Re: [android-building] Do any current branches currently build / boot for pandaboard

2,443 views
Skip to first unread message

Jean-Baptiste Queru

unread,
Oct 10, 2012, 12:32:28 PM10/10/12
to android-...@googlegroups.com
Only the master branch. There's been an issue with graphics since the
most recent drop of the kernel/libraries, and I've just merged a
change that should fix that (literally minutes ago). I haven't had
time to try it but there's a reasonable chance that it'll work.

JBQ

On Wed, Oct 10, 2012 at 7:52 AM, Weston Weems <wwe...@gmail.com> wrote:
> Do any branches or head currently build and boot with appropriate graphics
> drivers?
>
> It looks like the the nature of how the sgx driver was loaded may have
> changed, and in the several ics/jb builds I've attempted in the past couple
> days resulted in lots of different problems. (kernel modules not loading,
> pvrinit vs pvrsrvctl etc)
>
> Weston
>
> --
> You received this message because you are subscribed to the "Android
> Building" mailing list.
> To post to this group, send email to android-...@googlegroups.com
> To unsubscribe from this group, send email to
> android-buildi...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/android-building?hl=en



--
Jean-Baptiste M. "JBQ" Queru
Technical Lead, Android Open Source Project, Google.

Questions sent directly to me that have no reason for being private
will likely get ignored or forwarded to a public forum with no further
warning.

Jean-Baptiste Queru

unread,
Oct 10, 2012, 12:52:01 PM10/10/12
to android-...@googlegroups.com
Definitely stick with master. PandaBoard doesn't work with any other branches.

On the kernel front, omap-android-panda-3.0, which should be commit
4a094a55476cad51c85928e7c5766d71944e87ea

JBQ

On Wed, Oct 10, 2012 at 9:41 AM, Weston Weems <wwe...@gmail.com> wrote:
> Just to clarify (so I can test)
>
> I just need to make sure to repo sync master, and then make sure the kernels
> I am building are master from kernel/omap? or should I still be grabbing
> omap-android-panda-3.0?
>
> Weston

Adrien Grassein

unread,
Oct 11, 2012, 3:16:14 AM10/11/12
to android-...@googlegroups.com
Hi,

The pandaboard still not boot with master of device/ti/panda (kernel problem on I2c commit 6c8e76e158a9bc849ec946ab030595e06ad89210).

After revert this commit, there is a kernel panic.
Is the kernel booting on your board or not? 

Thanks,
Adrien

2012/10/10 Weston Weems <wwe...@gmail.com>
Definitely fixed the graphics driver problem... I'm slapping together my changes in the kernel now...

Jean-Baptiste Queru

unread,
Oct 11, 2012, 8:50:41 AM10/11/12
to android-...@googlegroups.com
I have evidence that the latest kernel boots on my boards, but I'm
hitting other issues (no display, no adb).

JBQ

Adrien Grassein

unread,
Oct 11, 2012, 9:04:20 AM10/11/12
to android-...@googlegroups.com
Do you have a Pandaboard or a Pandaboard ES (OMAP 4430 or OMAP 4460)?

2012/10/11 Jean-Baptiste Queru <j...@android.com>

Jean-Baptiste Queru

unread,
Oct 11, 2012, 9:25:03 AM10/11/12
to android-...@googlegroups.com
One of each. I'm seeing the same result on both.

JBQ

On Thu, Oct 11, 2012 at 6:04 AM, Adrien Grassein

Adrien Grassein

unread,
Oct 11, 2012, 9:32:40 AM10/11/12
to android-...@googlegroups.com
So it's very strange.

I checkouted the latest Android version (branch master on .repo/manifest commit f46b8df32efc4979bdca52bf52e83fd816819f60)
On device/ti/panda I am on 5f567f13b76dc6b2e1bfe3d279fbac0427704a09.

It's allright or not?

Thanks,
Adrien

2012/10/11 Jean-Baptiste Queru <j...@android.com>

Adrien Grassein

unread,
Oct 18, 2012, 3:16:13 AM10/18/12
to android-...@googlegroups.com
Could you copy/paste your manifest please. I'am on the top of Android master and my pandaboard still not boot :(


Thanks,
Adrien

2012/10/17 Dan Murphy <murp...@gmail.com>
I have booted the lastest master on a Panda ES.

It is running right now.

Dan


On Thursday, October 11, 2012 11:27:46 AM UTC-5, G. Blake Meike wrote:
Apropos: I have an ES.  Is there any plan to support it, ever?  No biggie, just curious.

thanks
  Blake Meike

Adrien Grassein

unread,
Oct 22, 2012, 4:11:59 AM10/22/12
to android-...@googlegroups.com
My pandaboard boots now (with graphics). This is my flashing script :

device/ti/panda/usbboot device/ti/panda/bootloader.bin

./out/host/linux-x86/bin/fastboot oem format
./out/host/linux-x86/bin/fastboot flash xloader device/ti/panda/xloader.bin
./out/host/linux-x86/bin/fastboot flash bootloader device/ti/panda/bootloader.bin
./out/host/linux-x86/bin/fastboot erase cache
./out/host/linux-x86/bin/fastboot -p panda flash userdata
./out/host/linux-x86/bin/fastboot -p panda flash cache
./out/host/linux-x86/bin/fastboot -p panda flashall


Thanks,
Adrien


2012/10/19 Alex <ame...@online.de>
I think you need to flash the cache partition.

Am Freitag, 19. Oktober 2012 19:26:03 UTC+2 schrieb Brad Arnold:
I'm not even that far. Mine won't boot at all.

My output is (sorry for the long body, for some reason google groups won't take this as an attachment):

     ...

[    5.051513] EXT4-fs (mmcblk0p6): VFS: Can't find ext4 filesystem
...
[    5.108612] fs_mgr: Cannot mount filesystem on /dev/block/platform/omap/omap_hsmmc.0/by-name/cache at /cache
[    5.119842] init: fs_mgr_mount_all returned an error 
...
[    5.397674] Kernel panic - not syncing: Attempted to kill init!

Brad Arnold

unread,
Oct 23, 2012, 7:24:58 PM10/23/12
to android-...@googlegroups.com
I don't know the specific commit that would have broke it, but these caught my eye:

commit 551ca77e35098272abfe9484457827c661d23ef0
Merge: 2266e49 328a484
Author: Elliott Hughes <e...@google.com>
Date:   Fri Oct 19 13:37:24 2012 -0700

    Merge "Do all the zoneinfo.* file generation in Java."

commit 328a48458880ed12f2e7d1632b98c5df8d211be7
Author: Elliott Hughes <e...@google.com>
Date:   Fri Oct 19 13:03:52 2012 -0700

    Do all the zoneinfo.* file generation in Java.
    
    Bug: 7012465
    Change-Id: Ia5d5ad75db94d680a6b34ccc26eb68d5bd2bfc29

commit 2266e4956f4b245903acd66f0cc32ec5caf5abae
Merge: 1022507 eb06129
Author: Elliott Hughes <e...@google.com>
Date:   Fri Oct 19 13:01:04 2012 -0700

    Merge "Switch bionic over to using libcore's copy of libcore.util.ZoneInfo."

commit eb06129c5c569897a897f9ec9ee300560ec3b436
Author: Elliott Hughes <e...@google.com>
Date:   Fri Oct 19 12:05:24 2012 -0700

    Switch bionic over to using libcore's copy of libcore.util.ZoneInfo.
    
    Bug: 7012465
    Change-Id: I1225494c5d77a20fd48be1e904d8695ef95860e3


But that might not be exactly it. There's also an entry in the git log about "Pull bionic's time zone data from the new single 'tzdata' file." It sounds like those zoneinfo.* files are depreciated for some uses?

What I do know is that earlier last week the code would build and make working img files. But now if you do a completely fresh repo init and sync, then make, you'll not end up with any "bionic/libc/zoneinfo/zoneinfo.*" files or "out/target/product/panda/system/usr/share/zoneinfo/zoneinfo.*" files. These zoneinfo.* files missing from the system.img prevents android from booting past the splash-screen. (and logcat on the device shows complaints relating to zoneinfo).

If I place zoneinfo.* files (from an old build) in "out/target/product/panda/system/usr/share/zoneinfo/zoneinfo.*" and re-run make to re-generate the .img files, I end up with a build that will boot.

Brad

On Tue, Oct 23, 2012 at 11:29 AM, Weston Weems <wwe...@gmail.com> wrote:
Got checkin hash? I'd be willing to look into it.

Weston


On Monday, October 22, 2012 10:16:14 AM UTC-7, Brad Arnold wrote:
Oh, I see there was a commit in bionic by Elliott Hughes last week updating the zoneinfo.* file generation code. I think there is a bug here.

On fresh working directories, the zoneinfo files never get generated. I suspect that's the source of all my problems.

Thoughts?

Brad

On Monday, October 22, 2012 10:40:29 AM UTC-6, Weston Weems wrote:
You know... I found for whatever reason timezone info database was infact NOT populated... so I had to find a zoneinfo.dat, zoneinfo.idx, zoneinfo.version to toss in /usr/share/zoneinfo, and now the pid exited stuff is not an issue. Its booting just fine.

I'll report back on the stability of this build (seeing as the last few syncs seemed to have a decent number of bugs in libdvm, and libhwui.

Weston

Brad Arnold

unread,
Oct 26, 2012, 1:37:54 PM10/26/12
to android-...@googlegroups.com
I've noticed that with master on the panda lots of apps that are not included with AOSP don't run at all. I've assumed that this may be due to an API which isn't exactly consistent with the last released version. I suspect master wouldn't pass the CTS (with any target device). Have you tried running other apps on the build you created? Is it better?

I was actually going to try this myself too, but it takes me so long to build on my home machine.

Brad

On Fri, Oct 26, 2012 at 10:12 AM, Weston Weems <wwe...@gmail.com> wrote:
Not out of the box, but it was rather easy to get it going... I just made sure hardware/ti/panda was current from head, kernel was at android-omap-panda-3.0, and that the sgx binaries matched. Everything else was simple sync from 4.1.2_r1.

Things are a LOT more stable than the head branch. finally! =)

Weston



On Thursday, October 25, 2012 4:13:45 AM UTC-7, michael wrote:
Hi Weston,
Does branch 4.1.2_r1 work on pandaboard for you?
Now im building 4.1.2_r1, i hope that can boot up successfully on pandaboard.

在 2012年10月24日星期三UTC+8下午11时56分00秒,Weston Weems写道:
Yep looks like the functionality is half there... it definitely doesnt create zoneinfo stuff, and tz initialization stuff on the java side expects them to be there. I guess this is the danger in dealing with the head. I'm gonna just grab 4.1.2_r1 and make it work for panda...

Weston

Alexander Meinke

unread,
Oct 30, 2012, 3:21:59 PM10/30/12
to android-...@googlegroups.com
Heyde,

it is quite interesting to see, that the i2c problem mentioned by @p fu is not specific to omap4430. I managed to boot my Panda by changing the SYSC register offset in drivers/i2c/busses/i2c-omap.c of the linux kernel from 0x10 to 0x20.
I think the problem lies in the code using/surrounding SYSC register (i2c subsys/drvier), because newer linux kernels (I tested version 3.6) have the register offset 0x10 and boot well on the same board!


Pandaboard Rev A2 / OMAP4430 ES2.1


Am 30.10.2012 um 14:21 schrieb p fu <fupin...@gmail.com>:

I can not boot my pandaboard ES(4460), with newest aosp code, and after add debug in kernel i found some thing wrong with i2c, who can give me some help?

<6>omap_i2c omap_i2c.1: bus 1 rev4.0 at 400 kHz

<4>Skipping twl internal clock init and using bootloader value (unknown osc rate)

<3>omap_i2c omap_i2c.1: controller timed out

<3>twl: i2c_write failed to transfer all messages (addr 0x0049, reg 3, len 4)

<3>omap_i2c omap_i2c.1: controller timed out

<3>twl: i2c_write failed to transfer all messages (addr 0x0049, reg 6, len 4)

<3>omap_i2c omap_i2c.1: controller timed out

<3>twl: i2c_write failed to transfer all messages (addr 0x0049, reg 0, len 4)

<6>twl6030: PIH (irq 39) chaining IRQs 368..387

<3>omap_i2c omap_i2c.1: controller timed out

<3>twl: i2c_read failed to transfer all messages (addr 0x0048, reg 29, len 1)

<3>twl6030: I2C err reading REG_VBATMIN_HI_CFG_STATE: -110

Brad Arnold

unread,
Nov 5, 2012, 10:29:29 AM11/5/12
to android-...@googlegroups.com
There's nothing wrong with these ones: https://developers.google.com/android/nexus/drivers#panda

(I've also run master using these)


On Sat, Nov 3, 2012 at 6:37 AM, AndroidNewb <neelima.h...@gmail.com> wrote:

Dan, Can you give me the link to the right graphics binaries you have used?
Thanks

On Wednesday, October 17, 2012 10:10:31 AM UTC-4, Dan Murphy wrote:
I have booted the lastest master on a Panda ES.

It is running right now.

Dan

On Thursday, October 11, 2012 11:27:46 AM UTC-5, G. Blake Meike wrote:
Apropos: I have an ES.  Is there any plan to support it, ever?  No biggie, just curious.

thanks
  Blake Meike

Brad Arnold

unread,
Nov 5, 2012, 6:50:33 PM11/5/12
to android-...@googlegroups.com
Do you have three files in: out/target/product/panda/system/usr/share/zoneinfo/ ?


On Mon, Nov 5, 2012 at 4:48 PM, Duncan Cunningham <dunc...@gmail.com> wrote:
I haven't had any luck getting my pandaboard to boot master.

If I use a build from master, when booting up it will just hang on Uncompressing Linux... done, booting the kernel.  No other output is display over serial.  I found some posts claiming that if you hit enter a bunch of times, the kernel will boot.  This doesn't do anything for me.  Then found some other posts claiming I need to use a version of the kernel that uses 1.8@785978 of the SGX driver.  So I built the kernel from this commit: 52f476403350050beb0dff135a55c06c9e7a82a9 and got the 1.8@785978 SGX binaries.  Now when it boots, I just get the below error repeated infinitely.  I've double checked and there isn't a SGX version mismatches.  The pandaboard doesn't display anything on screen.

[ 2895.670898] init: untracked pid 9335 exited
[ 2900.918670] PVR_K:(Error): BridgedDispatchKM: Driver initialisation not completed yet. [4836, drivers/gpu/pvr/bridged_pvr_bridge.c]
[ 2900.966613] init: untracked pid 9350 exited

Brad Arnold

unread,
Nov 6, 2012, 3:22:06 PM11/6/12
to android-...@googlegroups.com
Not sure about the panda version and kernel stuff then.

But it does sound (at least to me) that you're missing some zoneinfo files. Can you run ps aux or top or something to figure out what task your PID actually correlates to? (run ps aux, wait for next untracked PID printout, reference back to ps ax output). This should at least point you in the right direction.


On Tue, Nov 6, 2012 at 1:51 PM, Duncan Cunningham <dunc...@gmail.com> wrote:
The only file I have in that directory is tzdata.  I'm not for sure if the problem I am running in to is related to the zoneinfo files being removed.  If I use the master version of the kernel, the kernel won't boot at all.  Also I am using the regular Pandaboard, not the ES.

Brad Arnold

unread,
Nov 6, 2012, 4:43:55 PM11/6/12
to android-...@googlegroups.com
I've seen that before when I forgot to extract the graphics binaries.

Are you certain the the latest kernel + latest binaries don't work on the rev A? Because they work great for me on rev B / ES.


On Tue, Nov 6, 2012 at 3:40 PM, Duncan Cunningham <dunc...@gmail.com> wrote:
Looks like zygote is the untracked pid exiting.

Gabriel Beddingfield

unread,
Nov 8, 2012, 7:06:33 AM11/8/12
to android-...@googlegroups.com


On Nov 5, 2012 4:49 PM, "Duncan Cunningham" <dunc...@gmail.com> wrote:
>
> I haven't had any luck getting my pandaboard to boot master.
>
> If I use a build from master, when booting up it will just hang on Uncompressing Linux... done, booting the kernel.  No other output is display over serial.  I found some posts

This is a known issue and seems to be related to I2S init. Seems to affect "older" boards (don't know if that means a specific rav). I know that Dan is working on a solution.

-gabriel

Brad Arnold

unread,
Nov 8, 2012, 10:59:45 AM11/8/12
to android-...@googlegroups.com
I was able to make master work (as of about a week ago) as long as I inserted the three zoneinfo files (this is on rev B1). However, I decided I didn't want to run master, I wanted to run 4.1 (which I find to be more stable). So here's my process start to finish:

$ mkdir WORKING_DIRECTORY
$ cd WORKING_DIRECTORY
$ repo init -u https://android.googlesource.com/platform/manifest -b android-4.1.2_r1
$ repo sync
$ tar xzvf imgtec-panda-20120807-c4e99e89.tgz
$ chmod +x extract-imgtec-panda.sh
$ ./extract-imgtec-panda.sh
$ cd device/ti/panda/
$ git checkout aosp/master
$ cd ../../../hardware/ti/omap4xxx/
$ git checkout aosp/master
$ source build/envsetup.sh
$ lunch full_panda-userdebug
$ make

and then when it's all done:

$ device/ti/panda/usbboot device/ti/panda/bootloader.bin
power up board with no sd card, let it boot from usb, then insert SD card
$ fastboot oem format
$ fastboot flash xloader device/ti/panda/xloader.bin
$ fastboot flash bootloader device/ti/panda/bootloader.bin
reboot into fastboot by holding the GPIO button
$ fastboot flash cache
$ fastboot flash userdata
$ fastboot flashall


I've bolded the parts that are different from the typical documentation that is floating around. (including device/ti/panda/README)

No promises that this works with any revision other than B1. But I would assume it works with B2 as well, because I believe the only difference is the DDR manufacturer. 

Cheers,

Brad


On Wed, Nov 7, 2012 at 9:17 PM, Jay Yang <zwy...@marvell.com> wrote:
Hi Arnold,
How do you make it ? Could you describe it in detail? I use the latest system and kernel on rev ES,it still has some problems:
1、If I don't hit Enter key,the system will suspend in the process of boot up,and show "Uncompressing Linux... done, booting the kernel. "
2、If I hit Enter key continuously during the system boot up,the system will dead and the screen will block at Android starting-up frame,finally.

Thanks

Jay Yang

在 2012年11月7日星期三UTC+8上午5时50分59秒,Brad Arnold写道:

Ilya Tikhonovsky

unread,
Nov 9, 2012, 2:48:52 AM11/9/12
to android-...@googlegroups.com
Hi Brad,

I also have the same problem with booting pandaboard es Rev. B
I repeated all the steps you mentioned and got the same result.
The boot process hangs at the same stage - blank screen with no other log records.
Also I see i2c error in the log.

Could you please suggest me how can I troubleshoot the boot process or solve this problem.

------------------------------------------------------------------------------
Texas Instruments X-Loader 1.41 (Oct 21 2011 - 09:28:33)
OMAP4460: 1.2 GHz capable SOM
Starting OS Bootloader from MMC/SD1 ...


U-Boot 1.1.4-gedeced79 (Feb  6 2012 - 09:27:11)

Load address: 0x80e80000
DRAM:  1024 MB
Flash:  0 kB
Using default environment

In:    serial
Out:   serial
Err:   serial

efi partition table:
     256     128K xloader
     512     256K bootloader
    2048       8M recovery
   18432       8M boot
   34816     512M system
 1083392     256M cache
 1607680   14598M userdata
Net:   KS8851SNL
Hit any key to stop autoboot:  0 
kernel   @ 80008000 (3691628)
ramdisk  @ 81000000 (167091)
I2C read: I/O error

Starting kernel ...

Uncompressing Linux... done, booting the kernel.
------------------------------------------------------------------------------

Regards,
Tim.

Brad Arnold

unread,
Nov 9, 2012, 11:03:02 AM11/9/12
to android-...@googlegroups.com
Where did you see that output? The RS232 port?

That I2C error doesn't seem to matter, I get the exact same output as you from u-boot (including the I2C error), but the kernel boots fine. That message is a red herring.

It seems like the kernel doesn't properly have stdout set up. But that's odd if you're using the kernel from master and the same hardware rev as me. I would expect you'd see some output from the kernel.

Here's a thought: what program are you using to view the serial output? Maybe the kernel is set up to use flow control, but your serial terminal isn't supporting it? I'm using screen (ie: "sudo screen /dev/ttyUSB0 115200"). 

Brad

Ilya Tikhonovsky

unread,
Nov 9, 2012, 3:12:07 PM11/9/12
to android-...@googlegroups.com
Brad,

I'm using minicom. I set it up as described in this article http://omappedia.org/wiki/PandaBoard_Ubuntu_Minicom
and I had no problem with the terminal and linaro android image. Also the device works fine with ubuntu.

I'd say that the kernel hangs because the activity leds don't blink and no signal on hdmi and dvi ports.

Brad Arnold

unread,
Nov 9, 2012, 3:52:57 PM11/9/12
to android-...@googlegroups.com
I'm out of ideas. The latest kernel works fine for me. :/

Brad Arnold

unread,
Nov 13, 2012, 10:37:18 AM11/13/12
to android-...@googlegroups.com
I've not tried to get video acceleration working yet (though it's on my to-do list). If you have any luck, please post back :) .

I have had it working on other branches, but I don't remember which. It may have been 4.1.1, or it may have been master (a few weeks ago).

Brad


On Tue, Nov 13, 2012 at 3:21 AM, Don Han <ples...@gmail.com> wrote:
Brad

this worked for me, thanks.
but i don't know how to get hw video acceleration working.
i added ducati decoder to /etc/media_codecs.xml, and copied ducati-m3.bin to /system/vendor/firmware.
it is supposed that /dev/rpmsg-omx1 be created during kernel startup. but it's not there.

any sugestions from you?

Don

Jean-Baptiste Queru

unread,
Dec 3, 2012, 10:20:13 PM12/3/12
to android-...@googlegroups.com
I've merged in a bunch of changes, both in user-space and in the
kernel, and in my latest tests the master branch boots (but hangs) on
PandaBoard, and seems fine on PandaBoard ES.

JBQ

On Mon, Nov 26, 2012 at 11:34 AM, Alex <ame...@online.de> wrote:
> Mismatch between bootloader and kernel is only one of three causes mentioned
> at Common Problems Booting Linux.
>
> Am Montag, 26. November 2012 19:51:40 UTC+1 schrieb Andrey Vostrikov:
>>
>> Usually it indicates mismatch between bootloader and kernel.
>> Try first prebuilt images from top of master branch.
>>
>> On Monday, November 26, 2012 7:56:51 PM UTC+4, Janosch Maier wrote:
>>>
>>> I am building for the panda board from the main repo. The kernel is
>>> android-omap-panda-3.0.
>>>
>>> After flashing the panda board the kernel hungs. The last message I get
>>> on the serial console is ‘Booting kernel..’.

Jean-Baptiste Queru

unread,
Dec 5, 2012, 9:01:27 PM12/5/12
to android-...@googlegroups.com
Unfortunately, 4430 is likely to remain somewhat tricky. The
PandaBoard kernel is based on the one that is used in the Galaxy
Nexus, which uses a 4460, and it's very possible that there are other
changes lurking in there that make 4430 unhappy (we found one, but we
have to assume there are more).

JBQ

On Wed, Dec 5, 2012 at 10:46 AM, Alex <ame...@online.de> wrote:
> Thank you for the bunch of changes JBQ. At least the prebuilt kernel now
> boots without the need to build the kernel from source with different SYSC
> offset that marked kernel dirty.
> Sadly the system hang most of the time after the line "acc_release". If
> Dexopt is done which include rebooting three to five times, the last line is
> "eth0: no IPv6 routers present". Sometimes it really come up to the lock
> screen and further (see attached screencap0). The system seem stable until
> class_start main in init.rc. In my opinion, mediaserver make the system
> hang, but needs more investigation.
> I further need to change init.rc to start console as root, because I need to
> restart adbd. My newest bootlog (attached) indicate a problem with
> persistent, writeable memory (storage). That means, if I flash userdata and
> cache and got to the settings app changing time or language it will
> definitely hang on next reboot. Flashing partitions again make the system
> boot.
>
> Pandaboard Rev A2 / OMAP4430 ES2.1
>
Reply all
Reply to author
Forward
0 new messages