  Set card->sync_irq for enabling the missing
sync_stop PCM operation.

Link: https://lore.kernel.org/r/20191210063454.31603-54-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: vx222: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:51 +0000 (07:34 +0100)]
ALSA: vx222: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the threaded
interrupt handler.  Set card->sync_irq for enabling the missing
sync_stop PCM operation.

Link: https://lore.kernel.org/r/20191210063454.31603-53-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: pcxhr: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:50 +0000 (07:34 +0100)]
ALSA: pcxhr: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the threaded
interrupt handler.  Set card->sync_irq for enabling the missing
sync_stop PCM operation.

Link: https://lore.kernel.org/r/20191210063454.31603-52-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: mixart: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:49 +0000 (07:34 +0100)]
ALSA: mixart: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the threaded
interrupt handler.  Set card->sync_irq for enabling the missing
sync_stop PCM operation.

Link: https://lore.kernel.org/r/20191210063454.31603-51-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: lx6464es: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:48 +0000 (07:34 +0100)]
ALSA: lx6464es: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the threaded
interrupt handler.  Set card->sync_irq for enabling the missing
sync_stop PCM operation.

Link: https://lore.kernel.org/r/20191210063454.31603-50-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: wss: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:47 +0000 (07:34 +0100)]
ALSA: wss: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-49-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: wavefront: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:46 +0000 (07:34 +0100)]
ALSA: wavefront: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-48-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: sb: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:45 +0000 (07:34 +0100)]
ALSA: sb: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-47-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: opti9xx: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:44 +0000 (07:34 +0100)]
ALSA: opti9xx: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-46-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: opl3sa2: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:43 +0000 (07:34 +0100)]
ALSA: opl3sa2: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-45-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: msnd: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:42 +0000 (07:34 +0100)]
ALSA: msnd: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-44-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: gus: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:41 +0000 (07:34 +0100)]
ALSA: gus: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-43-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: es18xx: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:40 +0000 (07:34 +0100)]
ALSA: es18xx: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-42-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: es1688: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:39 +0000 (07:34 +0100)]
ALSA: es1688: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-41-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ad1816a: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:38 +0000 (07:34 +0100)]
ALSA: ad1816a: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-40-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ymfpci: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:37 +0000 (07:34 +0100)]
ALSA: ymfpci: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-39-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: via82xx: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:36 +0000 (07:34 +0100)]
ALSA: via82xx: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-38-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: trident: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:35 +0000 (07:34 +0100)]
ALSA: trident: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-37-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: sonicvibes: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:34 +0000 (07:34 +0100)]
ALSA: sonicvibes: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-36-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: sis7019: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:33 +0000 (07:34 +0100)]
ALSA: sis7019: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-35-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: rme9652: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:32 +0000 (07:34 +0100)]
ALSA: rme9652: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-34-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: rme96: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:31 +0000 (07:34 +0100)]
ALSA: rme96: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-33-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: rme32: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:30 +0000 (07:34 +0100)]
ALSA: rme32: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-32-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: riptide: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:29 +0000 (07:34 +0100)]
ALSA: riptide: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-31-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: oxygen: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:28 +0000 (07:34 +0100)]
ALSA: oxygen: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-30-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: nm256: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:27 +0000 (07:34 +0100)]
ALSA: nm256: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.  It's cleared and reset dynamically at IRQ re-acquiring for
the PM resume, too.

Link: https://lore.kernel.org/r/20191210063454.31603-29-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: maestro3: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:26 +0000 (07:34 +0100)]
ALSA: maestro3: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-28-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: lola: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:25 +0000 (07:34 +0100)]
ALSA: lola: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-27-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: korg1212: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:24 +0000 (07:34 +0100)]
ALSA: korg1212: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-26-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: intel8x0: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:23 +0000 (07:34 +0100)]
ALSA: intel8x0: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.  It's cleared and reset dynamically at IRQ re-acquiring for
the PM resume, too.

Link: https://lore.kernel.org/r/20191210063454.31603-25-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ice1724: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:22 +0000 (07:34 +0100)]
ALSA: ice1724: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-24-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ice1712: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:21 +0000 (07:34 +0100)]
ALSA: ice1712: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-23-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: hda: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:20 +0000 (07:34 +0100)]
ALSA: hda: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.  It's cleared and reset dynamically at IRQ re-acquiring for
the PM resume, too.

Link: https://lore.kernel.org/r/20191210063454.31603-22-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: fm801: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:19 +0000 (07:34 +0100)]
ALSA: fm801: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-21-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: es1968: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:18 +0000 (07:34 +0100)]
ALSA: es1968: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-20-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: es1938: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:17 +0000 (07:34 +0100)]
ALSA: es1938: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.  It's cleared and reset dynamically at IRQ re-acquiring for
the PM resume, too.

Link: https://lore.kernel.org/r/20191210063454.31603-19-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ens137x: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:16 +0000 (07:34 +0100)]
ALSA: ens137x: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-18-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: emu10k1: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:15 +0000 (07:34 +0100)]
ALSA: emu10k1: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-17-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: echoaudio: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:14 +0000 (07:34 +0100)]
ALSA: echoaudio: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.  It's cleared and reset dynamically at IRQ re-acquiring for
the PM resume, too.

Link: https://lore.kernel.org/r/20191210063454.31603-16-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ctxfi: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:13 +0000 (07:34 +0100)]
ALSA: ctxfi: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-15-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: cs5535audio: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:12 +0000 (07:34 +0100)]
ALSA: cs5535audio: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-14-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: cs46xx: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:11 +0000 (07:34 +0100)]
ALSA: cs46xx: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-13-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: cs4281: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:10 +0000 (07:34 +0100)]
ALSA: cs4281: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-12-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: cmipci: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:09 +0000 (07:34 +0100)]
ALSA: cmipci: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-11-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ca0106: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:08 +0000 (07:34 +0100)]
ALSA: ca0106: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-10-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: bt87x: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:07 +0000 (07:34 +0100)]
ALSA: bt87x: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-9-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: azt3328: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:06 +0000 (07:34 +0100)]
ALSA: azt3328: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-8-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: aw2: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:05 +0000 (07:34 +0100)]
ALSA: aw2: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-7-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: au88x0: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:04 +0000 (07:34 +0100)]
ALSA: au88x0: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-6-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: atiixp: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:03 +0000 (07:34 +0100)]
ALSA: atiixp: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-5-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: als300: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:02 +0000 (07:34 +0100)]
ALSA: als300: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-4-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ali5451: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:01 +0000 (07:34 +0100)]
ALSA: ali5451: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation, as well as removing the superfluous synchronize_irq()
call.

Link: https://lore.kernel.org/r/20191210063454.31603-3-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ad1889: Support PCM sync_stop
Takashi Iwai [Tue, 10 Dec 2019 06:34:00 +0000 (07:34 +0100)]
ALSA: ad1889: Support PCM sync_stop

The driver invokes snd_pcm_period_elapsed() simply from the interrupt
handler.  Set card->sync_irq for enabling the missing sync_stop PCM
operation.

Link: https://lore.kernel.org/r/20191210063454.31603-2-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: xen: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:45 +0000 (07:11 +0100)]
ALSA: xen: Drop superfluous ioctl PCM ops

PCM core deals the empty ioctl field now as default(*).
Let's kill the redundant lines.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-24-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: x86: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:44 +0000 (07:11 +0100)]
ALSA: x86: Drop superfluous ioctl PCM ops

PCM core deals the empty ioctl field now as default(*).
Let's kill the redundant lines.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-23-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: usb: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:43 +0000 (07:11 +0100)]
ALSA: usb: Drop superfluous ioctl PCM ops

PCM core deals the empty ioctl field now as default(*).
Let's kill the redundant lines.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-22-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: spi: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:42 +0000 (07:11 +0100)]
ALSA: spi: Drop superfluous ioctl PCM ops

PCM core deals the empty ioctl field now as default(*).
Let's kill the redundant lines.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-21-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: sparc: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:41 +0000 (07:11 +0100)]
ALSA: sparc: Drop superfluous ioctl PCM ops

PCM core deals the empty ioctl field now as default(*).
Let's kill the redundant lines.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-20-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: sh: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:40 +0000 (07:11 +0100)]
ALSA: sh: Drop superfluous ioctl PCM ops

PCM core deals the empty ioctl field now as default(*).
Let's kill the redundant lines.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-19-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: ppc: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:39 +0000 (07:11 +0100)]
ALSA: ppc: Drop superfluous ioctl PCM ops

PCM core deals the empty ioctl field now as default(*).
Let's kill the redundant lines.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-18-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: pcmcia: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:38 +0000 (07:11 +0100)]
ALSA: pcmcia: Drop superfluous ioctl PCM ops

PCM core deals the empty ioctl field now as default(*).
Let's kill the redundant lines.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-17-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: pci: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:37 +0000 (07:11 +0100)]
ALSA: pci: Drop superfluous ioctl PCM ops

PCM core deals the empty ioctl field now as default(*).
Let's kill the redundant lines.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-16-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: trident: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:36 +0000 (07:11 +0100)]
ALSA: trident: Drop superfluous ioctl PCM ops

snd_trident_ioctl() does nothing but calling the default handler.
Now PCM core accepts NULL as the default ioctl ops(*), so let's drop
altogether.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-15-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoALSA: asihpi: Drop superfluous ioctl PCM ops
Takashi Iwai [Tue, 10 Dec 2019 06:11:35 +0000 (07:11 +0100)]
ALSA: asihpi: Drop superfluous ioctl PCM ops

snd_card_asihpi_playback_ioctl() and snd_card_asihpi_capture_ioctl()
do nothing but calling the default handler.
Now PCM core accepts NULL as the default ioctl ops(*), so let's drop
altogether.

(*) commit fc033cbf6fb7 ("ALSA: pcm: Allow NULL ioctl ops")

Link: https://lore.kernel.org/r/20191210061145.24641-14-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>