Debian And Uboot

[svs57]

I already tried with the kernel marked with sdboot (that is the one build as you suggested), and the result was unsuccesfull
[div align=\"right\"][a href=\"index.php?act=findpost&pid=161524\"][{POST_SNAPBACK}][/a][/div]

I can't test it. I have only SL750. I changed only type of proc. in kernel config.
It's not difficult to crosscompile kernel.
You need PC with debian, cross (http://www.freaknet.org/martin/crosstool/packages/) and kernel sources from www.kernel.org.
On latest kernel (2.6.22-rc1) I didn't apply any patches.

# ARCH=arm CROSS_COMPILE=armv4t-crosstool-linux-gnueabi- make menuconfig
# ARCH=arm CROSS_COMPILE=armv4t-crosstool-linux-gnueabi- make
# cd arch/arm/boot/
# mkimage -A arm -O linux -T kernel -C none -a 0xa0008000 -e 0xa0008000 -n "Linux 2.6" -d zImage kernel.img

Put kernel.img  and modules on SD and flash using pdaXrom emergency system.

[pkwong]

can u post your .config?

[svs57]

I used it for 2.6.22-rc1

[pkwong]

do u have any experience in compiling 2.6.22-rc2?
i tried few times for my c1000
but they either doesn't boot up at all or can't mount the sd
mmcblk0p1 is not a valid block device...or something like that...
even if it boots, it takes a LONG time
i wonder if it's my cross-compiler's problem or my config...

the cross-compiler is built on gentoo on PC, using armv5te-iwmmxt-linux-gnueabi
do i need to build the kernel without eabi?

also, are there any size limits for the kernel?

i'll post the config here in case anyone is interested

[Capn_Fish]

I don't think EABI works, but I'm not that knowledgeable on the subject, so I may be wrong.

[svs57]

do u have any experience in compiling 2.6.22-rc2?
i tried few times for my c1000
but they either doesn't boot up at all or can't mount the sd
mmcblk0p1 is not a valid block device...or something like that...
even if it boots, it takes a LONG time
i wonder if it's my cross-compiler's problem or my config...

the cross-compiler is built on gentoo on PC, using armv5te-iwmmxt-linux-gnueabi
do i need to build the kernel without eabi?

also, are there any size limits for the kernel?

i'll post the config here in case anyone is interested
[div align=\"right\"][a href=\"index.php?act=findpost&pid=161943\"][{POST_SNAPBACK}][/a][/div]

2.6.22-rc2 works on my SD 4gb.
Yesterday I tried gentoo with this kernel on 512mb SD and got the same problem as you.
I install 2.6.20.2 kernel and 512 SD work.
I guess it's 2.6.22-rc kernel bug.
It strange 4G SD works with 2.6.22.rc  but 512mb doesn't.

[pkwong]

i was having this problem on gentoo also
it's fine when i switch back to debian :-)
however, i'm still having problem with suspend...do you?
i cannot resume after apm -s..
i followed your config for power mangement but still no luck

[svs57]

i was having this problem on gentoo also
it's fine when i switch back to debian :-)
however, i'm still having problem with suspend...do you?
i cannot resume after apm -s..
i followed your config for power mangement but still no luck
[div align=\"right\"][a href=\"index.php?act=findpost&pid=162018\"][{POST_SNAPBACK}][/a][/div]

With kernel >= 2.6.21  power button doesn't work as suspend.
I do suspend from console:
apm -s
But  power button works as resume.

[pkwong]

are there any ways to change the resume key?
i tried the Home key and it didn't work
and my power button seems to have some problem too...

[svs57]

are there any ways to change the resume key?
i tried the Home key and it didn't work
and my power button seems to have some problem too...
[div align=\"right\"][{POST_SNAPBACK}][/a][/div]

Try this patch
[a href=\"http://www.rpsys.net/openzaurus/patches/input_power-r7.patch]http://www.rpsys.net/openzaurus/patches/input_power-r7.patch[/url]

[pkwong]

Thanks!!
but the file it patches (drivers/input/power.c) seems not to exist in 2.6.22

a check in the kernel 2.6.22-rc1 changelog gives

    Input: remove no longer used power.c handler
   
    Delete the never-compiled source file drivers/input/power.c, and
    remove its entry from the corresponding Makefile, as there is no
    Kconfig file that refers to the config option INPUT_POWER
   
so i patched the file in 2.6.21, put it to 2.6.22, manually patched Kconfig and Makefile to make it compile...hope this is not too stupid
however, still didn't work..

[svs57]

Thanks!!
but the file it patches (drivers/input/power.c) seems not to exist in 2.6.22

a check in the kernel 2.6.22-rc1 changelog gives

    Input: remove no longer used power.c handler
   
    Delete the never-compiled source file drivers/input/power.c, and
    remove its entry from the corresponding Makefile, as there is no
    Kconfig file that refers to the config option INPUT_POWER
   
so i patched the file in 2.6.21, put it to 2.6.22, manually patched Kconfig and Makefile to make it compile...hope this is not too stupid
however, still didn't work..
[div align=\"right\"][a href=\"index.php?act=findpost&pid=162327\"][{POST_SNAPBACK}][/a][/div]

I also did similar. Then i back to 2.6.21. With this patch suspend button works!

[pkwong]

great! now everything works...rtc, suspend/resume...
did u enable the dynamic tick timer?
does it really save power?

[adf]

great! now everything works...rtc, suspend/resume...
did u enable the dynamic tick timer?
does it really save power?
[div align=\"right\"][a href=\"index.php?act=findpost&pid=162357\"][{POST_SNAPBACK}][/a][/div]

Very impressive!
How's the performance?
Any chance we'll see a how-to and installation files?

[svs57]

great! now everything works...rtc, suspend/resume...
did u enable the dynamic tick timer?
does it really save power?
[div align=\"right\"][{POST_SNAPBACK}][/a][/div]

Yes.
echo 1 > /sys/devices/system/timer/timer0/dyn_tick
Dynamic tick timer really save power when cpu is idle.
I did some hacked change in kernel to enable CONFIG_TIMER_STATS and install Intel powertop to monitor what program "eat" battery. [a href=\"http://www.linuxpowertop.org/]http://www.linuxpowertop.org/[/url]
With dynamic tick timer enable it 5 times less (on cpu idle)

Also I set ondemand governor cpufreq to save power.
# cpufreq-info
cpufrequtils 002: cpufreq-info © Dominik Brodowski 2004-2006
Report errors and bugs to linux@brodo.de, please.
analyzing CPU 0:
  driver: pxa25x
  CPUs which need to switch frequency at the same time: 0
  hardware limits: 99.5 MHz - 398 MHz
  available frequency steps: 99.5 MHz, 199 MHz, 299 MHz, 398 MHz
  available cpufreq governors: userspace, powersave, ondemand, conservative, performance
  current policy: frequency should be within 99.5 MHz and 398 MHz.
                  The governor "ondemand" may decide which speed to use
                  within this range.
  current CPU frequency is 99.5 MHz (asserted by call to hardware).