Joseph Myers, le mar. 03 avril 2018 00:10:30 +0000, a ecrit: > On Mon, 2 Apr 2018, Samuel Thibault wrote: > > Others really pose problem, like ./sysdeps/mach/hurd/bits/fcntl.h's > > l_type/l_whence being int instead of short. > > Where something is problematic to fix, because a fix would break the ABI > or needs some external feature, there is an xfail mechanism internal to > the conform tests.
Ok, thanks for the info, I could work it out :) There are some ABI-compatible changes which can be made. One of them is struct shmid_ds's shm_segsz which is int but should be size_t. That is defined as int in both ./sysvipc/sys/shm.h and ./sysdeps/gnu/bits/shm.h. I have checked that the kfreebsd port uses its own shm.h that defines it to size_t. I'd tend to think that we do not have any 64bit port using ./sysvipc/sys/shm.h or ./sysdeps/gnu/bits/shm.h, and thus we can just fix them into size_t since on 32bit ports it is ABI-compatible with int? (except the bitness, but I guess it does not problems in practice?) Samuel
