Bug 224947
| Summary: | audio/libsidplay: fails to build with clang 6.0 (blocks 8 ports) | ||
|---|---|---|---|
| Product: | Ports & Packages | Reporter: | Jan Beich <jbeich> |
| Component: | Individual Port(s) | Assignee: | Jan Beich <jbeich> |
| Status: | Closed FIXED | ||
| Severity: | Affects Only Me | CC: | dim, ehaupt, tijl |
| Priority: | --- | Keywords: | needs-patch |
| Version: | Latest | Flags: | tijl:
maintainer-feedback+
jbeich: maintainer-feedback? (dim) jbeich: maintainer-feedback? (ehaupt) |
| Hardware: | Any | ||
| OS: | Any | ||
| See Also: |
https://bugzilla.mozilla.org/show_bug.cgi?id=1329520 https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=225307 |
||
| Bug Depends on: | |||
| Bug Blocks: | 224669 | ||
Most configure_error bustage in bug 224669 is actually due to C++11 attributes poisoning. So this is caused by a bunch of GNU configure declaring their own version of exit(), without any attributes, then our stdlib.h declaring it with attributes, right? In that case using USE_CXXSTD=gnu++98 might also be an acceptable workaround, since that was the original standard with which those configure scripts were executed. In C++11 or later, I would indeed say that declaring functions inconsistently should be a cause for errors, or at least warnings. It is a pity you cannot turn these warnings off, though. (In reply to Dimitry Andric from comment #2) > So this is caused by a bunch of GNU configure declaring their own > version of exit(), without any attributes, then our stdlib.h > declaring it with attributes, right? Tijl, do you know why autoconf doesn't pick up the system declaration? The following snippet suggests the loop should terminate on the first iteration # Some people use a C++ compiler to compile C. Since we use `exit', # in C++ we need to declare it. In case someone uses the same compiler # for both compiling C and C++ we need to have the C++ compiler decide # the declaration of exit, since it's the most demanding environment. [...] for ac_declaration in \ '' \ 'extern "C" void std::exit (int) throw (); using std::exit;' \ 'extern "C" void std::exit (int); using std::exit;' \ 'extern "C" void exit (int) throw ();' \ 'extern "C" void exit (int);' \ 'void exit (int);' do cat >conftest.$ac_ext <<_ACEOF /* confdefs.h. */ _ACEOF cat confdefs.h >>conftest.$ac_ext cat >>conftest.$ac_ext <<_ACEOF /* end confdefs.h. */ $ac_declaration #include <stdlib.h> int main () { exit (42); ; return 0; } _ACEOF > In that case using USE_CXXSTD=gnu++98 might also be an acceptable > workaround, since that was the original standard with which those > configure scripts were executed. Requires checking consumers for regressions. https://gcc.gnu.org/wiki/Cxx11AbiCompatibility > In C++11 or later, I would indeed say that declaring functions > inconsistently should be a cause for errors, or at least warnings. > It is a pity you cannot turn these warnings off, though. Why Clang accepts #include <stdlib.h> extern "C" void exit(int); but rejects extern "C" void exit(int); #include <stdlib.h> Is such an inconsistent behavior really standardized? It looks like you can just add USES=autoreconf to the port Makefile to regenerate configure. The included configure has been generated with an old version of autoconf. Indeed, USES=autoreconf works around Clang quirk but requires adjusting any patches against configure or Makefile.in. audio/libsidplay still fails. Not sure about the fix: s/sbyte/ubyte/ or static_cast or -Wno-error=narrowing (e.g., ports r458130). samples.cpp:80:2: error: constant expression evaluates to 128 which cannot be narrowed to type 'sbyte' (aka 'signed char') [-Wc++11-narrowing] 0x80,0x91,0xa2,0xb3,0xc4,0xd5,0xe6,0xf7, ^~~~ samples.cpp:80:2: note: insert an explicit cast to silence this issue 0x80,0x91,0xa2,0xb3,0xc4,0xd5,0xe6,0xf7, ^~~~ static_cast<sbyte>( ) [...] build log: https://ptpb.pw/NPiS (In reply to Jan Beich from comment #5) > Indeed, USES=autoreconf works around Clang quirk but requires adjusting any > patches against configure or Makefile.in. audio/libsidplay still fails. Not > sure about the fix: s/sbyte/ubyte/ or static_cast or -Wno-error=narrowing > (e.g., ports r458130). > > samples.cpp:80:2: error: constant expression evaluates to 128 which cannot > be narrowed to type 'sbyte' (aka 'signed char') [-Wc++11-narrowing] > 0x80,0x91,0xa2,0xb3,0xc4,0xd5,0xe6,0xf7, > ^~~~ > samples.cpp:80:2: note: insert an explicit cast to silence this issue > 0x80,0x91,0xa2,0xb3,0xc4,0xd5,0xe6,0xf7, > ^~~~ > static_cast<sbyte>( ) Yes, this is pretty annoying, inserting all those static casts would be horrendous. If the warning can be turned off, so much the better. A commit references this bug: Author: jbeich Date: Fri Jan 26 14:39:16 UTC 2018 New revision: 459997 URL: https://svnweb.freebsd.org/changeset/ports/459997 Log: audio/libsidplay: unbreak build with Clang 6 (C++14 by default) configure:4036: c++ -o conftest -O2 -pipe -O3 -fno-strict-aliasing -fno-omit-frame-pointer -march=native -fstack-protector -fstack-protector conftest.cc >&5 In file included from conftest.cc:32: In file included from /usr/include/c++/v1/stdlib.h:94: /usr/include/stdlib.h:97:1: error: function declared '[[noreturn]]' after its first declaration _Noreturn void exit(int); ^ /usr/include/sys/cdefs.h:280:22: note: expanded from macro '_Noreturn' #define _Noreturn [[noreturn]] ^ conftest.cc:11:6: note: declaration missing '[[noreturn]]' attribute is here void exit (int); ^ [...] configure:4053: error: cannot compute sizeof (char), 77 samples.cpp:80:2: error: constant expression evaluates to 128 which cannot be narrowed to type 'sbyte' (aka 'signed char') [-Wc++11-narrowing] 0x80,0x91,0xa2,0xb3,0xc4,0xd5,0xe6,0xf7, ^~~~ samples.cpp:80:2: note: insert an explicit cast to silence this issue 0x80,0x91,0xa2,0xb3,0xc4,0xd5,0xe6,0xf7, ^~~~ static_cast<sbyte>( ) PR: 224947 Reported by: pkg-fallout (blocks 8 ports) Submitted by: tijl (autoreconf) Approved by: portmgr blanket Changes: head/audio/libsidplay/Makefile head/audio/libsidplay/files/ head/audio/libsidplay/files/patch-src_samples.cpp (In reply to commit-hook from comment #7) Oops, -O3 .. -fno-omit-frame-pointer -march=native are my local CFLAGS I forgot to strip from commit message. |
FreeBSD -CURRENT plans to update base Clang to 6.0.0 soon. While testing this port failed to build. The underlying issue appears to be Clang's idiosyncrasy with C++11 attributes. GCC 6+ is unaffected. $ cat a.cc extern "C" void exit(int); #include <stdlib.h> int main() { return 0; } $ c++ a.cc In file included from a.cc:3: In file included from /usr/include/c++/v1/stdlib.h:94: /usr/include/stdlib.h:97:1: error: function declared '[[noreturn]]' after its first declaration _Noreturn void exit(int); ^ /usr/include/sys/cdefs.h:278:22: note: expanded from macro '_Noreturn' #define _Noreturn [[noreturn]] ^ a.cc:1:17: note: declaration missing '[[noreturn]]' attribute is here extern "C" void exit(int); ^ 1 error generated. =========================================================================== =======================<phase: configure >============================ ===> Configuring for libsidplay-1.36.59_1 [...] checking whether exception-handling is supported... no [...] checking size of char... configure: error: cannot compute sizeof (char), 77 See `config.log' for more details. ===> Script "configure" failed unexpectedly. To reproduce: $ poudriere jail -cj clang6-amd64 -a amd64 -v projects/clang600-import -m svn+https $ poudriere bulk -Ctj clang6-amd64 audio/libsidplay http://package18.nyi.freebsd.org/data/headamd64PR224669-default/2018-01-02_08h32m49s/logs/errors/libsidplay-1.36.59_1.log