7 Mar
2007
7 Mar
'07
11:57 a.m.
hi list,
i just tried, obviously the rt-8 patch from the 2.6.20 branch applies
to the 2.6.20.1 kernel too!
cheers,
doc
7 Mar
7 Mar
3:22 p.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Dragan Noveski wrote:
hi list,
i just tried, obviously the rt-8 patch from the 2.6.20 branch applies
to the 2.6.20.1 kernel too!
same here! works amazingly well!
I had some minor issues with add-on modules (ipw3945 and some thinkpad
specials) but all i needed to do was to s/SA_SHIRQ/IRQF_SHARED/g in the
respective module sources.
I did not yet dig into chrt and IRQs + ACPI as I have two hours of
xrun-free jack UA-25 USB with just over two hours after compiling the
kernel ;-) - suspend + wireless work to, btw!
jackd -R -P 80 -d alsa -d hw:0 -r 48000 -p 64 -n 3 (internal intel 82801G)
jackd -R -P 80 -d alsa -d hw:0 -r 48000 -p 128 -n 2 (USB UA-25)
- -p64 on USB works as well but causes xruns when new clients connect! I
can live with that; some chrt might do the trick.. - I'm just about to
conclude the day with documenting what I've been doing the last two
nights: http://sputnik/wiki/wiki:x60sid
keep on rockin'
#robin
PS. can anyone shed some light on the CONFIG_NO_HZ option - seems like
it saves quite some power. Does it affect realtime performance. can it
be disabled at runtime?
PPS. i've usually disabled freq. scaling during audio-work as there were
x-runs when the cpu speed changed. this does no longer seem to be the
case.. is it still a good habit to shut down powernowd ?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFF7xDeeVUk8U+VK0IRAk1TAJsFYr0OmwLQY/qWjnJI9ELlFvbqqgCgqCHA
WfYPgBkS/OX0Va6G4funtyI=
=IGCN
-----END PGP SIGNATURE-----
3:33 p.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Robin Gareus wrote:
I'm just about to
conclude the day with documenting what I've been doing the last two
nights: http://sputnik/wiki/wiki:x60sid
that won't work for you guys of course:
http://mir.dnsalias.com/wiki/wiki:x60sid
it is :)
cheers,
robin
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFF7xOYeVUk8U+VK0IRAp7tAKC2wKnknoIkavacL98RokC7td7hUACeJOt7
miQC4cUqdr4osYmm1nqHNlE=
=5uzB
-----END PGP SIGNATURE-----
3:44 p.m.
On Wednesday 07 March 2007, Robin Gareus wrote:
Robin Gareus wrote:
Hi,
it doesn't seem as if you have tweaked your IRQ prios [which is the reason one
installs a -rt kernel in the first place [besides other improvements] ;) ]..
Read this [incomplete, partly out of date and somewhat misinformative] page of
mine for some info, because i'm a lazy bugger:
http://tapas.affenbande.org/wordpress/?page_id=40
Regards,
Flo
--
Palimm Palimm!
http://tapas.affenbande.org
I'm just about to
conclude the day with documenting what I've been doing the last two
nights: http://sputnik/wiki/wiki:x60sid
that won't work for you guys of course:
http://mir.dnsalias.com/wiki/wiki:x60sid
it is :)
3:51 p.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Florian Schmidt wrote:
On Wednesday 07 March 2007, Robin Gareus wrote:
i'll read it after dinner, gotta reboot now anyway.
I've still got the 64studio rtirq init.d script but it does not seem to
do anything with this 2.6.20.1-rt8 kernel.
$ cat /proc/interrupts
CPU0 CPU1
0: 22525075 0 IO-APIC-edge timer
1: 65615 0 IO-APIC-edge i8042
8: 0 0 IO-APIC-edge rtc
9: 41535 0 IO-APIC-fasteoi acpi
12: 306777 0 IO-APIC-edge i8042
20: 590376 0 IO-APIC-fasteoi uhci_hcd:usb1, yenta,
i915@pci:0000:00:02.0
21: 507439 0 IO-APIC-fasteoi uhci_hcd:usb2, ohci1394,
HDA Intel, ipw3945
22: 259302 0 IO-APIC-fasteoi uhci_hcd:usb3, sdhci:slot0
23: 543 0 IO-APIC-fasteoi uhci_hcd:usb4, ehci_hcd:usb5
218: 11034 0 PCI-MSI-edge eth1
219: 87243 0 PCI-MSI-edge libata
NMI: 0 0
LOC: 1154790 12345512
ERR: 0
MIS: 0
I have not tried jackdmp either, should I?
till later,
robin
Robin Gareus wrote:
Hi,
it doesn't seem as if you have tweaked your IRQ prios [which is the reason one
installs a -rt kernel in the first place [besides other improvements] ;) ]..
Read this [incomplete, partly out of date and somewhat misinformative] page of
mine for some info, because i'm a lazy bugger:
http://tapas.affenbande.org/wordpress/?page_id=40 I'm just about to
conclude the day with documenting what I've been doing the last two
nights: http://sputnik/wiki/wiki:x60sid
that won't work for you guys of
course:
http://mir.dnsalias.com/wiki/wiki:x60sid
it is :) -----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFF7xe4eVUk8U+VK0IRAhA/AJ9xHzbkC01lV1e1cD18RYfuLjfF5wCbBPvI
j8+0+rreyymYMXKg/ELbaK8=
=w4dY
-----END PGP SIGNATURE-----
4 p.m.
On 3/7/07, Robin Gareus <robin(a)gareus.org> wrote:
$ cat /proc/interrupts
20: 590376 0 IO-APIC-fasteoi uhci_hcd:usb1, yenta,
i915@pci:0000:00:02.0
21: 507439 0 IO-APIC-fasteoi uhci_hcd:usb2, ohci1394,
HDA Intel, ipw3945
22: 259302 0 IO-APIC-fasteoi uhci_hcd:usb3, sdhci:slot0
23: 543 0 IO-APIC-fasteoi uhci_hcd:usb4, ehci_hcd:usb5
For best results make sure the USB sound card is plugged into the port
using IRQ 23 and increase the priority of its handler.
Lee
7:49 p.m.
Lee Revell wrote:
On 3/7/07, Robin Gareus <robin(a)gareus.org>
wrote:
thanks. using usb3 IRQ22 (cause then the cable/port is on the left
side.. whatever) the sd-card slot is idle as I do not even posses any
SD-cards ! I'm gonna blacklist the module - guess that fam and friends
would love to poll it :)
cheers,
robin
$ cat /proc/interrupts
20: 590376 0 IO-APIC-fasteoi uhci_hcd:usb1, yenta,
i915@pci:0000:00:02.0
21: 507439 0 IO-APIC-fasteoi uhci_hcd:usb2, ohci1394,
HDA Intel, ipw3945
22: 259302 0 IO-APIC-fasteoi uhci_hcd:usb3, sdhci:slot0
23: 543 0 IO-APIC-fasteoi uhci_hcd:usb4,
ehci_hcd:usb5
For best results make sure the USB sound card is plugged into the port
using IRQ 23 and increase the priority of its handler.
Lee 
8 Mar
8 Mar
5:08 a.m.
On Wed, March 7, 2007 19:51, Robin Gareus wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Florian Schmidt wrote:
This latest one should be used to work with 2.6.20-rt due to IRQ thread
naming and priority changes:
http://www.rncbc.org/jack/rtirq-20070101.tar.gz
Cheers.
--
rncbc aka Rui Nuno Capela
rncbc(a)rncbc.org
On Wednesday 07 March 2007, Robin Gareus wrote:
i'll read it after dinner, gotta reboot now anyway.
I've still got the 64studio rtirq init.d script but it does not seem to
do anything with this 2.6.20.1-rt8 kernel.
Robin Gareus wrote:
Hi,
it doesn't seem as if you have tweaked your IRQ prios [which is the
reason one installs a -rt kernel in the first place [besides other
improvements] ;) ]..
Read this [incomplete, partly out of date and somewhat misinformative]
page of mine for some info, because i'm a lazy bugger:
http://tapas.affenbande.org/wordpress/?page_id=40
I'm just about to
conclude the day with documenting what I've been doing the last two
nights: http://sputnik/wiki/wiki:x60sid
that won't work for you guys of course:
http://mir.dnsalias.com/wiki/wiki:x60sid
it is :)
7 Mar
7 Mar
3:52 p.m.
On Wednesday 07 March 2007, Florian Schmidt wrote:
On Wednesday 07 March 2007, Robin Gareus wrote:
Erm,
sorry, i might have misread your posts..
Flo
--
Palimm Palimm!
http://tapas.affenbande.org
Robin Gareus wrote:
Hi,
it doesn't seem as if you have tweaked your IRQ prios [which is the reason
one installs a -rt kernel in the first place [besides other improvements]
;) ]..
Read this [incomplete, partly out of date and somewhat misinformative] page
of mine for some info, because i'm a lazy bugger:
http://tapas.affenbande.org/wordpress/?page_id=40
Regards,
Flo I'm just about to
conclude the day with documenting what I've been doing the last two
nights: http://sputnik/wiki/wiki:x60sid
that won't work for you guys of course:
http://mir.dnsalias.com/wiki/wiki:x60sid
it is :)
3:35 p.m.
On 3/7/07, Robin Gareus <robin(a)gareus.org> wrote:
PS. can anyone shed some light on the CONFIG_NO_HZ
option - seems like
it saves quite some power. Does it affect realtime performance. can it
be disabled at runtime?
Should not affect realtime performance - it was judged stable enough
for mainline and I've seen no reports on LKML of it causing problems.
PPS. i've usually disabled freq. scaling during
audio-work as there were
x-runs when the cpu speed changed. this does no longer seem to be the
case.. is it still a good habit to shut down powernowd ?
JACK was fixed to use gettimeofday() as the clock source so disabling
frequency scaling should no longer be needed.
Lee
3:41 p.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Lee Revell wrote:
On 3/7/07, Robin Gareus <robin(a)gareus.org>
wrote:
cool. I am already 10 mins into the compiling a tickless kernel (branch
prediction pipelined it) and will report back if there's trouble!
Robin
PS. can anyone shed some light on the
CONFIG_NO_HZ option - seems like
it saves quite some power. Does it affect realtime performance. can it
be disabled at runtime?
Should not affect realtime performance - it was judged stable enough
for mainline and I've seen no reports on LKML of it causing problems.
PPS. i've usually disabled freq. scaling
during audio-work as there were
x-runs when the cpu speed changed. this does no longer seem to be the
case.. is it still a good habit to shut down powernowd ?
JACK was fixed to use gettimeofday() as the clock source so disabling
frequency scaling should no longer be needed. -----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFF7xVUeVUk8U+VK0IRAnHRAKCkGnlqn2lr+8Y7Ns8S0ngemYaSqACfahyz
EVfSyQxrd9KpA9KPc6Px79o=
=Ukej
-----END PGP SIGNATURE-----
7:21 p.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Robin Gareus wrote:
Lee Revell wrote:
cool. I am already 10 mins into the compiling a tickless kernel (branch
prediction pipelined it) and will report back if there's trouble!
there are ;-)
well no jackd + ardour work great! low latency, no xruns.. USB -p128 -n2
the system works fine but for suspend (uswsusp) - I have not yet tried
suspend2 or various patches and am back to a preemptive 2.6.18-SMP for
day to day use :(
that yields occasional xruns with -p128 on USB. "-p256 -n 2" plays now
x-run free for 25 mins, but I guess I'm just lucky ;) - the internal
soundcard is crap but ran xrun free with -p64 -n3 on this non-rt patched
but preemtive 2.6.18.
back to 2.6.20.1: The idle power consumption with CONFIG_NO_HZ is only
slightly less: 13.2W avg - as opposed to 13.8W avg with 2.6.20-rt8 w/o
CONFIG_NO_HZ) (that was with active wireless and display brightness 4/7
, similar conditions.)
Did any of you guys get suspend/hibernate working with a 2.6.20.1-rt8 or
later? :) on a dual-core & intel-945GM that is. - on a related OT issue:
hdaps (acceleration sensor) only reports zero values with both patched
hdaps_protect-2.6.18.3-2 and vanilla 2.6.20.1 - but works fine under
windoof.
cheers,
robin
On 3/7/07, Robin Gareus <robin(a)gareus.org>
wrote:
PS. can anyone shed some light on the
CONFIG_NO_HZ option - seems like
it saves quite some power. Does it affect realtime performance. can it
be disabled at runtime?
Should not affect realtime performance - it was judged stable enough
for mainline and I've seen no reports on LKML of it causing problems. PPS.
i've usually disabled freq. scaling during audio-work as there were
x-runs when the cpu speed changed. this does no longer seem to be the
case.. is it still a good habit to shut down powernowd ?
JACK was fixed to use
gettimeofday() as the clock source so disabling
frequency scaling should no longer be needed. -----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFF70jZeVUk8U+VK0IRAvykAJ9xLse3IQIZHED/A0RQtp3ETcoVlwCcClxZ
yIQVmvaFEGyOEEc+ZFBgqqk=
=qgKh
-----END PGP SIGNATURE-----
8:15 p.m.
Robin Gareus <robin(a)gareus.org> writes:
causes xruns when new clients connect
This is a current limitation of JACK and there's no way you could fix
that. There are plans to fix this issue.
--
Esben Stien is b0ef@e s a
http://www. s t n m
irc://irc. b - i . e/%23contact
sip:b0ef@ e e
jid:b0ef@ n n
6544
days inactive
6545
days old
linux-audio-user@lists.linuxaudio.org
12 comments
6 participants
participants (6)
-
Dragan Noveski -
Esben Stien -
Florian Schmidt -
Lee Revell -
Robin Gareus -
Rui Nuno Capela