i#10073: generic EFI for wic
v4 cover i#10073: generic EFI for wic
Hi,
This patchset is an implementation of generic EFI approach for wic images.
Instead of introducing yet another wic plugin it uses existing APIs from EFI_PROVIDER classes to populate ${WORKDIR}/bootfs directory with EFI artifacts and bootloader configuration. This directory can be used by wic rootfs plugin to put into boot partition of the image.
Example kickstart file and 2 test cases were added to wic test suite to better illustrate the approach.
I personally like this approach much better than duplicating oe image creation functionality in wic plugins. This way we can have the code in one place and bootloaders can be configured the same way for oe and wic images.
Changes in v2:
- added patch: fixed default value of GRUB_ROOT
- fixed typo in commit message: timage_types_wic > image_types_wic
Changes in v3:
- scheduled do_prepare_wic_build as earlier as possible
Changes in v4:
- fixed build on platforms that don't support EFI
- fixed non-gpl3 build
- fixed build failure in systemd-boot code caused by rescheduled do_populate_bootfs to run earlier
The following changes since commit 920e1592fba8fd714b8fdf87f9792ebfa3377793:
selftest: fix test_unsupported_subcommand test case (2017-05-17 13:35:28 +0000)
are available in the git repository at:
git://git.yoctoproject.org/poky-contrib ed/wic/wip
http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=ed/wic/wip
Ed Bartosh (7):
systemd-boot: create output dir if it doesn't exist
efi: add efi_bootfs_populate API
image_types_wic: add do_populate_bootfs task
image_types_wic: merged 2 tasks
oe-selftest: add wic tests for generic EFI
grub-efi: fixed default value of GRUB_ROOT
image_types_wic: schedule prepare_wic_build correctly
meta-selftest/wic/test_generic_efi.wks.in | 9 ++++ meta/classes/grub-efi.bbclass | 14 +++-- meta/classes/image_types_wic.bbclass | 85 +++++++++++++++++++++++-------- meta/classes/systemd-boot.bbclass | 8 +++ meta/lib/oeqa/selftest/wic.py | 36 +++++++++++++ 5 files changed, 126 insertions(+), 26 deletions(-) create mode 100644 meta-selftest/wic/test_generic_efi.wks.in
Signature · 2 lines
Regards, Ed
PATCH v4 1/7 systemd-boot: create output dir if it doesn't exist
build_efi_cfg function creates configuration files for systemd-boot entries in 'S' directory. This directory may not exist when api is called, which breaks the build.
Creating the directory if it doesn't exist should fix this issue.
--- meta/classes/systemd-boot.bbclass | 2 ++ 1 file changed, 2 insertions(+)
meta/classes/systemd-boot.bbclass +2 -0
@@ -99,6 +99,8 @@ python build_efi_cfg() { bb.fatal('OVERRIDES not defined') entryfile = "%s/%s.conf" % (s, label) + if not os.path.exists(s): + os.makedirs(s) d.appendVar("SYSTEMD_BOOT_ENTRIES", " " + entryfile) try: entrycfg = open(entryfile, "w")
Signature · 1 line
2.12.0
PATCH v4 2/7 efi: add efi_bootfs_populate API
In reply to Ed BartoshAdded API to populate ${WORKDIR}/bootfs directory with EFI artifacts to both EFI provider classes(grub-efi and systemd-boot).
This API will be used to prepare artifacts for the wic image build.
[YOCTO #10073]
--- meta/classes/grub-efi.bbclass | 6 ++++++ meta/classes/systemd-boot.bbclass | 6 ++++++ 2 files changed, 12 insertions(+)
meta/classes/grub-efi.bbclass +6 -0
@@ -71,6 +71,12 @@ efi_hddimg_populate() { efi_populate $1 } +efi_bootfs_populate() { + bootfs_dir="${WORKDIR}/bootfs" + efi_populate $bootfs_dir + cp --dereference ${DEPLOY_DIR_IMAGE}/${KERNEL_IMAGETYPE} $bootfs_dir/vmlinuz +} + python build_efi_cfg() { import sys
meta/classes/systemd-boot.bbclass +6 -0
@@ -62,6 +62,12 @@ efi_hddimg_populate() { efi_populate $1 } +efi_bootfs_populate() { + bootfs_dir="${WORKDIR}/bootfs" + efi_populate $bootfs_dir + cp --dereference ${DEPLOY_DIR_IMAGE}/${KERNEL_IMAGETYPE} $bootfs_dir/vmlinuz +} + python build_efi_cfg() { s = d.getVar("S") labels = d.getVar('LABELS')
Signature · 1 line
2.12.0
PATCH v4 3/7 image_types_wic: add do_populate_bootfs task
In reply to Ed BartoshThis task generates root partition UUID and calls efi_bootfs_populate API of current EFI provider to populate bootfs directory with EFI artifacts.
[YOCTO #10073]
--- meta/classes/image_types_wic.bbclass | 37 ++++++++++++++++++++++++++++++++++++ 1 file changed, 37 insertions(+)
meta/classes/image_types_wic.bbclass +37 -0
@@ -115,3 +115,40 @@ python do_rootfs_wicenv () { addtask do_rootfs_wicenv after do_image before do_image_wic do_rootfs_wicenv[vardeps] += "${WICVARS}" do_rootfs_wicenv[prefuncs] = 'set_image_size' + +# Set variables for efi_bootfs_populate +GRUB_CFG ?= "${WORKDIR}/grub_wic.cfg" +SYSTEMD_BOOT_CFG ?= "${WORKDIR}/loader_wic.conf" +GRUB_GFXSERIAL ?= "1" +LABELS_WIC ?= "boot install" +LABELS ?= "${LABELS_WIC}" + +EFI_PROVIDER ?= "systemd-boot" +EFI_CLASS = "${@bb.utils.contains("MACHINE_FEATURES", "efi", "${EFI_PROVIDER}", "", d)}" +inherit ${EFI_CLASS} + +python do_populate_bootfs() { + def populate_bootfs(partuuid): + # remove bootfs dir as it may have files from previous build + bootfs = os.path.join(d.getVar("WORKDIR"), 'bootfs') + if os.path.exists(bootfs): + import shutil + shutil.rmtree(bootfs) + + d.setVar("APPEND", "root=PARTUUID=%s" % partuuid) + bb.build.exec_func('build_efi_cfg', d) + + bb.build.exec_func('efi_bootfs_populate', d) + + if d.getVar('USING_WIC'): + # Generate parition UUID + from uuid import uuid4 + partuuid = str(uuid4()) + d.setVar("ROOTFS_PARTUUID", partuuid) + + if d.getVar("EFI_CLASS"): + populate_bootfs(partuuid) +} + +addtask do_populate_bootfs before do_image_wic +do_populate_bootfs[depends] = "${MLPREFIX}${EFI_PROVIDER}:do_deploy virtual/kernel:do_deploy"
Signature · 1 line
2.12.0
PATCH v4 4/7 image_types_wic: merged 2 tasks
In reply to Ed BartoshMerged do_write_wks_template and do_populate_bootfs into new task do_prepare_wic_build to be able to write the same partition UUID into bootloader configuration files and kickstart file.
[YOCTO #10073]
--- meta/classes/image_types_wic.bbclass | 53 +++++++++++++++++++----------------- 1 file changed, 28 insertions(+), 25 deletions(-)
meta/classes/image_types_wic.bbclass +28 -25
@@ -43,27 +43,6 @@ do_image_wic[depends] += "wic-tools:do_populate_sysroot" WKS_FILE_DEPENDS ??= '' DEPENDS += "${@ '${WKS_FILE_DEPENDS}' if d.getVar('USING_WIC') else '' }" -python do_write_wks_template () { - """Write out expanded template contents to WKS_FULL_PATH.""" - import re - - template_body = d.getVar('_WKS_TEMPLATE') - - # Remove any remnant variable references left behind by the expansion - # due to undefined variables - expand_var_regexp = re.compile(r"\${[^{}@\n\t :]+}") - while True: - new_body = re.sub(expand_var_regexp, '', template_body) - if new_body == template_body: - break - else: - template_body = new_body - - wks_file = d.getVar('WKS_FULL_PATH') - with open(wks_file, 'w') as f: - f.write(template_body) -} - python () { if d.getVar('USING_WIC'): wks_file_u = d.getVar('WKS_FULL_PATH', False)
@@ -90,7 +69,6 @@ python () { # file in process_wks_template as well, so just put it in # a variable and let the metadata deal with the deps. d.setVar('_WKS_TEMPLATE', body) - bb.build.addtask('do_write_wks_template', 'do_image_wic', None, d) } #
@@ -127,7 +105,11 @@ EFI_PROVIDER ?= "systemd-boot" EFI_CLASS = "${@bb.utils.contains("MACHINE_FEATURES", "efi", "${EFI_PROVIDER}", "", d)}" inherit ${EFI_CLASS} -python do_populate_bootfs() { +python do_prepare_wic_build() { + # Prepare required artifacts for the wic image build: + # - Populate {WORKDIR}/bootfs directory with EFI content + # - Write wks.in template into the .wks file + def populate_bootfs(partuuid): # remove bootfs dir as it may have files from previous build bootfs = os.path.join(d.getVar("WORKDIR"), 'bootfs')
@@ -140,6 +122,23 @@ python do_populate_bootfs() { bb.build.exec_func('efi_bootfs_populate', d) + def write_wks_template(template_body, wks_file): + """Write out expanded template contents to WKS_FULL_PATH.""" + import re + + # Remove any remnant variable references left behind by the expansion + # due to undefined variables + expand_var_regexp = re.compile(r"\${[^{}@\n\t :]+}") + while True: + new_body = re.sub(expand_var_regexp, '', template_body) + if new_body == template_body: + break + else: + template_body = new_body + + with open(wks_file, 'w') as f: + f.write(template_body) + if d.getVar('USING_WIC'): # Generate parition UUID from uuid import uuid4
@@ -148,7 +147,11 @@ python do_populate_bootfs() { if d.getVar("EFI_CLASS"): populate_bootfs(partuuid) + + template = d.getVar("_WKS_TEMPLATE") + if template: + write_wks_template(template, d.getVar('WKS_FULL_PATH')) } -addtask do_populate_bootfs before do_image_wic -do_populate_bootfs[depends] = "${MLPREFIX}${EFI_PROVIDER}:do_deploy virtual/kernel:do_deploy" +addtask do_prepare_wic_build before do_image_wic +do_prepare_wic_build[depends] = "${MLPREFIX}${EFI_PROVIDER}:do_deploy virtual/kernel:do_deploy"
Signature · 1 line
2.12.0
PATCH v4 5/7 oe-selftest: add wic tests for generic EFI
In reply to Ed BartoshAdded test_generic_efi_grub_qemu and test_generic_efi_systemd_qemu test cases and wks template to build and boot generic EFI images in qemu.
[YOCTO #10073]
--- meta-selftest/wic/test_generic_efi.wks.in | 9 ++++++++ meta/lib/oeqa/selftest/wic.py | 36 +++++++++++++++++++++++++++++++ 2 files changed, 45 insertions(+) create mode 100644 meta-selftest/wic/test_generic_efi.wks.in
meta-selftest/wic/test_generic_efi.wks.in new file +9 -0
@@ -0,0 +1,9 @@ +# short-description: Create an EFI disk image +# long-description: Creates a partitioned EFI disk image that the user +# can directly dd to boot media. + +part /boot --source rootfs --rootfs-dir=${WORKDIR}/bootfs --fstype=vfat --label boot --active --align 1024 + +part / --source rootfs --fstype=ext4 --label root --align 1024 --uuid ${ROOTFS_PARTUUID} + +bootloader --ptable gpt
meta/lib/oeqa/selftest/wic.py +36 -0
@@ -789,3 +789,39 @@ part /etc --source rootfs --ondisk mmcblk0 --fstype=ext4 --exclude-path bin/ --r # 8 blocks is 4K (physical sector size) self.assertEqual(dest_stat.st_blocks, 8) os.unlink(dest) + + @only_for_arch(['i586', 'i686', 'x86_64']) + def test_generic_efi_grub_qemu(self): + """Test generic efi (grub-efi) image in qemu""" + config = 'IMAGE_FSTYPES = "wic"\nMACHINE_FEATURES_append = " efi"\n'\ + 'EFI_PROVIDER = "grub-efi"\n'\ + 'WKS_FILE = "test_generic_efi.wks.in"\n' + + self.append_config(config) + self.assertEqual(0, bitbake('core-image-minimal ovmf').status) + self.remove_config(config) + + with runqemu('core-image-minimal', ssh=False, + runqemuparams='ovmf', image_fstype='wic') as qemu: + cmd = "grep sda. /proc/partitions |wc -l" + status, output = qemu.run_serial(cmd) + self.assertEqual(1, status, 'Failed to run command "%s": %s' % (cmd, output)) + self.assertEqual(output, '2') + + @only_for_arch(['i586', 'i686', 'x86_64']) + def test_generic_efi_systemd_qemu(self): + """Test generic efi (systemd-boot) image in qemu""" + config = 'IMAGE_FSTYPES = "wic"\nMACHINE_FEATURES_append = " efi"\n'\ + 'EFI_PROVIDER = "systemd-boot"\n'\ + 'WKS_FILE = "test_generic_efi.wks.in"\n' + + self.append_config(config) + self.assertEqual(0, bitbake('core-image-minimal ovmf').status) + self.remove_config(config) + + with runqemu('core-image-minimal', ssh=False, + runqemuparams='ovmf', image_fstype='wic') as qemu: + cmd = "grep sda. /proc/partitions |wc -l" + status, output = qemu.run_serial(cmd) + self.assertEqual(1, status, 'Failed to run command "%s": %s' % (cmd, output)) + self.assertEqual(output, '2')
Signature · 1 line
2.12.0
PATCH v4 6/7 grub-efi: fixed default value of GRUB_ROOT
In reply to Ed BartoshCurrently GRUB_ROOT assigned a value of literal "${ROOT}" by default if ROOT variable is not set. This causes kernel commandline to look like this: "linux /vmlinuz LABEL=boot {ROOT}"
Used weak assignments of ROOT and GRUB_ROOT variables to make sure GRUB_ROOT is "" by default.
Allowed empty GRUB_ROOT to be able to use only APPEND variable for both EFI bootloaders (systemd-boot and grub-efi).
--- meta/classes/grub-efi.bbclass | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
meta/classes/grub-efi.bbclass +4 -4
@@ -27,7 +27,8 @@ GRUB_TIMEOUT ?= "10" GRUB_OPTS ?= "serial --unit=0 --speed=115200 --word=8 --parity=no --stop=1" EFIDIR = "/EFI/BOOT" -GRUB_ROOT ?= "${ROOT}" +ROOT ??= "" +GRUB_ROOT ??= "${ROOT}" APPEND ?= "" # Need UUID utility code.
@@ -121,8 +122,6 @@ python build_efi_cfg() { cfgfile.write('timeout=50\n') root = d.getVar('GRUB_ROOT') - if not root: - bb.fatal('GRUB_ROOT not defined') if gfxserial == "1": btypes = [ [ " graphics console", "" ],
@@ -146,7 +145,8 @@ python build_efi_cfg() { lb = "install-efi" cfgfile.write('linux /vmlinuz LABEL=%s' % (lb)) - cfgfile.write(' %s' % replace_rootfs_uuid(d, root)) + if root: + cfgfile.write(' %s' % replace_rootfs_uuid(d, root)) append = localdata.getVar('APPEND') initrd = localdata.getVar('INITRD')
Signature · 1 line
2.12.0
PATCH v4 7/7 image_types_wic: schedule prepare_wic_build correctly
In reply to Ed BartoshScheduled prepare_wic_build only if wic build enabled. Added dependencies to kernel and efi bootloader deploy tasks only if EFI is enabled.
This should fix build failure on machines without
EFI functionality.
--- meta/classes/image_types_wic.bbclass | 27 ++++++++++++++------------- 1 file changed, 14 insertions(+), 13 deletions(-)
meta/classes/image_types_wic.bbclass +14 -13
@@ -69,6 +69,11 @@ python () { # file in process_wks_template as well, so just put it in # a variable and let the metadata deal with the deps. d.setVar('_WKS_TEMPLATE', body) + + bb.build.addtask('do_prepare_wic_build', 'do_image_wic', None, d) + if d.getVar('EFI_CLASS'): + d.appendVarFlag('do_prepare_wic_build', 'depends', + '%s%s:do_deploy virtual/kernel:do_deploy' % (d.getVar('MLPREFIX'), d.getVar('EFI_CLASS'))) } #
@@ -139,19 +144,15 @@ python do_prepare_wic_build() { with open(wks_file, 'w') as f: f.write(template_body) - if d.getVar('USING_WIC'): - # Generate parition UUID - from uuid import uuid4 - partuuid = str(uuid4()) - d.setVar("ROOTFS_PARTUUID", partuuid) + # Generate parition UUID + from uuid import uuid4 + partuuid = str(uuid4()) + d.setVar("ROOTFS_PARTUUID", partuuid) - if d.getVar("EFI_CLASS"): - populate_bootfs(partuuid) + if d.getVar("EFI_CLASS"): + populate_bootfs(partuuid) - template = d.getVar("_WKS_TEMPLATE") - if template: - write_wks_template(template, d.getVar('WKS_FULL_PATH')) + template = d.getVar("_WKS_TEMPLATE") + if template: + write_wks_template(template, d.getVar('WKS_FULL_PATH')) } - -addtask do_prepare_wic_build before do_image_wic -do_prepare_wic_build[depends] = "${MLPREFIX}${EFI_PROVIDER}:do_deploy virtual/kernel:do_deploy"
Signature · 1 line
2.12.0
Re: [PATCH v4 7/7] image_types_wic: schedule prepare_wic_build correctly
Quoted 21 lines · Ed Bartosh
Scheduled
prepare_wic_buildonly if wic build enabled. Added dependencies to kernel and efi bootloader deploy tasks only if EFI is enabled.This should fix build failure on machines without
EFI functionality.Signed-off-by Ed Bartosh <[email protected]>---
meta/classes/image_types_wic.bbclass | 27 ++++++++++++++----------
---1 file changed, 14 insertions(+), 13 deletions(-)meta/classes/image_types_wic.bbclass +0
-0@@ -69,6 +69,11 @@ python () { # file in process_wks_template as well, so just putit in
Quoted 11 lines
# a variable and let the metadata deal with the
deps.
d.setVar('_WKS_TEMPLATE', body)
+
+ bb.build.addtask('do_prepare_wic_build', 'do_image_wic',
None, d)
+ if d.getVar('EFI_CLASS'):
+ d.appendVarFlag('do_prepare_wic_build', 'depends',
+ '%s%s:do_deploy
virtual/kernel:do_deploy' % (d.getVar('MLPREFIX'),
d.getVar('EFI_CLASS')))Ed,
Have you tested this with any layers? I tied recently with meta- intel and tripped over an issue with the rmc-boot not having an actual target for the EFI_CLASS to have a deploy task caused a failure.
Sau!
Quoted 36 lines
}
#meta/classes/image_types_wic.bbclass +0
-0@@ -139,19 +144,15 @@ python do_prepare_wic_build() {with
open(wks_file, 'w')as f:
f.write(template_body)
- ifd.getVar('USING_WIC'):
- # Generate parition UUID
- from uuid import uuid4
- partuuid = str(uuid4())
- d.setVar("ROOTFS_PARTUUID", partuuid)
+ # Generate parition UUID
+ from uuid import uuid4
+ partuuid = str(uuid4())
+ d.setVar("ROOTFS_PARTUUID", partuuid)
- ifd.getVar("EFI_CLASS"):
- populate_bootfs(partuuid)
+ ifd.getVar("EFI_CLASS"):
+ populate_bootfs(partuuid)
- template =d.getVar("_WKS_TEMPLATE")
- if template:
- write_wks_template(template,d.getVar('WKS_FULL_PATH')) + template =d.getVar("_WKS_TEMPLATE")
+ if template:
+ write_wks_template(template,d.getVar('WKS_FULL_PATH')) }meta/classes/image_types_wic.bbclass +0
-3- -addtask do_prepare_wic_build before do_image_wic -do_prepare_wic_build[depends] ="${MLPREFIX}${EFI_PROVIDER}:do_deploy virtual/kernel:do_deploy" --
2.12.0
Re: [PATCH v4 7/7] image_types_wic: schedule prepare_wic_build correctly
Quoted 28 lines · Saul
Quoted 27 lines · Ed Bartosh
Scheduled
prepare_wic_buildonly if wic build enabled. Added dependencies to kernel and efi bootloader deploy tasks only if EFI is enabled.This should fix build failure on machines without
EFI functionality.Signed-off-by Ed Bartosh <[email protected]>---
meta/classes/image_types_wic.bbclass | 27 ++++++++++++++----------
---1 file changed, 14 insertions(+), 13 deletions(-)meta/classes/image_types_wic.bbclass +0
-0@@ -69,6 +69,11 @@ python () { # file in process_wks_template as well, so just putit in
# a variable and let the metadata deal with the deps. d.setVar('_WKS_TEMPLATE', body) + + bb.build.addtask('do_prepare_wic_build', 'do_image_wic', None, d)
Quoted 8 lines
Quoted 5 lines
+ if
d.getVar('EFI_CLASS'):
+ d.appendVarFlag('do_prepare_wic_build', 'depends', + '%s%s:do_deploy virtual/kernel:do_deploy' % (d.getVar('MLPREFIX'),d.getVar('EFI_CLASS')))Ed,
Have you tested this with any layers?
I tested it with poky layers as I do for all my patches.
Quoted 3 lines
I tied recently with meta-
intel and tripped over an issue with the rmc-boot not having an actual target for theEFI_CLASSto have a deploy task caused a failure.
I also tested it on qemuarm machine that doesn't support EFI. It worked just fine.
I'll try to test it with meta-intel. Can you provide more detailed instructions how to reproduce this issue? Which image did you try to build?
--
Regards,
Ed
Re: [PATCH v4 7/7] image_types_wic: schedule prepare_wic_build correctly
Quoted 4 lines · Ed Bartosh
I'll try to test it with meta-intel. Can you provide more detailed instructions how to reproduce this issue? Which image did you try to build?
I just hit this.
image_types_wic does this:
+ d.appendVarFlag('do_prepare_wic_build', 'depends',
+ '%s%s:do_deploy virtual/kernel:do_deploy' %
(d.getVar('MLPREFIX'), d.getVar('EFI_CLASS')))
Where:
meta/classes/live-vm-common.bbclass:EFI_CLASS = "${@bb.utils.contains("MACHINE_FEATURES", "efi", "${EFI_PROVIDER}", "", d)}"
With meta-intel, EFI_PROVIDER is rmc-boot, which is a class but not a recipe.
Ross
Quoted 17 lines · Ed Bartosh
Hi,
This patchset is an implementation of generic EFI approach for wic images.
Instead of introducing yet another wic plugin it uses existing APIs from
EFI_PROVIDERclasses to populate ${WORKDIR}/bootfs directory with EFI artifacts
and bootloader configuration. This directory can be used by wic rootfs plugin
to put into boot partition of the image.Example kickstart file and 2 test cases were added to wic test suite to better
illustrate the approach.
Quoted 6 lines
I personally like this approach much better than duplicating oe image creation
functionality in wic plugins. This way we can have the code in one place and
bootloaders can be configured the same way for oe and wic images.
Ed,
I have been looking at this set of changes over the last few days, and while I like the basic approach, I don't think it goes far enough. I am looking at how one would create a generic bootfs via a DEPLOY_DIR and an image-bootfs.bb type of recipe that does the bootfs construction instead of the existing systemd-boot or grub-efi bbclass construction.
This will allow for more flexibility in creating labels for bootloaders, different types of boot loaders without having to create classes. We should then be able to construct the equivlant of hddimg or iso images via a recipe that depends on the various do_deploy() tasks of the images dependent items (syslinux, grub, virtual/kernel) and configuration.
I think we are still too locked into prepare_wic_build() calling build_efi_cfg and efi_bootfs_populate, which means wic still needs knowledge of the boot loaders instead of just having a recipe populate a bootfs partition and wic creating the partition and copying files into it.
Maybe I am missing something, so please correct me if I am wrong. I have an example of what I am thinking of in contrib/sgw/wic_wip
Thanks
Sau!
Quoted 46 lines
Changes in v2:
- added patch: fixed default value ofGRUB_ROOT
- fixed typo in commit message:timage_types_wic>image_types_wicChanges in v3:
- scheduleddo_prepare_wic_buildas earlier as possibleChanges in v4:
- fixed build on platforms that don't support EFI
- fixed non-gpl3 build
- fixed build failure in systemd-boot code caused by rescheduleddo_populate_bootfsto run earlierThe following changes since commit
920e1592fba8fd714b8fdf87f9792ebfa3377793:selftest: fix
test_unsupported_subcommandtest case (2017-05-17 13:35:28 +0000)are available in the git repository at:
git://git.yoctoproject.org/poky-contribed/wic/wip
http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=ed/wic/wipEd Bartosh (7):
systemd-boot: create output dir if it doesn't exist
efi: addefi_bootfs_populateAPI
image_types_wic: adddo_populate_bootfstask
image_types_wic: merged 2 tasks
oe-selftest: add wic tests for generic EFI
grub-efi: fixed default value ofGRUB_ROOT
image_types_wic: scheduleprepare_wic_buildcorrectlymeta-selftest/wic/test_generic_efi.wks.in | 9 ++++ meta/classes/grub-efi.bbclass | 14 +++-- meta/classes/image_types_wic.bbclass | 85
+++++++++++++++++++++++--------
meta/classes/systemd-boot.bbclass | 8 +++ meta/lib/oeqa/selftest/wic.py | 36 +++++++++++++5 files changed, 126 insertions(+), 26 deletions(-)create mode 100644 meta-selftest/wic/test_generic_efi.wks.in
--
Regards,
Ed
Quoted 33 lines · Saul
Quoted 17 lines · Ed Bartosh
Hi,
This patchset is an implementation of generic EFI approach for wic images.
Instead of introducing yet another wic plugin it uses existing APIs from
EFI_PROVIDERclasses to populate ${WORKDIR}/bootfs directory with EFI artifacts
and bootloader configuration. This directory can be used by wic rootfs plugin
to put into boot partition of the image.Example kickstart file and 2 test cases were added to wic test suite to better
illustrate the approach.Quoted 6 lines
I personally like this approach much better than duplicating oe image creation
functionality in wic plugins. This way we can have the code in one place and
bootloaders can be configured the same way for oe and wic images.Ed,
I have been looking at this set of changes over the last few days, and while I like the basic approach, I don't think it goes far enough. I am looking at how one would create a generic bootfs via a
DEPLOY_DIRand an image-bootfs.bb type of recipe that does the bootfs construction instead of the existing systemd-boot or grub-efi bbclass construction.
Quoted 6 lines
This will allow for more flexibility in creating labels for bootloaders, different types of boot loaders without having to create classes. We should then be able to construct the equivlant of hddimg or iso images via a recipe that depends on the various
do_deploy()tasks of the images dependent items (syslinux, grub, virtual/kernel) and configuration.
I like the idea in general. In practice it could be hard to synchronize variables between a recipe and .wks file
Obvious example of this would be generation of partition UUID and writing it into bootloader configs and .wks.in
I spent quite a bit of time on this. The only way I found to get them synchronized is to merge both tasks (generation of .wks file from .wks.in and generation of bootloader config) into one:
http://lists.openembedded.org/pipermail/openembedded-core/2017-May/136551.html
If we find a way to solve this then I'm all for your proposal.
Quoted 5 lines
I think we are still too locked into
prepare_wic_build()callingbuild_efi_cfgandefi_bootfs_populate, which means wic still needs knowledge of the boot loaders instead of just having a recipe populate a bootfs partition and wic creating the partition and copying files into it.
Well, I tend to disagree here. Yes, my code expects that build_efi_cfg and efi_bootfs_populate APIs are provided by bootloader class. However, it doesn't need to know any bootloader-specific internals. It calls those 2 APIs of current EFI provider class and expects fully populated bootloader artifacts in bootfs directory.
BTW, even if we use recipes to generate bootfs we'll most probably end up using some classes to avoid duplication of code in recipes. Those classes will probably contain less code than currently, but we'll need them unless we want to duplicate generation of bootloader configs and copying bootloader files in all bootf recipes.
Regards,
Ed