Kaiser Kernel Thread - ONLY FOR KERNEL AND BUGS

Krazy-Killa · Dec 22, 2010

I have no clue... can you RE-download from the git ? to a different dir ?
i've followed you're instructions as per PM to the letter, and downloaded from the SF.net git...
I have no clue.
i'll give you what i have:

Code:

The msm7k governor is based on the conservative governor
with some hacks specific to the msm7k.
Symbol: CPU_FREQ_DEFAULT_GOV_MSM7K [=y]
Prompt: msm7k
Defined at drivers/cpufreq/Kconfig:105
Depends on: <choice>
Location:
  -> CPU Frequency scaling
  -> CPU Frequency scaling (CPU_FREQ [=y])
  -> Default CPUFreq governor (<choice> [=y])
   Selects: CPU_FREQ_GOV_MSM7K && CPU_FREQ_GOV_PERFORMANCE

file:
kernel/drivers/cpufreq/cpufreq_msm7k.c

Hint:

Code:

/*  drivers/cpufreq/cpufreq_msm7k.c
 *
 *  based on:
 *  drivers/cpufreq/cpufreq_conservative.c
 *
 *  Copyright (C)  2001 Russell King
 *            (C)  2003 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>.
 *                      Jun Nakajima <jun.nakajima@intel.com>
 *            (C)  2004 Alexander Clouter <alex-kernel@digriz.org.uk>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

And... at the end:

Code:

MODULE_AUTHOR ("Alexander Clouter <alex-kernel@digriz.org.uk>");
MODULE_DESCRIPTION ("'cpufreq_conservative' - A dynamic cpufreq governor for "
                "Low Latency Frequency Transition capable processors "
                "optimised for use in a battery environment");
MODULE_LICENSE ("GPL");

Doesn't match... must be a hack of some sort... problem is, WHERE THE HELL DID I GET THIS ???

I have this too, though I'm still making heads or tails of all this, plus trying to figure out how to compile, luckally my laptop is already configured for compiling as I have to compile my wireless adapter drivers everytime I update my linux kernel.

n2rjt · Dec 29, 2010

I'm still quite happy with the kernel in post #63. It has never yet crashed for me. If I pull the battery or open the SIM card door, it recovers without loss.
If I shut down "cleanly", it does have some corruption, usually in Dalvik cache.
It is definitely more stable than anything I build myself from the git repo.
Perhaps I am using an older toolchain? I use the one that comes with the Fresh Froyo repo, in android/prebuilt/linux-x86/toolchain/arm-eabi-4.4.0/bin.

l1q1d · Dec 29, 2010

is there any change on the kernel to commit to the git?

Krazy-Killa · Dec 29, 2010

I've been trying to build my own kernel lately, and so far all I get are crashes during boot. Don't even get the press center to open menu part.

n2rjt · Dec 29, 2010

Krazy-Killa said:
I've been trying to build my own kernel lately, and so far all I get are crashes during boot. Don't even get the press center to open menu part.

Do you have latest bootenv and tinboot?

Sent from my Android on HTC Kaiser/Kaiser/Polaris using XDA App

Krazy-Killa · Dec 29, 2010

n2rjt said:
Do you have latest bootenv and tinboot?

Sent from my Android on HTC Kaiser/Kaiser/Polaris using XDA App

Um, as those don't ring a bell, it'll be safe to say that I don't.

I mean it compiles without a problem, just when I try to use the zImage for HaRET it crashes before the press center for menu.

Also out of curiousity, how do you build a kernel for NAND flash?

n2rjt · Dec 30, 2010

Oh you are using haret. Maybe the log file will give a clue. Or maybe you need an updated initrd.lzma from bootenv.

In same sourceforge place as the kernel you can find tinboot and bootenv. Bootenv has the initrd, tinboot builds an nbh file from the parts.

79bass · Jan 3, 2011

New Nand Driver

Hi,

It was a while I wasn't updating my Kaiser... I never had time, but I was waiting the next data corruption event to erasing everything and setting up the new initrd, and new juicy stuffs.

Ok, downloaded new kernel (29-11 and other from forum) and new atools.
First big problem, I can't set up the right panel anymore. Use to have panel 1, but now ot the setting are not applied anymore or something wrong with the kernel. nor panel 1, 2 or 3 are right, nor "Auto" option is working.
Then can't install system on NAND anymore, (device is not existing). So it is not a matter of unapplied update (I can't with no nand partitions), I guess it's a kernel problem.

I've browsed a little in the forum but haven't found a solution if not downgrade

kernel.
I'm not an expert on mobile device.. I can't have the device ID of my hardware (something like lspci doesn't exist?).

Thanks

mr_ppp · Jan 3, 2011

79bass said:
Hi,

It was a while I wasn't updating my Kaiser... I never had time, but I was waiting the next data corruption event to erasing everything and setting up the new initrd, and new juicy stuffs.

Ok, downloaded new kernel (29-11 and other from forum) and new atools.
First big problem, I can't set up the right panel anymore. Use to have panel 1, but now ot the setting are not applied anymore or something wrong with the kernel. nor panel 1, 2 or 3 are right, nor "Auto" option is working.
Then can't install system on NAND anymore, (device is not existing). So it is not a matter of unapplied update (I can't with no nand partitions), I guess it's a kernel problem.

I've browsed a little in the forum but haven't found a solution if not downgrade kernel.
I'm not an expert on mobile device.. I can't have the device ID of my hardware (something like lspci doesn't exist?).

Thanks

I've found myself if you have the simple battery option ticked it ignores panel type no matter what you do, it took me ages to realise this, changed it to advanced and entered my battery mah and now it works, might be the same problem? Never had a problem installing on NAND though!

79bass · Jan 3, 2011

Whaaa, that was quick...
The battery option fixed the problem.
I was using atools from git.
I used the simple setting because once upon a time was a fair way to adjust the battery but now this option should be fixed or removed... In fact it screw everything with new kernel in my device.

thanks for the tips

Best regards
Francesco

vinny13 · Jan 4, 2011

clemsyn said:
I think I'll upload the kernel for you guys to test out.. It's about the same as the kernel on page 12 except for CPU PM and I also decreased the nShortOpCaches to 4 to help decrease data corruption when turning off the device (I think ). I'll upload it soon.

Update:

Uploaded, Pls try it and let me know how it goes

Can someone convert this to a Vogue kernel for me? Atools won't work... Can't get edited kernels to properly install to the nand for unknown reasons...

dark_prince · Jan 5, 2011

vinny13 said:
Can someone convert this to a Vogue kernel for me? Atools won't work... Can't get edited kernels to properly install to the nand for unknown reasons...

I cant do the conversion in Hex Editor since those two are totally different (POLARIS and KAISER's worked for me by just changing the POLA100 with KAIS**** in the 3rd line). I'm really sorry of not being any help :/ (I feel it since I'm looking for the right support and trying to figure out stuff all on myself)

Atools isnt working for me too (python installer is deep-sh** with AMD's processors), and no one out there is really into making nbh's for us.

Atools is a multiplatform tool so it works on ALL systems!

If you are a stupid windows user you can't use it so don't ask!

For all clever user (Linux/Mac/Win)

This is punch line for me. Ofcourse I fall in the stupid category coz I cant run Atools good as you "CLEVER: people do.

The only hex edited version I was able to do on myself was a random shot to DZO's repository with some hex editing and Fat Free Froyo (the only wifi working combo). Rest all tests with about dozen kernels and half dozen distros out of eclairs and froyos FAILED. (Wifi failed to start).

My Device is KAIS130 HTC TyTn II, Panel Type 1. Now I will be trying Donuts soon when I'm piled up of pathetic WM bugs.

kthxbai

peter s · Jan 8, 2011

recommended toolchain

@l1qu1d:
could you please share the git for your htc-vogue 2.6.25 modifications (source) and the toolchain you used for your releases? (forget the latter if it was arm-2008q1-126-arm-none-linux-gnueabi-i686-pc-linux-gnu)

@all
am i the only one with 2.6.25 and sd-card mounted twice (once on /sdcard by /init (initrd) and once on /mnt/sdcard by /system/sysinit.rc (or vold respectively)?

thanks!
peter

Krazy-Killa · Jan 8, 2011

The /sdcard is a symlink to /mnt/sdcard. You can not mount a single device more than once. But you can create "shortcuts" to the mounting directory.

Though I wouldn't mind giving the 2.6.25 kernel a looking at, as I really like the stability.

Sent from my AT&T Tilt using XDA App

peter s · Jan 8, 2011

Krazy-Killa said:
The /sdcard is a symlink to /mnt/sdcard. You can not mount a single device more than once. But you can create "shortcuts" to the mounting directory

right but wrong

/sdcard is created on boot by /init from initrd. /system/sysinit.rc creates /mnt and /mnt/sdcard and attempts to ln -s /mnt/sdcard /sdcard - but fails, since it is already there and in use as a mount point for /dev/mmcblk0p1
the reason why dual mount is possible: there is a second logical device (/dev/block/vold:bla) using the same physical device...

i would like to fix the init script in initrd, hence my questions above.

Krazy-Killa · Jan 8, 2011

I have no problems with accessing the sdcard however, but I see what you mean.

My phiosophy on stuff like this is... If it ain't broken don't fix it.

What are you trying to do? I mean the sdcard mounting works fine.

Technically you can just modify the sysinit.rc file of the build you're using to do a symlink instead of mounting sincethe kernel already did it, so this is not a kernel related problem but a specific build problem.

Sent from my AT&T Tilt using XDA App

peter s · Jan 8, 2011

Krazy-Killa said:
Technically you can just modify the sysinit.rc file of the build you're using to do a symlink instead of mounting sincethe kernel already did it,

not quite, since the vold mount from /system/sysinit.rc is hotplug-compatible while the mount from the initrd init script is not.
plus, /sdcard is only there for backward compatibility - /mnt/sdcard seems to be the preferred mountpoint. there is even some attempted setprop-magic in the sysinit.rc to let vold know the card is mounted already. i assume this fails because of the different mountpoints (or possibly due to the different device names). i would like to experiment with this and other init related stuff.

Krazy-Killa said:
so this is not a kernel related problem but a specific build problem.

precisely it is neither the one, nor the other. to me it seems this is a missing (small) update/fix in initrd or its init script respectively. therefore it should be relevant for all android builds at least from froyo onward.

apart from all this, i consider it bad to have one physical medium mounted twice.

aceoyame · Jan 9, 2011

I've been gone for awhile. I was wondering if there were any kernels I should be looking out for in here or if clemsyn's changes were finally commited to git.

Krazy-Killa · Jan 9, 2011

Nope, no new kernels since your last post. Not sure about it being committed to git though.

Sent from my AT&T Tilt using XDA App

l1q1d · Jan 10, 2011

i didn't use a complete toolchain but only the android ndk (from android developer google site).
For the 2.6.25 sources look in the git (my sig)

Kaiser Kernel Thread - ONLY FOR KERNEL AND BUGS

Senior Member

Senior Member

Retired Recognized Developer

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Attachments

Senior Member

Member

Senior Member

Member

Senior Member

Member

Retired Recognized Developer

Senior Member

Retired Recognized Developer

Similar threads

Top Liked Posts