summaryrefslogtreecommitdiffstats
path: root/conf
Commit message (Collapse)AuthorAgeFilesLines
* gnutls: Disable integrated assembler on aarch64Khem Raj2023-09-081-0/+3
| | | | | | | | integrated assembler crashes latest clang on aarch64, so lets disable it for now Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit 9d89acd175067a828313b8b2a215e6f7595f61b5)
* aufs-util,libhugetlbfs,libc-bench: Use libgcc unwinder on glibc/armKhem Raj2023-09-081-2/+3
| | | | | | | | | these packages do static linking in parts and when libc is glibc then it has linked in some portions of libgcc since we can not use clang to compile glibc yet. Therefore use libgcc where needed for now Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit bf842214267663966ffa34e170368d55fcebd883)
* nonclangable: Use libgcc for unwinder as well when using for builtinsKhem Raj2023-09-081-9/+11
| | | | | | | | | | | | | | | clang 13 does not like to mix rtlib and unwindlib when using libgcc Fixes clang-13: error: --rtlib=libgcc requires --unwindlib=libgcc secondly compiler-rt does not havw 128 bit ( tf ) functions on x86 so fallback to libgcc for those use libatomic for x86 since 64bit atomics are not in compiler-rt Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit 96face19ae51510d5e5923e575496844b5b85e62)
* poke: Do not use -O2 with clangKhem Raj2023-09-081-0/+3
| | | | | | | | | clang13 has a regression which segfaults the compiler while compiling poke, its reported upstream, until it is fixed do not use -O2 compiler switch Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit d01e38f98bb03d9afa72e9dfc10eb936fc6a4c29)
* nonclangable: pass -rtlib only when using clang with gnu runtime hereKhem Raj2023-09-081-9/+9
| | | | | | | | The defaults wirh using RUNTIME = "llvm" is already pointing to compiler-rt there is no need to clutter cmdline Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit 785ce0f2df336ea2c7ef40a3e64eb84656f59eeb)
* mongodb: Use llvm runtime when using clangKhem Raj2023-09-081-0/+1
| | | | | | | | | | | Fixes linker errors like src/mongo/platform/overflow_arithmetic.h:63: error: undefined reference to '__mulodi4' src/mongo/platform/overflow_arithmetic.h:63: error: undefined reference to '__mulodi4' src/mongo/platform/overflow_arithmetic.h:63: error: undefined reference to '__mulodi4' src/mongo/platform/overflow_arithmetic.h:63: error: undefined reference to '__mulodi4' Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit 0c02a3503f6b240ec34c68bc9c00407bae2b0fa4)
* libc-bench: It works fine with new way of choosing llvm runtimeKhem Raj2023-09-081-3/+0
| | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit 55978074f349330367ffbbf991e3287141f8fe4d)
* nss: Pass options to select runtime for native partsKhem Raj2023-09-081-1/+1
| | | | | | | | Since clang maybe configured to use llvm runtime, we need to ensure to use gnu runtime in native builds Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit 563e11548bfe95444d447e058001d3d1da5fe8e8)
* layer.conf: Prefer llvm libunwind when RUNTIME = "llvm"Khem Raj2023-09-081-0/+1
| | | | | | | Avoids conflicts with libunwind recipe from OE-Core Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit 5be1042c67063354217154e3c35ed83408d916ce)
* clang: Upgrade to trunk/13.0 future seriesKhem Raj2023-09-081-1/+1
| | | | | | | | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com> (cherry picked from commit 76e37fafa29e9b2097eb823537b17d6cc638c1d5) Conflicts: - recipes-devtools/clang/common.inc: caused by 3880a395f870ecd8da73d2de1e2b88a855df2ef0, the patches have been kept. The patch introduced in fdb936e80b30f40cbf7ef9a8e124b58eec95f45c is part of LLVM 13 so it has been removed.
* Merge remote-tracking branch 'upstream/dunfell' into clang12Khem Raj2022-02-041-2/+2
|\
| * chromium,gn-native: Remove dependency on native libc++Khem Raj2021-08-131-2/+2
| | | | | | | | | | | | | | native versions of libcxx and compiler-rt are only available in honister and newer Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | intel-mediasdk,optee-os-stm32mp: Use dynamic-layersKhem Raj2022-02-031-0/+4
| | | | | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | chromium,gn: Do not depend on libc++ native recipesKhem Raj2021-08-131-2/+2
| | | | | | | | | | | | | | | | | | Master meta-browser has moved away from gcc completely even for native portions, but those changes require master branch of meta-clang as well for older layers where older than clang 13 is used, lets undo those changes Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | nonclangable: Remove no-integrated-as workarounds for opensslBernhard Rosenkränzer2021-07-141-11/+0
| | | | | | | | | | | | The problem is fixed and the workaround actually breaks things Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | valgrind: Allow building with clangBernhard Rosenkränzer2021-07-141-1/+0
| | | | | | | | | | | | | | | | Allow building valgrind with clang -- merge a patch from the Android NDK to fix the valgrind+clang aarch64 combo Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | nonclangable.conf: Remove workaround for a clang bug that has been fixedBernhard Rosenkränzer2021-07-141-3/+0
| | | | | | | | Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | Allow building cpufrequtils with clangBernhard Rosenkränzer2021-07-141-5/+0
| | | | | | | | | | | | | | Make sure invalid assembly code isn't included even in unreachable code; remove TOOLCHAIN=gcc workaround. Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | Allow building ppp with clangBernhard Rosenkränzer2021-07-141-4/+0
| | | | | | | | | | | | | | ppp's use of nested functions and gcc specific flags has been fixed. The workaround is no longer necessary. Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | Allow mongodb to be built with clangBernhard Rosenkränzer2021-07-141-2/+0
| | | | | | | | | | | | | | Building it with gcc was a workaround for a bug that has been fixed a long time ago. The workaround is no longer necessary. Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | libffi: Allow building with clang integrated-asBernhard Rosenkränzer2021-07-141-5/+0
| | | | | | | | | | | | | | The problem preventing this from succeeding earlier has already been fixed in libffi. There's no reason to keep the workaround. Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | Document why cpufrequtils isn't built with clangBernhard Rosenkränzer2021-07-141-0/+4
| | | | | | | | Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | clang: Allow building babeltrace2 with clangBernhard Rosenkränzer2021-07-141-2/+0
| | | | | | | | | | | | | | babeltrace2 failed to build with clang because of a bug in babeltrace. Fixing babeltrace is a better fix than forcing it to build with gcc. Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | clang: Fix pixman on 32-bit armBernhard Rosenkränzer2021-07-141-5/+5
| | | | | | | | | | | | Pixman needs workarounds for clang on 32-bit ARM Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | Drop unnecessary TOOLCHAIN-*="gcc" workaroundsBernhard Rosenkränzer2021-07-141-5/+8
| | | | | | | | Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | Build u-boot-seco-imx with gccBernhard Rosenkränzer2021-07-141-0/+1
| | | | | | | | | | | | | | | | | | Makefiles in u-boot-seco-imx hardcode gcc use - make sure the requested compiler is available even with TOOLCHAIN="clang" Fixes https://git.ostc-eu.org/OSTC/OHOS/meta-clang/-/issues/2 Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | Fix Avenger96 build with TOOLCHAIN="clang"Bernhard Rosenkränzer2021-07-141-0/+5
| | | | | | | | | | | | | | | | | | The problem is Makefiles in u-boot and trusted-firmware hardcoding compiler calls. This should be fixed at some point; until then, use the gcc toolchain for these components. Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | Dunfell compatibilityBernhard Rosenkränzer2021-07-141-1/+1
| | | | | | | | Signed-off-by: Bernhard Rosenkränzer <bernhard.rosenkraenzer.ext@huawei.com>
* | replace --rtlib with -rtlibKhem Raj2021-05-191-26/+26
| | | | | | | | | | | | This is advertised option from clang Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | openssh: Link with compiler-rt when using clangKhem Raj2021-04-231-0/+1
| | | | | | | | | | | | | | | | clang -ftrapv, needs compiler-rt [1] [1] https://github.com/openssh/openssh-portable/blob/master/INSTALL#L7 Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | layer.conf: Add 3.4 as supported releaseKhem Raj2021-04-151-1/+1
| | | | | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | layer.conf: Drop gatesgarthKhem Raj2021-04-151-1/+1
| | | | | | | | | | | | it has its own branch Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | iwd,ell: Mark gcc onlyKhem Raj2021-04-081-0/+6
| | | | | | | | | | | | They have started using nested functions lately. Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | nettle: Disable internal asm on ppc64Khem Raj2021-03-011-0/+7
| | | | | | | | | | | | | | generated asm files use .0: label which is not understood with internal asm, gnu as does fine job Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | clang: Update to 12.0.0 RC2Khem Raj2021-03-011-1/+1
| | | | | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | layer.conf: remove bcc and bpftrace from NON_MULTILIB_RECIPESKai Kang2021-02-051-2/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Because bcc and bpftrace have been set in NON_MULTILIB_RECIPES, 'bcc' rprovides 'lib32-bcc'. So when multilib is enabled, it is 'bcc' rather than 'lib32-bcc' installed to lib32 image such as lib32-core-image-minimal. Then the runtime dependency 64-bit package python3-core is installed accordingly. It causes 32-bit python modules which contains .so library do not work any more. For example, it fails to run command 'dnf': | ImportError: /usr/lib/python3.9/site-packages/libdnf/_error.so: wrong ELF class: ELFCLASS32 Remove bcc and bpftrace from NON_MULTILIB_RECIPES to make sure they are not installed to lib32 images and mess images up. Signed-off-by: Kai Kang <kai.kang@windriver.com>
* | clang/llvm: Upgrade to 11.1.0-rc2Khem Raj2021-01-261-1/+1
| | | | | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | qemu: Link with latomic on clang/x86Khem Raj2021-01-091-0/+1
| | | | | | | | | | | | | | clang delegates atomics to runtime library instead of builtins so link in libatomic on x86/gnu Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | ppp: Mark non-clangableKhem Raj2021-01-091-0/+4
| | | | | | | | | | | | | | New version of ppp uses gcc specific options and nested functions which clang does not support Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | pulseaudio: Remove -Qunused-arguments with clangKhem Raj2020-12-231-2/+2
| | | | | | | | | | | | | | | | -Qunused-arguments means its silencing warning about compiling arm/neon check code on x86 and tricking the build system to assume that compiler can support neon on x86 Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | redis: Mark non-clangable as of nowKhem Raj2020-12-141-0/+4
| | | | | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | luajit: Add -no-integrated-asKhem Raj2020-12-141-0/+8
| | | | | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | nonclangable.conf: link python3-numpy with python3-numpyKhem Raj2020-11-181-0/+1
| | | | | | | | | | | | | | This helps in addressing an issue where clang is used and it expects symbols in builtins library which are only provided by compiler-rt Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | clang: Update to latest on release/11.x branchKhem Raj2020-11-151-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The minor release is bumped to 11.0.1 Changes is this bump are * 0874e7ef66c Allow init_priority values <= 100 and > 65535 within system headers. * 02004c9e7c9 [ADT] Fix for ImmutableMapRef * 85ce339f1bd [cmake] Fix build of attribute plugin example on Windows * ef4ffcafbb2 [DAE] MarkLive in MarkValue(MaybeLive) if any use is live * 3c687677678 DeferredDiagnosticsEmitter crashes * 701addff1b7 [clang][Sema] Fix PR47676: Handle dependent AltiVec C-style cast * 5ad2592b5dc [X86] Fix cpu name typos * 8fc424f26bf Add release tarballs for libclc * 83716db47f9 [ELF] -r: don't crash when a non-SHF_LINK_ORDER orphan is added before a SHF_LINK_ORDER orphan * 1ff84a04aeb BPF: fix incorrect DAG2DAG load optimization * 38399ced95b [ConstantFold] Make areGlobalsPotentiallyEqual less aggressive. * b59b6b662bb Import github action definitions from release/10.x branch * 3039965645c Bump version to 11.0.1 Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | lto: Disable LTO for cairo and libidn2 with clangKhem Raj2020-11-131-0/+8
| | | | | | | | | | | | There are compile/link time errors Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | layer.conf: Add hardknott to supported releasesKhem Raj2020-11-131-1/+1
| | | | | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | qtbase: Use compiler-rt on riscv32 when using clangKhem Raj2020-11-131-0/+4
| | | | | | | | | | | | | | | | clang emits __lshrti3 expecting it from builtins library but libgcc does not have it, since we have compiler-rt support it because its compiled with -fforce-enable-int128 so link with it Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | tesseract: Remove -Qunused-arguments from CFLAGSKhem Raj2020-10-201-0/+1
| | | | | | | | | | | | | | | | | | | | | | This helps configure checks for sse where clang sends out a unused option warning but because we use -Qunused-arguments, this warning is suppressed and -Werror does not notice this warning and checks succeed for no reason on architectures who dont have SSE. Therefore removing this option helps compiler to emit the warning and rightly detect that these options can not be enabled for non-x86 architectures Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | libffi: Do not use internal assembler on mips64Khem Raj2020-09-101-0/+1
| | | | | | | | | | | | | | It uses GNU asm which clang's internal assembler gets confused with and can not interpret .eh_frame section permissions and type Signed-off-by: Khem Raj <raj.khem@gmail.com>
* | layer.conf: Drop zeus and dunfell from LAYERSERIES_COMPATKhem Raj2020-08-271-1/+1
| | | | | | | | Signed-off-by: Khem Raj <raj.khem@gmail.com>