On Fri, 2011-07-22 at 11:42 +0200, Uwe Kleine-König wrote: > Hello, > > On Thu, Jul 21, 2011 at 11:32:52PM +0200, Ben Hutchings wrote: > > On Thu, 2011-07-21 at 16:25 +0200, Uwe Kleine-König wrote: > > > # initramfs-generators > > > initramfs-fallback: linux-initramfs-tool > > > initramfs-tools: initramfs-tools (>= 0.99) > > > diff --git a/config/featureset-rt/config b/config/featureset-rt/config > > > new file mode 100644 > > > index 0000000..8136f16 > > > --- /dev/null > > > +++ b/config/featureset-rt/config > > > @@ -0,0 +1,5 @@ > > > +# CONFIG_PREEMPT_VOLUNTARY is not set > > > +CONFIG_PREEMPT_RT_FULL=y > > > +CONFIG_SCHED_TRACER=y > > > +CONFIG_MISSED_TIMER_OFFSETS_HIST=y > > > +CONFIG_WAKEUP_LATENCY_HIST=y > > > > I notice that you leave Xen (and other pvops) support enabled. I don't > > think this makes sense, as virtualisation will add back the latency and > > jitter that RT is trying to avoid. > hmm, it might not make much sense to run Xen on top of an rt kernel, > that's right. I'm not sure that disabling Xen (and others) is needed > though. There is no exact definition of rt in general, it always depends > on your use-case and then you have to verify/test if the machine you > want to use is capable to fulfill your requirements. So IMHO it's OK to > say: my machine is able to serve my rt requirements as long as it > doesn't run (as) a Xen guest. Don't do it then. Or only do it when > you're not recording audio. > > I'm sure there are many things more that increase the maximal latency, I > don't want (and cannot) go through all of them and choose if they are > still OK or crossed the border. True. > I could still be conviced to disable these, but I don't find how to get > rid of the "This kernel also runs on a Xen hypervisor. It supports both > privileged (dom0) and unprivileged (domU) operation." in the > description. Hints welcome. I think you would need to add a 'amd64_description' section in debian/config/amd64/rt/defines, overriding the one in debian/config/amd64/defines. Not sure. > > > diff --git a/config/featureset-rt/defines b/config/featureset-rt/defines > > > new file mode 100644 > > > index 0000000..caf2baf > > > --- /dev/null > > > +++ b/config/featureset-rt/defines > > > @@ -0,0 +1,7 @@ > > > +[abi] > > > +ignore-changes: * > > > + > > > +[description] > > > +part-long-rt: This kernel includes the PREEMPT_RT patch set > > > > This should be a whole sentence, ending in a full stop (period). > I guess just adding a period is good enough? Yes. Ben. -- Ben Hutchings Absolutum obsoletum. (If it works, it's out of date.) - Stafford Beer
Attachment:
signature.asc
Description: This is a digitally signed message part