Open CarlWachter opened 4 months ago
Why is this issue new?
Can't we undefine HAVE_FCNTL
if that is a problem?
It's related to the new kernel version. Set FD_CLOEXEC just in case the kernel does not support O_CLOEXEC
, as far as i understand it O_CLOEXEC
is now supported, as it is exported by cbindgen.
Can't we undefine
HAVE_FCNTL
if that is a problem? It is also used for a lot of other things, so removing it entirely may cause issues.
My suggestion would be to remove the previously referenced code and see if that solves it, but I don't have the time to deal with rebuilding the toolchain right now, so I'll just leave it for now.
as far as i understand it
O_CLOEXEC
is now supported, as it is exported by cbindgen.
It's only a stub, so there is no meaningful functionality behind it.
My suggestion would be to remove the previously referenced code
Why not solve the linking issue instead, so the kernel stub gets called?
Currently Fortran support is broken, due to an issue linking fcntl. This is due to the following implementation in gcc: https://github.com/hermit-os/gcc/blob/17f4b3f33eb235780db88692508c3bb13b3a32d0/libbacktrace/posix.c#L82C3-L82C8