| Age | Commit message (Collapse) | Author | Files |
|---|
|
Currently, in order to figure out variable values when run within the
eSDK, runqemu does not use the standard SDK method nor is it able to run
bitbake (since the eSDK environment isn't initialised like the normal
OE build environment). runqemu really ought to be fixed, but the quick
workaround is to set DEPLOY_DIR_IMAGE in the environment so that runqemu
can find image files.
Fixes [YOCTO #10447].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If CMAKE_SYSTEM_NAME is defined, CMake assumes we're cross-compiling,
which is not necessarily the case.
Signed-off-by: Kyle Russell <bkylerussell@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
The name of postinst scripts created by pixbufcache class
contains "useradd" in it. Remove it to avoid confusion.
As suggested by RP.
Signed-off-by: Maxin B. John <maxin.john@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Without this, eSDK builds are failing due to qemu-helper-native's dependency on this
task. It makes sense to allow this to execute in eSDK contexts (its a non-sstate task
intentionally).
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
populate_sdk_ext sets TMPDIR to a known static value with '/tmp' directory
name and that name is hard coded in a few places (e.g., in
meta-environment-extsdk.bb that writes the eSDK environment variables).
Distros that do not reset TCLIBCAPPEND (poky does) end up getting
TMPDIR = /tmp-${TCLIBCAPPEND} via defaultsetup.conf and that breaks
the functionality in eSDK that expects everything is in /tmp.
To get TMPDIR consistent, we also need to reset TCLIBCAPPEND in
populate_sdk_ext.bbclass.
Fixes: [YOCTO #11298]
Signed-off-by: Mikko Ylinen <mikko.ylinen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
It should be -IOEINCDIR, not -I-IOEINCDIR.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Currently there are multiple issues with useradd:
* If base-passwd rebuilds, it wipes out recipe specific user/group additions
to sysroots and causes errors
* If recipe A adds a user and recipe B depends on A, it can't see any of the
users/groups A adds.
This patch changes base-passwd so it always works as a postinst script
within the sysroot and copies in the master files, then runs any
postinst-useradd-* scripts afterwards to add additional user/groups.
The postinst-useradd-* scripts are tweaked so that if /etc/passwd doesn't exist
they just exit, knowning they'll be executed later. We also add a dummy entry to
the dummy passwd file from pseudo so we can avoid this too.
There is a problem where if recipe A adds a user and recipe B depends on A but
doesn't care about users, it may not have a dependency on the useradd/groupadd
tools which would therefore not be available in B's sysroot. We therefore also
tweak postinst-useradd-* scripts so that if the tools aren't present we simply
don't add users. If you need the users, you add a dependency on the tools in the
recipe and they'll be added.
We add postinst-* to SSTATE_SCAN_FILES since almost any postinst script of this
kind is going to need relocation help.
We also ensure that the postinst-useradd script is written into the sstate
object as the current script was only being added in a recipe local way.
Thanks to Peter Kjellerstedt <pkj@axis.com> and Patrick Ohly for some pieces
of this patch.
[Yocto #11124]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Commit 51edde653707e7a3cd2186082458f01f32cd1996 makes a wrong assumption
that SSTATE_MIRRORS have write permissions.
A mirror is by definition outside of it's user control. In my use case
it happens I does not have permissions to update the access time of the
dereferenced symbolic-link file.
Checked if file is writable before changing its atime.
Thanks to Paulo Neves for the patch.
[YOCTO #11307]
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Native sysroot of qemu-helper contains all required tools
(qemu-system and tunctl atm) for runqemu to work. It's not
removed by rm_dir and should always exist. It makes sense
to write it into qemuboot.conf to make runqemu to use it
as a default directory for native tools.
This should also speed up runqemu as it doesn't need to run
to run 'bitbake qemu-helper -e' to get its native sysroot.
[YOCTO #11266]
[YOCTO #11193]
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
On the do_uboot_mkimage task from kernel-uimage.bbclass, in case
KEEPUIMAGE is different than the default "yes" value, the uboot-mkimage
command fails because the path of the created uImage does not exist.
On this task, we are under the BUILDDIR so there is no folder arch/<ARCH>/boot.
Add the ${B} (for kernel build directory) as prefix to this folder fixes the problem.
Signed-off-by: Mylène Josserand <mylene.josserand@free-electrons.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
* Fixed:
The netmap has been blacklisted in
meta-networking/recipes-kernel/netmap/netmap_git.bb, but lib32-netmap still can
be built (suppose it doesn't depend on another broken recipe netmap-modules, it
is a little complicated, will talk below):
$ bitbake lib32-netmap
This is because of the old code masks on bb.event.ConfigParsed which can only
handle global blacklist, netmap sets blacklist in the recipe, so it can't be
handled, and lib32-netmap can be built. which was incorrect:
blacklist_multilib_eventhandler[eventmask] = "bb.event.ConfigParsed"
Move multilib code into multilib.bbclass can fix the problem easily:
$ bitbake lib32-netmap
ERROR: Nothing PROVIDES 'lib32-netmap'
ERROR: lib32-netmap was skipped: Recipe is blacklisted: BROKEN: <foo>
* Not fixed
Another problem is netmap-modules has also been blacklisted in the recipe, and
the recipe inherits module.bbclass, so multilib.bbclass doesn't handle it as the
code shows:
# There should only be one kernel in multilib configs
# We also skip multilib setup for module packages.
provides = (e.data.getVar("PROVIDES") or "").split()
if "virtual/kernel" in provides or bb.data.inherits_class('module-base', e.data):
raise bb.parse.SkipPackage("We shouldn't have multilib variants for the kernel")
And netmap-modules provides lib32-netmap-modules which is handled in
multilib_global.bbclass, so bitbake lib32-netmap-modules can't show
the blacklist message:
$ bitbake netmap-modules
ERROR: Nothing PROVIDES 'netmap-modules'
ERROR: netmap-modules was skipped: Recipe is blacklisted: BROKEN: <foo>
ERROR: netmap-modules was skipped: We shouldn't have multilib variants for the kernel
$ bitbake lib32-netmap-modules
ERROR: Nothing PROVIDES 'lib32-netmap-modules'. Close matches:
netmap-modules
netmap-modules
lib32-fbset-modes
Note the different messages between netmap-modules and lib32-netmap-modules.
This is because multilib.bbclass doesn't handle the "module" recipe so
there is no PN called lib32-netmap-modules, therefore blacklist.bbclass can't
handle it.
Note, there are two "netmap-modules" which needs to be fixed later.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
There seems to be little advantage to letting distro features affect
native builds. There is a significant disadvantage: a change to
DISTRO_FEATURES will trigger a lot of unnecessary native tasks. In a
test like this:
$ bitbake core-image-minimal
# append " systemd" to DISTRO_FEATURES
$ bitbake core-image-minimal
The latter build takes 44 minutes (28%) of cpu-time less with this
patch (skipping 135 native tasks). Sadly wall clock time was not
affected as glibc remains the bottleneck.
Set native distro features to DISTRO_FEATURES_NATIVE appended with
an intersection of DISTRO_FEATURES and DISTRO_FEATURES_FILTER_NATIVE.
Current default values (baitbake.conf) are
* DISTRO_FEATURES_FILTER_NATIVE ?= "api-documentation" (as gtk-doc-native
has much less dependencies when built without it)
* DISTRO_FEATURES_NATIVE ?= "x11" (to enable native UIs even if target
does not containe them)
Do the variable setting in native_virtclass_handler() because otherwise
it could still be overridden by appends and the feature backfilling.
Shuffle the early returns so DISTRO_FEATURES gets set as long as
the packagename ends with "-native".
Add similar variables for nativesdk.
To make nativesdk work we need to enable the locale options so
nativesdk-glibc-locales can build and to avoid the init manager check
in the nativesdk case so add those fixes.
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Fixed when the image is large and not enough memory:
grep: memory exhausted
Aborted
[YOCTO #11073]
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Dependencies on svn:// urls were failing as the cleandirs on do_fetch was destroying
any sysroot setup by the extend_recipe_sysroot function.
Add code so that if the task do_fetch, we move the cleandirs to a separate function before
the extend_recipe_sysroot prefunc else we'd wipe out the sysroot we just created.
This allows fetcher do_populate_sysroot dependencies to work correctly again.
I did try various other approaches and a seperate function with cleandirs was the
cleanest way to add this without code duplication or too much complexity.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Use the TaskArtifacts event to scan the SDK and ESDK manifests
to cleanly collect the respective artifact files.
The previous method was broken when the SDK file deployment moved
from the do_populate_sdk[_ext] tasks to their sstate tasks. That
method is disabled (but not yet removed) in preparation for the
rest of refactor work for the parent #10283 work.
[YOCTO #10850]
Signed-off-by: David Reyna <David.Reyna@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Save a file per task listing sha256sums for each file staged, i.e.
the output of the task. Some caveats:
1) This only covers sstate tasks since it uses SSTATEPOSTUNPACKFUNCS,
however those are generally the most interesting in terms of output
anyway.
2) The signature is taken before applying any relocations, so any
relocated files will actually have different signatures, but that's
churn that you probably won't want to see here.
3) At the moment if you run the same build twice without sstate you will
very likely see changes in the output for certain tasks due to things
like timestamps being present in the binary output. Fixing that is
a general Linux ecosystem problem - see this page for our efforts to
resolve it on our side:
https://wiki.yoctoproject.org/wiki/Reproducible_Builds
NOTE: you need to set your BUILDHISTORY_FEATURES value to include
"task" to enable collection of these signatures as it is is disabled by
default.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Append to the value with appendVarFlag() instead of setting it outright,
so that we can also append to it in other places. Accordingly, this
varflag is pipe-separated (since we want to be able to exclude any
string fragment, in this case including the leading space), thus put a
leading pipe character to play nicely with any existing value.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If we want to determine what changed since the last build, one angle
from which to look at it is to check the signatures. However, if we
don't actually have the signatures from the last build we don't have
anywhere to start. Save the signatures on each build in order to give us
the starting point.
NOTE: you need to set your BUILDHISTORY_FEATURES value to include
"task" to enable collection of these signatures as it is is disabled by
default.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Enabling or disabling buildhistory caused a rebuild of images, which
is undesirable. For example, it prevented image reuse from a main
build with buildhistory in a following oe-selftest where buildhistory
must be disabled.
The reason are the additional ROOTFS_POSTUNINSTALL_COMMAND and
ROOTFS_POSTUNINSTALL_COMMAND entries. Those need to be excluded both
via vardepvalueexclude and vardepsexclude.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
do_fetch"
There seems to be an issue with the patch, revert for now.
This reverts commit cd5e0a32184d98beb8d81e6b5527166d3ca4fb3c.
|
|
Dependencies on svn:// urls were failing as the cleandirs on do_fetch was destroying
any sysroot setup by the extend_recipe_sysroot function.
Add code so that if the task do_fetch, we move the cleandirs to the extend_recipe_sysroot
task else we'd wipe out the sysroot we just created.
This allows fetcher do_populate_sysroot dependencies to work correctly again.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Some TARGET_* flags are being exported in bitbake.conf currently, so
they are impacting all the tasks of a cross-canadian recipe even they
are not in use at all.
This can lead a lot of churn when the cross-canadian sysroot are shared
by machines while they have defined different TARGET_* flags.
Fix it by overriding with BUILDSDK_* flags.
Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Some TARGET_* flags are being exported in bitbake.conf currently, so
they are impacting all the tasks of a nativesdk recipe even they are
not in use at all.
This can lead a lot of churn when the nativesdk sysroot are shared by
machines while they have defined different TARGET_* flags.
Fix it by overriding with BUILDSDK_* flags.
Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Some TARGET_* flags are being exported in bitbake.conf currently, so
they are impacting all the tasks of a crosssdk recipe even they are
not in use at all.
This can lead a lot of churn when the crosssdk sysroot are shared by
machines while they have defined different TARGET_* flags.
Fix it by overriding with BUILD_* flags.
Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Some TARGET_* flags are being exported in bitbake.conf currently, so
they are impacting all the tasks of a cross recipe even they are not
in use at all.
This can lead a lot of churn when the cross sysroot are shared by
machines while they have defined different TARGET_* flags.
And sometimes it even causes "Taskhash mismatch" errors.
Fix it by overriding with BUILD_* flags.
Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We never need these tasks as dependencies of other sstate tasks since
they're only ever needed to build artefacts so we can always skip them
and save some time/space.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Two identical builds can end up having deploy/licenses folders
that differ. This is observed in cases where there are several
different license files of the same name in different folders,
i.e. "COPYING". In those case we have to differentiate the files
somehow and we do it via file expensions such as COPYING.0, COPYING.1.
However, which file will get which extension is presently random.
This means, for example, that COPYING.0 in one build is the same
as COPYING.1 in the other (and vice versa).
Although there is mothing wrong with this, for the sake of binary
reproducibility it is preferable to have a deterministic outcome.
Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Emit CONFFILES variable in pkgdata, or else the get_conffiles function
will return 'None' for some packages instead of the expected value. This
is especially true for optional module packages.
Signed-off-by: David Vincent <freesilicon@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
opkg uses empty lines as separator for next package and if an ipk file was
packaged with empty lines in DESCRIPTION opkg won't be able to handle such ipk
file, this happens at execution time.
This commit will replace empty lines in DESCRIPTION with a '.' when generating
an ipk package to avoid this issue.
[YOCTO #10677]
Signed-off-by: Mariano Lopez <mariano.lopez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
linux-yocto.inc
Add kernel_link_images task in kernel.bbclass instead of adding it in
linux-yocto.inc, or else the recipes inheriting kernel.bbclass might
run into implicit dependency issues.
Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
FILESDIR is long dead and unreferenced pretty much anywhere now, drop these
sanity references too.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
It defaults to ${INITRAMFS_IMAGE}-${MACHINE} if INITRAMFS_IMAGE is not
empty.
This allows the end users to be able to override the initramfs image
name with a customized value.
Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
In a addtask statement, do_strip should be strip.
Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This change is similar to the recently introduced
change to the distutils.bbclass fixing shebang
line in python scripts for nativesdk class builds.
v2: Rebased on top of new head.
Cc: XE-Linux <xe-linux-external@cisco.com>
Signed-off-by: Serhii Popovych <spopovyc@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Commit 11229de (distutils: Replacing path to native python by path
to python in the image to support python packages with
console-script setup) replaces path to python-native with path
to the python interpreter in the target image.
However on nativesdk builds ${bindir} expanded to the
${SDKPATHNATIVE}${prefix_nativesdk} making shebang line
pointing to the absolute path to env(1) in SDKPATHNATIVE
which may not be present if coreutils isn't part of nativesdk.
On the other hand commit cf63d90 (bitbake.conf: Define USRBINPATH
globally instead of individually) introduces USRBINPATH variable
which has correct value regarding build class and intended for
this use case.
v2: Rebased on top of new head.
Cc: XE-Linux <xe-linux-external@cisco.com>
Signed-off-by: Serhii Popovych <spopovyc@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Use the new executable argument to bb.utils.which() to ensure that the symlinks
point to executable files and not for example directories with the right name
which happened to be on $PATH.
[ YOCTO #11256 ]
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Having arm 32/64 bit headers coexisting turns out to be tricky. Unfortunately
our wrapper works using wordsize.h and this differs on arm so we can't use it.
Therefore replicate the logic here for arm. I did look into writing our
own wordsize.h but we also need to remap kernel headers on arm and
since wordsize.h comes from libc, that doesn't work for kernel headers.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
I suspect this was a typo and that TARGET_SYS makes more sense here. Its
also the only remaining user of MULTIMACH_HOST_SYS in OE-Core. Change it.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When we get to version 12 we have a hard break as we can't convert to newer
versions. There is no point in running the old conversions on an old tmpdir
only to hit that block. Remove all the old conversions to avoid that and
make things clearer.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
"bitbake -c populate_dsk_ext" must not trigger do_rm_work, because it
is impossible to declare that the additional tasks activated by "-c
populate_dsk_ext" must run before do_rm_work. When do_populate_dsk_ext
and do_rm_work are both active, the resulting race condition breaks
do_populate_dsk_ext.
The existing bitbake dependencies can't be used for that, because
"addtask populate_dsk_ext before do_rm_work" would then always execute
populate_dsk_ext also in normal builds.
do_populate_dsk_ext triggers do_rm_work indirectly through the
dependency on do_build of the SDK_TARGETs. Using the new
do_build_without_rm_work instead (when available, with do_build as
before if not) avoids the problem.
However, one has to be careful to not trigger do_rm_work in the same
build in some other way. "bitbake core-image-sato:do_populate_sdk_ext
core-image-sato:do_build" still fails, for example. Doing one after
the other works.
Fixes: [YOCTO 11042]
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Some classes, for example populate_sdk_ext.bbclass, must be able to
trigger a full build of a recipe without also triggering
do_rm_work. They cannot depend on do_build anymore, because that would
trigger do_rm_work.
Instead, do_build_without_rm_work can be used. It has the exact same
dependencies as do_build, minus do_rm_work and do_rm_work_all.
This may also be useful in a test build of a recipe where one wants
to preserve the work directory without having to modify configuration
settings:
bitbake foobar:do_build_without_rm_work
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This aligns with the behavior of grub-efi and gummiboot, and is needed to fix
auto-boot of intel-corei7-64 for non-GPLv3 builds.
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Previously, QEMU_USE_KVM=True enabled the use of kvm only when "x86"
was in the MACHINE name. That is too limiting, because for example
intel-corei7-64 can also use kvm but it wasn't possible to enable that
without changing OE-core.
That traditional usage is still supported. In addition, QEMU_USE_KVM
can be set to a list of space-separated MACHINE names for which kvm is
to be enabled.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
[YOCTO #11180]
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Missing runtime dependencies should result in an error rather than a
warning. Indeed, if RPM is listed in PACKAGE_CLASSES, it will throw
an error rather than install packages with missing dependencies. This
functionality should be consistent across package types. This patch
ensures that an error will be thrown.
[YOCTO #10949]
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
In entire meta/classes/ directory, replace shell tests of the form
"if test -? ..." with POSIX tests of the form "if [ -? ...
Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
For aesthetic style reasons, use "grep -q" instead of ">/dev/null".
Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When performing a file system check, the image created with mkfs will
trigger Pass 3A ('Optimizing directories') which turns the file system
into state "changed" (EXT2_FLAG_CHANGED).
This will let fsck request a reboot by setting the return code flag "2".
The result of this is that each ext-image built with oe-core will
trigger a reboot during the first time an fsck is triggered.
A common case where this might occur is when fsck detects having
a future superblock write time. This always happens when booting a
newly created ext4 rootfs with a target that does not have a recent time
set.
This patch moves the initial fsck run that performs the optimization
from the target to the host system and thus prevents the target from
performing an avoidable reboot.
Signed-off-by: Enrico Jorns <ejo@pengutronix.de>
Signed-off-by: Ulrich Ölmann <uol@pengutronix.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
