Re: [PATCH] ld: Allow R_X86_64_GOTPCREL for call *__tls_get_addr@GOTPCREL(%rip)

[Fangrui Song <maskray@google.com>]

On Fri, Jan 6, 2023 at 9:04 AM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Thu, Jan 5, 2023 at 1:06 PM Fangrui Song via Binutils
> <binutils@sourceware.org> wrote:
> >
> > _Thread_local int a;
> > int main() { return a; }
> >
> > % gcc -fno-plt -fpic a.c -fuse-ld=bfd -Wa,-mrelax-relocations=no
> > /usr/bin/ld.bfd: /tmp/ccSSBgrg.o: TLS transition from R_X86_64_TLSGD to R_X86_64_GOTTPOFF against `a' at 0xd in section `.text' failed
> > /usr/bin/ld.bfd: failed to set dynamic section sizes: bad value
> > collect2: error: ld returned 1 exit status
> >
> > This commit fixes the issue.
> >
> >     PR ld/24784
> >     * bfd/elf64-x86-64.c (elf_x86_64_check_tls_transition): Allow
> >       R_X86_64_GOTPCREL.
> > ---
> >  bfd/elf64-x86-64.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/bfd/elf64-x86-64.c b/bfd/elf64-x86-64.c
> > index 914f82d0151..095fe2e0fe6 100644
> > --- a/bfd/elf64-x86-64.c
> > +++ b/bfd/elf64-x86-64.c
> > @@ -1241,7 +1241,7 @@ elf_x86_64_check_tls_transition (bfd *abfd,
> >           if (largepic)
> >             return r_type == R_X86_64_PLTOFF64;
> >           else if (indirect_call)
> > -           return r_type == R_X86_64_GOTPCRELX;
> > +           return (r_type == R_X86_64_GOTPCRELX || r_type == R_X86_64_GOTPCREL);
> >           else
> >             return (r_type == R_X86_64_PC32 || r_type == R_X86_64_PLT32);
> >         }
> > --
> > 2.39.0.314.g84b9a713c41-goog
> >
>
> Since the new TLS sequence was added after R_X86_64_GOTPCRELX was
> required for call, R_X86_64_GOTPCREL should be invalid in this TLS sequence.
>
> --
> H.J.

I have multiple arguments (albeit no single one is very strong) that
this 1-deletion-1-addition change provides benefits for users (IMHO
with no burden to binutils at all).

Some projects may add -Wa,-mrelax-relocations=no to work around older
GNU ld. Then the project's toolchain requirement may increase and no
longer need to work around older GNU ld.
But a distribution may for some reason use a global -fno-plt (e.g.
Arch Linux) and then run into this TLS GD/LD->IE/LE optimization
issue.

rust src/ci/docker/host-x86_64/*musl/Dockerfile
openjdk/jdk19u make/autoconf/flags-cflags.m4 (this file appears to be
copied into quite a few projects)
Linux kernel arch/x86/boot/compressed/Makefile (not a good example as
it doesn't use TLS AFAICT)

R_X86_64_GOTPCREL isn't purely usefull. It may help linker design: for
R_X86_64_GOTPCRELX/R_X86_64_REX_GOTPCRELX, the linker can make a
decision upfront whether a GOT entry is needed
(this affects the size of .got, which may affect section layout and
whether other relocations may overflow).
This may increase risk of 32-bit relocation overflow.
R_X86_64_GOTPCREL can mitigate the risk while being aware to the user.

rustc somehow disables x86 relaxed relocations and defaults to `-Z
plt=no` and now relies on llvm-project to work around the GNU ld
compatibility issue.




-- 
宋方睿