Help with Installation
Hi all,
I'm trying to get trough my first installation of cpufreq and I was wondering if someone could point me to some more documentation on how to install it. I got the cpufreq-LINUX_2_4-20040402.tar.gz tarball for the 2.4.25 kernel.
I'm trying this out on my IBM ThinkPad T41 which I suppose would be using the speedstep-centrino driver. The question is : how is the cpufreq package included in the configuring and compile of the kernel ?
I've been trying to find some sort of an INSTALL file without luck, does anyone have or know of such a description ?
Signature · 1 line
Ozeki San
Quoted 6 lines · Ozeki San
Hi all,
I'm trying to get trough my first installation of cpufreq and I was wondering if someone could point me to some more documentation on how to install it. I got the cpufreq-LINUX_2_4-20040402.tar.gz tarball for the 2.4.25 kernel.
Quoted 6 lines
I'm trying this out on my IBM ThinkPad T41 which I suppose would be using the speedstep-centrino driver. The question is : how is the cpufreq package included in the configuring and compile of the kernel ?
I've been trying to find some sort of an INSTALL file without luck, does anyone have or know of such a description ?
tar xzvfp cpufreq-...tar.gz
cd cpufreq
./patchin.sh /path/to/kernel/sources
Do *not* remove the cpufreq directory that was unpacked. then
cd /path/to/kernel/sources
make menuconfig (or your preferred *config).
Under 'Processor type and feature' you will have a new 'CPU Frequency scaling' menu.
Under it:
'CPU Frequency scaling' -> yes
'CPU frequency table helpers' -> yes
Perhaps 'Support for governing from userspace' -> yes (or else most daemon may not work),
Then probably
'Intel Pentium-M Enhanced SpeedStep' -> yes or module, as you like.
Cheers,
Signature · 4 lines
Bruno Ducrot -- Which is worse: ignorance or apathy? -- Don't know. Don't care.
Quoted 5 lines · On Sat, 3 Apr 2004 16:50:45 +0200 Bruno Ducrot wrote:
tar xzvfp cpufreq-...tar.gz
cd cpufreq
./patchin.sh /path/to/kernel/sourcesDo *not* remove the cpufreq directory that was unpacked.
Why did you set up such a strange procedure?
What are the advantages over a simple patch like the old ones found in
http://www.codemonkey.org.uk/projects/cpufreq/ ?
I'd say a gzipped (or bzip2ed) patch file would be much more easy to apply (especially with automatic tools such as Debian kernel-package, that is, make-kpkg), but of course I may be wrong...
Could you please explain the rationale behind such a choice?
Signature · 4 lines
| GnuPG Key ID = DD6DFCF4 | You're compiling a program
Francesco | Key fingerprint = | and, all of a sudden, boom!
Poli | C979 F34B 27CE 5CD8 DC12 | -- from APT HOWTO,
| 31B5 78F4 279B DD6D FCF4 | version 1.8.0_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq
Quoted 12 lines · Francesco Poli
Quoted 5 lines · On Sat, 3 Apr 2004 16:50:45 +0200 Bruno Ducrot wrote:
tar xzvfp cpufreq-...tar.gz
cd cpufreq
./patchin.sh /path/to/kernel/sourcesDo *not* remove the cpufreq directory that was unpacked.
Why did you set up such a strange procedure?
What are the advantages over a simple patch like the old ones found in
http://www.codemonkey.org.uk/projects/cpufreq/ ?
Quoted 5 lines
I'd say a gzipped (or bzip2ed) patch file would be much more easy to apply (especially with automatic tools such as Debian kernel-package, that is, make-kpkg), but of course I may be wrong...
Could you please explain the rationale behind such a choice?
We don't have the (massive) resources to be able to generate a patch for every single kernel version which someone may want to apply the cpufreq code to from the CVS.
Signature · 4 lines
Russell King
Linux kernel 2.6 ARM Linux - http://www.arm.linux.org.uk/
maintainer of: 2.6 PCMCIA - http://pcmcia.arm.linux.org.uk/
2.6 Serial coreQuoted 3 lines · On Sat, 3 Apr 2004 17:22:42 +0100 Russell King wrote:
We don't have the (massive) resources to be able to generate a patch for every single kernel version which someone may want to apply the cpufreq code to from the CVS.
Well, I think that a patch against the latest 2.4.x kernel and one against the latest 2.4.x pre-release (2.4.x-prex or 2.4.x-rcx) would suffice.
Adapting one such patch in order to apply it to some less recent 2.4.x kernel version is not so difficult.
At least, that was my personal experience: I successfully applied cpufreq-2.4.22-pre6-1.gz (by hand-editing the diff file) to 2.4.22, 2.4.23, 2.4.24 and 2.4.25.
Adapting a hypothetical cpufreq-2.4.25-1.gz to the same kernel versions should not be more challenging...
What's your opinion?
Signature · 4 lines
| GnuPG Key ID = DD6DFCF4 | You're compiling a program
Francesco | Key fingerprint = | and, all of a sudden, boom!
Poli | C979 F34B 27CE 5CD8 DC12 | -- from APT HOWTO,
| 31B5 78F4 279B DD6D FCF4 | version 1.8.0_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq
Quoted 11 lines · Francesco Poli
Quoted 5 lines · On Sat, 3 Apr 2004 16:50:45 +0200 Bruno Ducrot wrote:
tar xzvfp cpufreq-...tar.gz
cd cpufreq
./patchin.sh /path/to/kernel/sourcesDo *not* remove the cpufreq directory that was unpacked.
Why did you set up such a strange procedure?
What are the advantages over a simple patch like the old ones found in
http://www.codemonkey.org.uk/projects/cpufreq/ ?
The reason is that there are too few cpufreq developers. I'd done the patches found at that site [and a lot more in the months before], but I lack the time to continue providing these patches.
Quoted 5 lines
I'd say a gzipped (or bzip2ed) patch file would be much more easy to apply (especially with automatic tools such as Debian kernel-package, that is, make-kpkg), but of course I may be wrong...
Could you please explain the rationale behind such a choice?
Number of active developers and time these can [not want!] spend on this project.
Quoted 6 lines
Well, I think that a patch against the latest 2.4.x kernel and one against the latest 2.4.x pre-release (2.4.x-prex or 2.4.x-rcx) would suffice.
Adapting one such patch in order to apply it to some less recent 2.4.x kernel version is not so difficult.
Well, it might be usefull for you and others if someone (you?) could provide such up-to-date patches, and keeps 2.4. cpufreq up-to-date with recent 2.6. developments, as far as the latter is possible. However, none of the people currently active in cpufreq development seem to have sufficient time to do so. In case _you_ want to do it [as you said, it shouldn't be so difficult], please do so.
Dominik
_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq
Quoted 6 lines · On Sun, 4 Apr 2004 23:11:44 +0200 Dominik Brodowski wrote:
Well, it might be usefull for you and others if someone (you?) could provide such up-to-date patches, and keeps 2.4. cpufreq up-to-date with recent 2.6. developments, as far as the latter is possible. However, none of the people currently active in cpufreq development seem to have sufficient time to do so. In case _you_ want to do it [as you said, it shouldn't be so difficult], please do so.
Perhaps it would be useful to setup a little script that
1) downloads the latest 2.4.x kernel (maybe via rsync)
2) modifies it using latest cpufreq from
http://ftp.linux.org.uk/pub/linux/cpufreq/
3) creates the patch against the unmodified kernel
4) gzips it and moves the resulting file in the right place
Such a script should be run daily (via cron) on ftp.linux.org.uk
This could perhaps work and provide up-to-date patches, as long as 2.4.x cpufreq at http://ftp.linux.org.uk/pub/linux/cpufreq/ is kept up-to-date with 2.6.x
Perhaps patchin.sh should be modified in order to use cp (or mv) rather than ln -s
What do you think?
Signature · 4 lines
| GnuPG Key ID = DD6DFCF4 | You're compiling a program
Francesco | Key fingerprint = | and, all of a sudden, boom!
Poli | C979 F34B 27CE 5CD8 DC12 | -- from APT HOWTO,
| 31B5 78F4 279B DD6D FCF4 | version 1.8.0_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq
Quoted 3 lines · Francesco Poli
1) downloads the latest 2.4.x kernel (maybe via rsync)
2) modifies it using latest cpufreq from
http://ftp.linux.org.uk/pub/linux/cpufreq/
This would be a manual step.
Quoted 1 line
Such a script should be run daily (via cron) on ftp.linux.org.uk
You're making the very big assumption that ftp.linux.org.uk _has_ the space to do this. The fact is that disk space is short, and the replacement server has been lost by the parcel carrier, and the state of parcel carriers over here means that it's only possible to recover a fraction of the value. Yes, we know in hind sight that it would've been better to have put it into a car and driven it to the hosting facility, but hind sight is such a wonderful thing.
So, _we_ _do_ _not_ _currently_ _have_ _the_ _resources_ _to_ _host_ _such_ _a_ _service_. As I've said _three_ times in this thread so far.
The fact is that right now there is only 10MB free for the anonymous FTP archive.
Signature · 4 lines
Russell King
Linux kernel 2.6 ARM Linux - http://www.arm.linux.org.uk/
maintainer of: 2.6 PCMCIA - http://pcmcia.arm.linux.org.uk/
2.6 Serial coreQuoted 4 lines · On Mon, 5 Apr 2004 22:06:54 +0100 Russell King wrote:
Quoted 2 lines
2) modifies it using latest cpufreq from
http://ftp.linux.org.uk/pub/linux/cpufreq/This would be a manual step.
Why? Did I misunderstand something?
Wouldn't some shell commands like
tar xzvfp cpufreq-...tar.gz
cd cpufreq
./patchin.sh /path/to/kernel/sources
suffice?
I am quoting Bruno Ducrot's instructions... but maybe I don't catch something important...
Quoted 9 lines
Quoted 2 lines
Such a script should be run daily (via cron) on ftp.linux.org.uk
You're making the very big assumption that ftp.linux.org.uk _has_ the space to do this. [...]
So, _we_ _do_ _not_ _currently_ _have_ _the_ _resources_ _to_ _host_ _such_ _a_ _service_. As I've said _three_ times in this thread so far.
I'm sorry: I didn't realize you were /so/ short of space... :-( I thought you meant you couldn't generate and store one daily patch for /every/ 2.4.x kernel version (that is 25 daily patches + a zillion of patches against /all/ the pre-releases).
I now realize that you don't even have enough resources for one daily patch against /one/ kernel version.
Quoted 3 lines
The fact is that right now there is only 10MB free for the anonymous FTP archive.
That's really a pity... :-(
Signature · 4 lines
| GnuPG Key ID = DD6DFCF4 | You're compiling a program
Francesco | Key fingerprint = | and, all of a sudden, boom!
Poli | C979 F34B 27CE 5CD8 DC12 | -- from APT HOWTO,
| 31B5 78F4 279B DD6D FCF4 | version 1.8.0_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq
Quoted 9 lines · Francesco Poli
Wouldn't some shell commands like
tar xzvfp cpufreq-...tar.gz
cd cpufreq
./patchin.sh /path/to/kernel/sourcessuffice?
I am quoting Bruno Ducrot's instructions... but maybe I don't catch something important...
That assumes that the CVS has been kept up to date with the latest 2.4 kernel.
Quoted 1 line
I'm sorry: I didn't realize you were /so/ short of space... :-(
Well, last nights generation of the CVS snapshot has failed due to lack of disk space.
As I've said, this was meant to be solved by a replacement machine, and had that replacement machine not been lost by the parcel carrier, it would've been on line by now.
Signature · 4 lines
Russell King
Linux kernel 2.6 ARM Linux - http://www.arm.linux.org.uk/
maintainer of: 2.6 PCMCIA - http://pcmcia.arm.linux.org.uk/
2.6 Serial coreQuoted 6 lines · On Wed, 7 Apr 2004 08:11:33 +0100 Russell King wrote:
Quoted 6 lines
Wouldn't some shell commands like
tar xzvfp cpufreq-...tar.gz
cd cpufreq
./patchin.sh /path/to/kernel/sources
Quoted 6 lines
Quoted 3 lines
suffice?
I am quoting Bruno Ducrot's instructions... but maybe I don't catch something important...That assumes that the CVS has been kept up to date with the latest 2.4 kernel.
I see: so some manual adjustments could be necessary and that is why you said this should be a manual operation.
TNX for the explanation!
Signature · 4 lines
| GnuPG Key ID = DD6DFCF4 | You're compiling a program
Francesco | Key fingerprint = | and, all of a sudden, boom!
Poli | C979 F34B 27CE 5CD8 DC12 | -- from APT HOWTO,
| 31B5 78F4 279B DD6D FCF4 | version 1.8.0_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq
Quoted 9 lines · Francesco Poli
Quoted 6 lines · On Sun, 4 Apr 2004 23:11:44 +0200 Dominik Brodowski wrote:
Well, it might be usefull for you and others if someone (you?) could provide such up-to-date patches, and keeps 2.4. cpufreq up-to-date with recent 2.6. developments, as far as the latter is possible. However, none of the people currently active in cpufreq development seem to have sufficient time to do so. In case _you_ want to do it [as you said, it shouldn't be so difficult], please do so.
Quoted 7 lines
Perhaps it would be useful to setup a little script that
1) downloads the latest 2.4.x kernel (maybe via rsync)
2) modifies it using latest cpufreq from
http://ftp.linux.org.uk/pub/linux/cpufreq/
3) creates the patch against the unmodified kernel
4) gzips it and moves the resulting file in the right place
IMHO, such a script can only work well if it is "being watched", e.g. someone (you?) verifies the results equal what you expect. Also, differences in the base kernel need to be tracked which might make adjustments of the cpufreq files / patch necessary.
Quoted 5 lines
Such a script should be run daily (via cron) on ftp.linux.org.uk
This could perhaps work and provide up-to-date patches, as long as 2.4.x cpufreq at http://ftp.linux.org.uk/pub/linux/cpufreq/ is kept up-to-date with 2.6.x
But assuming that this CVS export is up-to-date [which it isn't], why is your script necessary?
Quoted 2 lines
Perhaps
patchin.shshould be modified in order to use cp (or mv) rather than ln -s
Modifying the patchin.sh script to become more user friendly is something which can be done -- but _please_ be aware that 2.4. cpufreq is, at best, in "beta" state and not as up to date as 2.6.
Dominik
_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq
Quoted 32 lines · Dominik Brodowski
Quoted 9 lines · Francesco Poli
Quoted 6 lines · On Sun, 4 Apr 2004 23:11:44 +0200 Dominik Brodowski wrote:
Well, it might be usefull for you and others if someone (you?) could provide such up-to-date patches, and keeps 2.4. cpufreq up-to-date with recent 2.6. developments, as far as the latter is possible. However, none of the people currently active in cpufreq development seem to have sufficient time to do so. In case _you_ want to do it [as you said, it shouldn't be so difficult], please do so.
Quoted 7 lines
Perhaps it would be useful to setup a little script that
1) downloads the latest 2.4.x kernel (maybe via rsync)
2) modifies it using latest cpufreq from
http://ftp.linux.org.uk/pub/linux/cpufreq/
3) creates the patch against the unmodified kernel
4) gzips it and moves the resulting file in the right placeIMHO, such a script can only work well if it is "being watched", e.g. someone (you?) verifies the results equal what you expect. Also, differences in the base kernel need to be tracked which might make adjustments of the cpufreq files / patch necessary.
Quoted 5 lines
Such a script should be run daily (via cron) on ftp.linux.org.uk
This could perhaps work and provide up-to-date patches, as long as 2.4.x cpufreq at http://ftp.linux.org.uk/pub/linux/cpufreq/ is kept up-to-date with 2.6.x
But assuming that this CVS export is up-to-date [which it isn't], why is your script necessary?
Quoted 7 lines
Quoted 2 lines
Perhaps
patchin.shshould be modified in order to use cp (or mv) rather than ln -sModifying the
patchin.shscript to become more user friendly is something which can be done -- but _please_ be aware that 2.4. cpufreq is, at best, in "beta" state and not as up to date as 2.6.
I can host this if that can help. But do not expect though that such patch will be updated so much though.
Signature · 4 lines
Bruno Ducrot -- Which is worse: ignorance or apathy? -- Don't know. Don't care.
Quoted 3 lines · Bruno Ducrot
I can host this if that can help. But do not expect though that such patch will be updated so much though.
Done. It's under ftp://ftp.poupinou.org/cpufreq/
and it's called cpufreq-24-2004040601.diff,
with .gz and .bz2 extensions.
Signature · 4 lines
Bruno Ducrot -- Which is worse: ignorance or apathy? -- Don't know. Don't care.
Quoted 3 lines · On Tue, 6 Apr 2004 10:47:23 +0200 Bruno Ducrot wrote:
Done. It's under ftp://ftp.poupinou.org/cpufreq/
and it's called cpufreq-24-2004040601.diff,
with .gz and .bz2 extensions.
Wow! It seems that this is just what I was dreaming about... ;-) I will try and apply this as soon as 2.4.26 final is out!
TNX 1.0E+06 :-)))
How was it done? Is the procedure executable by an automatic script?
Signature · 4 lines
| GnuPG Key ID = DD6DFCF4 | You're compiling a program
Francesco | Key fingerprint = | and, all of a sudden, boom!
Poli | C979 F34B 27CE 5CD8 DC12 | -- from APT HOWTO,
| 31B5 78F4 279B DD6D FCF4 | version 1.8.0_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq
Quoted 2 lines · Francesco Poli
How was it done?
With my fingers typing on my keyboard. Note that you have to learn dvorak typing though.
Quoted 1 line
Is the procedure executable by an automatic script?
Probably.
Signature · 4 lines
Bruno Ducrot -- Which is worse: ignorance or apathy? -- Don't know. Don't care.
Quoted 4 lines · On Mon, 5 Apr 2004 23:11:31 +0200 Dominik Brodowski wrote:
IMHO, such a script can only work well if it is "being watched", e.g. someone (you?) verifies the results equal what you expect. Also, differences in the base kernel need to be tracked which might make adjustments of the cpufreq files / patch necessary.
OK, it's quite clear I misunderstood the role of
http://ftp.linux.org.uk/pub/linux/cpufreq/ tar archives... :-( I thought that these archives were there to provide the latest cpufreq backport for 2.4.x kernels, and that one (following Bruno Ducrot's instructions) would be able to use them to modify a 2.4.x kernel (and be happy).
That is why I said that a kernel patch would be easier to apply.
If those archives are online for completely different purposes (which ones?), I apologize for all the noise.
Now the question is: I would like to run the latest 2.4.x kernel with cpufreq support, what do I do?
Quoted 9 lines
Quoted 6 lines
Such a script should be run daily (via cron) on ftp.linux.org.uk
This could perhaps work and provide up-to-date patches, as long as 2.4.x cpufreq at http://ftp.linux.org.uk/pub/linux/cpufreq/ is kept up-to-date with 2.6.x
But assuming that this CVS export is up-to-date [which it isn't], why is your script necessary?
It would generate a kernel patch whose application is straightforward... right?
Quoted 7 lines
Quoted 3 lines
Perhaps
patchin.shshould be modified in order to use cp (or mv) rather than ln -sModifying the
patchin.shscript to become more user friendly is something which can be done -- but _please_ be aware that 2.4. cpufreq is, at best, in"beta" state and not as up to date as 2.6.
Of course, cpufreq cannot be backported from 2.6.x to 2.4.x by an automatic script. That's a manual operation in any case. And one in which I cannot help much[1], I'm afraid...
Notes:
[1] IANAKH: I am not a kernel hacker :-(
Signature · 4 lines
| GnuPG Key ID = DD6DFCF4 | You're compiling a program
Francesco | Key fingerprint = | and, all of a sudden, boom!
Poli | C979 F34B 27CE 5CD8 DC12 | -- from APT HOWTO,
| 31B5 78F4 279B DD6D FCF4 | version 1.8.0_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq
Quoted 7 lines · Francesco Poli
Quoted 4 lines · On Mon, 5 Apr 2004 23:11:31 +0200 Dominik Brodowski wrote:
IMHO, such a script can only work well if it is "being watched", e.g. someone (you?) verifies the results equal what you expect. Also, differences in the base kernel need to be tracked which might make adjustments of the cpufreq files / patch necessary.
Quoted 7 lines
OK, it's quite clear I misunderstood the role of
http://ftp.linux.org.uk/pub/linux/cpufreq/ tar archives... :-( I thought that these archives were there to provide the latest cpufreq backport for 2.4.x kernels, and that one (following Bruno Ducrot's instructions) would be able to use them to modify a 2.4.x kernel (and be happy).
That is why I said that a kernel patch would be easier to apply.
You _are_ able to use them 2.4. kernels, but the 2.4. backport may contain errors and bugs which are fixed in 2.6. If I'd find out a way so that the day had 72 hours for me, I'd update the 2.4. backport...
Dominik
_______________________________________________
Cpufreq mailing list
[email protected]
http://www.linux.org.uk/mailman/listinfo/cpufreq