On 03/19/2012 05:27 PM, Gary Thomas wrote:
On 2012-03-19 14:49, jfabernathy wrote:
On 03/19/2012 04:25 PM, Gary Thomas wrote:
On 2012-03-19 13:25, jfabernathy wrote:
On 03/19/2012 03:13 PM, Gary Thomas wrote:
On 2012-03-19 13:04, jfabernathy wrote:
On 03/19/2012 03:03 PM, Gary Thomas wrote:
On 2012-03-19 12:59, jfabernathy wrote:
On 03/19/2012 02:55 PM, Gary Thomas wrote:
On 2012-03-19 10:02, jfabernathy wrote:
Just did a git pull today on poky and meta-intel. I tried to
rebuild meta-sugarbay again and got a build error. Not sure
why yet. Not critical, just a spot check of things.
------------------
NOTE: package mesa-dri-2_7.11-r13.1: task do_configure:
Succeeded
NOTE: Running task 1723 of 5148 (ID: 1928,
/home/jim/poky/meta/recipes-graphics/mesa/mesa-dri_7.11.bb,
do_compile)
NOTE: package mesa-dri-2_7.11-r13.1: task do_compile: Started
ERROR: Function failed: do_compile (see
/build/sugarbay-master/tmp/work/x86_64-poky-linux/mesa-dri-2_7.11-r13.1/temp/log.do_compile.25687
for further information)
ERROR: Logfile of failure stored in:
/build/sugarbay-master/tmp/work/x86_64-poky-linux/mesa-dri-2_7.11-r13.1/temp/log.do_compile.25687
Log data follows:
| DEBUG: SITE files ['endian-little', 'common-linux',
'common-glibc', 'bit-64', 'x86_64-linux', 'common']
| ERROR: Function failed: do_compile (see
/build/sugarbay-master/tmp/work/x86_64-poky-linux/mesa-dri-2_7.11-r13.1/temp/log.do_compile.25687
for further information)
| NOTE: make -j 8
| make[1]: Entering directory
`/build/sugarbay-master/tmp/work/x86_64-poky-linux/mesa-dri-2_7.11-r13.1/Mesa-7.11/src'
| Making sources for autoconf
| make[2]: Entering directory
`/build/sugarbay-master/tmp/work/x86_64-poky-linux/mesa-dri-2_7.11-r13.1/Mesa-7.11/src/mapi/glapi'
| make[2]: *** No rule to make target
`/build/sugarbay-master/tmp/sysroots/x86_64-linux/usr/lib/x86_64-poky-linux/gcc/x86_64-poky-linux/4.6.3/include/stddef.h',
needed by
`glapi_dispatch.o'. Stop.
| make[2]: Leaving directory
`/build/sugarbay-master/tmp/work/x86_64-poky-linux/mesa-dri-2_7.11-r13.1/Mesa-7.11/src/mapi/glapi'
| make[1]: *** [subdirs] Error 1
| make[1]: Leaving directory
`/build/sugarbay-master/tmp/work/x86_64-poky-linux/mesa-dri-2_7.11-r13.1/Mesa-7.11/src'
| make: *** [default] Error 1
| ERROR: oe_runmake failed
NOTE: package mesa-dri-2_7.11-r13.1: task do_compile: Failed
ERROR: Task 1928
(/home/jim/poky/meta/recipes-graphics/mesa/mesa-dri_7.11.bb,
do_compile) failed with exit code '1'
What's your build host?
Was this build from scratch? There was a discussion of much the
same error on the meta-ti list last week where the host GCC was
updated between two subsequent builds that gave this error.
Building on latest release Ubuntu 11.10 (Linux ubuntu-x64
3.0.0-16-generic #29-Ubuntu SMP)
gcc is 4.6.1
Was your build from scratch? Do you use [extant] sstate cache?
I had a preview successful build and just reran that after
bitbake -c cleanall core-image-sato
Do you have rm_work in your local.conf?
No to question on rm_work in local.conf
Try this:
% bitbake -c cleansstate mesa-dri
% bitbake mesa-dri
This work without errors. I'm now doing bitbake core-image-sato to
see if it completes.
BTW, 'bitbake -c cleanall core-image-sato' doesn't really do much
- images don't have much (any?)
stuff that needs to get cleaned. You can run 'bitbake
core-image-sato' as many times as you like
and it will completely rebuild the image every time.
So if I do a git pull and want to retest with new stuff, just
bitbake core-image-sato or should I do something with cleansstate??
If you run just 'bitbake core-image-sato', any packages (recipes)
which have been changed that are used by the image will be rebuilt.
You should not have to clean anything for this to work properly.
After I fixed mesa-dri and started bitbake core-image-sato, it ran
for a long time and had a build failure similar to the first one on
kexec-tools. I did the bitbake -c cleansstate
kexec-tools trick again and restarted bitbake core-image-sato. Not
sure why this is, but it seems to work.
These errors should not happen - you should be able to update the
yocto meta-data and rebuild an image any time and not suffer such
problems.
Did you change anything else, especially with your host tools,
between the time your previous build completed and you did this
pull and rebuild? For example, running the Ubuntu package updater?
I never got a complete good build after I updated some unknown items in
Ubuntu 11.10 dev system and did a recent git pull until I did a
completely new bitbake core-image-sato to a new build directory. That
completed correctly.
So it seems we may need some Rules for users to follow when rebuilding
with the latest versions of Yocto. It seems there are some dependency on
certain key packages on the dev system not changing. Completely
rebuilding seems to be the safest method and given the time wasted when
things fail, might be the fastest.
Jim A
_______________________________________________
yocto mailing list
yocto@yoctoproject.org
https://lists.yoctoproject.org/listinfo/yocto
