[PATCH] pipe_base.h: use tsc_khz for tsc-to-nanoseconds calculations

This fixes a clock drift issue that is measurable with clocktest. The change was suggested by Jan a while ago [0], I tested it again on 4.19.89. [0] https://www.xenomai.org/pipermail/xenomai/2019-May/040929.html Signed-off-by: Bart Vissers --- arch/x86/include/asm/ipipe_base.h | 4 ++-- 1 file

Re: Large ToD drift with clocktest

a regression regarding the > TSC-to-nanoseconds > > calculations. Maybe check the parameters Xenomai is using > internally, if they > > are differing. > > > > Jan > > > > > > > > Thanks again, > > > Bart

Re: Large ToD drift with clocktest

Jan > > > > > Thanks again, > > Bart > > > > On Tue, 14 May 2019 at 09:14, Jan Kiszka > <mailto:jan.kis...@web.de>> wrote: > > > > On 13.05.19 18:41, Bart Vissers via Xenomai wrote: > > > Hi all, > > > > >

Re: Large ToD drift with clocktest

ork due to too much drift wrt to slave clocks. We don't really care about the system time, but I added ntp logs because I thought it might help finding the cause. Thanks again, Bart On Tue, 14 May 2019 at 09:14, Jan Kiszka wrote: > On 13.05.19 18:41, Bart Vissers via Xenomai wrote: > &

Large ToD drift with clocktest

Hi all, When running clocktest I get the following output: $ /usr/xenomai/bin/clocktest == Testing built-in CLOCK_REALTIME (0) CPU ToD offset [us] ToD drift [us/s] warps max delta [us] --- -- -- 01553975.5 4

Re: latency test

I've done some testing with the X-window suggestions at: https://gitlab.denx.de/Xenomai/xenomai/wikis/Troubleshooting#the_latency_test_shows_high_latencies My results on an Intel PC with Xenomai 3.0.7, Linux 4.9, Debian 9.7 with xfce4, worst-case latencies: - Using Intel driver: 39.8 us - Using fb