today, arago fails to build for MACHINE=am180x-dev when it used to work in the past, is there a way to fix this?

This looks to be a result of kernel.org changing the tarballs around.  In this case is looks like you were able to download the new tarball, but the checksums have been changed from the ones OE was expecting.  The quick fix would be to edit the arago-oe-dev/recipes/linux-libc-headers/linux-libc-headers_2.6.31.bb recipe and change the checksum values to:

SRC_URI[md5sum] = "451474ed7a118c02865b0b463cd6c161"

SRC_URI[sha256sum] = "6bc7de7234c12a65d0fed957a4b00b9ba685ff0f06bbcf85900973472eebcbd8"

and be sure to bump the PR number.  You could also fix this if you have an old copy of the original tarball in your DL_DIR that you could share with others that need it.

Chase

Chase,

Thank you for responding so quickly.

We deleted the files that you specified and the build went further before running into a 2nd problem.

I captured the error and then decided to start over. arago-tmp and download directories are removed and another build started.

The 2nd problem appears to be real. For some reason the recipe is looking for a file that does not exist in our toolchain from SDK 5.02.

What it is looking for does not exist in any of the SDK 5.02 toolchains on our other development machines either.

here is the build session failure for the 2nd case:

=====================================

lcardona@egc003:~/oe/arago$ bitbake arago-base-image

NOTE: Handling BitBake files: \ (7675/7675) [100 %]

Parsing of 7675 .bb files complete (7218 cached, 457 parsed). 7896 targets, 426 skipped, 585 masked, 0 errors.

Build Configuration:

BB_VERSION        = "1.10.2"

METADATA_BRANCH   = "master"

METADATA_REVISION = "c017010"

TARGET_ARCH       = "arm"

TARGET_OS         = "linux-gnueabi"

MACHINE           = "am180x-evm"

DISTRO            = "angstrom"

DISTRO_VERSION    = "2011.09"

TARGET_FPU        = "soft"

NOTE: Resolving any missing task queue dependencies

NOTE: Preparing runqueue

NOTE: Executing runqueue

NOTE: Running task 164 of 3848 (ID: 128, /home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb, do_install)

NOTE: package external-toolchain-arago-1.0-r10: task do_install: Started

ERROR: TaskFailed event exception, aborting

ERROR: Build of /home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb do_install failed

ERROR: Task 128 (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb, do_install) failed with exit code 1

ERROR: '/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb' failed

ERROR: Function do_install failed

NOTE: Task failed: ('function do_install failed', '/home/lcardona/oe/arago-tmp/work/armv5te-arago-linux-gnueabi/external-toolchain-arago-1.0-r10/temp/log.do_install.25143')

ERROR: Logfile of failure stored in: /home/lcardona/oe/arago-tmp/work/armv5te-arago-linux-gnueabi/external-toolchain-arago-1.0-r10/temp/log.do_install.25143

Log data follows:

| cp: warning: source file `/home/lcardona/ti-sdk-am180x-evm-05.02.00.00/linux-devkit/arm-arago-linux-gnueabi/usr/include/netdb.h' specified more than once

| cp: warning: source file `/home/lcardona/ti-sdk-am180x-evm-05.02.00.00/linux-devkit/arm-arago-linux-gnueabi/usr/include/syscall.h' specified more than once

| cp: warning: source file `/home/lcardona/ti-sdk-am180x-evm-05.02.00.00/linux-devkit/arm-arago-linux-gnueabi/usr/include/sysexits.h' specified more than once

| cp: warning: source file `/home/lcardona/ti-sdk-am180x-evm-05.02.00.00/linux-devkit/arm-arago-linux-gnueabi/usr/include/syslog.h' specified more than once

| cp: cannot stat `/home/lcardona/ti-sdk-am180x-evm-05.02.00.00/linux-devkit/arm-arago-linux-gnueabi/usr/lib/libsupc*': No such file or directory

| ERROR: Function do_install failed

NOTE: package external-toolchain-arago-1.0-r10: task do_install: Failed

ERROR: TaskFailed event exception, aborting

ERROR: Build of /home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb do_install failed

ERROR: Task 128 (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb, do_install) failed with exit code 1

ERROR: '/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb' failed

ERROR: '/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb' failed

lcardona@egc003:~/oe/arago$ ^C

lcardona@egc003:~/oe/arago$

Jeffrey,

I am passing this on to the Arago maintainer.  There is a separate toolchain that we use during our SDK builds and it looks like there is something in how this toolchain is packaged that is different than how our SDKs package the toolchain.  Short term the toolchain is supposed to be released to ti.com as an individual download until we can fix this in the SDK.

Chase

Chase,

Thank you for replying.

There is a secondary issue.

Our company is using Arago to develop our custom root file system for our product.

We are locked to kernel 2.6.37 because that is what Arago had months ago and we have crafted additional drivers for our proprietary hardware.

Our previous builds used that kernel that was available in Arago more than a few months ago and this worked just fine.

Then, there was a regression from 2.6.37 back to 2.6.32, but the Arago git archive is not tagged with a releases anymore.

How do we maintain order if we just follow the Arago "getting started" web page that says to just clone whatever they have.

Next, we put our overlays as arago-custom only to find out months down the road that we've built the whole project on quick-sand and nothing builds from scratch anymore.

Our overlays consist of adding our special applications, additional libraries, additional configuration, etc.

How is this intended to work for projects that now relay on an unversioned Arago git archive?

Is the user expected to clone once and then be forever responsible to maintain consistency themselves and perhaps back in changes from the Arago upstream archive?

At this point, we are very concerned because there is no clear direction of how to fix this situation.

Any suggestions?

Jeffrey,

The regression from 2.6.37 was not for the am180x-evm device.  My current nightly builds show the kernel version as 2.6.37.  Can you confirm that when you build virtual/kernel that you are seeing the kernel version being built as 2.6.32?  This is different than the linux-libc-headers package version.

As for the other questions:

1.  We do generally recommend that you take the latest Arago sources because that reflects what we are currently building and staging for our Sitara Linux SDK.  We do have SDK trees that we tag with each release.  You can find them at:

http://arago-project.org/git/projects/?p=arago-amsdk.git;a=summary

http://arago-project.org/git/projects/?p=arago-oe-amsdk.git;a=summary

and they replace the arago and arago-oe-dev repositories respectively.  In general though, once you start your project and get close to production I would expect you to stabilize and tag your own local trees.  This is much like if you were pulling from any open source project repository where you would be responsible for either pulling from a tag (again, see the trees I provided above) or at some point freezing your pulls until you release your product and then updating.

2.  I understand the download issue is frustrating.  A lot of this came about from when kernel.org was hacked and many of the tarballs were re-generated with changed md5sums.  This is not an Arago issue but instead is an issue that has impacted all of openembedded.

So overall you should consider Arago as the "mainline" of TI development.  When pulling from mainline you will eventually need to stabilize/freeze the sources.  You can use the AMSDK trees to refer to the tags for our releases as a baseline as well.  I think some of the confusion you are having is that the "getting started" page is getting started with Arago development, and that usually entails starting with the latest sources.

Chase

Jeffrey,

On the toolchain side I was able to build the full target file system with the toolchain included in the AMSDK 05.03.  Can you try updating to that SDK?  The component I was unable to build was to repackage the toolchain in the SDK for deployment in an SDK.  I know that sounds circular, so what I'm really saying is that the method we use to put the toolchain in the SDK will not work using the toolchain already in the SDK.  This does not hamper target side development so you should be OK.

What I did was just add the linux-devkit/bin directory to my path and I could do actions like bitbake arago-tisdk-image which rebuilds the SDK target image.

Chase

Chase,

I agree with the recommendation and came to the same conclustion myself. I'm currently dowloading all the 5.03.02 stuff in order to recover our build process.

This was a real catch 22 in that the 2011-July Arago snapshot could not build from scratch because some of the packages became unfetchable.

I could not compile with the latest Arago because the SDK-5.02 toolchain did not support the build requirement.

Consequently all our developers world-wide are awaiting instructions on how recover the OE build process.

There ought to be additional instructions on setting up the Arago build environment that gives developers some insight on these issues.

Thank you very much for your help and assistance. It is very much appreciated.

I'll let you know what our results are. My expectations are that we can duplicate what you have already verified.

Regards,

Jeff Franks

Chase,

The arago-tisdk-image using the SDK 5.03.02 got past the point that it broke previously.

However, the Pam module seems unfetchable from any provider. I've checked all of them with the web browser and they are truely unreachable.

We have our conf.local file set to a single CPU so that it is easier to sort out the error messages.

Here is the latest session failure:

-------

NOTE: Running task 299 of 6022 (ID: 3077, /home/lcardona/oe/arago-oe-dev/recipes/xorg-proto/xproto_7.0.20.bb, do_compile)
NOTE: package xproto-1_7.0.20-r2.0: task do_compile: Started
NOTE: package xproto-1_7.0.20-r2.0: task do_compile: Succeeded
NOTE: Running task 300 of 6022 (ID: 3067, /home/lcardona/oe/arago-oe-dev/recipes/xorg-proto/xproto_7.0.20.bb, do_install)
NOTE: package xproto-1_7.0.20-r2.0: task do_install: Started
NOTE: package xproto-1_7.0.20-r2.0: task do_install: Succeeded
NOTE: Running task 304 of 6022 (ID: 3068, /home/lcardona/oe/arago-oe-dev/recipes/xorg-proto/xproto_7.0.20.bb, do_populate_sysroot)
NOTE: package xproto-1_7.0.20-r2.0: task do_populate_sysroot: Started
NOTE: package xproto-1_7.0.20-r2.0: task do_populate_sysroot: Succeeded
NOTE: Running task 320 of 6022 (ID: 4487, virtual:native:/home/lcardona/oe/arago-oe-dev/recipes/xorg-proto/xproto_7.0.20.bb, do_configure)
NOTE: package xproto-native-1_7.0.20-r2.0: task do_configure: Started
--2012-02-23 15:54:59--  http://kernel.org/pub/linux/libs/pam/library/Linux-PAM-1.1.1.tar.bz2
Resolving kernel.org... NOTE: package xproto-native-1_7.0.20-r2.0: task do_configure: Succeeded
NOTE: Running task 330 of 6022 (ID: 2139, /home/lcardona/oe/arago-oe-dev/recipes/pam/libpam_1.1.1.bb, do_fetch)
NOTE: package libpam-1.1.1-r2: task do_fetch: Started
NOTE: fetch http://kernel.org/pub/linux/libs/pam/library/Linux-PAM-1.1.1.tar.bz2
149.20.4.69
Connecting to kernel.org|149.20.4.69|:80... connected.
HTTP request sent, awaiting response... 404 Not Found
2012-02-23 15:55:04 ERROR 404: Not Found.

--2012-02-23 15:55:05--  http://mirrors.openembedded.org//Linux-PAM-1.1.1.tar.bz2
Resolving mirrors.openembedded.org... NOTE: fetch http://mirrors.openembedded.org//Linux-PAM-1.1.1.tar.bz2
82.197.159.157
Connecting to mirrors.openembedded.org|82.197.159.157|:80... failed: No route to host.
--2012-02-23 15:55:08--  http://sources.openembedded.org//Linux-PAM-1.1.1.tar.bz2
Resolving sources.openembedded.org... 140.211.169.152
Connecting to sources.openembedded.org|140.211.169.152|:80... NOTE: fetch http://sources.openembedded.org//Linux-PAM-1.1.1.tar.bz2
connected.
HTTP request sent, awaiting response... 404 Not Found
2012-02-23 15:55:09 ERROR 404: Not Found.

--2012-02-23 15:55:09--  http://www.angstrom-distribution.org/unstable/sources/Linux-PAM-1.1.1.tar.bz2
Resolving www.angstrom-distribution.org... NOTE: fetch http://www.angstrom-distribution.org/unstable/sources/Linux-PAM-1.1.1.tar.bz2
188.40.83.200
Connecting to www.angstrom-distribution.org|188.40.83.200|:80... failed: Connection refused.
--2012-02-23 15:55:10--  http://arago-project.org/files/sources/Linux-PAM-1.1.1.tar.bz2
Resolving arago-project.org... 184.73.161.183
Connecting to arago-project.org|184.73.161.183|:80... NOTE: fetch http://arago-project.org/files/sources/Linux-PAM-1.1.1.tar.bz2
connected.
HTTP request sent, awaiting response... 404 Not Found
2012-02-23 15:55:10 ERROR 404: Not Found.

ERROR: TaskFailed event exception, aborting
ERROR: Build of /home/lcardona/oe/arago-oe-dev/recipes/pam/libpam_1.1.1.bb do_fetch failed
ERROR: Task 2139 (/home/lcardona/oe/arago-oe-dev/recipes/pam/libpam_1.1.1.bb, do_fetch) failed with exit code 1
ERROR: '/home/lcardona/oe/arago-oe-dev/recipes/pam/libpam_1.1.1.bb' failed
NOTE: Task failed: Fetch failed: Unable to fetch URL http://kernel.org/pub/linux/libs/pam/library/Linux-PAM-1.1.1.tar.bz2 from any source.
NOTE: package libpam-1.1.1-r2: task do_fetch: Failed
ERROR: TaskFailed event exception, aborting
ERROR: Build of /home/lcardona/oe/arago-oe-dev/recipes/pam/libpam_1.1.1.bb do_fetch failed
ERROR: Task 2139 (/home/lcardona/oe/arago-oe-dev/recipes/pam/libpam_1.1.1.bb, do_fetch) failed with exit code 1
ERROR: '/home/lcardona/oe/arago-oe-dev/recipes/pam/libpam_1.1.1.bb' failed
ERROR: '/home/lcardona/oe/arago-oe-dev/recipes/pam/libpam_1.1.1.bb' failed

Jeffrey,

I found these sources at http://pkgs.fedoraproject.org/repo/pkgs/pam/Linux-PAM-1.1.1.tar.bz2/9b3d952b173d5b9836cbc7e8de108bee/

You can download the tarball there into your DL_DIR and create the .md5 file for it or modify the libpam_1.1.1.bb recipe to use this SRC_URI instead.

Chase

Chase,

Yes, that worked for that item.

I ran into seven more and I'm only half-way into the build for arago-tisdk-image and now working on only 4 hours of sleep.

Now, I've run into another non-fetchable but it is more complicated.

The URL is in arago-oe-dev/recipes/opkg-utils/opkg-utils_svn.bb but this is broke.

Any suggestions for this one?

--jlf

Looks like in oe-core (the next generation of OE) they moved to a git repository hosted on yoctoproject.org.  You can see the recipe at http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-devtools/opkg-utils.  You could probably just pack port that one into your OE.

Chase

Chase,

Yes, that seems like it worked. The build is continuing. About 448 files are now in the download area and growing.

So much of the web changed from the last time that arago-oe-dev was released (December 2011) that it is impossible to build arago-tisdk-image without changes to the .bb files.

The build process is so brittle because of the dependencies on resources that cannot be counted on, i.e., they might not be there, the website may be down, etc. This kills future builds as these dependencies cannot be satisfied without human intervention.

Is it possible that the arago project archive the downloads and make these available for a build from scratch?

Eliminating resource dependencies that you have no control over will allow builds go to completion out-of-the-box.

What prevents this from happening?

--jlf

Jeffrey,

We are looking into this, but there we need to make sure we can do this legally.  It is something we are investigating and hope to have in the future but I can't give you an exact timeline.

Chase

Chase,

We are getting closer to a complete build but ran into another non-fetchable that I cannot figure out.

The URL from arago-oe-dev/recipes/rt-tests/rt-tests_git.bb does not resolve.

I could not locate this anywhere on the web.

Do you have any suggestions?

--jlf

Jeffrey,

Check http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-rt/rt-tests

Chase

Chase,

The recipe was copied into the appropriate arago-oe-dev folder and the other ones were move out.

This also ends up unfetchable.

Any suggestions?

--jlf

Jeffrey,

Are you able to do "git clone git://github.com/clrkwllms/rt-tests.git" manually?  I was able to do that.

Chase

Chase,

Yes I was able to do that command manually.

However, when I restarted the build, it tries to do it again like it never "sees" that the rt-tests directory is already there in downloads.

Here is the session for the manual git command, followed by contents of the .bb file, followed by another build session failure

------------ manual git command -------------

lcardona@egc003:~/oe/downloads$ ls *rt-tests*
ls: cannot access *rt-tests*: No such file or directory

lcardona@egc003:~/oe/downloads$ git clone git://github.com/clrkwllms/rt-tests.git
Initialized empty Git repository in /home/lcardona/oe/downloads/rt-tests/.git/
remote: Counting objects: 1822, done.
remote: Compressing objects: 100% (751/751), done.
remote: Total 1822 (delta 1051), reused 1820 (delta 1049)
Receiving objects: 100% (1822/1822), 342.76 KiB | 191 KiB/s, done.
Resolving deltas: 100% (1051/1051), done.

lcardona@egc003:~/oe/downloads$ ls *rt-tests*
git_github.com.clrkwllms.rt-tests.git_5f1e84f8b015df3ff950056494134eca3f640d70.tar.gz.lock

rt-tests:
COPYING  doc  Makefile  rt-tests.spec-in  scripts  src

----------  rt-test_0.83.bb ------------

DESCRIPTION = "Real-Time preemption testcases"
HOMEPAGE = "https://rt.wiki.kernel.org/index.php/Cyclictest"
SECTION = "tests"
DEPENDS = "linux-libc-headers eglibc"
LICENSE = "GPLv2 & GPLv2+"
LIC_FILES_CHKSUM = "file://COPYING;md5=751419260aa954499f7abaabaa882bbe \
                    file://src/cyclictest/cyclictest.c;beginline=7;endline=9;md5=ce162fe491d19d2ec67dff6dbc938d50 \
                    file://src/pi_tests/pi_stress.c;beginline=6;endline=19;md5=bd426a634a43ec612e9fbf125dfcc949"
# Version v0.83
SRCREV = "5f1e84f8b015df3ff950056494134eca3f640d70"

# git -> 0.83 needs a PE bump
PE = "1"

SRC_URI = "git://github.com/clrkwllms/rt-tests.git"

S = "${WORKDIR}/git"

CFLAGS += "-I${S}/src/include -D_GNU_SOURCE -Wall -Wno-nonnulli ${LDFLAGS}"

# calling 'uname -m' is broken on crossbuilds
EXTRA_OEMAKE = "NUMA=0"

do_install() {
        oe_runmake install DESTDIR=${D} SBINDIR=${sbindir} MANDIR=${mandir} \
                           INCLUDEDIR=${includedir}
}

---------- another build attempt from where we failed previously ------------

 lcardona@egc003:~/oe/arago$ bitbake arago-tisdk-image
NOTE: Handling BitBake files: \ (7674/7674) [100 %]
Parsing of 7674 .bb files complete (7217 cached, 457 parsed). 7895 targets, 426 skipped, 585 masked, 0 errors.

Build Configuration:
BB_VERSION        = "1.10.2"
METADATA_BRANCH   = "master"
METADATA_REVISION = "c017010"
TARGET_ARCH       = "arm"
TARGET_OS         = "linux-gnueabi"
MACHINE           = "am180x-evm"
DISTRO            = "angstrom"
DISTRO_VERSION    = "2011.09"
TARGET_FPU        = "soft"

NOTE: Resolving any missing task queue dependencies
ERROR: Multiple .bb files are due to be built which each provide virtual/libc (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
ERROR: Multiple .bb files are due to be built which each provide virtual/arm-arago-linux-gnueabi-libc-for-gcc (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
ERROR: Multiple .bb files are due to be built which each provide virtual/libiconv (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
--2012-02-27 11:54:07--  http://www.angstrom-distribution.org/unstable/sources/git_github.com.clrkwllms.rt-tests.git_5f1e84f8b015df3ff950056494134eca3f640d70.tar.gz
Resolving www.angstrom-distribution.org... NOTE: Preparing runqueue
ERROR: Multiple .bb files are due to be built which each provide virtual/libc (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
ERROR: Multiple .bb files are due to be built which each provide virtual/arm-arago-linux-gnueabi-libc-for-gcc (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
ERROR: Multiple .bb files are due to be built which each provide virtual/libiconv (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
NOTE: Executing runqueue
140.211.169.179
Connecting to www.angstrom-distribution.org|140.211.169.179|:80... NOTE: Running task 3004 of 6037 (ID: 2846, /home/lcardona/oe/arago-oe-dev/recipes/rt-tests/rt-tests_0.83.bb, do_fetch)
connected.
HTTP request sent, awaiting response... 404 Not Found
2012-02-27 11:54:07 ERROR 404: Not Found.

Cloning into /home/lcardona/oe/downloads/git/github.com.clrkwllms.rt-tests.git...
NOTE: package rt-tests-1_0.83-r0: task do_fetch: Started
NOTE: fetch http://www.angstrom-distribution.org/unstable/sources/git_github.com.clrkwllms.rt-tests.git_5f1e84f8b015df3ff950056494134eca3f640d70.tar.gz
rsync: failed to connect to github.com: Connection refused (111)
rsync error: error in socket IO (code 10) at clientserver.c(122) [Receiver=3.0.7]
fatal: Could not run rsync to get refs
ERROR: TaskFailed event exception, aborting
ERROR: Build of /home/lcardona/oe/arago-oe-dev/recipes/rt-tests/rt-tests_0.83.bb do_fetch failed
ERROR: Task 2846 (/home/lcardona/oe/arago-oe-dev/recipes/rt-tests/rt-tests_0.83.bb, do_fetch) failed with exit code 1
ERROR: '/home/lcardona/oe/arago-oe-dev/recipes/rt-tests/rt-tests_0.83.bb' failed
NOTE: Task failed: Fetch failed: Unable to fetch URL git://github.com/clrkwllms/rt-tests.git from any source.
NOTE: package rt-tests-1_0.83-r0: task do_fetch: Failed
ERROR: TaskFailed event exception, aborting
ERROR: Build of /home/lcardona/oe/arago-oe-dev/recipes/rt-tests/rt-tests_0.83.bb do_fetch failed
ERROR: Task 2846 (/home/lcardona/oe/arago-oe-dev/recipes/rt-tests/rt-tests_0.83.bb, do_fetch) failed with exit code 1
ERROR: '/home/lcardona/oe/arago-oe-dev/recipes/rt-tests/rt-tests_0.83.bb' failed
ERROR: '/home/lcardona/oe/arago-oe-dev/recipes/rt-tests/rt-tests_0.83.bb' failed
lcardona@egc003:~/oe/arago$

--------------

--jlf

Jeffrey,

It seems like you are having an issue connecting to github.com.  I just ran this with no issue in oe-core after deleting my checkout and forcing a reclone.  Do you perhaps not have a proxy setup correctly?

Chase

Chase,

We don't have a proxy, it is just a router between the virtual Linux (network bridged) and the ISP connection.

The base machine is Windows 7 running VMplayer as specified in the setup manual.

It works for everything else for this particular arago-tisdk-image build process, and nothing has changed in the network or connection setup.

Do you have any ideas about this?

--jlf

Jeffrey,

Add protocol=git to the SRC_URI line.  i.e.

SRC_URI = "git://github.com/clrkwllms/rt-tests.git;protocol-git"

Chase

Chase,

...;protocol=git" resolved that issue.

Next download = iozone3 and that went okay.

Next download = hdparam-9.35.tar.gz became unfetchable.

I was able to download it manually outside of OE to a different area, and the md5sum was good according to the .bb file.

The following is the content of the .bb file and the build session at this point:

---- hdparam_9.35.bb ------------

DESCRIPTION = "hdparm is a Linux shell utility for viewing \
and manipulating various IDE drive and driver parameters."
SECTION = "console/utils"
PRIORITY = "optional"
LICENSE = "BSD"

PACKAGES += "wiper"

FILES_wiper = "${bindir}/wiper.sh"
FILES_${PN} = "${base_sbindir} ${mandir}"

RDEPENDS_wiper = "bash gawk stat"

SRC_URI = "${SOURCEFORGE_MIRROR}/hdparm/hdparm-${PV}.tar.gz "
SRC_URI[md5sum] = "520996cc36b69212c3907df351296702"
SRC_URI[sha256sum] = "b778e5f42d918226892417986dc135e783e7a1b7b3986a4b1f637bcf51366f5f"

do_install () {
 install -d ${D}/${base_sbindir} ${D}/${mandir}/man8 ${D}/${bindir}
 oe_runmake 'DESTDIR=${D}' 'sbindir=${base_sbindir}' install
 mv ${D}${base_sbindir}/hdparm ${D}${base_sbindir}/hdparm.${PN}
 cp ${S}/wiper/wiper.sh ${D}/${bindir}
}

pkg_postinst_${PN} () {
 update-alternatives --install ${base_sbindir}/hdparm hdparm hdparm.${PN} 100
}

pkg_prerm_${PN} () {
 update-alternatives --remove hdparm hdparm.${PN}
}

-------- build session for arago-tisdk-image ----------

lcardona@egc003:~/oe/arago$ bitbake arago-tisdk-image
NOTE: Handling BitBake files: \ (7674/7674) [100 %]
Parsing of 7674 .bb files complete (7216 cached, 458 parsed). 7895 targets, 426 skipped, 585 masked, 0 errors.

Build Configuration:
BB_VERSION        = "1.10.2"
METADATA_BRANCH   = "master"
METADATA_REVISION = "c017010"
TARGET_ARCH       = "arm"
TARGET_OS         = "linux-gnueabi"
MACHINE           = "am180x-evm"
DISTRO            = "angstrom"
DISTRO_VERSION    = "2011.09"
TARGET_FPU        = "soft"

NOTE: Resolving any missing task queue dependencies
ERROR: Multiple .bb files are due to be built which each provide virtual/libc (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
ERROR: Multiple .bb files are due to be built which each provide virtual/arm-arago-linux-gnueabi-libc-for-gcc (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
ERROR: Multiple .bb files are due to be built which each provide virtual/libiconv (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
FATAL: hdparm-9.35: http://downloads.sourceforge.net/hdparm/hdparm-9.35.tar.gz cannot check archive integrity
ERROR: Function base_do_fetch failed
ERROR: TaskFailed event exception, aborting
ERROR: Build of /home/lcardona/oe/arago-oe-dev/recipes/hdparm/hdparm_9.35.bb do_fetch failed
NOTE: Preparing runqueue
ERROR: Multiple .bb files are due to be built which each provide virtual/libc (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
ERROR: Multiple .bb files are due to be built which each provide virtual/arm-arago-linux-gnueabi-libc-for-gcc (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
ERROR: Multiple .bb files are due to be built which each provide virtual/libiconv (/home/lcardona/oe/arago/recipes/meta/external-toolchain-arago.bb /home/lcardona/oe/arago-oe-dev/recipes/eglibc/eglibc_2.12.bb).
 This usually means one provides something the other doesn't and should.
NOTE: Executing runqueue
ERROR: Task 2801 (/home/lcardona/oe/arago-oe-dev/recipes/hdparm/hdparm_9.35.bb, do_fetch) failed with exit code 1
ERROR: '/home/lcardona/oe/arago-oe-dev/recipes/hdparm/hdparm_9.35.bb' failed
NOTE: Running task 3013 of 6037 (ID: 2801, /home/lcardona/oe/arago-oe-dev/recipes/hdparm/hdparm_9.35.bb, do_fetch)
NOTE: package hdparm-9.35-r0: task do_fetch: Started
NOTE: The checksums for '/home/lcardona/oe/downloads/hdparm-9.35.tar.gz' did not match.
Expected MD5: '520996cc36b69212c3907df351296702' and Got: '78d74b8d059dfa3fa0ad31dff9f5ab14'
Expected SHA256: 'b778e5f42d918226892417986dc135e783e7a1b7b3986a4b1f637bcf51366f5f' and Got: '99c6b63b6dce9b3abfa1e323a74fccfd85b42d509e9d6b24a22bd9a0b0c7e4a7'
NOTE: Your checksums:
SRC_URI[md5sum] = "78d74b8d059dfa3fa0ad31dff9f5ab14"
SRC_URI[sha256sum] = "99c6b63b6dce9b3abfa1e323a74fccfd85b42d509e9d6b24a22bd9a0b0c7e4a7"

FATAL: hdparm-9.35: http://downloads.sourceforge.net/hdparm/hdparm-9.35.tar.gz cannot check archive integrity
ERROR: '/home/lcardona/oe/arago-oe-dev/recipes/hdparm/hdparm_9.35.bb' failed
lcardona@egc003:~/oe/arago$

-------------

--jlf

Jeffrey,

The recipe md5sum and sha256sum do not seem to match the values for the tarball at /home/lcardona/oe/downloads/hdparm-9.35.tar.gz.  Either your tarball is wrong or the recipe needs updating.  The output shows that you got md5sum and sha256sum values that do not match the expected values in the recipe.

Chase

Chase,

Oops, operator error. There must have been some previous attempt that had a corrupted download.

I removed hdparam-9.35.* and fired off another build.

The build is proceeding. Hopefully it will complete.

I'll let you know what the result is.

Thanks,

--jlf

Chase,

We got past hdparm-9.35.tar.gz and downloaded and compiled a lot more files.

We have now have 699 items in the oe download area.

However, a compile problem has occured on eglibc. I've tried doing a clean on that recipe, which did not help.

Then I deleted the eglibc downloaded file and a clean on the recipe, which did not help.

I've attached a zip file that has the compile time failure.

Any suggestions?

--jlf

Jeffrey,

It looks like you are seeing two recipes, external-toolchain-arago and eglibc providing the same libraries.  You should NOT be building eglibc your self since that should come from the arago toolchain.  Did you clean your arago-tmp directory before you switched the toolchain when you moved to the latest SDK?  If not then you should rm the arago-tmp directory (You can leave your downloads directory alone)

Chase

Chase,

Yes, the download and arago-tmp were scratched before starting on this journey with the new toolchain.

I'll do as you say and leave the downloaded files and scratch arago-tmp and restart the build.

I hope that works and will let you know as soon as there is another result.

--jlf

Chase,

I did what you said, but have the same results with eglibc trying to compile and failing.

Any suggestions?

--jlf

Jeffrey,

It looks like you have a conflict between your toolchain which should be providing eglibc and eglibc itself.  Can you verify that you have set TOOLCHAIN_BRAND = arago and TOOLCHAIN_TYPE = external?  Try setting those in your local.conf.

If that still doesn't work you may try asking this question at arago-devel@gforge.ti.com

I just did a clean build of am180x-evm using the AMSDK toolchain with no issues.  I don't see the multiple providers messages you do

Chase

Chase,

local.conf was already set the way you suggested.

I don't recall seeing multiple providers messages when I started the original build. Now, I cannot get rid of it.

I've started totally over again from the point of setting up the OE environment including git from arago, editing the conf.local and setenv.

I have mulitple smp set to 4.

The build started without multiple providers messages. So, somewhere along the line something got broke in the previous scenarios.

The first non-fetchable was for arago/recipes/bluez because of a checksum error. I fixed the URL in the .bb file, removed the bluez* from dowloads and restarted the build.

It failed shortly thereafter on the linux davinci psp patch. I set the smp back to 1 and restarted the build. It failed again same way.

Then I did a clean on that .bb file and restarted the build (single smp) and the build is continuing and there are no multiple providers messages.

I'll continue on and put the changes into the .bb files to overcome the non-fetchables and watch out for the multiple providers messages. They are gone for now.

I'll let you know how things go from this point using only single smp.

--jlf

Chase,

As was said before, the local.conf file already had these entries.

The build completed !!!

The difference this time was to use only one cpu core.

This took a lot longer to build and it went to completion. There were only two .bb modifications needed to get to fix unfetchable URLs.

The multiple providers error never came back after starting from scratch and setting local.conf to use only one cpu core.

It looks like using multiple cores on a virgin build from scratch including downloads is what caused most of the problems with the two exceptions noted below.

Thank you so much for your help on this.

lcardona@egc003:~/oe/arago$ git status
# On branch master
# Changed but not updated:
#   (use "git add <file>..." to update what will be committed)
#   (use "git checkout -- <file>..." to discard changes in working directory)
#
# modified:   recipes/bluez/bluez4_4.95.bb
#

cardona@egc003:~/oe/arago-oe-dev$ git status
# On branch master
# Changed but not updated:
#   (use "git add <file>..." to update what will be committed)
#   (use "git checkout -- <file>..." to discard changes in working directory)
#
# modified:   recipes/rt-tests/rt-tests_git.bb
#

Jeffrey,

I'm glad to hear that.  I'm looking at this closer but I do clean builds often with 24 cores.  However, I do not often clean my downloads directory so maybe that has something to do with it.

I'm building a list of missing packages to try and get those up on the Arago mirror for now.

Chase

Jeffrey,

There you have your problem, in the rt-tests recipe that you ported from OE-Core:

DEPENDS = "linux-libc-headers eglibc"

It specifically depends on eglibc, which throws a wrench in the external toolchain support in Arago/Classic-OE. Please replace that with virtual/libc and everything will work.

Denys