11 Feb
2005
11 Feb
'05
10:32 p.m.
Those of you who enjoy working with bleeding-edge kernels, please try
Andrew's latest 2.6.11-rc3-mm2 patchset. I am hopeful that this may
be included in the base kernel, soon. It would help for some of you
to build this kernel, run it and report on any problems (or successes)
with the realtime LSM.
This version no longer supports the `allcaps' option (too much risk),
so you need to run `jackd' rather than `jackstart' with it, unless you
have a very recent version of JACK. (Even so, using `jackd' is
preferable.)
You'll need the 2.6.10 sources from <http://kernel.org/>, then the
2.6.11-rc3 patchset, followed by 2.6.11-rc3-mm2 (in that order).
Configure the realtime-lsm under `security'. It can either be
built-in or loaded as a module (y or m). When the module is active,
its parameters can now be changed dynamically via sysfs like this...
# echo 1 > /sys/module/realtime/parameters/any
# echo 29 > /sys/module/realtime/parameters/gid
# echo 1 > /sys/module/realtime/parameters/mlock
If you try it, please let me know how well it works.
--
joq
12 Feb
12 Feb
10:44 p.m.
New subject: [linux-audio-user] Re: [linux-audio-dev] realtime-lsm in Andrew Morton's latest kernel patchset
On Sat, Feb 12, 2005 at 01:35:09PM EST, Jack O'Quin wrote:
Those of you who enjoy working with bleeding-edge kernels, please try
Andrew's latest 2.6.11-rc3-mm2 patchset. I am hopeful that this may
be included in the base kernel, soon. It would help for some of you
to build this kernel, run it and report on any problems (or successes)
with the realtime LSM.
Is the patch that is in the mm tree available anywhere? I would like to
build some test kernels for Slackware but don't want to build mm
kernels, so the patch would be great in this instance.
--
Luke Yelavich
http://www.audioslack.com
luke(a)audioslack.com
11:04 p.m.
New subject: [linux-audio-user] Re: [linux-audio-dev] realtime-lsm in Andrew Morton's latest kernel patchset
Luke Yelavich wrote:
Is the patch that is in the mm tree available
anywhere? I would like to
build some test kernels for Slackware but don't want to build mm
kernels, so the patch would be great in this instance.
http://www.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.11-rc3/2…
Seems to fit well against 2.6.11-rc3-bk9 too.
cheers, Cal
13 Feb
13 Feb
12:11 a.m.
On 11 February 2005 at 20:35, "Jack O'Quin" <joq(a)io.com> wrote:
Those of you who enjoy working with bleeding-edge
kernels, please try
Andrew's latest 2.6.11-rc3-mm2 patchset. I am hopeful that this may
be included in the base kernel, soon. It would help for some of you
to build this kernel, run it and report on any problems (or successes)
with the realtime LSM.
What version of ALSA does this use?
Thanks....
--
Kevin
12:42 a.m.
New subject: [linux-audio-user] realtime-lsm in Andrew Morton's latest kernel patchset
On Sat, 2005-02-12 at 19:06 -0800, Kevin Cosgrove wrote:
On 11 February 2005 at 20:35, "Jack
O'Quin" <joq(a)io.com> wrote:
You really don't want to run the -mm kernel, unless you really like to
test every bleeding edge feature the kernel developers came up with this
week. IOW, unless you enjoy your machine locking up.
This is important for political reasons, not technical. It means the
realtime LSM is likely to end up in the mainline kernel. It doesn't
really need more testing, everyone knows it works.
Just apply the broken out patch if you need the realtime LSM.
Lee
Those of you who enjoy working with bleeding-edge
kernels, please try
Andrew's latest 2.6.11-rc3-mm2 patchset. I am hopeful that this may
be included in the base kernel, soon. It would help for some of you
to build this kernel, run it and report on any problems (or successes)
with the realtime LSM.
What version of ALSA does this use?
1:37 a.m.
On Sat, 2005-02-12 at 19:06 -0800, Kevin Cosgrove
wrote:
> On 11 February 2005 at 20:35, "Jack O'Quin" <joq(a)io.com> wrote:
>
> > Those of you who enjoy working with bleeding-edge kernels, please try
> > Andrew's latest 2.6.11-rc3-mm2 patchset. I am hopeful that this may
> > be included in the base kernel, soon. It would help for some of you
> > to build this kernel, run it and report on any problems (or successes)
> > with the realtime LSM.
>
> What version of ALSA does this use?
1.0.8
Lee Revell <rlrevell(a)joe-job.com> writes:
You really don't want to run the -mm kernel,
unless you really like to
test every bleeding edge feature the kernel developers came up with this
week. IOW, unless you enjoy your machine locking up.
This is important for political reasons, not technical. It means the
realtime LSM is likely to end up in the mainline kernel. It doesn't
really need more testing, everyone knows it works.
Mostly true. But, Chris Wright and I did make two small changes
lately. They work fine for me, but that's not the same as having 30
or 40 people try it on different systems. Since this version *may*
wind up going into 2.6.11 or 2.6.12, I really *would* like for more
people to test it.
The politics is important, too. It wouldn't hurt for us to
demonstrate to the kernel developers that a number of us really do
care about this feature. Lee and Paul and I have been hammering the
point home, but a few more voices would not hurt.
Lee is right about -mm kernels. Don't try it if you aren't
comfortable working with development kernels. Most likely, nothing
bad will happen to your system. But, there could be problems with
certain devices or configurations. Reporting them to the kernel
developers is a bit help. But, don't run it in any production
systems.
--
joq
2:31 a.m.
Jack O'Quin wrote:
Is the LSM patch, (in the same form as the one in the -mm kernel) easily
available to patch and test against the mainline kernel?
-Reuben
On Sat,
2005-02-12 at 19:06 -0800, Kevin Cosgrove wrote:
>On 11 February 2005 at 20:35, "Jack O'Quin" <joq(a)io.com> wrote:
>
>
>>Those of you who enjoy working with bleeding-edge kernels, please try
>>Andrew's latest 2.6.11-rc3-mm2 patchset. I am hopeful that this may
>>be included in the base kernel, soon. It would help for some of you
>>to build this kernel, run it and report on any problems (or successes)
>>with the realtime LSM.
>
>What version of ALSA does this use?
1.0.8
Lee Revell <rlrevell(a)joe-job.com> writes:
You really don't want to run the -mm kernel,
unless you really like to
test every bleeding edge feature the kernel developers came up with this
week. IOW, unless you enjoy your machine locking up.
This is important for political reasons, not technical. It means the
realtime LSM is likely to end up in the mainline kernel. It doesn't
really need more testing, everyone knows it works.
Mostly true. But, Chris Wright and I did make two small changes
lately. They work fine for me, but that's not the same as having 30
or 40 people try it on different systems. Since this version *may*
wind up going into 2.6.11 or 2.6.12, I really *would* like for more
people to test it.
The politics is important, too. It wouldn't hurt for us to
demonstrate to the kernel developers that a number of us really do
care about this feature. Lee and Paul and I have been hammering the
point home, but a few more voices would not hurt.
Lee is right about -mm kernels. Don't try it if you aren't
comfortable working with development kernels. Most likely, nothing
bad will happen to your system. But, there could be problems with
certain devices or configurations. Reporting them to the kernel
developers is a bit help. But, don't run it in any production
systems.
12:26 p.m.
Reuben Martin <reuben.m(a)gmail.com> writes:
Is the LSM patch, (in the same form as the one in the
-mm kernel)
easily available to patch and test against the mainline kernel?
http://www.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.11-rc3/2…
That should work against recent kernels. There could be problems with
versions earlier than 2.6.10.
I get good results with 2.6.10, BTW. Some of Ingo Molnar's RT
preemption patches have been merged into the mainline. It has much
better realtime latency performance than any stable kernel I've tried.
--
joq
3:01 a.m.
On Sat, Feb 12, 2005 at 11:39:42PM -0600, Jack O'Quin wrote:
The politics is important, too. It wouldn't hurt for us to
demonstrate to the kernel developers that a number of us really do
care about this feature. Lee and Paul and I have been hammering the
point home, but a few more voices would not hurt.
Ok. Let me see if get this right (I'm sort of new to mixing linux and
audio but not new to linux or audio). It is the realtime-lsm patch
that is to be tested, right? If so, and you/we need more people on the
"I'm-happy-with-this-patch". I'm one of the happy users. I use it on a
(crappy) laptop and it runs fine. Or should I report my success to the
kernel list in some way?
My system:
hardware: ACER Ferrari 3000
kernel: 2.6.10-gentoo-r6
realtime-lsm: 0.8.5 (-r1 in gentoo)
alsa: 1.0.6
- Bengan --------------------------------------------------------------
- KTHNOC/SUNET/NORDUnet | http://www.sunet.se/~bengan | +46 8 7906586 -
12:40 p.m.
Bengt Gördén <bengan(a)sunet.se> writes:
Ok. Let me see if get this right (I'm sort of new
to mixing linux and
audio but not new to linux or audio). It is the realtime-lsm patch
that is to be tested, right? If so, and you/we need more people on the
"I'm-happy-with-this-patch". I'm one of the happy users. I use it on a
(crappy) laptop and it runs fine. Or should I report my success to the
kernel list in some way?
Glad to hear it. This is enough.
My system:
hardware: ACER Ferrari 3000
kernel: 2.6.10-gentoo-r6
realtime-lsm: 0.8.5 (-r1 in gentoo)
alsa: 1.0.6
That's good to know.
At this point, I'm mainly interested in results using 2.6.11-rc3-mm2,
because that is the exact version Andrew is merging with the mainline
kernel. Only those comfortable with and interested in building
development kernels should try that.
The main reason we asked Andrew to include this LSM in the kernel is
that it's difficult to keep patches working as each new kernel version
comes out. Once in the kernel tree, that happens automatically.
Many kernel developers opposed this. It *is* something of a hack.
But, no one has been able to come up with a better solution, and this
one at least looks safe. I expect it to be replaced by something
better in the 2.8 timeframe.
--
joq
1:11 p.m.
New subject: [linux-audio-user] realtime-lsm in Andrew Morton's latest kernel patchset
On Sun, 2005-02-13 at 10:42 -0600, Jack O'Quin wrote:
At this point, I'm mainly interested in results
using 2.6.11-rc3-mm2,
because that is the exact version Andrew is merging with the mainline
kernel. Only those comfortable with and interested in building
development kernels should try that.
OS - gentoo (~x86)
kernel - 2.6.11-rc3-RT-V0.7.38-10 (ingo's realtime V0.7.38-10,
realtime-lsm 0.8.5
hardware - Dell Inspiron 5100, p4 2.4ghz, 640mb, 7200rpm, ati m9 64mb
soundcards - Intel 82801DB-ICH4, edirol UA-25 USB
alsa 1.0.8
realtime-lsm working great, some odd lockups with RT patches.
-ry
--
Ryan Gallagher <ruinaudio(a)comcast.net>
4:10 p.m.
New subject: [linux-audio-user] realtime-lsm in Andrew Morton's latest kernel patchset
On Sun, 2005-02-13 at 11:13 -0600, Ryan Gallagher wrote:
On Sun, 2005-02-13 at 10:42 -0600, Jack O'Quin
wrote:
That's expected behavior. I configure with PREEMPT_DESKTOP because
PREEMPT_RT is not stable enough yet.
Lee
At this point, I'm mainly interested in
results using 2.6.11-rc3-mm2,
because that is the exact version Andrew is merging with the mainline
kernel. Only those comfortable with and interested in building
development kernels should try that.
OS - gentoo (~x86)
kernel - 2.6.11-rc3-RT-V0.7.38-10 (ingo's realtime V0.7.38-10,
realtime-lsm 0.8.5
hardware - Dell Inspiron 5100, p4 2.4ghz, 640mb, 7200rpm, ati m9 64mb
soundcards - Intel 82801DB-ICH4, edirol UA-25 USB
alsa 1.0.8
realtime-lsm working great, some odd lockups with RT patches. 
7:12 p.m.
Jack O'Quin wrote:
...
Many kernel developers opposed this. It *is* something of a hack.
But, no one has been able to come up with a better solution, and this
one at least looks safe. I expect it to be replaced by something
better in the 2.8 timeframe.
Whoa... like in two or more years from now :-(
Is there any extra reasonable lobbying the rest of us
could do ?
I tried the patches but my system fell over, probably
for other reasons (nvidia etc). I hope to try again rsn.
FWIW the -mm2 patch would not apply over a Gentoo patched
kernel but a "normal" kernel, 2.6.10 + rc3 + mm2, would
not load the nvidia-kernel (amd64) Gentoo ebuilt module.
--markc
10:05 p.m.
Mark Constable <markc(a)renta.net> writes:
Jack O'Quin wrote:
That's why I am patiently working to get realtime-lsm merged soon.
...
Many kernel developers opposed this. It *is* something of a hack.
But, no one has been able to come up with a better solution, and this
one at least looks safe. I expect it to be replaced by something
better in the 2.8 timeframe.
Whoa... like in two or more years from now :-( Is there any extra reasonable lobbying the rest of us
could do ?
Looks like things are going OK right now. No need to rock the boat
unless something comes unglued again.
I tried the patches but my system fell over, probably
for other reasons (nvidia etc). I hope to try again rsn.
FWIW the -mm2 patch would not apply over a Gentoo patched
kernel but a "normal" kernel, 2.6.10 + rc3 + mm2, would
not load the nvidia-kernel (amd64) Gentoo ebuilt module.
Sorry to hear that. Thanks for trying.
--
joq
11:12 p.m.
New subject: [linux-audio-user] realtime-lsm in Andrew Morton's latest kernel patchset
On Sun, 2005-02-13 at 20:07 -0600, Jack O'Quin wrote:
Mark Constable <markc(a)renta.net> writes:
Congratulations for seeing this through. After endless and often
frustrating debate on LKML I have gained a real appreciation of the kind
of scrutiny that Linux kernel code is subjected to. If they add a
feature that exposes any kind of API whatsoever they are committed to
supporting it forever. It's not easy to convince the kernel developers
that you have solved a problem they are not aware existed, but we seem
to have done it.
I think the argument that convinced them is that the API (none from the
user POV and only 3 parameters from the admin POV) is so simple it can
be easily and transparently replaced by a better solution anytime.
Lee
Jack O'Quin wrote:
That's why I am patiently working to get realtime-lsm merged soon. ...
Many kernel developers opposed this. It *is* something of a hack.
But, no one has been able to come up with a better solution, and this
one at least looks safe. I expect it to be replaced by something
better in the 2.8 timeframe.
Whoa... like in two or more years from now :-(
14 Feb
14 Feb
2:03 a.m.
Mark Constable wrote:
FWIW the -mm2 patch would not apply over a Gentoo
patched
kernel but a "normal" kernel, 2.6.10 + rc3 + mm2, would
not load the nvidia-kernel (amd64) Gentoo ebuilt module.
You don't need mm2 to test out the realtime-lsm. Just apply the broken
out patch. I applied it to Gentoo-Sources 2.6.10-r7 and it works just fine.
-Reuben
6:05 p.m.
New subject: [linux-audio-user] realtime-lsm in Andrew Morton's latestkernelpatchset
El dom, 13-02-2005 a las 20:12, Mark Constable escribió:
[snip]
I tried the patches but my system fell over, probably
for other reasons (nvidia etc). I hope to try again rsn.
I heard that there were some problems with the propietary nvidia driver
with the 2.6.8 kernel, but not the 2.6.10. Anyway, there was also a
problem with old cards being unsopported (including some Geforce 4)
again with the same driver. I think they fixed it in the last version,
but I'm not sure.
FWIW the -mm2 patch would not apply over a Gentoo
patched
kernel but a "normal" kernel, 2.6.10 + rc3 + mm2, would
not load the nvidia-kernel (amd64) Gentoo ebuilt module.
--markc
Just my 2 pence
Damian-
6:25 a.m.
New subject: [linux-audio-user] realtime-lsm in Andrew Morton's latest kernel patchset
Jack O'Quin wrote:
At this point, I'm mainly interested in results using 2.6.11-rc3-mm2,
because that is the exact version Andrew is merging with the mainline
kernel. Only those comfortable with and interested in building
development kernels should try that.
The main reason we asked Andrew to include this LSM in the kernel is
that it's difficult to keep patches working as each new kernel version
comes out. Once in the kernel tree, that happens automatically.
Many kernel developers opposed this. It *is* something of a hack.
But, no one has been able to come up with a better solution, and this
one at least looks safe. I expect it to be replaced by something
better in the 2.8 timeframe.
OK. I'm quite happy with Ingo's RT patch /and/ the realtime-lsm module. I
am currently running 2.6.11-rc4-RT-V0.7.39-01 on my laptop (P4/UP) and on
my desktop (P4/SMP-HT). And this is all about PREEMPT_RT=y.
FYI the realtime-lsm module is now being provided by the broken out patch
from 2.6.11-rc3-mm2, and it builds and works fine, as ever expected :)
To speak the truth :) I can't live without this patch combination anymore.
Its been a long way, not without drawbacks. On my hardware setup, nowadays
I already consider those pretty stable to my standards -- already in
production status to me.
That's mostly why I don't have a point to test on 2.6.11-rc3-mm2, but I
took the rt-lsm.patch from it, if that's of any help.
Cheers.
--
rncbc aka Rui Nuno Capela
rncbc(a)rncbc.org
13 Feb
13 Feb
2:34 a.m.
On 12 February 2005 at 23:42, Lee Revell <rlrevell(a)joe-job.com> wrote:
Just apply the broken out patch if you need the
realtime LSM.
I have that patch, or is it realcap, applied to my 2.6.7 kernel.
Mandrake 10.1 comes with ALSA 1.0.6. Both the kernel and the ALSA
version are known to be less than stellar. I'm looking to move to at
least 2.6.8 and 1.0.7 when I can get my hands on something stable.
Maybe Mandrake 10.2 (now in beta I think) will be of help.
TTFN....
--
Kevin
7317
days inactive
7320
days old
linux-audio-user@lists.linuxaudio.org
18 comments
11 participants
participants (11)
-
Bengt Gördén -
Cal -
Jack O'Quin -
Kevin Cosgrove -
Lee Revell -
Lisandro Damián Nicanor Perez Meyer -
Luke Yelavich -
Mark Constable -
Reuben Martin -
Rui Nuno Capela -
Ryan Gallagher