Re: [PATCH v3] x86: Generate PLT32 relocation for ".long foo@PLT - .L4"

On Thu, Jul 30, 2026 at 3:14=E2=80=AFPM Jan Beulich <[email protected]> wro=
te:
>
> On 27.07.2026 10:44, H.J. Lu wrote:
> > On Mon, Jul 27, 2026 at 3:58=E2=80=AFPM H.J. Lu <[email protected]> w=
rote:
> >> On Mon, Jul 27, 2026 at 3:46=E2=80=AFPM Jan Beulich <[email protected]=
> wrote:
> >>> On 27.07.2026 05:47, H.J. Lu wrote:
> >>>> On Sat, Jul 25, 2026 at 10:35=E2=80=AFAM H.J. Lu <[email protected]=
m> wrote:
> >>>>>
> >>>>> Add BFD_RELOC_X86_64_PLT32 and use it to generate R_X86_64_PLT32 fo=
r
> >>>>> directives like ".long foo@PLT - .L4" so that PLT entries are used =
to
> >>>>> resolve the PC32 relocation against function symbols for such direc=
tives.
> >>>>>
> >>>>> bfd/
> >>>>>
> >>>>> PR gas/34423
> >>>>> * elf64-x86-64.c (x86_64_reloc_map): Add BFD_RELOC_X86_64_PLT32.
> >>>>> * reloc.c (bfd_reloc_code_real): Add BFD_RELOC_X86_64_PLT32.
> >>>>> * bfd-in2.h: Regenerated.
> >>>>> * libbfd.h: Likewise.
> >>>>>
> >>>>> gas/
> >>>>>
> >>>>> PR gas/34423
> >>>>> * config/tc-i386.c (x86_cons): Return BFD_RELOC_X86_64_PLT32 for
> >>>>> directives like ".long foo@PLT - .L4".
> >>>>> (tc_gen_reloc): Generate R_X86_64_PLT32 for BFD_RELOC_X86_64_PLT32
> >>>>> with addend computed like R_X86_64_PC32.
> >>>>> * testsuite/gas/i386/reloc64.l: Updated.
> >>>>> * testsuite/gas/i386/reloc64.s: Replace ".long xtrn@plt - ." with
> >>>>> ".long xtrn@plt - _start".
> >>>>> * testsuite/gas/i386/ilp32/reloc64.l: Updated.
> >>>>> * testsuite/gas/i386/ilp32/reloc64.s: Replace ".long xtrn@plt - ."
> >>>>> with ".long xtrn@plt - ptr".
> >>>>> * testsuite/gas/i386/x86-64-jump-table.d: New file.
> >>>>> * testsuite/gas/i386/x86-64-jump-table.d: Likewise.
> >>>>> * testsuite/gas/i386/x86-64-jump-table.s: Likewise.
> >>>>> * testsuite/gas/i386/x86-64.exp: Run x86-64-jump-table.
> >>>>> * testsuite/gas/i386/ilp32/x86-64-jump-table.d: New file.
> >>>>>
> >>>>> ld/
> >>>>>
> >>>>> PR gas/34423
> >>>>> * testsuite/ld-x86-64/pr34423.c: New file.
> >>>>> * testsuite/ld-x86-64/x86-64-jump-table.s: Likewise.
> >>>>> * testsuite/ld-x86-64/x86-64.exp: Run gas/34423 tests.
> >>>>
> >>>> Add i386 support like LLVM assembler in the v2 patch.
> >>>
> >>> Why would S_IS_LOCAL() be the criteria? There isn't anything wrong wi=
th a
> >>
> >> I will remove it.
> >
> > Removed in v3.
> >
> >>> global, at the very least a hidden/protected one, is there? Instead d=
on't
> >>> you want to verify that the subtrahend is in the current section (whi=
ch
> >>> of course we may not know until after having parsed all input)?
> >>
> >> It will be reverified later :
> >>
> >> [hjl@gnu-tgl-3 pic-1]$ cat bar.s
> >>         .section .rodata
> >>         .p2align 2
> >> .L4:
> >>         .long foo1@PLT - bar
> >> [hjl@gnu-tgl-3 pic-1]$ ./as  -o bar.o bar.s
> >> bar.s: Assembler messages:
> >> bar.s:4: Error: can't resolve foo1 - bar
> >> [hjl@gnu-tgl-3 pic-1]$
> >>
> >>> Surprising code like
> >>>
> >>> +      if (fixp->fx_r_type =3D=3D BFD_RELOC_X86_64_PC32_TO_PLT32)
> >>> +       code =3D BFD_RELOC_X86_64_PC32_TO_PLT32;
> >>> +      else if (fixp->fx_r_type =3D=3D BFD_RELOC_386_PC32_TO_PLT32)
> >>> +       code =3D BFD_RELOC_386_PLT32;
> >>>
> >>> imo wants at least a brief comment as to the different behavior for i=
386
> >>> vs x86-64.
> >>>
> >>
> >> I will add some comments.
> >
> > Comments are added in v3.  The difference is explicit addend and implic=
it
> > addend.
> >
> > Here is the v3 patch.
>
> Thanks, but once again I have to ask: Why did this need rushing in? I ask=
ed

We can get real usage today.

> before that you please allow at least a week between submission and
> committing. Perhaps except when a change is entirely uncontroversial, or
> really urgent to fix e.g. recently introduced breakage (neither of which =
is
> the case here).
>
> As to the change itself: We have had support for .slong for a long time.
> Imo
>
>         .long foo@PLT - .L4
>
> is a bogus construct on 64-bit, and
>
>         .slong foo@PLT - .L4

We have

.long foo - .L4

It is natural to add

.long foo@PLT - .L4

> should be demanded there.
>
> Then: Why do gas/testsuite/gas/i386/reloc{32,64}.s need fiddling with? .
> should be as suitable to use there as _start / ptr.

Since

.long xtrn@plt - .

works now.  I change it to " .long xtrn@plt - _start" and added
new working ones in plt.s

> Jan


--=20
H.J.