Re: [Rosegarden-user] System timer resolution is too low

[Ken R. <ke...@re...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Mon, Jan 29, 2007 at 10:54:16AM +0000, Chris Cannam wrote:
> On Sunday 28 Jan 2007 02:09, Ken Restivo wrote:
> > This guide says, use the RTC timer, it's the best option for Rosegarden:
> > http://tapas.affenbande.org/wordpress/?page_id=40
> >
> > Which is correct? Either:
> > 	1) RTC is the best timing option for using Rosegarden with JACK, or
> > 	2) RTC will lock up your system, don't use it with Rosegarden.
> 
> Michael is right here, I think -- they're both true.  With caveats.
> 
> RTC should be a far better timing option than the default 250Hz system timer.  
> If configured properly it should also be a slightly better timing option than 
> a 1000Hz system timer on an RT kernel, although the system timer still ought 
> to be good enough for decent MIDI use.  Left to their own devices they'll 
> both drift away from the JACK clock source, but Rosegarden doesn't leave them 
> to their own devices, it keeps adjusting the skew to maintain sync.  The 
> results should be OK either way.
> 
> (Your problem with MIDI Thru events piling up and not being processed in time 
> is a slightly separate issue from which timer offers better timing, of 
> course.)
> 
> However I still don't believe that the RTC timer is reliable enough to be a 
> safe recommendation.  It does seem to work OK on some systems, but even with 
> fairly recent kernels I've seen quite a few more or less random lockups.  If 
> you can work out how to get enough information to enable anyone on alsa-devel 
> or the kernel lists to fix this completely, you're a better man than I.
> 

I'm definitely not a better man than you, but Ingo might well be. And it seems he may have fixed this problem?

https://sourceforge.net/mailarchive/forum.php?thread_id=31523098&forum_id=42046

> Note that the problems with RTC do appear to be somewhere in the way that the 
> ALSA drivers use it -- not in Rosegarden and not necessarily in the RTC 
> driver itself.  Other software that uses RTC might be fine.
> 


My admittedly sketchy understanding of this thread, is that the RTC lockup problem was *not* a Rosegarden problem, but rather a bug in the RT kernel itself.  Or are these two different issues?

I've been running this patch for over 8 hours now with no lockups. I still have tons of other system problems, but the lockup one appears to be gone.

- -ken
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)

iD8DBQFFwHfJe8HF+6xeOIcRAiRTAKDBNv3RoWQk/52jI97YASsDdGTt7wCeKa3P
BLrPtdJeLKoHSgglOP8pQfc=
=g807
-----END PGP SIGNATURE-----