Some of this is taken directly from the 4Front Technologies documentation, so portions © 4Front Technologies.
ioctls listed in current OSS v4.0 documentation
| Still missing | |||||
| ioctl | compatibility | summary | notes | argument | |
| SNDCTL_DSP_COOKEDMODE | ossv4 | {Dis,en}able “on fly” format conversions | 2006-08-15 This one is a toughy. The FreeBSD sound system uses a 2-stage buffer scheme where the primary is closest to the user, and the secondary is a chunk (typically) DMA'd from the soundcard. (This approach supports virtual channels and in-kernel format conversion allowing mixdown of multiple processes' output to a single hardware channel. Cool stuff, really.) However, this ioctl is meant to give processes direct access to the hardware buffer, giving the application the most control possible (ex: minimizing latency by avoiding in-kernel processing). To faithfully support this will require some massaging, which I still intend to do. But since this ioctl doesn't appear to be very popular yet (no source code hits after a cursory Googling), I believe it's lower in the priority list than remaining mixer-releated work. | int enabled | |
| Still missing, and not fully documented | |||||
| ioctl | compatibility | summary | notes | argument | |
| SNDCTL_MIDIINFO | ossv4 | Returns information about a MIDI device | No sample available. “Under construction.” | oss_midi_info | |
| SNDCTL_MIX_ENUMINFO | ossv4 | Not fully documented. See $OSS/sndkit/samples/mixext.readme. oss_mixer_enuminfo is provided in soundcard.h, and can inspect ossmix.c (example program) for hints. | oss_mixer_enuminfo | ||
| SNDCTL_MIX_EXTINFO | ossv4 | Not fully documented. See $OSS/sndkit/samples/mixext.readme. | oss_mixext | ||
| Null stub (commented, but just returns EINVAL) | |||||
| ioctl | compatibility | summary | notes | argument | |
| SNDCTL_CARDINFO | ossv4? | Returns information about a sound card or pseudo device | “The usability of this call is very limited. It's provided only for completeness of the API. OSS API doesn't have any concept of card. Any information returned by this ioctl calld is reserved exclusively for the utility programs included in the OSS package. Applications should not try to use for this information in any ways.” | oss_card_info | |
| SNDCTL_GETLABEL | ossv4 | Returns label assigned to this device | Under construction. From Hannu@4Front: For example ossxmix (just like some HW mixer consoles) can show variable “labels” for certain controls. By default the application name (say quake) is shown as the label but applications may change the labels themselves. | oss_label_t | |
| SNDCTL_SETLABEL | ossv4 | Sets the label for this device | See SNDCTL_GETLABEL. | oss_label_t | |
| SNDCTL_GETSONG | ossv4 | Returns the current song name (if given) related with this device | Under construction. From Hannu@4Front: Audio players like xmms and ossplay can tell the song or file name they are playing. This is shown in /dev/sndstat. More importantly the song name may be obtained by the application located at the downstream side of a loopback device. | oss_longname_t | |
| SNDCTL_SETSONG | ossv4 | Sets the song name related with this device. | See SNDCTL_GETSONG. | oss_longname_t | |
| SNDCTL_SETNAME | ossv4 | Sets the name of this device (if changeable) | Under construction. From Hannu@4Front: This call is used to change the device name reported in /dev/sndstat and ossinfo. So instead of using some generic “OSS loopback audio (MIDI) driver” the device may be given a meaningfull name depending on the current context (for example “OSS virtual wave table synth” or “VoIP link to London”). | oss_longname_t | |
| SNDCTL_DSP_GET_CHNORDER | ossv4-draft | Get the channel ordering of a multi channel device | Under construction; “returns the interleaving order of a multichannel stream” | unsigned long long map | |
| SNDCTL_DSP_READCTL | ossv4 | Reads the S/PDIF interface status. | 4Front advises avoiding this and WRITECTL. | oss_digital_control | |
| SNDCTL_DSP_WRITECTL | ossv4 | Alters the S/PDIF interface setup. | See READCTL | oss_digital_control ctl | |
| Preliminary support added | |||||
| ioctl | compatibility | summary | notes | argument | |
| SNDCTL_DSP_CURRENT_IPTR | ossv4 | Returns current recording position | Same caveat as SNDCTL_DSP_CURRENT_OPTR. | oss_count_t | |
| SNDCTL_DSP_CURRENT_OPTR | ossv4 | Returns current playback position | This operation queries the pre-feeder software buffer, not the actual hardware buffer. As such, the current pointer location isn't exactly accurate. This could be fixed by: (a) translating from hardware pointer location to this buffer; (b) fixing the COOKEDMODE ioctl command and using the hardware buffer pointer locations. (Just using the hardware locations is feasible, but not very useful to a userland application.) | oss_count_t | |
| SNDCTL_DSP_GETERROR | ~ossv3.8 | Returns audio device error information | FreeBSD impl. should work as advertised. | audio_errinfo | |
| SNDCTL_DSP_GETIPEAKS | ossv4 | Returns peak levels for all recording channels | Currently does next to nothing. Appears to be intended for devices with their own peak meters, so drivers must be twiddled. I'm considering doing analysis in-kernel as an off-by-default option, but I imagine any applications needing this functionality would also perform analysis themselves if the hardware wasn't available. | oss_peaks_t | |
| SNDCTL_DSP_GETOPEAKS | ossv4 | The peak levels for all playback channels | Same restrictions as SNDCTL_DSP_GETIPEAKS. | oss_peaks_t | |
| SNDCTL_DSP_GET_PLAYTGT_NAMES | ossv4 | Returns labels for the currently available output routings | Works, but is missing corresponding *SET* ioctl cmds. | oss_mixer_enuminfo | |
| SNDCTL_DSP_GET_PLAYTGT | ossv4 | Returns the current output routing | For now, each device node only corresponds to a single channel, so this will always return 0. (Same for SNDCTL_DSP_SET_PLAYTGT, too.) | int route | |
| SNDCTL_DSP_GETPLAYVOL | ossv4 | Returns the current audio playback volume | int level | ||
| SNDCTL_DSP_GET_RECSRC_NAMES | ossv4 | Returns labels for the currently available recording sources | See GET_PLAYTGT_NAMES. | oss_mixer_enuminfo | |
| SNDCTL_DSP_GET_RECSRC | ossv4 | Returns the current recording source | See SNDCTL_DSP_GET_PLAYTGT. | int route | |
| SNDCTL_DSP_GETRECVOL | ossv4 | Returns the current audio recording level | int level | ||
| SNDCTL_DSP_HALT_INPUT | ossv4 | Aborts audio recording operation | Quoth OSS, “In the earlier OSS versions this ioctl call was called as SNDCTL_DSP_RESET_INPUT. This name is still valid but for clarity reasons it is not recommended in new programs.” Usually used when changing audio format while recording is in progress. | NULL | |
| SNDCTL_DSP_HALT_OUTPUT | ossv4 | Aborts audio playback operation | Similar notes as HALT_INPUT. | NULL | |
| SNDCTL_DSP_LOW_WATER | ossv4 | Sets the trigger threshold for select() | This is initially set to the device fragment size, but what happens when that is changed *after* calling this? Does a change in fragment size overwrite this? | int nbytes | |
| SNDCTL_DSP_POLICY | ossv4 | Sets the timing policy of an audio device | User picks integer 0..10 where 0 uses minimum buffer sizes & 10 uses maximum sizes. Tradeoffs are latency vs. CPU usage. Quoth OSS, “easier to use version of SNDCTL_DSP_SETFRAGMENT.” | int policy | |
| SNDCTL_DSP_SET_PLAYTGT | ossv4 | Sets the current output routing | See GET_PLAYTGT_NAMES. | int route | |
| SNDCTL_DSP_SETPLAYVOL | ossv4 | Changes the current audio playback volume | int level | ||
| SNDCTL_DSP_SET_RECSRC | ossv4 | Sets the current recording source | See GET_RECSRC_NAMES. | int route | |
| SNDCTL_DSP_SETRECVOL | ossv4 | Changes the current audio recording level | “This ioctl call replaces the SOUND_MIXER_IGAIN/RECLEV method used with earlier OSS versions.” | int level | |
| SNDCTL_DSP_SILENCE | ossv4 | Clears the playback buffer with silence | “Can be used to implement playback pause.” Resume w/ SNDCTL_DSP_SKIP. (So sound samples are just lost?) | NULL | |
| SNDCTL_DSP_SKIP | ossv4 | Discards all samples in the playback buffer | “This ioctl call discards all unplayed samples in the playback buffer by moving the current write position immediately before the point where the device is currently reading the samples. This feature can be used to emulate resume function after playback pause.” | NULL | |
| SNDCTL_DSP_SYNCGROUP | ossv4 | Creates a synchronization group. | oss_syncgroup group | ||
| SNDCTL_DSP_SYNCSTART | ossv4 | Starts all devices added to a synchronization group. | BUG: Playback starts on all channels, but gets interrupted (audio pops once like when resetting a channel), though all channels still stay in sync. Still can't track down the cause of this, but will come back to it later. | int id | |
| SNDCTL_AUDIOINFO | ossv4 | Return information about an audio device | There are two problems: 1. Channel assignment/allocation is non-deterministic, so opening an audio device node may not result in getting the corresponding hardware or virtual channel. 2. Is meant to query drivers directly to get rates, capabilities, etc., while _getcaps includes software-emulated formats. | oss_audioinfo | |
| SNDCTL_MIXERINFO | ossv4 | Returns information about a mixer device | Ioctl call supported by all OSS devices. (See SNDCTL_AUDIOINFO.) | oss_mixerinfo | |
| SNDCTL_SYSINFO | ossv4 | Return information about OSS and the devices | For now, only gathers information about audio devices. (Is supposed to compile information about MIDI and mixer devices, too.) | oss_sysinfo | |
| Already supported | |||||
| ioctl | compatibility | summary | notes | argument | |
| SNDCTL_DSP_CHANNELS | all versions | Set the number of audio channels | API doesn't limit number of channels; depends on device. (ex 1=mono, 2=stereo) | ||
| SNDCTL_DSP_GETBLKSIZE | all versions | Get the current fragment size | Obsoleted by SNDCTL_DSP_GETOSPACE SNDCTL_DSP_GETISPACE | int fragsize | |
| SNDCTL_DSP_GETCAPS | ~ossv3.5 | Returns the capabilities of an audio device | See Audio device capabilties | int caps | |
| SNDCTL_DSP_GETFMTS | all versions | Returns a list of natively supported sample formats | Supposed to report only native formats, but FreeBSD also reports emulated formats by default (toggled via sysctl). | ||
| SNDCTL_DSP_GETIPTR | all versions | Returns the current recording pointer | Useful only w/ mmap. Obsoleted by SNDCTL_DSP_CURRENT_IPTR. | count_info | |
| SNDCTL_DSP_GETISPACE | all versions | Returns the number of currently recorded bytes | Can be used to determine # of bytes that can be recorded w/o blocking. OSS docs state it's “out of fashion.” | audio_buf_info | |
| SNDCTL_DSP_GETODELAY | all versions | Returns the playback buffering delay | int delay | ||
| SNDCTL_DSP_GETOPTR | all versions | Returns the current playback pointer | Similar comments to GETIPTR. | count_info | |
| SNDCTL_DSP_GETOSPACE | all versions | Returns the number of currently recorded bytes | Can be used to determine # of bytes that can play w/o blocking. Quoth OSS docs, “This ioctl is out of fashion.” | audio_buf_info | |
| SNDCTL_DSP_GETTRIGGER | ??? | Returns the current trigger bits | Obsolete. “Will be removed from the API in distant future.” | ||
| SNDCTL_DSP_HALT | all versions | Aborts audio recording and/or playback operation | Quoth OSS, though I think there may be a copy/paste mistake somewhere, “There is no need to use this call after opening the device. In the earlier OSS versions this ioctl call was called as SNDCTL_DSP_RESET_OUTPUT.” | NULL | |
| SNDCTL_DSP_SETFRAGMENT | all versions | Sets the buffer size hint | User can hint at max # of fragments, size of fragments | int frag = (max_fragments « 16) | (size_selector) | |
| SNDCTL_DSP_POST | all versions | Forces audio playback to start | According to OSS, useless. FreeBSD just calls chn_start. | NULL | |
| SNDCTL_DSP_SETDUPLEX | all versions | Turns on the duplex mode | NULL | ||
| SNDCTL_DSP_SETFMT | all versions | Select the sample format | Supported formats TODO: compare to presently supported formats (expecting no major differences) | int format | |
| SNDCTL_DSP_SETSYNCRO | ??? | Slaves the audio device to the /dev/sequencer driver | OSS: “obsolete”. | NULL | |
| SNDCTL_DSP_SETTRIGGER | all versions | Starts audio recording and/or playback in sync | OSS: “However in some cases it's necessary to start it manually. There are exactly three cases when it's necessary. 1) When using mmap() this is the only way to start the device. 2) When recording and playback needs to be started exactly at the same moment. 3) When doing normal recording but when even the first read cannot block.” | int mask | |
| SNDCTL_DSP_SPEED | all versions | Set the sampling rate | int rate | ||
| SNDCTL_DSP_SUBDIVIDE | pre-ossv4 | Requests the device to use smaller fragments | “Obsolete call. Not implemented in OSS 4.0 and later. Use SNDCTL_DSP_SETFRAGMENT instead.” | ||
| SNDCTL_DSP_SYNC | all versions | Suspend the application until all samples have been played | NULL | ||
| Missing, but obsolete | |||||
| ioctl | compatibility | summary | notes | argument | |
| SNDCTL_DSP_BIND_CHANNEL | unofficial, obsolete | Route stereo output to the specified channels | Obsoleted by SNDCTL_DSP_SET_PLAYTGT SNDCTL_DSP_GET_PLAYTGT SNDCTL_DSP_GET_PLAYTGT_NAMES | ||
| SNDCTL_DSP_GETCHANNELMASK | unofficial | Returns the bindings supported by the device | Same as SNDCTL_DSP_BIND_CHANNEL | ||
| SNDCTL_DSP_SAMPLESIZE | ??? | Sets the sample size | “This ioctl call is obsolete. You should use SNDCTL_DSP_SETFMT instead.” | ||