Chris Robinson
7d73ac4405
Avoid calling some AL functions from under the context lock
...
There are still some more occurances to clear out (deletion in gen error,
effects, some filters), which shall be coming up. There is a possibility for a
deadlock between the listlock and the global/context lock, if another attempt
to get the listlock is made while under the context lock.
2011-06-16 09:14:41 -07:00
Chris Robinson
c429a5be25
The DirectSound DLL is always loaded by the Win32 API
2011-06-16 06:17:57 -07:00
Chris Robinson
fba58fb1f0
Make sure user configuration is always initialized for Windows
2011-06-16 05:36:16 -07:00
Chris Robinson
dad7f1cf0f
Prefer DllMain when in Windows
2011-06-16 05:34:13 -07:00
Chris Robinson
8bee7a2b7f
Don't suspend individual contexts when updating
2011-06-16 05:29:38 -07:00
Chris Robinson
632b4e6332
Avoid some channel order duplication
2011-06-16 00:00:40 -07:00
Chris Robinson
f26798bd19
Remove some unnecessary casts
2011-06-15 23:52:21 -07:00
Chris Robinson
7f4dcefc80
Remove the databuffer functions and structs
...
It's been disabled for a while now, and not likely to be re-enabled.
2011-06-15 23:46:35 -07:00
Chris Robinson
d702cb9b49
Avoid an unnecessary lock when retrieving the current context
2011-06-15 23:22:34 -07:00
Chris Robinson
2081228055
Declare some variables where they're used
2011-06-15 06:45:51 -07:00
Chris Robinson
f85283c036
Don't prematurely set the HRTF flag
2011-06-15 02:00:41 -07:00
Chris Robinson
14d6ab384d
Use a flag to specify when a device is running
2011-06-15 01:59:07 -07:00
Chris Robinson
62036745f3
Separate config init from library init
...
OSX has a bug where the file descriptor resource limit can't be increased once
certain functions (such as fopen) are used. Since lib initialization happens in
a global constructor and includes opening and parsing config files, it prevents
apps from being able to increase the rlimit once the lib is loaded, which is
usually before main().
By delaying config init, it allows apps the opportunity to change the rlimit
before calling certain OpenAL functions.
2011-06-14 09:43:33 -07:00
Chris Robinson
264274685a
Throw an error when no global initialization is available
2011-06-14 08:47:53 -07:00
Chris Robinson
1babf28491
Unset the local context if it's being destroyed
2011-06-14 08:27:09 -07:00
Chris Robinson
5a15dc4ddf
Avoid multiple list searches when destroying devices and contexts
2011-06-14 07:04:08 -07:00
Chris Robinson
0f782b385b
Create TLS for contexts earlier
2011-06-14 06:57:51 -07:00
Chris Robinson
1534fcfafd
Remove some unneeded includes
2011-06-14 05:06:39 -07:00
Chris Robinson
1c7ed6e693
Avoid using function pointers directly in the ALSA backend
...
Also make it use the lib wrappers
2011-06-14 05:00:59 -07:00
Chris Robinson
0c5c56cf88
Avoid using function pointers directly in the PulseAudio backend
...
Also, switch to the lib wrapper functions
2011-06-14 04:45:52 -07:00
Chris Robinson
53c5275aa4
Use a proper enum for the probe type
2011-06-14 04:02:58 -07:00
Chris Robinson
c93d7a1721
Open the logfile in append/text mode
2011-06-14 03:04:00 -07:00
Chris Robinson
9c7dad1bdf
Combine multiple probe functions into one
2011-06-14 02:28:15 -07:00
Chris Robinson
bfe09b933e
Trace errors in the opensl callback
2011-06-14 01:51:44 -07:00
Chris Robinson
823f5779e1
Set format_pcm according to the device format
2011-06-14 01:48:13 -07:00
Chris Robinson
96a18ba1d5
Make fewer assumptions on the OpenSL buffer count
2011-06-12 19:36:28 -07:00
Chris Robinson
b8bb41107c
Don't be silent when an OpenSL error occurs
2011-06-12 18:10:01 -07:00
Chris Robinson
ccffb7e380
Add missing semicolon
2011-06-12 16:30:51 -07:00
Chris Robinson
39088b2035
Uninline the timeGetTime wrapper
2011-06-12 05:04:36 -07:00
Chris Robinson
8db05b8c17
Avoid defining a macro twice
2011-06-12 04:45:03 -07:00
Chris Robinson
ba069ded40
Add an OpenSL backend
...
Currently for Android's OpenSL ES implementation
2011-06-12 04:41:42 -07:00
Chris Robinson
6b8209fdc9
Cleanup dynamic loading in the PortAudio backend
2011-06-12 04:37:32 -07:00
Chris Robinson
ccf62fc525
Add wrappers to handle dyanmically loading libs
2011-06-12 04:36:24 -07:00
Chris Robinson
6bfab6427a
Avoid calling some function pointers directly
2011-06-12 03:55:02 -07:00
Chris Robinson
aab4f85383
Change HRTF table user input format
2011-06-03 03:15:13 -07:00
Chris Robinson
99b9208896
Improve default HRTF filter delays
...
They're now rounded from the source's float delays
2011-06-03 03:09:13 -07:00
Chris Robinson
7402e3743d
Readd the default HRTF filter (based on the diffuse set)
2011-06-03 01:47:33 -07:00
Chris Robinson
7ddfacb58f
Use a minimum phase HRTF data set
...
This reduces the coefficient size from 128 down to 32, with a set of delays
2011-06-03 01:06:00 -07:00
Chris Robinson
c7a80418d9
Separate device-type parameter checks for alcGetIntegerv
2011-05-29 23:20:33 -07:00
Chris Robinson
e98104dac8
Avoid holding locks once the context is removed from the lists
2011-05-29 23:03:59 -07:00
Chris Robinson
884f668855
Uninline a couple more functions
2011-05-29 03:28:34 -07:00
Chris Robinson
2bbe9d1b1f
Uninline a couple mutex wrapper functions
2011-05-29 03:08:26 -07:00
Chris Robinson
91c3cffe41
Protect the mixer while the device's context array is being updated
2011-05-29 03:04:32 -07:00
Chris Robinson
60a55aa897
Hold the global lock while the device is being updated
2011-05-29 02:56:00 -07:00
Chris Robinson
9325a78043
Use a separate lock for device and context list accesses
2011-05-29 02:26:46 -07:00
Chris Robinson
cd0e80126c
Require a lock for IsDevice calls
2011-05-28 22:21:11 -07:00
Chris Robinson
1778c0d7d8
Get the device format directly from the string
2011-05-28 21:37:33 -07:00
Chris Robinson
b07901f1a0
Set the subformat type when an unknown one is encountered
2011-05-28 21:21:37 -07:00
Chris Robinson
a15a1abb21
Trace the format that is set when reporting HRTF disabled
2011-05-28 20:40:46 -07:00
Chris Robinson
7da2eea8f0
Add a 5.1 device format that uses side channels instead of back
2011-05-28 19:35:32 -07:00
