yaxunl added a comment. In D133705#3793931 <https://reviews.llvm.org/D133705#3793931>, @MaskRay wrote:
> I know very little about HIP, but I am concerned with relying on extensions > as well. For example, I've seen `libc++.a.1` (we use this for the real > archive while `libc++.a` is a linker script) and `.la` (libtool). > A `.so` file is sometimes a linker script and it may reference an archive > file. For example glibc `libc.so` typically does something like `GROUP ( > /lib/x86_64-linux-gnu/libc.so.6 /usr/lib/x86_64-linux-gnu/libc_nonshared.a > AS_NEEDED ( /lib64/ld-linux-x86-64.so.2 ) )` > > An ELF linker doesn't really care what extensions are used for what kind of > input (archive,relocatable file,shared object). `.a` and `.so` are only used > for `-lfoo` lookup. For HIP, we only need to unbundle archive files. We always try to unbundle files passed by `-l` options. If it is not archive files, it is OK, since the unbundled will generate an empty file. For input files, the current patch only tries to unbundle '*.a' or '*.lib' files. If it is a concern that some archive files not with extension '*.a' or '*.lib' are missed, one solution might be to try to unbundle any files classified as 'Nothing' (which is how clang classifies archive files). CHANGES SINCE LAST ACTION https://reviews.llvm.org/D133705/new/ https://reviews.llvm.org/D133705 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits