I am attempting to build Linaro's latest stable AOSP build for the pandaboard (https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a...). It appears to build fine, but unfortunately when I load it on the SD card and attempt to boot the pandaboard, I get the message:
"ERROR: Did not find a cmdline Flattened Device Tree Could not find a valid device tree"
I've attached the full log up through this error. I have followed the directions from the page I linked above. Please let me know if you need more information from me about my build environment, other logs, etc. and I'll be happy to supply them.
Thanks!
-Aaron Lindsay
Hi Aaron,
I have tested build 184 and it boots up fine . It appears that you have picked the wrong bootloader while creating the image . The logs shows uboot version as "U-Boot 2012.04.01 (Aug 15 2012 - 16:44:44)" which surely won't work with AOSP kernel . We use a prebuilt uboot in these aosp builds which are picked from TI's repositories . Please use the tarballs and the linaro-android-media-create to create the images and let us know if you still run into the issue.
Regards, Vishal
On 17 August 2012 00:52, Aaron Lindsay aclindsa@gmail.com wrote:
I am attempting to build Linaro's latest stable AOSP build for the pandaboard ( https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a... ). It appears to build fine, but unfortunately when I load it on the SD card and attempt to boot the pandaboard, I get the message:
"ERROR: Did not find a cmdline Flattened Device Tree Could not find a valid device tree"
I've attached the full log up through this error. I have followed the directions from the page I linked above. Please let me know if you need more information from me about my build environment, other logs, etc. and I'll be happy to supply them.
Thanks!
-Aaron Lindsay
Vishal,
Ah, I'm building everything from source. so am just pulling in whatever u-boot version pinned-manifest.xml points to. I have tried copying over u-boot.{img|bin} from the boot.tar.bz2 and MLO available from Linaro's build to the boot.tar.bz2 created when I use the linaro_android_build_cmds.sh script. This appears to get me past the device tree message, but unfortunately, never makes it past "Uncompressing Linux... done, booting the kernel." I suspect this is due to some communication error between uboot and the kernel, but am not sure what.
Does Linaro's build process do anything else (besides using a pre-built uboot) differently from what I would get by simply running linaro_android_build_cmds.sh?
Thanks for the help!
-Aaron
On Fri, Aug 17, 2012 at 3:09 AM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
Hi Aaron,
I have tested build 184 and it boots up fine . It appears that you have picked the wrong bootloader while creating the image . The logs shows uboot version as "U-Boot 2012.04.01 (Aug 15 2012 - 16:44:44)" which surely won't work with AOSP kernel . We use a prebuilt uboot in these aosp builds which are picked from TI's repositories . Please use the tarballs and the linaro-android-media-create to create the images and let us know if you still run into the issue.
Regards, Vishal
On 17 August 2012 00:52, Aaron Lindsay aclindsa@gmail.com wrote:
I am attempting to build Linaro's latest stable AOSP build for the pandaboard (https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a...). It appears to build fine, but unfortunately when I load it on the SD card and attempt to boot the pandaboard, I get the message:
"ERROR: Did not find a cmdline Flattened Device Tree Could not find a valid device tree"
I've attached the full log up through this error. I have followed the directions from the page I linked above. Please let me know if you need more information from me about my build environment, other logs, etc. and I'll be happy to supply them.
Thanks!
-Aaron Lindsay
Hi Aaron,
On 17 August 2012 23:30, Aaron Lindsay aclindsa@gmail.com wrote:
Vishal,
Ah, I'm building everything from source. so am just pulling in whatever u-boot version pinned-manifest.xml points to. I have tried copying over u-boot.{img|bin} from the boot.tar.bz2 and MLO available from Linaro's build to the boot.tar.bz2 created when I use the linaro_android_build_cmds.sh script. This appears to get me past the device tree message, but unfortunately, never makes it past "Uncompressing Linux... done, booting the kernel." I suspect this is due to some communication error between uboot and the kernel, but am not sure what.
You need to pick the MLO also from the prebuilt boot.tar.bz2 . MLO and uboot together make the pair which works for aosp to boot. I am confused here . If you build from source these prebuilt MLO and uboot are part of the source and they get copied into the boottarball . I am not sure why is it not happening in your build.
Does Linaro's build process do anything else (besides using a pre-built uboot) differently from what I would get by simply running linaro_android_build_cmds.sh?
linaro_android_build_cmds.sh build should be able to build the exact images as generated on android build . Let me know if the right uboot binary is not copied when using linaro_android_build_cmds.sh .
Thanks for the help!
-Aaron
On Fri, Aug 17, 2012 at 3:09 AM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
Hi Aaron,
I have tested build 184 and it boots up fine . It appears that you have picked the wrong bootloader while creating the image . The logs shows
uboot
version as "U-Boot 2012.04.01 (Aug 15 2012 - 16:44:44)" which surely
won't
work with AOSP kernel . We use a prebuilt uboot in these aosp builds
which
are picked from TI's repositories . Please use the tarballs and the linaro-android-media-create to create the images and let us know if you still run into the issue.
Regards, Vishal
On 17 August 2012 00:52, Aaron Lindsay aclindsa@gmail.com wrote:
I am attempting to build Linaro's latest stable AOSP build for the pandaboard (
https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a... ).
It appears to build fine, but unfortunately when I load it on the SD card and attempt to boot the pandaboard, I get the message:
"ERROR: Did not find a cmdline Flattened Device Tree Could not find a valid device tree"
I've attached the full log up through this error. I have followed the directions from the page I linked above. Please let me know if you need more information from me about my build environment, other logs, etc. and I'll be happy to supply them.
Thanks!
-Aaron Lindsay
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hi Aaron
On 08/17/2012 09:07 PM, Vishal Bhoj wrote:
Hi Aaron,
On 17 August 2012 23:30, Aaron Lindsay aclindsa@gmail.com wrote:
Vishal,
This is ok to boot git://gitorious.org/pandaboard/u-boot-mainline.git. I have tested today
this is the branch origin/v2011.06-panda
And you have the source code of a bootloader that boot the omapzoom kernel. I'm still trying to understand the reason why the omapzoom kernel doesn't boot with the u-boot-arm branch or linaro uboot one.
Michael
Ah, I'm building everything from source. so am just pulling in whatever u-boot version pinned-manifest.xml points to. I have tried copying over u-boot.{img|bin} from the boot.tar.bz2 and MLO available from Linaro's build to the boot.tar.bz2 created when I use the linaro_android_build_cmds.sh script. This appears to get me past the device tree message, but unfortunately, never makes it past "Uncompressing Linux... done, booting the kernel." I suspect this is due to some communication error between uboot and the kernel, but am not sure what.
You need to pick the MLO also from the prebuilt boot.tar.bz2 . MLO and uboot together make the pair which works for aosp to boot. I am confused here . If you build from source these prebuilt MLO and uboot are part of the source and they get copied into the boottarball . I am not sure why is it not happening in your build.
Does Linaro's build process do anything else (besides using a pre-built uboot) differently from what I would get by simply running linaro_android_build_cmds.sh?
linaro_android_build_cmds.sh build should be able to build the exact images as generated on android build . Let me know if the right uboot binary is not copied when using linaro_android_build_cmds.sh .
Thanks for the help!
-Aaron
On Fri, Aug 17, 2012 at 3:09 AM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
Hi Aaron,
I have tested build 184 and it boots up fine . It appears that you have picked the wrong bootloader while creating the image . The logs shows
uboot
version as "U-Boot 2012.04.01 (Aug 15 2012 - 16:44:44)" which surely
won't
work with AOSP kernel . We use a prebuilt uboot in these aosp builds
which
are picked from TI's repositories . Please use the tarballs and the linaro-android-media-create to create the images and let us know if you still run into the issue.
Regards, Vishal
On 17 August 2012 00:52, Aaron Lindsay aclindsa@gmail.com wrote:
I am attempting to build Linaro's latest stable AOSP build for the pandaboard (
https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a... ).
It appears to build fine, but unfortunately when I load it on the SD card and attempt to boot the pandaboard, I get the message:
"ERROR: Did not find a cmdline Flattened Device Tree Could not find a valid device tree"
I've attached the full log up through this error. I have followed the directions from the page I linked above. Please let me know if you need more information from me about my build environment, other logs, etc. and I'll be happy to supply them.
Thanks!
-Aaron Lindsay
linaro-android mailing list linaro-android@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-android
Vishal,
On Fri, Aug 17, 2012 at 3:07 PM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
You need to pick the MLO also from the prebuilt boot.tar.bz2 . MLO and uboot together make the pair which works for aosp to boot. I am confused here . If you build from source these prebuilt MLO and uboot are part of the source and they get copied into the boottarball . I am not sure why is it not happening in your build.
I am copying over MLO in addition to uboot - sorry if my original wording was sloppy or ambiguous. I have now done a completely fresh check-out of the same revision (https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a...) which has again resulted in a uboot version of "U-Boot SPL 2012.04.01 (Aug 17 2012 - 18:42:11)"
Any ideas as to why running this script on my system wouldn't be pulling in the prebuilt bootloader as you say it should?
-Aaron
Hi Aaron,
On 20 August 2012 19:53, Aaron Lindsay aclindsa@gmail.com wrote:
Vishal,
On Fri, Aug 17, 2012 at 3:07 PM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
You need to pick the MLO also from the prebuilt boot.tar.bz2 . MLO and
uboot
together make the pair which works for aosp to boot. I am confused here
. If
you build from source these prebuilt MLO and uboot are part of the source and they get copied into the boottarball . I am not sure why is it not happening in your build.
I am copying over MLO in addition to uboot - sorry if my original wording was sloppy or ambiguous. I have now done a completely fresh check-out of the same revision ( https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a... ) which has again resulted in a uboot version of "U-Boot SPL 2012.04.01 (Aug 17 2012 - 18:42:11)"
The android-build page mentions the configs we set for the build where we have "USE_PREBUILT_UBOOT=true" which is a custom variable which is not picked while generating linaro_android_build_cmds.sh script.
Please export this variable and then trigger the build . It should work.
Any ideas as to why running this script on my system wouldn't be pulling in the prebuilt bootloader as you say it should?
-Aaron
On Tue, Aug 21, 2012 at 2:01 AM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
The android-build page mentions the configs we set for the build where we have "USE_PREBUILT_UBOOT=true" which is a custom variable which is not picked while generating linaro_android_build_cmds.sh script.
Please export this variable and then trigger the build . It should work.
Doh! I saw there was a chunk of exported variables in the script that matched those on the build webpage and didn't bother to double-check and make sure ALL of them were there. Please forgive my ignorance. I'll rebuild it with this set and see how it goes.
Might it be a good idea to put this variable in the build scripts for those builds that require it to boot?
Thanks!
-Aaron
On 21 August 2012 08:54, Aaron Lindsay aclindsa@gmail.com wrote:
On Tue, Aug 21, 2012 at 2:01 AM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
The android-build page mentions the configs we set for the build where we have "USE_PREBUILT_UBOOT=true" which is a custom variable which is not picked while generating linaro_android_build_cmds.sh script.
Please export this variable and then trigger the build . It should work.
Doh! I saw there was a chunk of exported variables in the script that matched those on the build webpage and didn't bother to double-check and make sure ALL of them were there. Please forgive my ignorance. I'll rebuild it with this set and see how it goes.
Aaron,
Its not your ignorance! From time to time we use "work around" env vars to quickly turn a build. These then get worked into the build. The process is not well documented though. Sorry for the confusion. Thanks for working with us on it!
Might it be a good idea to put this variable in the build scripts for those builds that require it to boot?
Yuppers! Feel like sending a patch in? :)
Thanks!
-Aaron
linaro-android mailing list linaro-android@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-android
Vishal, Zach,
On Tue, Aug 21, 2012 at 9:59 AM, Zach Pfeffer zach.pfeffer@linaro.org wrote:
On 21 August 2012 08:54, Aaron Lindsay aclindsa@gmail.com wrote:
On Tue, Aug 21, 2012 at 2:01 AM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
The android-build page mentions the configs we set for the build where we have "USE_PREBUILT_UBOOT=true" which is a custom variable which is not picked while generating linaro_android_build_cmds.sh script.
Please export this variable and then trigger the build . It should work.
Doh! I saw there was a chunk of exported variables in the script that matched those on the build webpage and didn't bother to double-check and make sure ALL of them were there. Please forgive my ignorance. I'll rebuild it with this set and see how it goes.
Hmm, I have now done a fresh build with this variable enabled, and confirmed on boot that it says "U-Boot 1.1.4-gc1cd80bc-dirty (Jan 24 2012 - 10:21:09)" rather than the version I was building before. However, booting the kernel still hangs after "Uncompressing Linux... done, booting the kernel." with no other messages or errors. This is, again, with an unmodified checkout of https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a....
I will gladly help debug this in more depth with a little push in the right direction.
Might it be a good idea to put this variable in the build scripts for those builds that require it to boot?
Yuppers! Feel like sending a patch in? :)
I will gladly send a patch in if we end up finding the solution!
-Aaron
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hi
On 08/22/2012 09:34 PM, Aaron Lindsay wrote:
Vishal, Zach,
On Tue, Aug 21, 2012 at 9:59 AM, Zach Pfeffer zach.pfeffer@linaro.org wrote:
On 21 August 2012 08:54, Aaron Lindsay aclindsa@gmail.com wrote:
On Tue, Aug 21, 2012 at 2:01 AM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
The android-build page mentions the configs we set for the build where we have "USE_PREBUILT_UBOOT=true" which is a custom variable which is not picked while generating linaro_android_build_cmds.sh script.
Please export this variable and then trigger the build . It should work.
Doh! I saw there was a chunk of exported variables in the script that matched those on the build webpage and didn't bother to double-check and make sure ALL of them were there. Please forgive my ignorance. I'll rebuild it with this set and see how it goes.
Hmm, I have now done a fresh build with this variable enabled, and confirmed on boot that it says "U-Boot 1.1.4-gc1cd80bc-dirty (Jan 24 2012 - 10:21:09)" rather than the version I was building before. However, booting the kernel still hangs after "Uncompressing Linux... done, booting the kernel." with no other messages or errors. This is, again, with an unmodified checkout of https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a....
I will gladly help debug this in more depth with a little push in the right direction.
Might it be a good idea to put this variable in the build scripts for those builds that require it to boot?
Yuppers! Feel like sending a patch in? :)
Please apply this patch to your build
Michael
I will gladly send a patch in if we end up finding the solution!
-Aaron
linaro-android mailing list linaro-android@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-android
- -- | Michael Nazzareno Trimarchi Amarula Solutions BV | | COO - Founder Cruquiuskade 47 | | +31(0)851119172 Amsterdam 1018 AM NL |
Michael,
I enabled DEBUG_LL and EARLY_PRINTK, and found it was an unhandled alignment trap. I got a few of these: [ 0.000000] Unhandled fault: alignment exception (0x801) at 0xc06ebeb7
So, I unset ALIGNMENT_TRAP in my config, and am now able to boot.
After a little digging, it looks like something similar to "EXTRA_CFLAGS=-mno-unaligned-access" might do the trick as well, but I haven't tested that yet.
It looks like some people have had trouble with this on a CodeSourcery compiler before [1,2], but I haven't yet found stuff about this on gcc (I'm using "arm-linux-gnueabi-gcc (Ubuntu/Linaro 4.6.2-14ubuntu2~ppa1) 4.6.2").
Has anyone else had similar issues?
-Aaron
[1] http://blog.galemin.com/2011/05/linux-kernel-2-6-39-codesourcery-2011-03-41-... [2] http://blog.galemin.com/2011/05/linux-kernel-2-6-39-codesourcery-2011-03-41-...
Aaron,
Can you check if the prebuild MLO and u-boot are getting used ? Can you compare the checksum for the MLO and u-boot in the generated boottarball and the one in device/ti/panda . They should match .
Also can you share the complete boot log from serial console from xloader and u-boot and the kernel.
On 23 August 2012 20:15, Aaron Lindsay aclindsa@gmail.com wrote:
Michael,
I enabled DEBUG_LL and EARLY_PRINTK, and found it was an unhandled alignment trap. I got a few of these: [ 0.000000] Unhandled fault: alignment exception (0x801) at 0xc06ebeb7
So, I unset ALIGNMENT_TRAP in my config, and am now able to boot.
After a little digging, it looks like something similar to "EXTRA_CFLAGS=-mno-unaligned-access" might do the trick as well, but I haven't tested that yet.
It looks like some people have had trouble with this on a CodeSourcery compiler before [1,2], but I haven't yet found stuff about this on gcc (I'm using "arm-linux-gnueabi-gcc (Ubuntu/Linaro 4.6.2-14ubuntu2~ppa1) 4.6.2").
Has anyone else had similar issues?
-Aaron
[1] http://blog.galemin.com/2011/05/linux-kernel-2-6-39-codesourcery-2011-03-41-... [2] http://blog.galemin.com/2011/05/linux-kernel-2-6-39-codesourcery-2011-03-41-...
linaro-android mailing list linaro-android@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-android
Vishal,
On Thu, Aug 23, 2012 at 10:49 AM, Vishal Bhoj vishal.bhoj@linaro.org wrote:
Can you check if the prebuild MLO and u-boot are getting used ? Can you compare the checksum for the MLO and u-boot in the generated boottarball and the one in device/ti/panda . They should match.
They match (also cross-checked with boot.tar.bz2 from https://android-build.linaro.org/builds/~linaro-android/panda-master-gcc44-a...).
Also can you share the complete boot log from serial console from xloader and u-boot and the kernel.
I've attached the log from the working kernel (i.e. CONFIG_ALIGNMENT_TRAP=n) as alignment_trap_fix.log and the non-working version as crashing.log
-Aaron
linaro-android@lists.linaro.org
-
Aaron Lindsay -
Michael Trimarchi -
Vishal Bhoj -
Zach Pfeffer