Per discussion on gentoo-dev, sys-libs/glog uses the "unwind" USE flag differently from other packages, and more in line with the "libunwind" flag from sys-libs/libcxx et al. Please consider renaming the flag to libunwind.
Currently: profiles/use.desc: unwind - Add support for call stack unwinding and function name resolution dev-cpp/glog/metadata.xml: <flag name="unwind">Use <pkg>sys-libs/libunwind</pkg> for stack unwinding instead of glibc/gcc (may be more reliable on x86_64)</flag> GLog src/utilities.h: https://github.com/google/glog/blob/9630e0e848da22e27b346c38d9b05f0a16cbf7b3/src/utilities.h#L86-L128 [[[ // There are three different ways we can try to get the stack trace: // // 1) The libunwind library. This is still in development, and as a // separate library adds a new dependency, but doesn't need a frame // pointer. It also doesn't call malloc. // // 2) Our hand-coded stack-unwinder. This depends on a certain stack // layout, which is used by gcc (and those systems using a // gcc-compatible ABI) on x86 systems, at least since gcc 2.95. // It uses the frame pointer to do its work. // // 3) The gdb unwinder -- also the one used by the c++ exception code. // It's obviously well-tested, but has a fatal flaw: it can call // malloc() from the unwinder. This is a problem because we're // trying to use the unwinder to instrument malloc(). // // 4) The Windows API CaptureStackTrace. ... #if defined(HAVE_LIB_UNWIND) # define STACKTRACE_H "stacktrace_libunwind-inl.h" #elif !defined(NO_FRAME_POINTER) # if defined(__i386__) && __GNUC__ >= 2 # define STACKTRACE_H "stacktrace_x86-inl.h" # elif defined(__x86_64__) && __GNUC__ >= 2 && HAVE_UNWIND_H # define STACKTRACE_H "stacktrace_x86_64-inl.h" # elif (defined(__ppc__) || defined(__PPC__)) && __GNUC__ >= 2 # define STACKTRACE_H "stacktrace_powerpc-inl.h" # elif defined(OS_WINDOWS) # define STACKTRACE_H "stacktrace_windows-inl.h" # endif #endif #if !defined(STACKTRACE_H) && defined(HAVE_EXECINFO_H) # define STACKTRACE_H "stacktrace_generic-inl.h" #endif ]]] (HAVE_LIB_UNWIND is defined when libunwind is used.) GLog src/stacktrace_libunwind-inl.h: https://github.com/google/glog/blob/f581614a78c68da17700b039ea69c774c1c6b892/src/stacktrace_libunwind-inl.h [[[ // Produce stack trace using libunwind ]]] GLog src/stacktrace_x86-inl.h: https://github.com/google/glog/blob/f278d734c1c02901631e04d1ac9099a4897ab75b/src/stacktrace_x86-inl.h [[[ // Produce stack trace ]]] GLog src/stacktrace_x86_64-inl.h (unwind.h required): https://github.com/google/glog/blob/f365379532ad3c0d86fa1b52434392df256166b4/src/stacktrace_x86_64-inl.h [[[ // Produce stack trace using libgcc ]]] GLog src/stacktrace_generic-inl.h (execinfo.h required): https://github.com/google/glog/blob/f365379532ad3c0d86fa1b52434392df256166b4/src/stacktrace_generic-inl.h [[[ // Portable implementation - just use glibc // // Note: The glibc implementation may cause a call to malloc. // This can cause a deadlock in HeapProfiler. ]]] sys-devel/clang and sys-devel/gcc provide unwind.h. sys-libs/glibc provides execinfo.h. sys-libs/musl (1.1.24) does not provide execinfo.h. sys-libs/uclibc-ng provides execinfo.h. So most architectures have non-libunwind fallback, but Musl ARM32/ARM64/MIPS/... have no fallback.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=a62ec2473825e637157e1dbfe9efe2d4a95d7c50 commit a62ec2473825e637157e1dbfe9efe2d4a95d7c50 Author: Arfrever Frehtes Taifersar Arahesis <Arfrever@Apache.Org> AuthorDate: 2021-04-01 23:00:00 +0000 Commit: Mike Gilbert <floppym@gentoo.org> CommitDate: 2021-04-02 15:00:48 +0000 dev-cpp/glog: Rename "unwind" USE flag to "libunwind". Enable it by default. Closes: https://bugs.gentoo.org/618190 Signed-off-by: Arfrever Frehtes Taifersar Arahesis <Arfrever@Apache.Org> Signed-off-by: Mike Gilbert <floppym@gentoo.org> dev-cpp/glog/glog-0.3.4-r1.ebuild | 8 ++++---- dev-cpp/glog/metadata.xml | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-)
