image.bbclass: Separate out image generation into a new task, do_image
PATCH 1/4 image.bbclass: Separate out image generation into a new task, do_image
I've heard complaints from people trying to create more interesting image types about how hard it is to understand the rootfs/image generation code and that its a pain to develop/test/debug.
Having looked at it myself, the internal construction of shell functions which then gets passed into a multiprocessing pool is rather convoluted and it places rather odd constraints on when variables are expanded. Its therefore no wonder people find it confusing/complex.
This patch starts the process of splitting this up by separating out image generation from the do_rootfs task into a new do_image task.
meta/classes/image.bbclass +10 -4
@@ -207,7 +207,6 @@ PACKAGE_EXCLUDE[type] = "list" fakeroot python do_rootfs () { from oe.rootfs import create_rootfs - from oe.image import create_image from oe.manifest import create_manifest # Handle package exclusions
@@ -244,9 +243,6 @@ fakeroot python do_rootfs () { # Generate rootfs create_rootfs(d) - - # generate final images - create_image(d) } do_rootfs[dirs] = "${TOPDIR}" do_rootfs[lockfiles] += "${IMAGE_ROOTFS}.lock"
@@ -254,6 +250,16 @@ do_rootfs[cleandirs] += "${S}" do_rootfs[umask] = "022" addtask rootfs before do_build +fakeroot python do_image () { + from oe.image import create_image + + # generate final images + create_image(d) +} +do_image[dirs] = "${TOPDIR}" +do_image[umask] = "022" +addtask do_image after do_rootfs before do_build + MULTILIBRE_ALLOW_REP =. "${base_bindir}|${base_sbindir}|${bindir}|${sbindir}|${libexecdir}|${sysconfdir}|${nonarch_base_libdir}/udev|/lib/modules/[^/]*/modules.*|" MULTILIB_CHECK_FILE = "${WORKDIR}/multilib_check.py" MULTILIB_TEMP_ROOTFS = "${WORKDIR}/multilib"
Hi Richard,
I have issues building a core-image-minimal Raspberry Pi 2 SD image after this series of patches using openembedded-core master (b3b679d5be86f73d1a06c7230cb00872f0a407b5) and meta-raspberrypi master (0d8e9271855ccb9bc21bdc0743f8e0f6ff0d2d5c).
| DEBUG: Executing python function set_image_size | DEBUG: Python function set_image_size finished | DEBUG: Executing shell function do_image_rpi-sdimg | /home/build/git/pi2/build/tmp/work/raspberrypi2-custom-linux-gnueabi/core-image-minimal/1.0-r0/temp/run.do_image_rpi-sdimg.8349: line 108: do_image_rpi-sdimg: command not found | WARNING: /home/build/git/pi2/build/tmp/work/raspberrypi2-custom-linux-gnueabi/core-image-minimal/1.0-r0/temp/run.do_image_rpi-sdimg.8349:1 exit 127 from | do_image_rpi-sdimg | ERROR: Function failed: do_image_rpi-sdimg (log file is located at /home/build/git/pi2/build/tmp/work/raspberrypi2-custom-linux-gnueabi/core-image-minimal/1.0-r0/temp/log.do_image_rpi-sdimg.8349)
openembedded-core 38237b7ac53c416f85c4a70a61acafc3404c8b5f - ok openembedded-core b3b679d5be86f73d1a06c7230cb00872f0a407b5 - fails
Any ideas?
Regards,
Jonathan
Quoted 55 lines · Richard Purdie
I've heard complaints from people trying to create more interesting image types about how hard it is to understand the rootfs/image generation code and that its a pain to develop/test/debug.
Having looked at it myself, the internal construction of shell functions which then gets passed into a multiprocessing pool is rather convoluted and it places rather odd constraints on when variables are expanded. Its therefore no wonder people find it confusing/complex.
This patch starts the process of splitting this up by separating out image generation from the
do_rootfstask into a newdo_imagetask.Signed-off-by Richard Purdie <[email protected]>meta/classes/image.bbclass +10
-4@@ -207,7 +207,6 @@ PACKAGE_EXCLUDE[type] = "list" fakeroot python do_rootfs () { from oe.rootfs import create_rootfs - from oe.image import create_image from oe.manifest import create_manifest # Handle package exclusions@@ -244,9 +243,6 @@ fakeroot python do_rootfs () { # Generate rootfs create_rootfs(d) - - # generate final images - create_image(d) } do_rootfs[dirs] = "${TOPDIR}" do_rootfs[lockfiles] += "${IMAGE_ROOTFS}.lock"@@ -254,6 +250,16 @@ do_rootfs[cleandirs] += "${S}" do_rootfs[umask] = "022" addtask rootfs before do_build +fakeroot python do_image () { + from oe.image import create_image + + # generate final images + create_image(d) +} +do_image[dirs] = "${TOPDIR}" +do_image[umask] = "022" +addtask do_image after do_rootfs before do_build + MULTILIBRE_ALLOW_REP =. "${base_bindir}|${base_sbindir}|${bindir}|${sbindir}|${libexecdir}|${sysconfdir}|${nonarch_base_libdir}/udev|/lib/modules/[^/]*/modules.*|" MULTILIB_CHECK_FILE = "${WORKDIR}/multilib_check.py" MULTILIB_TEMP_ROOTFS = "${WORKDIR}/multilib"
Quoted 19 lines · Jonathan Liu
I have issues building a core-image-minimal Raspberry Pi 2 SD image after this series of patches using openembedded-core master (
b3b679d5be86f73d1a06c7230cb00872f0a407b5) and meta-raspberrypi master
(0d8e9271855ccb9bc21bdc0743f8e0f6ff0d2d5c).Quoted 4 lines
DEBUG: Executing python function
set_image_size
DEBUG: Python functionset_image_sizefinished
DEBUG: Executing shell function do_image_rpi-sdimg
/home/build/git/pi2/build/tmp/work/raspberrypi2-custom-linux-gnueabi/core-image-minimal/1.0-r0/temp/run.do_image_rpi-sdimg.8349:
line 108: do_image_rpi-sdimg: command not foundQuoted 1 line
WARNING:
/home/build/git/pi2/build/tmp/work/raspberrypi2-custom-linux-gnueabi/core-image-minimal/1.0-r0/temp/run.do_image_rpi-sdimg.8349:1 exit 127 fromQuoted 2 lines
do_image_rpi-sdimg
ERROR: Function failed: do_image_rpi-sdimg (log file is located at
Quoted 7 lines
/home/build/git/pi2/build/tmp/work/raspberrypi2-custom-linux-gnueabi/core-image-minimal/1.0-r0/temp/log.do_image_rpi-sdimg.8349)openembedded-core
38237b7ac53c416f85c4a70a61acafc3404c8b5f- ok openembedded-coreb3b679d5be86f73d1a06c7230cb00872f0a407b5- failsAny ideas?
Try the patch I just sent to the OE-Core list, "Handle image types containing '-' correctly". Looks like image types containing the "-" character would cause problems but I think there is a workaround.
Cheers,
Richard