Re: [PATCH] Cygwin: dsp: Implement SNDCTL_DSP_SETFRAGMENT ioctl().
On Jan 31 20:20, Takashi Yano wrote: > On Tue, 31 Jan 2023 10:28:40 +0100 > Corinna Vinschen wrote: > > On Jan 30 22:09, Takashi Yano wrote: > > > Previously, SNDCTL_DSP_SETFRAGMENT was just a fake. In this patch, > > > it has been implemented to allow latency control in some apps. > > > > > > Reviewed-by: Corinna Vinschen > > > Signed-off-by: Takashi Yano > > > --- > > > winsup/cygwin/fhandler/dsp.cc | 78 - > > > winsup/cygwin/local_includes/fhandler.h | 3 + > > > 2 files changed, 42 insertions(+), 39 deletions(-) > > > > LGTM. Given how much I *don't* use the audio stuff in Cygwin, > > would you just like to take over maintainership for this code? > > BTW, should this pach be applied only for master branch? > What do you think? It's new functionality, so, yeah, main branch only. Thanks, Corinna
Re: [PATCH] Cygwin: dsp: Implement SNDCTL_DSP_SETFRAGMENT ioctl().
On Jan 31 20:18, Takashi Yano wrote: > On Tue, 31 Jan 2023 10:28:40 +0100 > Corinna Vinschen wrote: > > LGTM. Given how much I *don't* use the audio stuff in Cygwin, > > would you just like to take over maintainership for this code? > > Thanks. I could take care of it if you don't mind? Not at all! Thanks, Corinna
Re: [PATCH] Cygwin: dsp: Implement SNDCTL_DSP_SETFRAGMENT ioctl().
On Tue, 31 Jan 2023 10:28:40 +0100 Corinna Vinschen wrote: > On Jan 30 22:09, Takashi Yano wrote: > > Previously, SNDCTL_DSP_SETFRAGMENT was just a fake. In this patch, > > it has been implemented to allow latency control in some apps. > > > > Reviewed-by: Corinna Vinschen > > Signed-off-by: Takashi Yano > > --- > > winsup/cygwin/fhandler/dsp.cc | 78 - > > winsup/cygwin/local_includes/fhandler.h | 3 + > > 2 files changed, 42 insertions(+), 39 deletions(-) > > LGTM. Given how much I *don't* use the audio stuff in Cygwin, > would you just like to take over maintainership for this code? BTW, should this pach be applied only for master branch? What do you think? -- Takashi Yano
Re: [PATCH] Cygwin: dsp: Implement SNDCTL_DSP_SETFRAGMENT ioctl().
On Tue, 31 Jan 2023 10:28:40 +0100 Corinna Vinschen wrote: > LGTM. Given how much I *don't* use the audio stuff in Cygwin, > would you just like to take over maintainership for this code? Thanks. I could take care of it if you don't mind? -- Takashi Yano
Re: [PATCH] Cygwin: dsp: Implement SNDCTL_DSP_SETFRAGMENT ioctl().
On Jan 30 22:09, Takashi Yano wrote: > Previously, SNDCTL_DSP_SETFRAGMENT was just a fake. In this patch, > it has been implemented to allow latency control in some apps. > > Reviewed-by: Corinna Vinschen > Signed-off-by: Takashi Yano > --- > winsup/cygwin/fhandler/dsp.cc | 78 - > winsup/cygwin/local_includes/fhandler.h | 3 + > 2 files changed, 42 insertions(+), 39 deletions(-) LGTM. Given how much I *don't* use the audio stuff in Cygwin, would you just like to take over maintainership for this code? Thanks, Corinna
Re: winsup/cygwin/sys/termios.h bit rates extension
On Jan 29 22:57, Brian Inglis wrote: > FreeBSD, NetBSD, and Linux all bumped their serial bit rates to support > 500k(+500k)4000k, extending the rates to 3500k and 4000k, dropping 128k and > 256k, renumbering the extended baud rate indices under Linux, effectively > changing the ABI for any previously compiled serial application. > > See: > https://cygwin.com/git/?p=newlib-cygwin.git;a=blob;f=winsup/cygwin/include/sys/termios.h;hb=HEAD#l189 > > Patch would be like: > diff a/winsup/cygwin/include/sys/termios.h > b/winsup/cygwin/include/sys/termios.h > --- a/winsup/cygwin/include/sys/termios.h > +++ b/winsup/cygwin/include/sys/termios.h > @@ -190,19 +190,19 > #define CBAUDEX 0x0100f > #define B57600 0x01001 > #define B115200 0x01002 > -#define B128000 0x01003 > -#define B230400 0x01004 > +#define B230400 0x01003 > -#define B256000 0x01005 > -#define B460800 0x01006 > +#define B460800 0x01004 > -#define B50 0x01007 > +#define B50 0x01005 > -#define B576000 0x01008 > +#define B576000 0x01006 > -#define B921600 0x01009 > +#define B921600 0x01007 > -#define B100 0x0100a > +#define B100 0x01008 > -#define B1152000 0x0100b > +#define B1152000 0x01009 > -#define B150 0x0100c > +#define B150 0x0100a > -#define B200 0x0100d > +#define B200 0x0100b > -#define B250 0x0100e > +#define B250 0x0100c > -#define B300 0x0100f > +#define B300 0x0100d > +#define B350 0x0100e > +#define B400 0x0100f > > #define CRTSXOFF 0x04000 > #define CRTSCTS 0x08000 > > Is this acceptable, not really any issue for Cygwin, or an issue, and some > compatibility code would be required to do an internal upgrade, and return > an error for unsupported speeds, or should we add another bit to extend > CBAUD/CBAUDEX to 0x0101f, and use higher indices 0x01010/0x01011? We'd need a compat layer, depending on the version of Cygwin the executable has been created under (see include/sys/cygwin.h). Just extending CBAUD/CBAUDEX. isn't an option because all bits in cflags are taken, afaics. However, afaics our CBAUDEX is defined incorrectly. The BSDs define it as LINUX_CBAUDEX, because it's apparently not a BSD idea. And per Linux, it should only contain the mask bit which defines extended speeds, so #define CBAUD0x0100f #define CBAUDEX 0x01000 would be the right (i. e., Linux-compatible). Corinna