diff options
Diffstat (limited to 'src/output/oss_plugin.c')
| -rw-r--r-- | src/output/oss_plugin.c | 23 |
1 files changed, 22 insertions, 1 deletions
diff --git a/src/output/oss_plugin.c b/src/output/oss_plugin.c index bd3ccb774..d7df594d3 100644 --- a/src/output/oss_plugin.c +++ b/src/output/oss_plugin.c @@ -41,6 +41,15 @@ # include <sys/soundcard.h> #endif /* !(defined(__OpenBSD__) || defined(__NetBSD__) */ +/* We got bug reports from FreeBSD users who said that the two 24 bit + formats generate white noise on FreeBSD, but 32 bit works. This is + a workaround until we know what exactly is expected by the kernel + audio drivers. */ +#ifndef __linux__ +#undef AFMT_S24_PACKED +#undef AFMT_S24_NE +#endif + struct oss_data { int fd; const char *device; @@ -347,7 +356,7 @@ oss_setup_sample_rate(int fd, struct audio_format *audio_format, case SUCCESS: if (!audio_valid_sample_rate(sample_rate)) break; - + audio_format->sample_rate = sample_rate; return true; @@ -461,6 +470,12 @@ oss_setup_sample_format(int fd, struct audio_format *audio_format, break; audio_format->format = mpd_format; + +#ifdef AFMT_S24_PACKED + if (oss_format == AFMT_S24_PACKED) + audio_format->reverse_endian = + G_BYTE_ORDER != G_LITTLE_ENDIAN; +#endif return true; case ERROR: @@ -502,6 +517,12 @@ oss_setup_sample_format(int fd, struct audio_format *audio_format, break; audio_format->format = mpd_format; + +#ifdef AFMT_S24_PACKED + if (oss_format == AFMT_S24_PACKED) + audio_format->reverse_endian = + G_BYTE_ORDER != G_LITTLE_ENDIAN; +#endif return true; case ERROR: |