Re: [Announce] kernel-power v48
 

At second thought, I think that whole i2c access thing reminds me "issue of software quality" (people reading texrat blog will know what I mean). Even *if* by some unknown reason, BMI got a factor that prevent damage, we should not follow "wrong" path of coding things, i.e we should not accept wrong practices.

Also, we are not sure if someone got problems, but, we are also not sure that one day, someone won't get real software/hardware damage by that, and we shouldn't risk it.

So, thanks for pushing this further and explaining (both irc log and vi_ post)

Re: [Announce] kernel-power v48
 

in extras-devel yet?

Re: [Announce] kernel-power v48
 

No, you can take a look at this page, no need to ask in the future:
http://maemo.org/packages/repository...ree_armel/all/.

Re: [Announce] kernel-power v48
 

It would have been mentioned, and as well you would have received update notification in HAM/FAP (if you check the repositories at least once a week) and via the packages search as shown above.So, please install manually using the linked packages from the 1st post, and report wrong battery values afterwards.

Re: [Announce] kernel-power v48
 

@pali - I am unable to build kp48 from your git repository under scratchbox.

This are the steps i follow:

Please advice if I am missing something and what are the correct steps.

Thanks.

Re: [Announce] kernel-power v48
 

You need to download source code of linux kernel 2.6.28 and unpack it into same directory

Re: [Announce] kernel-power v48
 

I've tried that too, same result.

And which is the "same directory"?

Directory you run git clone... in?
kernel-power?
kernel-power/kernel-power-2.6.28?

And as I understood from your previous posts that you are using madde for your development - did you ever try to compile KP in scratchbox?

Re: [Announce] kernel-power v48
 

download linux kernel 2.6.28 tarball. you can use apt-get source kernel-power or kernel.org website. Then you will have tarball - unpack it - it will unpack into subfolder. Then move all files from this subfolder to kernel-power/kernel-power-2.6.28

I did not used scratchbox - but it must work. It is same as madde - it use dpkg-buildpackage which is same on all debian system.

Re: [Announce] kernel-power v48
 

...and the result:

Applying patch phys_to_page.diff
Applying patch ext4-data-corruption.diff
2 out of 2 hunks FAILED
1 out of 1 hunk FAILED
2 out of 2 hunks FAILED
6 out of 6 hunks FAILED
5 out of 5 hunks FAILED
9 out of 9 hunks FAILED
1 out of 1 hunk FAILED
6 out of 6 hunks FAILED
2 out of 2 hunks FAILED
8 out of 8 hunks FAILED
3 out of 3 hunks FAILED
1 out of 1 hunk FAILED
Patch ext4-data-corruption.diff does not apply (enforce with -f)
make: *** [source-copy-stamp] Error 1

Re: [Announce] kernel-power v48
 

You have bad tarball. use this:
ftp://ftp.kernel.org/pub/linux/kerne...-2.6.28.tar.gz

Re: [Announce] kernel-power v48
 

It is from maemo repository, why you say it is bad. I think you (or author) should re-write this patch so it could be applied against kernel in maemo repos, otherwise you won't be able to upload KP48 in extras-devel.

Re: [Announce] kernel-power v48
 

ok , so if using latest release of v48 (28-Jul-2011 18:46) what do we use to display temperature ? and can it be set to load on boot ?
wouldnt know how to stop bme to start bq but dont consider myself out of my depth with oc/power kernel etc :)

Re: [Announce] kernel-power v48
 

The funniest thing, is that kp47 is also based on non-maemo kernel. Maybe this is the case, why it can't be promoted to testing. Someone suggested it before - to re-write for maemo kernel...

---

As for battery temperature - AFAIK nothing, if You want to keep bme running. I hope I'm wrong, and someone will correct me (only thing I can think about - i2ctools?).

Re: [Announce] kernel-power v48
 

I do not understand why I could use maemo kernel tarball for kernel-power source dsc package... Do you know something about Debian packages?? I think this is stupid reason.

and for temp: the best will be to write library which can read battery properties from bq27x00_battery module via sysfs. this library will be loaded via LD_PRELOAD, so BME will be reading values from bq module instead directly i2c bus.

if somebody can help me with writing library we will have earlier working bq and bme together. (library needs to rewrite function ioctl)

Re: [Announce] kernel-power v48
 

Since updating from kp47 to 48, i get reproducible reboots. Happens when switching to bleeding edge wl1251 drivers using the respective status applet (status-area-wlan-driver-selector-applet; 0.2.2-3). When they are loaded and i have to select the WiFi AP.

Full syslog available on request if any1 interested, too long to post and analyzing is limited as i'm at the airport ATM. Only GLIB CRITICAL attached for now.

Is this happening for someone too?

Re: [Announce] kernel-power v48
 

if did not recompile wl1251 bleeding edge driver agains kernel-power v48 you could have problem...

Re: [Announce] kernel-power v48
 

Pali, what stupid reason could be the fact that patches do not apply against kernel source which was tested by nokia (stock kernel) and afterwards by community (KP up to version 46). And you have already been told that not using it prevents KP from corectly entering extras-devel.

And I don't get it why you insist on using kernel source tarball outside maemo repos. Will you please give some reasoning besides 'that is the way I like it, take it or leave it'

Re: [Announce] kernel-power v48
 

Thanks. I did not, cause the general opinion was that is wasn't necessary for 47, and no one mentioned the necessity for 48 also. What is the relevant change in 48 that we should recompile this time?

Re: [Announce] kernel-power v48
 

Maybe it will work or maybe not. Always is recomment to compile modules agains current version and not stripping version symbols (for ability to load it in other version).

Or maybe this is not problem with wl1251 drivers...

Re: [Announce] kernel-power v48
 

In Debian packaing manual is written to use orig tarball from upstream and all source changes should be in debian/patches/ (which will be applied when compiling/unpacking) - so this is reason.

I generated patch(es) of all changes in nokia stock kernel and it is in debian/patches/ (some of them was done by titan).

So kernel-power is nokia stock kernel (2.8.28 tarball + nokia patches) + our patches.

Re: [Announce] kernel-power v48
 

Maemo packaging is not exactly same as Debian packaging (in details). If You're already using Nokia stock Kernel as mainstream kernel+ patches to make it Nokia one + kp patches, why can't You just use Nokia stock kernel + kp patches, and finally, have correctly working repo maintainership? It's only ideological, or is there something meritocratic behind it?

Re: [Announce] kernel-power v48
 

Tigerite, can you point me to the appropriate literature that may explain what it is I must do to make compcache work on the N900 using the v48 power-kernel?

Re: [Announce] kernel-power v48
 

Here is bzr branch which contains initial code of library for BME:
https://code.launchpad.net/~pali/+junk/maemo_libbqioctl

Re: [Announce] kernel-power v48
 

The only further information i got out of it is; it's happening when connecting to the AP. Just loading the .ko via included load.sh seems ok. Hmm...

Re: [Announce] kernel-power v48
 

Why you don't count fact that patches don't apply against Nokia's .orig tarball as a problem?

Did you see this:
http://talk.maemo.org/showpost.php?p...postcount=1129

EDIT: Yes, I saw irc log.

Re: [Announce] kernel-power v48
 

Now I merged my git repository into garage kernel-power project. Now I'm on maintainer list of all kernel-power packages. And X-Fade find problem why I cannot promoted kernel-power to extras-testing.

I will build new deb package of prerelease v48 (now from garage git repo). It will be published again on my server. When all will be OK, I push v48 into extras-devel.

It may be problem that v48 still not be visible on maemo.org package interface, but I'm sure that v49 will be visible.

Re: [Announce] kernel-power v48
 

Great news Pali, and props to X-Fade for finding the issue.

Anyone seen tigrette around? I'd still like to try the ramzswap he was tinkering with.

Re: [Announce] kernel-power v48
 

I would literally stab a baby for some compcache right now.

Re: [Announce] kernel-power v48
 

Great news Pali and many thanks for your efforts!



Shows you are not a father or your kids are too old already :(


*looks at his newborn daughter next to him*

Re: [Announce] kernel-power v48
 

I didn't literally mean literally literally, I meant literally like figuratively literally.

Re: [Announce] kernel-power v48
 

I uploaded new build to my server. Now I'm going to test it on my N900.

Re: [Announce] kernel-power v48
 

@Pali: TBH I wonder how you are able to build kp48 at all, having in mind that patches up to (and including) phys_to_page.diff are against directory kernel-power-2.6.28 while those from ext4-data-corruption.diff to the end of series are against linux-2.6.28. And different directory name is the reason for wrong checksum.

Otherwise looks like pr13 , kp46 and kp47 are using one and the same source in .orig file. Just root directory is with different name(for kp47).

So I would recommend you to rename kp48 .orig directory to kernel-power-2.6.28 and to modify patches from ext4-data-corruption.diff to the end of series to be against kernel-power-2.6.28. That way I think everything should be ok.

Re: [Announce] kernel-power v48
 

Today I tested building and inspecting kernel-power and kernel sources/tarballs. Finaly I undertood where was problem. I forget one thing, that SDK kernel package does not respect debian packaing policy (or use old...). But this was my mistake do not check all.

Shortly:
We have only 2 kernel tarballs. One is from SDK kernel package and one orig 2.6.28 from kernel.org. Both are of course different, have different md5sums, but it is same. Really same!

Unpacked SDK tarball, unpacked 2.6.28 tarball and diff has no output. So both sources are same, but packed tarball different.

Problem:
SDK kernel dsc (debian source) package using diff file which should contains only debian directory. But it contains included hard diff which patch also whole kernel.
And I always used apt-get source (which download & unpack & apply debian diff file) and then I checked directory. I did not know that kernel SDK package also containts hard diff, so where was problem.

Now I know this, I can use tarball from SDK with correct md5sum and belive that package interface accept v48 :-)

---

@freemangordon:
Prefixes in diff files in debian/patches are ignored. Quilt use patch -p1 so it is not problem if patch start with incorrect directory.

If you want to test building, now use only garage git repo (updated info in first post too).

I buided new deb packages from clean clone from garage repo.

---

In last today build I did not found any problem. I think it is stable. If nobody found problem I will push this version into extras-devel (and promote into extras-testing) is 2 days.

Re: [Announce] kernel-power v48
 

@ pali

sounds awesome will be waiting to kernel-power48...please post a changelog once its in the devel repos.!!

Re: [Announce] kernel-power v48
 

changelog is on first page!

Re: [Announce] kernel-power v48
 

sorry my bad pali....was too eager for this version...thanks..seems really great work done by all involved in this...

Re: [Announce] kernel-power v48
 

@pali
I've been testing the new .debs since 22:00 GMT last night. Booted, charged, called, watched vids, browsed, used modest, csbwin, clock and a bunch of other apps.

No problems to report as of 14:00 GMT today.

Re: [Announce] kernel-power v48
 

when will it be available for normal users? can it play 720p videos and mkv files?

Re: [Announce] kernel-power v48
 

pali loads n loads of thanks for your great work ,dude .hope soon it comes to extras .bravo !

Re: [Announce] kernel-power v48
 

pali's server is down, must be very busy!!