Remove ccache-native's dependencies so that all native recipes can depend on
ccache-native and use it except ccache-native itself.
It has the following 4 dependencies:
- xz-native: It is intruduced by ccache-3.4.2.tar.xz, use ccache-3.4.2.tar.gz
to replace of it can fix the problem.
- zlib-native: Use --with-bundled-zlib to fix it.
- autotools-native: Set INHIBIT_AUTOTOOLS_DEPS and add a do_configure() to fix
the problem.
- quilt-native: Set PATCHTOOL = "patch" to fix it.
(From OE-Core rev: f3d8bda91ee186dae8847d6438c1dfd6f6d8ee4f)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* Redefine the reference path of libnfs.a to
../support/nfs/.libs/libnfs.a to fix below
error when run "make -C tests statdb_dump".
| make: *** No rule to make target '../support/nfs/libnfs.a', needed by 'statdb_dump'. Stop.
| make: *** No rule to make target '../../support/nfs/libnfs.a', needed by 'nsm_client'. Stop.
* The function generic_make_pathname is introduced in
nfs-utils 2.3.1.
Add the source file which defines function generic_make_pathname to
libnsm_a_SOURCES of libnsm.a to fix the undefined reference
when run "make -C tests statdb_dump"
| ../support/nsm/libnsm.a(file.o): In function `nsm_make_pathname':
| /usr/src/debug/nfs-utils/2.3.1-r0/nfs-utils-2.3.1/support/nsm/file.c:175: undefined reference to `generic_make_pathname'
| /usr/src/debug/nfs-utils/2.3.1-r0/nfs-utils-2.3.1/support/nsm/file.c:175: undefined reference to `generic_make_pathname'
| /usr/src/debug/nfs-utils/2.3.1-r0/nfs-utils-2.3.1/support/nsm/file.c:175: undefined reference to `generic_make_pathname'
| ../support/nsm/libnsm.a(file.o): In function `nsm_setup_pathnames':
| /usr/src/debug/nfs-utils/2.3.1-r0/nfs-utils-2.3.1/support/nsm/file.c:280: undefined reference to `generic_setup_basedir'
| collect2: error: ld returned 1 exit status
* After the logic of commit[dbb643e Removed missing-prototypes warnings.]
introduced, there comes below error when run
"make -C tests/nsm_client nsm_client"
| nlm_sm_inter_svc.c:20:1: error: no previous prototype for 'nlm_sm_prog_3' [-Werror=missing-prototypes]
It is because rpcgen doesn't generate -Wmissing-prototypes
free code for nlm_sm_inter_svc.c with below logic
in tests/nsm_client/Makefile.am
[snip]
GENFILES_SVC = nlm_sm_inter_svc.c
[snip]
$(GENFILES_SVC): %_svc.c: %.x $(RPCGEN)
test -f $@ && rm -rf $@ || true
$(RPCGEN) -m -o $@ $<
So add the patch to not fatalize -Wmissing-prototypes.
(From OE-Core rev: 18947cd4782934e9bc3ec5507db8db51258d9b2c)
Signed-off-by: Mingli Yu <Mingli.Yu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The unihash should be fetched using the task filename that includes the
multiconfig prefixes.
[YOCTO #13124]
(Bitbake rev: 5e7f4e77e27bceaf6c68137cacb4f8d7d7de49dd)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The only user of the configuration module is paprefs, which as of 1.0 uses GSettings not GConf.
Also GConf is unmaintained, so one less recipe using it is good.
(From OE-Core rev: 463e064a4f4a664ae1b871b98f26b561d1cbe98b)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Instead of assuming that the schemas are located in ${PN}, add a
GSETTINGS_PACKAGE variable for the package name and default it to ${PN}.
For recipe that can conditionally ship schemas, support GSETTINGS_PACKAGE being
empty gracefully by doing nothing.
(From OE-Core rev: c2b9c34ce4af151cc0422e14af775c6c962de051)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The magic multilibisation doesn't quite catch every instance, so replace
'pulseaudio' with ${PN} where needed to ensure the packaging is as intended.
(From OE-Core rev: 65de45912facaede0c40a3b881f4e095590ca205)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The intention is that PN is empty, enforce that so new files don't end up in PN
silently.
(From OE-Core rev: b302e1556535601dbb5a18d02e6988a17fda5fbc)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There are failures when multiple users run oe-selftest on the same
host:
PermissionError: [Errno 13] Permission denied: '/tmp/oe-saved-
tests/201812250324_qemu'
This is because /tmp/oe-saved-tests was created by user A, while user B tries
to write data in it, then the error will happen. This patch can fix the
problem.
Move the dumped data to ${LOG_DIR}/runtime-hostdump/ rather than
/tmp/oe-saved-tests/ to fix the problem.
(From OE-Core rev: e219fe5329599cd6c3682f521eaee3852a2c8980)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently SDKIMAGE_LINGUAS is broken for any inputs except "all".
In the non-"all" case, each enabled language package is installed via
pm.install("nativesdk-glibc-binary-localedata-%s.utf-8" % lang)
This will throw a python exception since pm.install() expects a list of
strings and not a string.
Fix the problem by constructing a list.
That way it is now also possible to call the package installer just
once.
Cc: "Burton, Ross" <ross.burton@intel.com>
Fixes: 67615e0175 ("rootfs_rpm.bbclass: migrate image creation to dnf")
(From OE-Core rev: 475a5d9ec21a329be973691734f9e8bcb332338c)
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The icecc setup for the SDK was broken in multilib configurations now
that each multilib environment runs the post-relocate scripts
separately. Including $TARGET_PREFIX in the icecc shim path and in the
toolchain environment name prevents the various multilib setups from
conflicting.
[YOCTO #13128]
(From OE-Core rev: 13704d7a9d147382169d0c4bd77cb4f5577c65b1)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If a given PN is listed in the icecream blacklist, there is a very good
chance that the native, nativesdk, and multilib variants should also be
skipped. Check the blacklist entries against BPN to cover them.
[YOCTO #13128]
(From OE-Core rev: 1863f695a1411e95e7e547a3eb3e7ef6604a93bf)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is needed by some init system services and if
init system is not sysvinit then we dont have it, therefore
its useful to have it provided via busybox as a backup
Enable CONFIG_SETSID to get setsid, needed by runit
(From OE-Core rev: 456010a042e17d2f5f1314248e2b9f1a3a1a8d93)
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The Yocto kernel expects a proper /dev/console to exist before running init,
otherwise booting will fail with the message:
Kernel panic - not syncing: /dev/console is missing or not a character device!
Please ensure your rootfs is properly configured
Add similar fix as was done for initramfs-framework and initramfs-live-boot
in commits
0352841cd92f6316bcac092e2fff9d28c352b36b
1b64664f0c388f41084f5db6e46e3e68c53fb6d9
(From OE-Core rev: 490d6fbd14805b6c72b525fbe8c9c6e08d796597)
Signed-off-by: Jacob Kroon <jacob.kroon@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We use subprocess.check_output() to run the command, which means that we need care
about the error, so the 2>/dev/null should not be used, otherwise it is hard to
debug when the error happens.
I guess it was copied from previous lines, but that command's error can be
ignored (excpet: pass):
try:
subprocess.check_output(cmd, shell=True, stderr=subprocess.STDOUT)
except subprocess.CalledProcessError:
# Can "fail" if internal headers/transient sources are attempted
pass
But we don't do this in the current location, so remove "2>/dev/null"
(From OE-Core rev: 017a53d2743be2b5d4965a39b4e126fb74f700ad)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
"dbus_daemon" is supposed to be set to the full dbus-daemon file
path, not just its directory.
(From OE-Core rev: fa95b51b4c867fdb8011881097a0160d7d3e4e71)
Signed-off-by: Jed <jed.openxt@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
While compiling with `-Werror=format-truncation=', it failed
[snip]
|super0.c:236:32: error: 'snprintf' output may be truncated
before the last format character [-Werror=format-truncation=]
| snprintf(nb, sizeof(nb), "%4d", d);
| ^
|super0.c:236:3: note: 'snprintf' output between 5 and 12 bytes
into a destination of size 11
| snprintf(nb, sizeof(nb), "%4d", d);
[snip]
(From OE-Core rev: cd5013a94609ef2f3e0d621339e22e4137d649d3)
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When UBOOT_DTB_BINARY is set to "", the keys for signed booting are
expected to be already present in U-Boot's DTB, so don't issue warnings
for this.
(From OE-Core rev: 04656f5df326a72747fc5878ce201b636a2a419c)
Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In distro_features_check.bbclass it checks whether items in
REQUIRED_DISTRO_FEATURES and CONFLICT_DISTRO_FEATURES exist in
DISTRO_FEATURES. But it only shows one required or conflict distro
feature when error occurs. Update to show them all at one time.
(From OE-Core rev: d0441c40afdba119a65189d6a5aca5c533f68279)
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add PACKAGECONFIG[manpages] for generating man pages.
(From OE-Core rev: 078b2389a76ca3eba2307c7aa20be095cf96e060)
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Drop 0001-Unset-need_charset_alias-when-building-for-musl.patch as it
had been fixed upstream.
(From OE-Core rev: 491c0ee55e7c6aa71ef5590cc38f866d9098bfe0)
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Drop 0001-GLES2-Get-sin-cos-out-of-vertex-shader.patch as it had been
merged upstream.
(From OE-Core rev: b46dbcad31c990b5556d61357e0a976948a5dede)
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Drop 0001-Fix-build-with-gperf-3.1.patch as it had been fixed upstream.
(From OE-Core rev: 0baf6799ae40461fbf3f18e098a3db448fec037b)
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The project pages and wiki has been moved to GitHub.
(From OE-Core rev: 45a5129663bd6f9e94fd5c674d8ce1dcc99195e0)
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Observing depsig.do_package for a package inbetween rebuilds indicated
that FILES_INFO was changing content order randomly. Force it to be
deterministic by sorting with respect to the keys when serializing.
Suggested-by: Joshua Watt <jpewhacker@gmail.com>
(From OE-Core rev: b0364be65bf0b3dee44f81379f4062e9f707c128)
Signed-off-by: Jacob Kroon <jacob.kroon@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
perl-cross is a build system overlay from buildroot project that aims to bring
a bit of sanity to cross-building perl. The advantage of using that is that we
can drop a lot of custom patches (that no one really understands), and simplify
the perl recipe as well. Also the build time goes down from several minutes to
about 30 seconds. The whole thing becomes maintainable again, in my opinion.
When rewriting the recipe I had two goals in mind:
1. Stay with upstream defaults as much as possible
2. Add custom patches only when their necessity was proven through testing.
http://arsv.github.io/perl-cross/
(From OE-Core rev: 52f2828314f851263ca3a6beb41ec936fab4d3ab)
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This was change to git recently to obtain openssl 1.1 compatible pre-release code
(before 7.9 was out), however tarballs are preferred, and with them upstream version
checks work (openssh uses a weird git tag scheme).
(From OE-Core rev: a921a96e84f4d0b3f3d29a571981de08eb148e15)
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
And fix the reported upstream check failures.
(From OE-Core rev: 63af0edcd954fac530ba17b04e5df5837cddc0a4)
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Various versions of ld have different defaults for this
(even between e.g. Ubuntu 16.04 and 18.04). This has the
consequence of putting either RPATH or RUNPATH into the binary,
depending on the linker, which have different priorities
vs LD_LIBRARY_PATH env var. Also, I think, the original
executable binary is looked up for the tags when resolving
a dependency of a shared library when using RPATH, but not
RUNPATH.
Let's make this deterministic.
(From OE-Core rev: 661d700ab7428b0ad982bfd2714cca134bad85a2)
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Some patch files create entirely new files, so their permissions are subject to
the host umask. If such a file is later installed into a package with no change
in permissions, it breaks the reproducibility of the package.
This was observed on libpam, for instance: The patch file
pam-security-abstract-securetty-handling.patch creates a new file
(tty_secure.c). This file is later copied into the -dbg package with no change
in permissions.
(From OE-Core rev: 2a2bbd755b330cd63f7f6e2f2b374a3ae065b37a)
Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* For changes, see:
https://www.tcpdump.org/libpcap-changes.txt
* Merge inc and bb and remove unnecessary flags.
* Remove all patches, they have either been upstreamed or fixed
differently.
* Compilation with bluez5 works just fine, enable it and remove bluez4
config.
* Backport a commit to fix musl builds.
(From OE-Core rev: 23fa5d49667c16f6b2763f4da4de63afa632e83d)
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
>From NEWS:
- bug fixes:
* do not autouninstall packages because of forcebest updates
- new features:
* support rpm's new '^' version separator
* support set/get_considered_list in bindings
* new experimental SOLVER_FLAG_ONLY_NAMESPACE_RECOMMENDED flag
(From OE-Core rev: 5b0d21e4b369821235d5b1d00a578b2dca5c5d09)
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
