On Tue, 2011-03-01 at 23:17 -0800, John Stultz wrote:

On Wed, 2011-03-02 at 14:23 +0800, Jeremy Chang wrote:

...

Hi, John: I have one question about the origin/dev/linaro.android branch right now for beagleboard xM. After tried, I still can not see the terminal log after the "Virtual kernel memory layout" message with EARLY_PRINTK enabled., even printk messages in omap3_beagle_init() can not be seen. I see With 2.6.36/37+: the console should be changed to 'ttyO2,115200n8' in u-boot. but the problem still exists for me. Could you give some clue for me how to solve this?

Huh. I'm not sure if I've played with early printk on beagle.

If not, mind sending me the .config you're using and I'll take a look?

Oops, sorry, that second line was confusing. I was initially going to ask if you were using the android_omap3_defconfig, but clearly that's not the case as EARLY_PRINTK isn't enabled there.

So do please send your config.

thanks -john

Hi John: Actually I didn't mean to enable EARLY_PRINTK in .config. Actually I only used the android_omap3_defconfig, and I found I can not see the terminal messages so I tried to enable EARLY_PRINTK to see if there're any suspicious.

The last line I can see is: Uncompressing Linux... done, booting the kernel.

Any clue on this for me?

Regards, -Jeremy

On Wed, Mar 2, 2011 at 3:30 PM, John Stultz john.stultz@linaro.org wrote:

On Tue, 2011-03-01 at 23:17 -0800, John Stultz wrote:

...

On Wed, 2011-03-02 at 14:23 +0800, Jeremy Chang wrote:

...

Hi, John:     I have one question about the origin/dev/linaro.android branch right now for beagleboard xM. After tried,  I still can not see the terminal log after the "Virtual kernel memory layout" message with EARLY_PRINTK enabled., even printk messages in omap3_beagle_init() can not be seen.     I see With 2.6.36/37+: the console should be changed to 'ttyO2,115200n8' in u-boot.  but the problem still exists for me.     Could you give some clue for me how to solve this?

Huh.  I'm not sure if I've played with early printk on beagle.

If not, mind sending me the .config you're using and I'll take a look?

Oops, sorry, that second line was confusing. I was initially going to ask if you were using the android_omap3_defconfig, but clearly that's not the case as EARLY_PRINTK isn't enabled there.

So do please send your config.

thanks -john

Hi, John: Thanks. Though the problem for me is still the same.

I tried in uboot with following commands. # mmc init # setenv bootargs 'console=ttyO2,115200n8 mem=256M root=/dev/mmcblk0p2 rw rootfstype=ext3 rootwait init=/init' # fatload mmc 1 0x82000000 uImage # bootm 0x82000000

Any idea about if there could be any other difference in between? Thanks.

Regards, -Jeremy

On Thu, Mar 3, 2011 at 7:30 AM, John Stultz john.stultz@linaro.org wrote:

On Thu, 2011-03-03 at 02:47 +0800, Jeremy Chang wrote:

...

Hi John:     Actually I didn't mean to enable EARLY_PRINTK in .config. Actually I only used the android_omap3_defconfig, and I found I can not see the terminal messages so I tried to enable EARLY_PRINTK to see if there're any suspicious.

The last line I can see is:     Uncompressing Linux... done, booting the kernel.

Huh. I'm not seeing that at all with the android_omap3_defconfig

Are your bootargs something like: 'console=ttyO2,115200n8 mem=256M root=/dev/mmcblk0p2 rw rootfstype=ext3 rootwait init=/init' ? thanks -john

On Thu, Mar 3, 2011 at 3:43 PM, John Stultz john.stultz@linaro.org wrote:

On Thu, 2011-03-03 at 15:27 +0800, Jeremy Chang wrote:

...

Hi, John:     Thanks.  Though the problem for me is still the same.

I tried in uboot with following commands.     # mmc init     # setenv bootargs 'console=ttyO2,115200n8 mem=256M root=/dev/mmcblk0p2 rw rootfstype=ext3 rootwait init=/init'     # fatload mmc 1 0x82000000 uImage     # bootm 0x82000000

Any idea about if there could be any other difference in between? Thanks.

I'm baffled. Do you have any further details on the beagleboard xm revision? I think mine is an A2.

From the box and the sticker on the board, it says Rev B. So it looks like the u-boot I used from rowboat[1] shows RevA as below messages,

OMAP34xx/35xx-GP ES2.0, CPU-OPP2 L3-165MHz OMAP3 Beagle board + LPDDR/NAND I2C: ready DRAM: 256 MiB NAND: 256 MiB In: serial Out: serial Err: serial Beagle xM Rev A

This is the suspicious difference.

Have you been able to boot the same kernel source using a different .config?

not yet. same for omap2plus_defconfig

Have you been able to boot other kernels ? Do you have the .config for those (along with the kernel revision)?

I can boot with the 2.6.32 kernel and the u-boot.bin, MLO from the rowboat[1], with console in u-boot set to ''ttyS2,115200n8"

Regards, -Jeremy

[1] http://code.google.com/p/rowboat/

thanks -john

On Thu, Mar 3, 2011 at 8:02 PM, Alexander Sack asac@linaro.org wrote:

On Thu, Mar 3, 2011 at 9:14 AM, Jeremy Chang jeremy.chang@linaro.org wrote:

...

On Thu, Mar 3, 2011 at 3:43 PM, John Stultz john.stultz@linaro.org wrote:

...

On Thu, 2011-03-03 at 15:27 +0800, Jeremy Chang wrote:

...

Hi, John:     Thanks.  Though the problem for me is still the same.

I tried in uboot with following commands.     # mmc init     # setenv bootargs 'console=ttyO2,115200n8 mem=256M root=/dev/mmcblk0p2 rw rootfstype=ext3 rootwait init=/init'     # fatload mmc 1 0x82000000 uImage     # bootm 0x82000000

Any idea about if there could be any other difference in between? Thanks.

I'm baffled. Do you have any further details on the beagleboard xm revision? I think mine is an A2.

From the box and the sticker on the board, it says Rev B.    So it looks like the u-boot I used from rowboat[1] shows RevA as below messages,

I tarred up the boot/vfat partition that i successfully used for booting our android beagle build:

http://people.linaro.org/~asac/tmp/beagle-boot.tar.gz

its  1. linaro u-boot with a good boot.scr that boots android from mmcblk0p2  2. uImage built from johns tree using android omap3 defconfig ...

see if that changes anything (i see output on terminal for whole boot process)

Hi, Alexander:

Thanks for the help.

It turns out that just after I changed to using the Android toolchain from 4.4.3 to 4.4.0 , this problem is solved. So this may be the toolchain and the kernel source code compatible issue.

Regards, -Jeremy

--

- Alexander

On 03/02/2011 06:47 PM, Somebody in the thread at some point said:

Hi Jeremy -

 Actually I didn't mean to enable EARLY_PRINTK in .config. Actually

I only used the android_omap3_defconfig, and I found I can not see the terminal messages so I tried to enable EARLY_PRINTK to see if there're any suspicious.

 The last line I can see is:
 Uncompressing Linux... done, booting the kernel.

 Any clue on this for me?

To debug this you do want EARLY_PRINTK enabled in your .config, and add this kernel commandline:

earlyprintk=1 earlycon=ttyO2,115200n81 loglevel=8

Without EARLY_PRINTK, nothing comes on the console until the console device is initialized, until then all the log info is buffered in memory and it's all dumped out at that point. So, if you blow an OOPS early in boot, before the console stuff is initialized, you won't hear anything about it.

With EARLY_PRINTK, every log message is printed to serial synchronously, so you can hear about your problem if the problem produced any log footprint, or at least get a clue about how far through the init process you got before the black smoke comes ^^

-Andy