Age | Commit message (Collapse) | Author |
|
Replaces usage of the deprecated oe.utils.getstatusoutput() with Python
subprocess calls.
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
nativesdk-gpgme fails package_qa when setting PACKAGE_DEBUG_SPLIT_STYLE
= "debug-with-srcpkg".
ERROR: nativesdk-gpgme-1.10.0-r0 do_package_qa: QA Issue: non debug package contains .debug directory: nativesdk-python3-gpg path /work/x86_64-nativesdk-oesdk-linux/nativesdk-gpgme/1.10.0-r0/packages-split/nativesdk-python3-gpg/usr/local/oecore-x86_64/sysroots/x86_64-oesdk-linux/usr/lib/python3.5/site-packages/gpg/.debug/_gpgme.cpython-35m-x86_64-linux-gnu.so [debug-files]
This turns out to be because the automatic moving of the -dbg package to
the beginning of the package list is disabled in that case, so the
python3-gpg packages that the recipe prepends to PACKAGES ends up before
the -dbg package.
It's not clear why the "and not split_source_package" was added when
debug-with-srcpkg was introduced. Presumably the intention was to
prevent the -dbg package to end up before the -src package, which we of
course need to. But at the same time, we still need -dbg packages to end
up before all other packages.
Using list.insert(0, ...) also means that if there happens to more than
one -dbg package, their relative ordering gets inverted in the new list.
This tries to fix these issues by sorting the packages by (priority,
original position), where priority is 10 for -src, 30 for -dbg and 50
for everything else. That guarantees that packages of the same "type"
preserve their relative ordering, while also ensuring that -dbg always
preceed other packages. This scheme is also quite extensible, and,
should the need arise, one could even expose the priorities as a knob
the recipe author could use to ensure specific orderings of packages
instead of the somewhat fragile and coarse-grained method of "prepend or
append, and ensure you do that in a proper order".
Probably the autodebug condition needs to stay, but I think the
split_source_package condition in the preceding elif should be removed,
so that that logic applies to all packages called -src, not just the one
we might have created a few lines above.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We want to preserve sparse files when building the system, add the option
to tar to ensure we do this when copying files.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Recent changes broke the preservation of hardlinks during processing due to a missing index.
Fix this, reducing the size of the git recipe packages in particular (it contains many hardlinks).
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
of setting them
* this allows to work around some of the issues with file-rdeps
* e.g. in my case I have /usr/lib/libmali.so.0.1 which according to rpmdeps
provides only following libmali*
libmali.so(LIBMALI_1.0)
libmali.so.0.1
but many users of this link with libmali.so directly and according to
rpmdeps require libmali.so which causes file-rdeps for a lot of
recipes
* I was using simple work around to just set:
RPROVIDES_${PN} = "libmali.so libGLESv2.so libEGL.so"
but that doesn't work well with multilib, because the value gets
expanded with MLPREFIX and
RPROVIDES_${PN} = "lib32-libmali.so lib32-libGLESv2.so lib32-libEGL.so"
doesn't help at all.
[YOCTO #9217] Many unsolveable QA warnings from build-deps and file-rdeps
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The variable is PACKAGE_WRITE_DEPS as seen in the default below the
comment.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Using 'sub' in one function just confuses things, standardise and fix
formatting of the parameters too.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
After this replacement, the parent exception handling works so we don't
need subprocess wrapping with bb.error in the underlying functions.
The underlying contexts also have better module handling so the imports
can be cleaned up.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Replace usage of oe.utils.getstatusoutput() with direct subprocess calls. The
call is just a wraper to the subprocess call of the same name and requires
the caller to handle exceptions themselves. We usually do this badly, failing
to show the output or the command or the return code.
Its much safer to rely on a call like subprocess.check_output() instead.
This also makes it easier to spot and remove cases where shell=True isn't
needed in a later cleanup.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This allows the calls to is_elf (which calls file) to happen in parallel
allowing a speedup of do_package and do_populate_sysroot for native
recipes.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
There are probably further cleanups needed here but this at least removes
the major code duplication between these two similar funcitons, keeping the
kernel module ".ko" extension check for efficiency to avoid opening and
reading file contents in the general case.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This function is a bit more invasive to add parallelism to but
allows the shlibs analysis to happen in multiple threads. In order
to return values correctly/safely the data types needed tweaking
to avoid lists and use immutable objects.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Adjust the code so that the splitting of debug symbols from files happens
in parallel. To to this we need to move some path handling code into the
main function and pass more parameters in.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We don't need any functionality from the shell here, its just extra fork
overhead. Therefore remove it and use subprocess directly.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The command for running dwarfsrcfiles is simple and does not need a subshell
for each execution. By expanding out this function to use check_output()
from subprocess and a list of arguments, the shell overhead can be dropped.
For recipes with lots of files this gives a significant saving.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
I think this lock dates from before we had sstate for do_packagedata.
Since WORKDIR is recipe specific and we write into WORKDIR, we no longer need
any write locks in the do_packagedata code itself, its handled by the sstate
task lock for the final copy in at the end. The final write lock can be simply
removed.
The only time we need read locking is when actually reading data from the
shared directory. We can therefore reduce the window the lock is held
significantly as well, hence improving the speed of packagedata tasks running
in parallel.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Some packages containing shared libraries might be registered
as shlib providers when they shouldn't (for example, the lib is for
their private use and must not generate any dependency).
EXCLUDE_FROM_SHLIBS is targeted at that, but it could be set
for entire recipe only.
This patch allows to set list of packages that are not candidates
for shlib providers.
Variable EXCLUDE_PACKAGES_FROM_SHLIBS is used, e.g.:
EXCLUDE_PACKAGES_FROM_SHLIBS = "glibc-ptest glibc-mtest"
Cc: Andrii Bordunov <aborduno@cisco.com>
Signed-off-by: Oleksii Konoplitskyi <okonopli@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Allow debugsource listing using dwarfsourcefiles to fail for static
libraries when the archive content is not as expected.
Signed-off-by: Ola x Nilsson <olani@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
PACKAGELOCK is there to protect readers of PKGDATA_DIR from writes and files
changing whilst they're being read. With various changes to the codebase,
the lock code has become confused as the files are now written by the sstate
code in do_packagedata, not in do_package directly any longer.
This change cleans up the code so read sites take the shared lock (anything in
do_package), write sites take the full lock (do_packagedata sstate).
The lock from do_package sstate is no longer needed since it doesn't write
outside WORKDIR.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
!= mingw*
"package.bbclass: Include dbgsrc for static libs" introduced a regression
on mingw, fix this by excluding on that TARGETOS.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The debugsource must be added from the package providing the static
lib, because any package using that lib does not have access to the
source code.
Fixes [YOCTO #12558]
Signed-off-by: Ola x Nilsson <olani@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
The isELF function works by running:
result = file <pathname>
if 'ELF' in result
By default 'file' will prepend the result with the path name of the file
that is being checked. This usually works fine, such as:
$ file /home/foo/openembedded-core/meta/classes/package.bbclass
/home/foo/openembedded-core/meta/classes/package.bbclass: Python script, ASCII text executable, with very long lines
However, if the path includes 'ELF', ELF will end up in the result, and then
the check will return positive.
$ file /home/ELF/openembedded-core/meta/classes/package.bbclass
/home/ELF/openembedded-core/meta/classes/package.bbclass: Python script, ASCII text executable, with very long lines
This will then result in the isELF coming back true, and possibly causing the
checks that use isELF, such as the 'is it already stripped' check, to do the
incorrect thing.
Adding the '-b' option to file will result in the path being omitted in the
result:
$ file /home/ELF/openembedded-core/meta/classes/package.bbclass
Python script, ASCII text executable, with very long lines
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
flag
Without this change, there will be two sstate index files in
tmp/sstate-control for any machine that contains a dash in the
name.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Apparently there are recipes in the wild which generate files with
filenames containing '$' characters - which cause errors during
packaging.
Instead of adding another special case to escape '$' characters when
constructing the command passed to oe.utils.getstatusoutput(), switch
to using single quotes to quote the path - and therefore make isELF()
consistent with the way filenames and paths are quoted by every other
caller of oe.utils.getstatusoutput() in oe-core.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
This allows catching errors in the scriptlets which would otherwise
go unnoticed, e.g. this sequence:
====
bogus_command
proper_command
====
would work just fine without any visible warnings or errors.
This was previously done only for rpm packages; this patch replaces
the rpm-specific tweak with one that works for all package types.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The shlibs detection/handling iterates the package list twice, but PRIVATE_LIBS
is only fetched in the first loop which means the second loop only considers the
value set.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
All typos were in comments but some of these comments end up in run/log files.
The typos can be annoying when searching the log files so they're worth
fixing with ispell.
Signed-off-by: Randy MacLeod <Randy.MacLeod@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
There are cases where its useful to allow only a select few files
to be excluded from the package stripping mechanism. Currently
this isn't possible so add a variable to allow this.
This is to be used sparingly as in general the core code should be
doing the right thing. This is better than the alternative of leaving the whole
package unstripped.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This function is a convenient and more readable shortcut for situations
when the postinst code always needs to run on target. All commands that
cannot be executed during cross-install and can only be run on target
should go into this function. They will only be executed on first boot
(if package was cross-installed) or immediately during package installation
on target.
Plain pkg_postinst() works as before: it is run during cross-install time,
it can contain a request to defer to first boot, and it is also run
during package installation on target.
Also fix the oeqa test for this functionality to use the new function
where appropriate.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Debugedit provided by rpm 4.14 is rewriting binaries in-place, and was
found to produce broken output at least for grub:
http://lists.openembedded.org/pipermail/openembedded-core/2017-November/143989.html
A replacement utility was suggested via private mail:
https://lists.fedorahosted.org/archives/list/elfutils-devel@lists.fedorahosted.org/message/VZP4G5N2ELYZEDAB3QYLXYHDGX4WMCUF/
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
If FILESYSTEM_PERMS_TABLES was not defined, the default was selected by the
packages.bbclass. This made it difficult for a recipe or layer to 'append'
to the default.
Copy the default into the bitbake.conf, allowing future _append and += style
actions.
Default was remove from package.bbclass. If a value is not set, only the
built-in perms fixes will be used.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
The silent rpmdeps failures fixed in a preceeding commit mean we
need to rerun all packaging.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Add a new file, fs-perms-persistent-log.txt, which treats /var/log
as a directory instead of a link.
Modify package.bbclass to use this file if VOLATILE_LOG_DIR is set to boolean
false value.
[YOCTO #6132]
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
During the transition to dnf and rpm4, the functionality to
automatically make RPM determine dependencies was lost.
Before the transition, an OE specific tool called rpmdeps-oecore had
been added to the rpm suit. It was based on the rpmdeps tool that is
part of rpm. For each file specified on its command line, it would
output the provides and requires that RPM could determine.
During the transition to rpm4, rpmdeps-oecore was replaced with the
standard rpmdeps. However, what no one noticed was that unless rpmdeps
is given options, e.g., -P or -R, to tell it what it should output, it
will not output anything. Thus, it would do all the work to determine
the requirements, but would keep silent about it. And since no output
from rpmdeps is expected unless there are requirements, there were no
warnings indicating that everything was not working as expected.
Porting the old rpmdeps-oecore to work with rpm4 is not really
possible since it relied on being able to access internals of RPM that
are no longer available. However, it turned out that rpmdeps had a
debug option, --rpmfcdebug, that would output exactly the information
that we need, albeit in a different format and to stderr. To make this
usable, rpmdeps has now received a new option, --alldeps, which sends
the information we need to stdout.
(From OE-Core rev: 958501b3d9201aaabb81ec644c6049e0c9b737e7)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The configuration variable PACKAGE_DEBUG_SPLIT_STYLE includes
the new mode debug-with-srcpkg that instructs the system to
remove the source files from the debug package but include them in
a separate package with a "-src" suffix in the name.
[YOCTO #9998]
Signed-off-by: Juan M Cruz Alcaraz <juan.m.cruz.alcaraz@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This reverts commit aea90e9ee6f34e7c1c08307080b1e29646668df6.
RP hadn't meant to merge this yet and its causing problems so delay it until its
ready.
|
|
During the transition to dnf and rpm4, the functionality to
automatically make RPM determine dependencies was lost.
Before the transition, an OE specific tool called rpmdeps-oecore had
been added to the rpm suit. It was based on the rpmdeps tool that is
part of rpm. For each file specified on its command line, it would
output the provides and requires that RPM could determine.
During the transition to rpm4, rpmdeps-oecore was replaced with the
standard rpmdeps. However, what no one noticed was that unless rpmdeps
is given options, e.g., -P or -R, to tell it what it should output, it
will not output anything. Thus, it would do all the work to determine
the requirements, but would keep silent about it. And since no output
from rpmdeps is expected unless there are requirements, there were no
warnings indicating that everything was not working as expected.
Porting the old rpmdeps-oecore to work with rpm4 is not really
possible since it relied on being able to access internals of RPM that
are no longer available. However, it turned out that rpmdeps had a
debug option, --rpmfcdebug, that would output exactly the information
that we need, albeit in a different format and to stderr. To make this
usable, rpmdeps has now received a new option, --alldeps, which sends
the information we need to stdout.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.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>
|
|
Changes to PRIVATE_LIBS should change the sstate checksum. To make
that happen, it needs to be listed in the list of package specific
variables, therefore add it.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
bb.data.expand(x, d) is deprecated API.
[YOCTO #10678]
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Changes to SKIP_FILEDEPS should change the sstate checksum. To make
that happen, it needs to be listed in the list of package specific
variables, therefore add it.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
There are several issues with rpmdeps after the rpm v5 -> v4 transition:
* _rpmfc_magic_path is an invalid option for rpm4
* --rpmpopt is an invalid option for rpm4
* we need to use the path to rpmrc since otherwise it poitns at the
original build path
* we need to set MAGIC in the environment so libmagic can find its
files.
This patch addresses those and ensures rpmdeps works in relocated builds
from sstate (or with rm_work).
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This ensures that remapping happens correctly, particularly
when package groups list noarch packages, and multilib is in use. Previously
this was masked by rpm doing *another* layer of remapping on top of this.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
rpm4 installs them in different locations than rpm5. This also replaces
our custom rpmdeps-oecore with standard rpmdeps; I'm not seeing a
significant performance penalty.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Add support to detect dll files as shared objects as well as process
Windows .dll and .exe files to determine the runtime libraries
dependencies.
This implementation is sufficient to detect and map runtime library
dependencies between packages. And does not implement any version naming
conventions that might apply for .dll files (e.g. lib*-x.dll).
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
USE_LDCONFIG could previously be set to 0 by distros which do not
require ldconfig or ld.so.conf on the target. Since more and more
recipes may need to respect that option, replace the ad-hoc variable
with a distro feature.
Distros which previously set:
USE_LDCONFIG = "0"
Should now instead use:
DISTRO_FEATURES_BACKFILL_CONSIDERED_append = " ldconfig"
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
This patch is comparatively large and invasive. It does only do one thing, switching the
system to build using recipe specific sysroots and where changes could be isolated from it,
that has been done.
With the current single sysroot approach, its possible for software to find things which
aren't in their dependencies. This leads to a determinism problem and is a growing issue in
several of the market segments where OE makes sense. The way to solve this problem for OE is
to have seperate sysroots for each recipe and these will only contain the dependencies for
that recipe.
Its worth noting that this is not task specific sysroots and that OE's dependencies do vary
enormously by task. This did result in some implementation challenges. There is nothing stopping
the implementation of task specific sysroots at some later point based on this work but
that as deemed a bridge too far right now.
Implementation details:
* Rather than installing the sysroot artefacts into a combined sysroots, they are now placed in
TMPDIR/sysroot-components/PACKAGE_ARCH/PN.
* WORKDIR/recipe-sysroot and WORKDIR/recipe-sysroot-native are built by hardlinking in files
from the sysroot-component trees. These new directories are known as RECIPE_SYSROOT and
RECIPE_SYSROOT_NATIVE.
* This construction is primarily done by a new do_prepare_recipe_sysroot task which runs
before do_configure and consists of a call to the extend_recipe_sysroot function.
* Other tasks need things in the sysroot before/after this, e.g. do_patch needs quilt-native
and do_package_write_deb needs dpkg-native. The code therefore inspects the dependencies
for each task and adds extend_recipe_sysroot as a prefunc if it has populate_sysroot
dependencies.
* We have to do a search/replace 'fixme' operation on the files installed into the sysroot to
change hardcoded paths into the correct ones. We create a fixmepath file in the component
directory which lists the files which need this operation.
* Some files have "postinstall" commands which need to run against them, e.g. gdk-pixbuf each
time a new loader is added. These are handled by adding files in bindir with the name
prefixed by "postinst-" and are run in each sysroot as its created if they're present.
This did mean most sstate postinstalls have to be rewritten but there shouldn't be many of them.
* Since a recipe can have multiple tasks and these tasks can run against each other at the same
time we have to have a lock when we perform write operations against the sysroot. We also have
to maintain manifests of what we install against a task checksum of the dependency. If the
checksum changes, we remove its files and then add the new ones.
* The autotools logic for filtering the view of m4 files is no longer needed (and was the model
for the way extend_recipe_sysroot works).
* For autotools, we used to build a combined m4 macros directory which had both the native and
target m4 files. We can no longer do this so we use the target sysroot as the default and add
the native sysroot as an extra backup include path. If we don't do this, we'd have to build
target pkg-config before we could built anything using pkg-config for example (ditto gettext).
Such dependencies would be painful so we haven't required that.
* PKDDATA_DIR was moved out the sysroot and works as before using sstate to build a hybrid copy
for each machine. The paths therefore changed, the behaviour did not.
* The ccache class had to be reworked to function with rss.
* The TCBOOTSTRAP sysroot for compiler bootstrap is no longer needed but the -initial data
does have to be filtered out from the main recipe sysroots. Putting "-initial" in a normal
recipe name therefore remains a bad idea.
* The logic in insane needed tweaks to deal with the new path layout, as did the debug source
file extraction code in package.bbclass.
* The logic in sstate.bbclass had to be rewritten since it previously only performed search and
replace on extracted sstate and we now need this to happen even if the compiled path was
"correct". This in theory could cause a mild performance issue but since the sysroot data
was the main data that needed this and we'd have to do it there regardless with rss, I've opted
just to change the way the class for everything. The built output used to build the sstate output
is now retained and installed rather than deleted.
* The search and replace logic used in sstate objects also seemed weak/incorrect and didn't hold
up against testing. This has been rewritten too. There are some assumptions made about paths, we
save the 'proper' search and replace operations to fixmepath.cmd but then ignore this. What is
here works but is a little hardcoded and an area for future improvement.
* In order to work with eSDK we need a way to build something that looks like the old style sysroot.
"bitbake build-sysroots" will construct such a sysroot based on everything in the components
directory that matches the current MACHINE. It will allow transition of external tools and can
built target or native variants or both. It also supports a clean task. I'd suggest not relying on
this for anything other than transitional purposes though. To see XXX in that sysroot, you'd have
to have built that in a previous bitbake invocation.
* pseudo is run out of its components directory. This is fine as its statically linked.
* The hacks for wayland to see allarch dependencies in the multilib case are no longer needed
and can be dropped.
* wic needed more extensive changes to work with rss and the fixes are in a separate commit series
* Various oe-selftest tweaks were needed since tests did assume the location to binaries and the
combined sysroot in several cases.
* Most missing dependencies this work found have been sent out as separate patches as they were found
but a few tweaks are still included here.
* A late addition is that extend_recipe_sysroot became multilib aware and able to populate multilib
sysroots. I had hoped not to have to add that complexity but the meta-environment recipe forced my
hand. That implementation can probably be neater but this is on the list of things to cleanup later
at this point.
In summary, the impact people will likely see after this change:
* Recipes may fail with missing dependencies, particularly native tools like gettext-native,
glib-2.0-native and libxml2.0-native. Some hosts have these installed and will mask these errors
* Any recipe/class using SSTATEPOSTINSTFUNCS will need that code rewriting into a postinst
* There was a separate patch series dealing with roots postinst native dependency issues. Any postinst
which expects native tools at rootfs time will need to mark that dependency with PACKAGE_WRITE_DEPS.
There could well be other issues. This has been tested repeatedly against our autobuilders and oe-selftest
and issues found have been fixed. We believe at least OE-Core is in good shape but that doesn't mean
we've found all the issues.
Also, the logging is a bit chatty at the moment. It does help if something goes wrong and goes to the
task logfiles, not the console so I've intentionally left this like that for now. We can turn it down
easily enough in due course.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Add a new variable to allow markup of postinstall (and preinst)
script dependnecies on native/cross tools.
If your postinstall can execute at rootfs creation time rather than on
target but depends on a native tool in order to execute, you need to
list that tool in PACKAGE_WRITE_DEPENDS.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
There were a few straggling expansion parameter removals left for
getVar/getVarFlag where the odd whitespace meant they were missed
on previous passes. There were also some plain broken ussages such
as:
d.getVar('ALTERNATIVE_TARGET', old_name, True)
path = d.getVar('PATH', d, True)
d.getVar('IMAGE_ROOTFS', 'True')
which I've corrected (they happend to work by luck).
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When calculating PKGSIZE we sum the size of files after doing lstat() so we
don't count directory metadata overhead, but were not correctly handling
hardlinks. This results in packages such as e2fsprogs-mke2fs having PKGSIZE of
1.5M when it's actually a single 300K binary with five hardlinks.
[ YOCTO #10423 ]
Signed-off-by: Ross Burton <ross.burton@intel.com>
|