Rerunning wic to build a new image (after making one already) hangs unless you manually remove the wic build directory. It appears that mcopy is left running at 100% cpu. Reverting 2d3c9f89852bd009fc7accee495ede2fbcadc3bb wic: bootimg-partition: Stop removing whole workdir Resolves the problem. Alexandre discussed this briefly on the list, but it appears the resolution was dropped.
Can you show me wic commandline? I can't reproduce this with "wic create directdisk -e core-image-minimal"
wic create ../meta-sdr/contrib/wks/sdimage-8G.wks -e gnuradio-demo-image -o images/$MACHINE Something like that. Also check with A Belloni, I'm pretty sure he has an idea. His contact info should be i the commit log.
Works just fine for me: $ time wic create ../../meta-sdr/contrib/wks/sdimage-8G.wks -e gnuradio-demo-image -o images/$MACHINE Checking basic build environment... Done. Creating image(s)... Info: The new image(s) can be found here: images/build/sdimage-8G-201504170100-mmcblk.direct The following build artifacts were used to create the image(s): ROOTFS_DIR: /home/ed/git/yocto/poky/build-gnuradio-demo-image/tmp/work/qemux86-poky-linux/gnuradio-demo-image/1.0-r0/rootfs BOOTIMG_DIR: KERNEL_DIR: /home/ed/git/yocto/poky/build-gnuradio-demo-image/tmp/deploy/images/qemux86 NATIVE_SYSROOT: /home/ed/git/yocto/poky/build-gnuradio-demo-image/tmp/sysroots/x86_64-linux The image(s) were created using OE kickstart file: ../../meta-sdr/contrib/wks/sdimage-8G.wks real 1m22.049s user 0m16.440s sys 0m40.332s $ ls -lh images/build/sdimage-8G-201504170100-mmcblk.direct -rw-r--r-- 1 ed users 7.0G Apr 17 01:01 images/build/sdimage-8G-201504170100-mmcblk.direct Please, provide more info. Wild guess: the reason could be caused by mcopy from your distro and was fixed by fa263f238bbddb00c9953994fb69cc358170e2ec wic: don't use host paths when looking for native tools Can you try if it's still reproduceable with current master?
Current master fails to run at all. I filed a bug for this. Did you run the command twice?
current master worked for me. yes, I run wic twice.
you should be able to run wic after applying this fix: http://lists.openembedded.org/pipermail/openembedded-core/2015-April/103837.html
Any news? Is it still reproducible?
This should be fixed with: http://lists.openembedded.org/pipermail/openembedded-core/2015-April/104199.html
OK, finally got a master build to where I could check. master looks OK now. I need to double check, but I think I need to make a backport request for fido to fix the issue there.
Did you need my patch or just plain master is enough?
Alexandre, there was a suggestion from Adrian to use API calls instead of shelling out "rm -f". Would you be willing to follow it? http://lists.openembedded.org/pipermail/openembedded-core/2015-April/104230.html
It looks like plain master is enough.
Ed, Does this still needinfo? It looks like Phlip answered.
Fixed in master.
Failed to verify because of bug 8309