the original ad-audiomp branch notes:
Add MP locking to the audio drivers.
Making the audio drivers MP safe is necessary before efforts
can be made to make the VM system MP safe.
The are two locks per device instance, an ISR lock and
a character device lock. The ISR lock replaces calls to
splaudio()/splx(), and will be held across calls to device
methods which were called at splaudio() before (e.g.
trigger_output). The character device lock is held across
calls to nearly all of the methods, excluding some only
used for initialization, e.g. get_locks.
Welcome to 5.99.57.
audio framework
Summary of changes:
* struct audio_params
- remove sw_code, factor, factor_denom, hw_sample_rate,
hw_encoding ,hw_precision, and hw_channels. Conversion information
is conveyed by stream_filter_list_t.
- change the type of sample_rate: u_long -> u_int
- add `validbits,' which represents the valid data size in
precision bits. It is required in order to distinguish 24/32bit
from 24/24bit or 32/32bit.
* audio_hw_if
- add two parameters to set_params()
stream_filter_list_t *pfil, stream_filter_list *rfil
A HW driver should set filter recipes for requested formats
- constify audio_params parameters of trigger_output() and
trigger_input(). They represent audio formats for the hardware.
- make open() and close() optional
- add int (AUMODE_PLAY or AUMODE_RECORD) and audio_params_t parameters
to round_blocksize()
* sw_code is replaced with stream_filter_t.
stream_filer_t converts audio data in an input buffer and writes
into another output buffer unlike sw_code, which converts data in
single buffer.
converters in dev/auconv.c, dev/mulaw.c, dev/aurateconv.c,
dev/tc/bba.c, dev/ic/msm6258.c, and arch/arm/iomd/vidcaudio.c are
reimplemented as stream_filter_t
* MI audio
- audiosetinfo() builds filter pipelines from stream_filter_list_t
filled by audio_hw_if::set_params()
- audiosetinfo() returns with EINVAL if mmapped and set_params()
requests filters
- audio_write(), audio_pint(), and audio_rint() invoke a filter
pipeline.
- ioctl() for FIONREAD, AUDIO_WSEEK, AUDIO_GETIOFFS,
AUDIO_GETOOFFS, and audio_prinfo::{seek,samples} for
AUDIO_GETINFO handle values for a buffer nearest to userland.
* add `struct device *' parameter to ac97_attach()
* all of audio HW drivers follow audio_hw_if and ac97 changes
audio device interface:
1) When attempting to match the appropriate mixer control, we weren't
checking the class label, but only the second level label, so for
devices that had both an "inputs.cd" and a "record.cd", for example,
we could never do the right thing except by chance. For this reason,
evidently, all the record masters were labeled (by the underlying
drivers) either "record.record" or "record.volume", to distinguish
from "outputs.master". We'll now accept "record.master", and document
that as the the new preferred way.
2) More importantly, the model was deficient. Selecting a port on many
chips completely disables most of the level controls, which doesn't play
nice with other applications which are trying to use the interface. Now,
selecting a port simply sets which mixer input control shall be changed,
setting state in the audio layer. In other words, the "mixerout" port
is really selected all the time, enabling the final stage mixer, and
setting "gain" sets the level of the appropriate input. It should be
possible for separate applications to each control the mic, dac, and cd
inputs at the same time using this interface, simply by reiterating their
port selections with each change, but applications that don't bother to
do that aren't any worse off than they were before.
The user is expected to set the master output with another application,
dedicated to that task. Though it is now meaningful to select "no port"
with the audio device interface, to manipulate the master output, there's
no particular reason for an audio device consumer to do that. (I added
the capability in order to restore the initial state of the audio device,
for testing purposes. It might also be useful to users of broken binary-
only applications.)
Observe that the mixer device interface (and so, "mixerctl") still
retains all capabilities, including the ability to set the actual input
port on the chip, overriding the level controls. No change is being made
to the mixer device interface. The mixer device simply presents all the
controls on the chip, with no attempt at abstraction, so there are no
bugs there.
The upshot is, that applications that have been trying to use the audio
device interface to change the volume, such as mplayer, now "just work".
I've tested these changes extensively with "eso" and "eap" since first
proposing them on tech-kern last January, and somewhat with "esm" and a
few others. This closes both PR kern/10221, and PR kern/17159.
* Separate the code to set the default parameters into a new function,
audio_set_defaults(). Make it use audiosetinfo(), which properly initializes
the block size and whatnot. Use this in both audioattach() and the
/dev/audio case of audio_open().
* Do not force a reinitialization when /dev/sound is opened.
* Do all of the block size sanity checks in auto_init_ringbuffer(), not in
both audio_calc_blksize() and audiosetinfo().
* Fix a bug in audiosetinfo() that caused the block size to not be recalculated
immediately if we set it to 0.
* For AUDIO_GET[IO]OFFS, modify the deltablks calculation so that it gives us
the number of block boundaries crossed.
http://mail-index.netbsd.org/tech-kern/2002/03/04/0005.html
auconv.c: Add conversion functions
audio.c: Sample alignment, calling conversion functions, etc.
audio_if.h: Add four hw_* members to "struct audio_params"
audiovar.h: Add conversion buffers, etc.
auich and uaudio: Add conversion request code to *_set_params().
The changes is to allow some limited mixer manipulation through
the audio device (instead of the mixer device).
This rendered 4 methods in audio_hw_if unused so garbage collect these.
- Change the way attach and open works to allow multiple audio
devices.
- Split the mulaw.c file into two to avoid dragging in mulaw
convertsion when they are not needed. Add 16 bit alaw/mulaw tables.
- Change the way audio properties are gotten.
- Recognize more versions os SoundBlaster.
- It is now possible to handle devices that want "looping" DMA,
e.g. the SoundBlaster correctly. The WSS and SB drivers use this.
To do this several new methods were introduced in audio_hw_if.
- Different silence handling (forced by previous change).
- The audio driver can now be mmap()-ed, but due to problems in
the VM system only for writing for now.
- The OSS (Linux) audio emulation takes advantage of some of the
new features.
fix semantics of AUDIO_WERROR (now returns a count).
Also repair audio driver back to "real-time mode" where user must
provide data quickly enough for real time audio or silence is played
until user catches up. Add optional "play all" mode (additional bit in
mode field) to play all samples provided, with silence when needed to
avoid buffer underrun.
jmcneill