summaryrefslogtreecommitdiff
path: root/package/ola
diff options
context:
space:
mode:
authorJörg Krause <joerg.krause@embedded.rocks>2016-09-22 01:31:43 +0200
committerThomas Petazzoni <thomas.petazzoni@free-electrons.com>2016-09-22 20:27:57 +0200
commite79272fa7ff3d66c18de3514b912cd9d68d121a4 (patch)
tree87b8ca1dfeab8068b695654248b597114b315b48 /package/ola
parent3f15f969aa7e5691abdbee81e9af21c3308f9a6a (diff)
package/ola: add patch to fix build with GCC 6
Fixes: http://autobuild.buildroot.net/results/2d5/2d55b5d88a06c7b8e6baeb96973009a451e992d9/ http://autobuild.buildroot.net/results/899/89922e61c583cf1d11bd0bafdd5586c35d8f6e15/ http://autobuild.buildroot.net/results/d5b/d5b8fe66ff8d9ea91e87ef6fbe8274f5e24aa7b0/ http://autobuild.buildroot.net/results/89b/89b136e6dced6ca9842a1f23141b0cb999f783da/ .. and many more. Building OLA with a GCC 6 cross-toolchain fails: ``` /usr/bin/arm-linux-g++ -DHAVE_CONFIG_H -I. -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -I./include -I./include -Wall -Wformat -W -isystem /usr/arm-buildroot-linux-gnueabihf/sysroot/usr/include -pthread -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os -pthread -c -o libs/acn/e131_transmit_test.o libs/acn/e131_transmit_test.cpp /usr/bin/arm-linux-g++ -DHAVE_CONFIG_H -I. -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -I./include -I./include -Wall -Wformat -W -isystem /usr/arm-buildroot-linux-gnueabihf/sysroot/usr/include -pthread -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os -pthread -c -o libs/acn/E131TestFramework.o libs/acn/E131TestFramework.cpp In file included from /opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0/ext/string_conversions.h:41:0, from /opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0/bits/basic_string.h:5402, from /opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0/string:52, from ./tools/ola_trigger/config.ypp:2: /opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0/cstdlib:75:25: fatal error: stdlib.h: No such file or directory #include_next <stdlib.h> ^ compilation terminated. ``` The C++ library in GCC 6 now provides its own `<stdlib.h>` header that wraps the C library header of the same name, so in `<cstdlib>` the header include ``` #include <stdlib.h> ``` has become ``` #include_next <stdlib.h> ``` `#include_next` is sensitive to the order of directories in the preprocessor's search path, so if that order is changed with `-isystem` then the compiler can't find the right header: ``` [1] /usr/arm-buildroot-linux-gnueabihf/sysroot/usr/include [2] /opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0 [..] End of search list. ``` `<cstdlib>` is located in [2] whereas `<stdlib.h>` (C library header) is in [1]. In this case, the `#include_next <stdlib.h>` statement in `<cstdlib>`, located in [2], is evaluated **after** the search path [1], so the compiler does not find the right system header. The problem here is that the OLA build system replaces the `-I` in the CFLAGS from libprotobuf with `-isystem` to fix some warnings treated as errors in the libprotobuf header files. `-isystem` should be used to suppress warnings in system headers only and the libprotobuf header files are not system files. The correct fix is to compile with less restrictions and remove `-Werror` for the build. This is already done by passing `--disable-fatal-warnings` to OLA. Fix the reordering of the GCC search paths by removing the replacement of `-I` with `-isystem`. Upstream status: https://github.com/OpenLightingProject/ola/pull/1126 Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Diffstat (limited to 'package/ola')
-rw-r--r--package/ola/0002-Remove-replacing-I-with-isystem.patch106
1 files changed, 106 insertions, 0 deletions
diff --git a/package/ola/0002-Remove-replacing-I-with-isystem.patch b/package/ola/0002-Remove-replacing-I-with-isystem.patch
new file mode 100644
index 000000000..010c99a3b
--- /dev/null
+++ b/package/ola/0002-Remove-replacing-I-with-isystem.patch
@@ -0,0 +1,106 @@
+From bbb03794def326c2e8ad2de523c5a61a4c8cb464 Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?J=C3=B6rg=20Krause?= <joerg.krause@embedded.rocks>
+Date: Thu, 22 Sep 2016 00:58:58 +0200
+Subject: [PATCH] Remove replacing -I with -isystem
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+Building OLA with a GCC 6 cross-toolchain fails:
+
+```
+/usr/bin/arm-linux-g++ -DHAVE_CONFIG_H -I. -D_LARGEFILE_SOURCE
+-D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -I./include -I./include
+-Wall -Wformat -W -isystem
+/usr/arm-buildroot-linux-gnueabihf/sysroot/usr/include -pthread
+-D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os
+-pthread -c -o libs/acn/e131_transmit_test.o
+libs/acn/e131_transmit_test.cpp
+/usr/bin/arm-linux-g++ -DHAVE_CONFIG_H -I. -D_LARGEFILE_SOURCE
+-D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -I./include -I./include
+-Wall -Wformat -W -isystem
+/usr/arm-buildroot-linux-gnueabihf/sysroot/usr/include -pthread
+-D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os
+-pthread -c -o libs/acn/E131TestFramework.o
+libs/acn/E131TestFramework.cpp
+In file included from
+/opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0/ext/string_conversions.h:41:0,
+ from
+/opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0/bits/basic_string.h:5402,
+ from
+/opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0/string:52,
+ from ./tools/ola_trigger/config.ypp:2:
+/opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0/cstdlib:75:25:
+fatal error: stdlib.h: No such file or directory
+ #include_next <stdlib.h>
+ ^
+compilation terminated.
+```
+
+The C++ library in GCC 6 now provides its own `<stdlib.h>` header that
+wraps the C library header of the same name, so in `<cstdlib>` the
+header include
+
+```
+#include <stdlib.h>
+```
+
+has become
+
+```
+#include_next <stdlib.h>
+```
+
+`#include_next` is sensitive to the order of directories in the
+preprocessor's search path, so if that order is changed with `-isystem`
+then the compiler can't find the right header:
+
+```
+[1] /usr/arm-buildroot-linux-gnueabihf/sysroot/usr/include
+[2] /opt/ext-toolchain/arm-buildroot-linux-gnueabihf/include/c++/6.1.0
+[..]
+End of search list.
+```
+
+`<cstdlib>` is located in [2] whereas `<stdlib.h>` (C library header) is
+in [1]. In this case, the `#include_next <stdlib.h>` statement in
+`<cstdlib>`, located in [2], is evaluated **after** the search path [1],
+so the compiler does not find the right system header.
+
+The problem is that the OLA build system replaces the `-I` in the CFLAGS
+from libprotobuf with `-isystem` to fix some warnings treated as errors
+in the libprotobuf header files.
+
+`-isystem` should be used to suppress warnings in system headers only
+and the libprotobuf header files are not system files.
+
+The correct fix is to compile with less restrictions and remove
+`-Werror` for the build.
+
+As using `-isystem` is reordering GCCs search path and using `-isystem`
+is really not necessary, remove the faulty replacement of `-I`.
+
+Upstream status: https://github.com/OpenLightingProject/ola/pull/1126
+
+Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
+---
+ config/ola.m4 | 3 ---
+ 1 file changed, 3 deletions(-)
+
+diff --git a/config/ola.m4 b/config/ola.m4
+index 2796cfb..d3b8cc8 100644
+--- a/config/ola.m4
++++ b/config/ola.m4
+@@ -24,9 +24,6 @@ AC_DEFUN([PROTOBUF_SUPPORT],
+ AC_REQUIRE_CPP()
+ PKG_CHECK_MODULES(libprotobuf, [protobuf >= $1])
+
+-# We want to replace -I with -isystem here to disable errors in the .h files
+-# See https://groups.google.com/forum/#!topic/open-lighting/39Mj0KXlCIk
+-libprotobuf_CFLAGS=`echo $libprotobuf_CFLAGS | sed 's/-I/-isystem /'`
+ AC_SUBST([libprotobuf_CFLAGS])
+
+ AC_ARG_WITH([protoc],
+--
+2.10.0
+