State of PandaBoard

3,378 views
Skip to first unread message

Jean-Baptiste Queru

unread,
Jul 10, 2012, 9:18:58 PM7/10/12
to android-...@googlegroups.com
There's a known issue with graphics on PandaBoard, where the display
is entirely black.

I suspect that's related to the changes that were done in Jelly Bean
around triple buffering.

My understanding is that someone is already working on updating the
graphics libraries. In the meantime, PandaBoards won't display much
:-(

JBQ

--
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,
Jul 11, 2012, 11:53:36 AM7/11/12
to android-...@googlegroups.com
I always build the PandaBoard kernel directly from AOSP, so the kernel
source is always available. The panda kernel hasn't changed since ICS:
I can't merge newer upstream kernels as that breaks the graphics
driver, and I can't update to a new graphics driver as I don't have
any newer graphics libraries that work on PandaBoard.

JBQ

On Wed, Jul 11, 2012 at 1:09 AM, Will Tisdale <willt...@gmail.com> wrote:
> Damn, thats a shame.
>
> I did a build yesterday and tried the 1.8@550175 blobs to match the kernel,
> but as you say, black screen of nothingness. :(
>
> When will you be pushing the panda kernel source?
>
>
> - Will
> --
> 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

Will Tisdale

unread,
Jul 11, 2012, 11:55:32 AM7/11/12
to android-...@googlegroups.com
Ah, fair enough. I'll await new graphics libs then. :)

Thanks for your reply.


- Will

Chris Bainbridge

unread,
Jul 12, 2012, 9:31:21 AM7/12/12
to android-...@googlegroups.com
I also did a build yesterday and discovered this black screen issue.
Is there a last known-good version? And if so, what precisely needs to
be checked out? Any idea how long before this issue is fixed (days?
weeks?).

Jean-Baptiste Queru

unread,
Jul 12, 2012, 11:15:39 AM7/12/12
to android-...@googlegroups.com
PandaBoard can't be used from tagged releases. Only the master branch
can be used.

JBQ

On Thu, Jul 12, 2012 at 5:48 AM, hagenp <hagen....@googlemail.com> wrote:
> Which AOSP build _will_ work for PandaBoard (rev. A2 / OMAP4430 ES2.1) at
> this time?
>
> Currently I'm trying to get AOSP "android-4.0.4_r2.1" (IMM76L) with
> imgtec-panda-imm76i-67545da7.tgz (which should also cover IMM76L) up and
> running, following the instructions in device/ti/panda/README.
> After proper udev setup, usbboot/fastboot work fine with my 4GiB SD card.
>
> The kernel boots, but prints these errors on the serial console (no graphics
> output):
>
>> [ 4.933013] PVR_K: (FAIL) SGXInit: Incompatible driver DDK revision
>> (279068)/device DDK revision (785978).
>> [ 4.943359] PVR_K:(Error): PVRSRVFinaliseSystem: Failed
>> PVRSRVDevInitCompatCheck call (device index: 0) [451,
>> drivers/gpu/pvr/pvrsrv.c]
>> [ 4.956298] PVR_K:(Error): BridgedDispatchKM: Initialisation failed.
>> Driver unusable. [4795, drivers/gpu/pvr/bridged_pvr_bridge.c]
>>
>
> Any idea how to proceed? Should the branch android-4.0.4_r2.1 not work with
> the IMM76L drivers? Or do I need to configure something?
> [After starting with full_panda-userdebug, I'm currently re-building
> everything on full_panda-eng, but have no high hopes for it, hence my
> question.]
> Thanks! -Hagen
>
> --
> 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 Queru

unread,
Jul 12, 2012, 11:16:34 AM7/12/12
to android-...@googlegroups.com
We have patches that should fix PandaBoard, as soon as I can get them
reviewed and merged.

As a fallback, I have enough information available to recreate a
last-known-good state. I haven't had time to do that, though.

JBQ

Will Tisdale

unread,
Jul 12, 2012, 11:18:44 AM7/12/12
to android-...@googlegroups.com
Good question. Probably none currently, because the JB code has been
merged into master, and it was only master which was working. (As JBQ
just posted).

But now because JB and pandaboard graphics are broken, you can't get a
working build from master, unless you happen to have an old, pre JB
tree kicking about. I guess you could build JB master without the
libs, but it wouldn't be ideal.

The other alternative is waiting for the new libs.


- Will

On 12 July 2012 13:48, hagenp <hagen....@googlemail.com> wrote:
> Which AOSP build _will_ work for PandaBoard (rev. A2 / OMAP4430 ES2.1) at
> this time?
>
> Currently I'm trying to get AOSP "android-4.0.4_r2.1" (IMM76L) with
> imgtec-panda-imm76i-67545da7.tgz (which should also cover IMM76L) up and
> running, following the instructions in device/ti/panda/README.
> After proper udev setup, usbboot/fastboot work fine with my 4GiB SD card.
>
> The kernel boots, but prints these errors on the serial console (no graphics
> output):
>
>> [ 4.933013] PVR_K: (FAIL) SGXInit: Incompatible driver DDK revision
>> (279068)/device DDK revision (785978).
>> [ 4.943359] PVR_K:(Error): PVRSRVFinaliseSystem: Failed
>> PVRSRVDevInitCompatCheck call (device index: 0) [451,
>> drivers/gpu/pvr/pvrsrv.c]
>> [ 4.956298] PVR_K:(Error): BridgedDispatchKM: Initialisation failed.
>> Driver unusable. [4795, drivers/gpu/pvr/bridged_pvr_bridge.c]
>>
>
> Any idea how to proceed? Should the branch android-4.0.4_r2.1 not work with
> the IMM76L drivers? Or do I need to configure something?
> [After starting with full_panda-userdebug, I'm currently re-building
> everything on full_panda-eng, but have no high hopes for it, hence my
> question.]
> Thanks! -Hagen
>

Jean-Baptiste Queru

unread,
Jul 13, 2012, 10:56:10 AM7/13/12
to android-...@googlegroups.com
We have changes in process that are expected to fix things. If those
don't work, we'll find a way to go back to the last-known-good state.

JBQ

On Fri, Jul 13, 2012 at 12:06 AM, Benedikt Dietrich
<beni...@googlemail.com> wrote:
>
>
>>
>> But now because JB and pandaboard graphics are broken, you can't get a
>> working build from master, unless you happen to have an old, pre JB
>> tree kicking about.
>>
>
> Is it somehow possible to go back in the master tree with repo and do a
> checkout of the status before the JB merge?
>
> Thanks,
> Benedikt

Jean-Baptiste Queru

unread,
Jul 13, 2012, 12:35:56 PM7/13/12
to android-...@googlegroups.com
Right now there's no easy way.

JBQ

On Fri, Jul 13, 2012 at 9:25 AM, hagenp <hagen....@googlemail.com> wrote:
> On Friday, 13 July 2012 16:56:10 UTC+2, Jean-Baptiste Queru wrote:
>>
>> We have changes in process that are expected to fix things. If those
>> don't work, we'll find a way to go back to the last-known-good state.
>
>
> Hmmm... what can I do to get a working Panda/ICS build?

Hagen Patzke

unread,
Jul 13, 2012, 12:53:48 PM7/13/12
to android-...@googlegroups.com

> > Hmmm... what can I do to get a working Panda/ICS build?

On Jul 13, 2012 6:36 PM, "Jean-Baptiste Queru" <j...@android.com> wrote:
> Right now there's no easy way.

Yes, I noticed.

To test a component for a customer, early next week I need a debuggable ICS for PandaBoard - but any sub-version will do. Have you e.g. got a repo branch/tag or a bunch of git IDs for me?

Jean-Baptiste Queru

unread,
Jul 13, 2012, 12:58:18 PM7/13/12
to android-...@googlegroups.com
If you hunt around in all projects for the "ics-plus-aosp" branch and
create a manifest for that (which should be mostly similar to what
master-dalvik had before the Jelly Bean push), that might work. Maybe.

JBQ

Martin Heller

unread,
Jul 20, 2012, 8:55:17 AM7/20/12
to android-...@googlegroups.com
Hello,

do you have an idea when AOSP builds for PandaBoard will be fixed?
And will the master branch (i.e. 4.1) be fixed or a working tagged
release of 4.0 will be made?

Thanks

Martin

Jean-Baptiste Queru

unread,
Jul 20, 2012, 10:39:04 AM7/20/12
to android-...@googlegroups.com
We're currently stuck on the code review of a platform change to
disable VSync on hardware that doesn't support it. Until that change
gets approved by the maintainer of that code, we can't move forward.

JBQ
Message has been deleted

Magnus Bäck

unread,
Aug 1, 2012, 2:29:50 PM8/1/12
to android-...@googlegroups.com
On Wednesday, August 01, 2012 at 10:03 EDT,
sebbi <sebi.s...@googlemail.com> wrote:

> I have a good checkout of the AOSP master before they merged it with JB.
>
> What do I have to provide to help others? Really like that copy, runs
> almost perfect on my Pandaboard ES with Wifi+BT running pretty good.

You can capture the state of your repository and produce a manifest file
with SHA-1s that others can sync against:

repo manifest -r -o my-working-source.xml

[...]

--
Magnus Bäck
ba...@google.com

Jean-Baptiste Queru

unread,
Aug 1, 2012, 5:36:47 PM8/1/12
to android-...@googlegroups.com
I've just added a branch ics-plus-aosp that matches the state of the
AOSP master before the Jelly Bean release. In quick testing, it seems
to work on PandaBoard.

JBQ

Jean-Baptiste Queru

unread,
Aug 7, 2012, 12:24:14 PM8/7/12
to android-...@googlegroups.com
I believe that they're on git.omapzoom.org, but I'm not sure. I got
those binaries directly from TI.

JBQ

On Tue, Aug 7, 2012 at 9:22 AM, Eric Pfaffhauser
<eric.pfa...@gmail.com> wrote:
> Branch ics-plus-aosp works great. Thanks JBQ!
>
> But: where can I find the sources/repos/configs you've used to create
> bootloader.bin and x-loader.bin?
>
> eRiC

Jean-Baptiste Queru

unread,
Aug 8, 2012, 11:30:34 AM8/8/12
to android-...@googlegroups.com
If this is similar to the kernel markers, gedeced79 means "Git commit
edeced79", so you need to look for revision edeced79

JBQ

On Wed, Aug 8, 2012 at 6:45 AM, hagenp <hagen....@gmail.com> wrote:
> On Tuesday, 7 August 2012 21:27:56 UTC+2, Keith Deacon wrote:
>>
>> The binaries are build from the source on git.omapzoom.org.
>> uboot -
>> http://git.omapzoom.org/?p=repo/u-boot.git;a=shortlog;h=refs/heads/omap4_dev
>> xloader -
>> http://git.omapzoom.org/?p=repo/x-loader.git;a=shortlog;h=refs/heads/omap4_dev
>
>
> bootloader.bin identifies itself as 'U-Boot 1.1.4-gedeced79 (Feb 6 2012 -
> 09:27:11)'.
> Any pointer where to find this exact git revision (gedeced79)? Thanks!
> -Hagen

Chris Bainbridge

unread,
Aug 22, 2012, 6:51:40 PM8/22/12
to android-...@googlegroups.com
I just did a Pandaboard build of master and it is booting ok.

On 22 August 2012 14:15, hagenp <hagen....@gmail.com> wrote:
> Just for info: PandaBoard kernel does not come up with a current master
> build:
>>
>> Texas Instruments X-Loader 1.41 (Oct 21 2011 - 09:28:33)
>> 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 3044M userdata
>> Net: KS8851SNL
>> Hit any key to stop autoboot: 0
>> kernel @ 80008000 (3691560)
>> ramdisk @ 81000000 (169139)
>> I2C read: I/O error
>>
>> Starting kernel ...
>>
>> Uncompressing Linux... done, booting the kernel.
>
>
> [no change for five minutes]
>
> $ repo manifest -r -o current.xml
> current.xml content pasted here (aosp-master-20120822GMT1312):
> http://pastebin.com/wLpugAc1

Jean-Baptiste Queru

unread,
Sep 28, 2012, 11:36:38 AM9/28/12
to android-...@googlegroups.com
For me (ES EB3), the kernel boots, but the graphics system fails to
initialize, in a way that looks like an incompatibility between
userland and the kernel.

I'm pretty sure that it was working when I prepared those files, but
that was weeks ago.

Do things still working when going back to the previous kernel and
graphics libraries?

JBQ

On Fri, Sep 28, 2012 at 7:24 AM, hagenp <hagen....@gmail.com> wrote:
> On Friday, 28 September 2012 08:20:38 UTC+2, Adrien Grassein wrote:
>>
>> Someone else have the same problem?
>
>
> Yes. I just checked, and with a current "master" build for full_panda-eng
> the kernel hangs (ES2.1).
>
> ("repo manifest -r " also throws errors, so I can't even upload a snapshot
> manifest. Tsk.)
>
> --
> 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 Queru

unread,
Sep 28, 2012, 11:47:51 AM9/28/12
to android-...@googlegroups.com
In bad news, it looks like the old kernel and graphics binaries also
fail (same errors). I'm guessing that something else changed since
early August.

JBQ

Adrien Grassein

unread,
Sep 28, 2012, 12:02:07 PM9/28/12
to android-...@googlegroups.com

Yes everithing is ok with previois kernel and drivers with android Master.

Jean-Baptiste Queru

unread,
Oct 4, 2012, 2:01:46 PM10/4/12
to android-...@googlegroups.com
I've tested locally with the previous kernel and the previous graphics
libraries, and things are working for me. At least we know that the
rest of the master branch is OK, so whatever issues we're having are
either the kernel, the graphics libraries, or both, or an
incompatibilty between those.

Now, surprisingly, I don't manage to get any logcat output at all from
the most recent kernel I submitted. That's definitely not expected,
and that would suggest that the kernel is bad. I'll try to regenerate
one from source. Has anyone rebuilt that same kernel from source and
gotten it to boot at all (regardless of graphics).

JBQ

On Tue, Oct 2, 2012 at 11:09 AM, Nater <sigrist....@gmail.com> wrote:
> I can confirm that Change-Id: I5f4559d3997d7d57ca93317b41e2b06e512d46bc
> (commit 76605d821bcc0db9cd897d40fff45b5a7fce2c32) will work for
> device/ti/panda .
>
>
> On Monday, October 1, 2012 1:16:49 AM UTC-5, Adrien Grassein wrote:
>>
>> For me : kernel with commit 4a094a55476cad51c85928e7c5766d71944e87ea does
>> not work and I5f4559d3997d7d57ca93317b41e2b06e512d46bc works.
>>
>> Adrien
>>
>> Le vendredi 28 septembre 2012 23:22:01 UTC+2, Mikey Walzy a écrit :
>>>
>>> Which previous kernel are you using? We have built Android with prebuilt
>>> kernel, works fine, but now we use the kernel from the kernel build page and
>>> it doesn't work. A linaro 2.6.35 will allow me to build but gets hung up
>>> after booting kernel. So does a 3.4. Please help. We would like to build
>>> android with kernel source rather than prebuilt and get it running on
>>> pandaboard, so that we can include custom hardware/firmware files but we've
>>> had no luck finding a kernel compatible with android. We are willing to
>>> revert to previous android build if necessary.
>>> Thanks,
>>> BT
>>>
>>> On Friday, September 28, 2012 11:10:43 AM UTC-5, Adrien Grassein wrote:
>>>>
>>>> Yes everithing is ok with previois kernel and drivers with android
>>>> Master.
>

Adrien Grassein

unread,
Oct 4, 2012, 2:39:58 PM10/4/12
to android-...@googlegroups.com
Yes, I rebuilt the kernel from source from both commit (the previous commit, and the new one). As expected, the 'previous' works and the new one not.
Also, I try to merge ti-omap4 branch onto the ti-omap4-panda branch, and this not works too. 

2012/10/4 Jean-Baptiste Queru <j...@android.com>

Jean-Baptiste Queru

unread,
Oct 4, 2012, 3:47:32 PM10/4/12
to android-...@googlegroups.com
I think that the situation with not having logcat is a red herring.
I've also had logcat work fine with the latest kernel, and I've had it
fail with the previous one.

I'm digging deeper.

JBQ

Jean-Baptiste Queru

unread,
Oct 4, 2012, 5:40:06 PM10/4/12
to android-...@googlegroups.com
And now I can't get any build to give me any adb connection. I have no
idea what's going on :-(

JBQ

On Thu, Oct 4, 2012 at 2:05 PM, Alex <ame...@online.de> wrote:
> I enabled 'kernel hacking - kernel low-level debugging functions + early
> printk' in kernel config mentioned on Common Problems Booting Linux. The
> output on serial link is attached.
Reply all
Reply to author
Forward
0 new messages