The old format tried to do something when the requested format was not
supported by the hardware, but did not actually select any other format.
Now we try to switch to any format supported by dmix, or any other
format when the plugin is not dmix.
The code to set the number of channels did not work when the requested
channel count was not available and when the min/max channel counts were
not identical.
Replacing the entire selection code with
snd_pcm_hw_params_set_channels_near() gives the same result in the cases
where it previously worked, and works in all other cases.
safe_strtoll() now accepts numbers in any base. It formerly assumed that
its input was a decimal number, which had the consequence that
hexadecimal or octal numbers would be parsed as strings when occurring
outside of parameter lists.
This obsoletes some workarounds in the file permission parsing code that
relied on this bug.
Introduce "max_periods" option to specify the max number of periods
per buffer to each plugin.
- When max_periods = -1, the fixed buffer size as the slave size is
used (old behavior).
- When max_periods = 0 (or 1), the number of periods is variable
between 2 and the slave buffer size.
- When max_periods greater than 2 is given, it specifies the max
periods of that pcm explicitly.
When no option is given in the PCM defintion, the value
"defaults.pcm.dmix_max_periods" is referred as default.
The default value is 0, as defined in alsa.conf.
You can override this in ~/.asoundrc or /etc/asound.conf as you like.
Fixed the noisy output of dmix with two (or less) periods.
The dmix tends to give noise or XRUN when running with two periods
because of its implementation nature. To avoid this, the start
position is aligned to the period size, so that the updates are synced
with interrupts of slave PCM.
snd_timer_tread struct is a bad design for 32/64bit compatibility,
and reading this struct on 32bit program returns zero. This results
in tight poll looping (bug#1938, #1945).
For avoiding this bug, now more bigger buffer is read to cover the
64bit tread struct, too. Also this optimizes the read without
checking -EAGAIN in the case both user-space and kernel have the same
tread size.
With the patch, dmix allows apps to use more flexible buffer sizes.
The max buffer size is unlimited, and the minimal buffer size is
(period size * 2). The buffer size is aligned to period size.
The period size is still bound to the period size of slave PCM.
To back to the old behavior (the fixed buffer size), you can set
defaults.pcm.dmix_variable_buffer false
in your configuration.
- Support multi-card/device for dmix/dsnoop/dshare plugins
The unique ipc key is calculated based on card/device/sub index
- Clean up and share the code among all d* plugins
- Refer the defaults.pcm.* configuration
The base ipc_key number, ipc_gid and ipc_perm are referred.
Add to the dmix plugin support for the S24_3LE sample format which is
used by 24-bit USB devices.
The optimized assembler version uses only 23 bits for sample data so
that the lowest bit can be used for synchronization because there is no
24-bit cmpxchg instruction.
The poll() call in the resource server for PCM direct plugins might be
interrupted with an signal. Handle the case gracefuly and don't exit
the server (which brings another problems - cannot connect to socket -
later).
- this patch adds support for suspend & result for dmix and other
direct plugins
- the timer detection / initialization (TREAD support) was redesigned and
the check for proper driver version was moved to the timer_hw.c
- Fix semaphore gid control
- Do semaphore discard for all semaphores (although currently defined as 1)
- Fix the wrong check of return value from shmctl
- Fixed unbalanced semaphores (which may cause deadlock)
- Do semaphore-up before blocking calls for communication with the server
- Don't discard semaphores on the client side
- Open slave PCMs always in non-blocking mode to avoid blocking by semaphore
with the secondary open
dmix transfers data in asynchronously so that it doesn't write on the
last active period. This will allow the arbitrary buffer size in
future, too.
The hwsync call to slave is removed from commit and hwsync callbacks.
This hack was added to fix the noisy output (typically on xmms/bmp)
but it doesn't happen any more. Now the hwsync behavior is as same as
on 1.0.8 (i.e. hwsync is called only when slowptr option is set).
- Fix CPU hang-up during snd_pcm_drain()
A new internal function snd_pcm_wait_nocheck() to force to call poll().
- Clean up, fix status() callback of dmix
More fixes for 100% CPU hang-up problem with dmix.
- Clear timer queues properly when XRUN happens.
- Don't check XRUN when the stream isn't really running.
- Do hwsync properly for the delayed start state.
