From: FFADO <ffa...@ff...> - 2012-07-17 13:06:25
|
#242: ntp interferes with ffado ------------------------+--------------------------------------------------- Reporter: ppalmers | Owner: ppalmers Type: bug | Status: new Priority: major | Milestone: FFADO 2.x Component: | Version: FFADO 2.0-rc2 (1.999.42) Resolution: | Keywords: Device_name: all | ------------------------+--------------------------------------------------- Comment (by jwoithe): Further to [comment:60 comment 60], I've checked in r2186. In light of the realisation that rate differences between CLOCK_MONOTONIC and CLOCK_MONOTONIC_RAW could cause inconsistent sleeps when times are taken from the latter but the former is used for sleep timing, we'll try using CLOCK_MONOTONIC for everything. The ABS_TIMER mode is also utilised where appropriate if flagged as available at compile time. Stefan: based on tests here and earlier ones done by yourself I expect that this change should permit correct operation on your "good-RTC" PC. The behaviour on the "bad-RTC" PC is unlikely to be different to that which you detailed in [comment:58 comment 58]. However I think we're approaching a solution which may allow this ticket to be closed off. Regarding the xruns flagged on the bad-RTC PC, I am beginning to think that we've done almost all we can do to deal with PCs whose RTCs are so badly out of whack. While you're still getting xruns flagged, it seems that FFADO is at least able to continue now. As to why these xruns are triggered I'm pretty much out of fresh ideas. A race condition during the setting of the clock (as you suggested) seems to be the most likely candidate at the moment, and if that's the case there's probably not a whole lot more FFADO can do. As always, thoughts and test results are welcome. -- Ticket URL: <http://subversion.ffado.org/ticket/242#comment:61> FFADO <http://subversion.ffado.org/index.fcgi> Free Firewire Audio Drivers |