[PATCH 3/3] LoongArch: Enable shared library and PIE support for loongarch*-elf
mintsuki
mintsuki@protonmail.com
Tue Sep 8 21:37:40 GMT 2026
loongarch*-elf withholds GENERATE_SHLIB_SCRIPT and GENERATE_PIE_SCRIPT,
so ld refuses -shared and -pie for it. The exclusion arrived with the
original LoongArch ld port, copied from the RISC-V emulparams, and no
reason was recorded for it either. loongarch*-elf does not set
EMBEDDED.
The rationale is the one given for the RISC-V patch in this series; it
applies here unchanged.
ld/
* emulparams/elf32loongarch-defs.sh (GENERATE_SHLIB_SCRIPT): Set
unconditionally rather than only for non-loongarch*-elf targets.
(GENERATE_PIE_SCRIPT): Likewise.
* emulparams/elf64loongarch-defs.sh (GENERATE_SHLIB_SCRIPT):
Likewise.
(GENERATE_PIE_SCRIPT): Likewise.
Assisted-by: Claude:claude-opus-5
Signed-off-by: Mintsuki <mintsuki@protonmail.com>
---
ld/emulparams/elf32loongarch-defs.sh | 11 ++---------
ld/emulparams/elf64loongarch-defs.sh | 11 ++---------
2 files changed, 4 insertions(+), 18 deletions(-)
diff --git a/ld/emulparams/elf32loongarch-defs.sh b/ld/emulparams/elf32loongarch-defs.sh
index f80f5742..05626db0 100644
--- a/ld/emulparams/elf32loongarch-defs.sh
+++ b/ld/emulparams/elf32loongarch-defs.sh
@@ -16,15 +16,8 @@ if test `echo "$host" | sed -e s/64//` = `echo "$target" | sed -e s/64//`; then
esac
fi
-# Enable shared library support for everything except an embedded elf target.
-case "$target" in
- loongarch*-elf)
- ;;
- *)
- GENERATE_SHLIB_SCRIPT=yes
- GENERATE_PIE_SCRIPT=yes
- ;;
-esac
+GENERATE_SHLIB_SCRIPT=yes
+GENERATE_PIE_SCRIPT=yes
IREL_IN_PLT=
TEXT_START_ADDR=0x10000
diff --git a/ld/emulparams/elf64loongarch-defs.sh b/ld/emulparams/elf64loongarch-defs.sh
index a8147bf7..9b9deabb 100644
--- a/ld/emulparams/elf64loongarch-defs.sh
+++ b/ld/emulparams/elf64loongarch-defs.sh
@@ -16,15 +16,8 @@ if test `echo "$host" | sed -e s/64//` = `echo "$target" | sed -e s/64//`; then
esac
fi
-# Enable shared library support for everything except an embedded elf target.
-case "$target" in
- loongarch*-elf)
- ;;
- *)
- GENERATE_SHLIB_SCRIPT=yes
- GENERATE_PIE_SCRIPT=yes
- ;;
-esac
+GENERATE_SHLIB_SCRIPT=yes
+GENERATE_PIE_SCRIPT=yes
# In all cases, the number is big-endian.
# LoongArch nop is 'andi $r0,$r0,0'.
--
2.55.0
On Tuesday, September 8th, 2026 at 11:34 PM, mintsuki <mintsuki@protonmail.com> wrote:
> These three tests match .rela.dyn entries with the symbol index of
> r_info written out in full, and that index depends on the default hash
> style. A GNU hash table requires the undefined symbols to precede the
> defined ones in .dynsym, so an undefined weak in these tests sorts to
> index 2 with --hash-style=both or gnu, and to the end of the table with
> sysv. The expectations were written against the linux default of
> "both"; the default for a bare-metal target is sysv, where the same
> three relocations come out correct but against different indices.
>
> The tests are unsupported on loongarch*-elf today, so this changes
> nothing yet; it is needed by the following patch, which enables shared
> library and PIE support for that target and so lets them run. Pinning
> the hash style rather than relaxing the expectation keeps them strict,
> and matches what a good number of tests elsewhere in the testsuite
> already do.
>
> ld/
> * testsuite/ld-loongarch-elf/relr-data-shared.d: Pass
> --hash-style=both.
> * testsuite/ld-loongarch-elf/relr-discard-shared.d: Likewise.
> * testsuite/ld-loongarch-elf/relr-got-shared.d: Likewise.
>
> Assisted-by: Claude:claude-opus-5
> Signed-off-by: Mintsuki <mintsuki@protonmail.com>
> ---
> ld/testsuite/ld-loongarch-elf/relr-data-shared.d | 2 +-
> ld/testsuite/ld-loongarch-elf/relr-discard-shared.d | 2 +-
> ld/testsuite/ld-loongarch-elf/relr-got-shared.d | 2 +-
> 3 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/ld/testsuite/ld-loongarch-elf/relr-data-shared.d b/ld/testsuite/ld-loongarch-elf/relr-data-shared.d
> index 37e4c0da..81eaa577 100644
> --- a/ld/testsuite/ld-loongarch-elf/relr-data-shared.d
> +++ b/ld/testsuite/ld-loongarch-elf/relr-data-shared.d
> @@ -1,5 +1,5 @@
> #source: relr-data.s
> -#ld: -shared -z pack-relative-relocs -T relr-relocs.ld
> +#ld: -shared -z pack-relative-relocs --hash-style=both -T relr-relocs.ld
> #readelf: -rW
>
> Relocation section '\.rela\.dyn' at offset 0x[0-9a-f]+ contains 6 entries:
> diff --git a/ld/testsuite/ld-loongarch-elf/relr-discard-shared.d b/ld/testsuite/ld-loongarch-elf/relr-discard-shared.d
> index 8bfd8ba5..7456793c 100644
> --- a/ld/testsuite/ld-loongarch-elf/relr-discard-shared.d
> +++ b/ld/testsuite/ld-loongarch-elf/relr-discard-shared.d
> @@ -1,5 +1,5 @@
> #source: relr-discard.s
> -#ld: -shared -z pack-relative-relocs -T relr-discard.ld
> +#ld: -shared -z pack-relative-relocs --hash-style=both -T relr-discard.ld
> #readelf: -rW
>
> Relocation section '\.rela\.dyn' at offset 0x[0-9a-f]+ contains 1 entry:
> diff --git a/ld/testsuite/ld-loongarch-elf/relr-got-shared.d b/ld/testsuite/ld-loongarch-elf/relr-got-shared.d
> index 169e0e5d..1c0d2b91 100644
> --- a/ld/testsuite/ld-loongarch-elf/relr-got-shared.d
> +++ b/ld/testsuite/ld-loongarch-elf/relr-got-shared.d
> @@ -1,5 +1,5 @@
> #source: relr-got.s
> -#ld: -shared -z pack-relative-relocs -T relr-relocs.ld
> +#ld: -shared -z pack-relative-relocs --hash-style=both -T relr-relocs.ld
> #readelf: -rW
>
> Relocation section '\.rela\.dyn' at offset 0x[0-9a-f]+ contains 3 entries:
> --
> 2.55.0
>
>
> On Tuesday, September 8th, 2026 at 11:28 PM, mintsuki <mintsuki@protonmail.com> wrote:
>
> > riscv*-elf is the only RISC-V target for which ld refuses -shared and
> > -pie. elf32lriscv-defs.sh has withheld GENERATE_SHLIB_SCRIPT and
> > GENERATE_PIE_SCRIPT from it since be1dce26256a ("RISC-V: Disable shared
> > library support for embedded elf."), whose commit message is a ChangeLog
> > entry and records no reason. Nothing in the RISC-V backend depends on
> > the exclusion, and riscv*-elf does not even set EMBEDDED.
> >
> > Other bare-metal ELF targets do generate the scripts. aarch64*-elf was
> > given PIE support deliberately, in 3af09e8db308 ("Support -pie for
> > aarch64*-elf targets."), and arm-elf and x86_64-elf have both. The
> > general question was raised in f2aaebdb9797 ("GENERATE_SHLIB_SCRIPT vs.
> > EMBEDDED.") and settled the other way: that patch was reverted in
> > 8a46e447a88d, the author concluding that the ELF and program headers are
> > still written to the file, so an embedded ELF target producing an ET_DYN
> > image is not the mistake he had taken it for.
> >
> > ld/
> > * emulparams/elf32lriscv-defs.sh (GENERATE_SHLIB_SCRIPT): Set
> > unconditionally rather than only for non-riscv*-elf targets.
> > (GENERATE_PIE_SCRIPT): Likewise.
> >
> > Assisted-by: Claude:claude-opus-5
> > Signed-off-by: Mintsuki <mintsuki@protonmail.com>
> > ---
> > ld/emulparams/elf32lriscv-defs.sh | 11 ++---------
> > 1 file changed, 2 insertions(+), 9 deletions(-)
> >
> > diff --git a/ld/emulparams/elf32lriscv-defs.sh b/ld/emulparams/elf32lriscv-defs.sh
> > index 01655616..5c0dac2a 100644
> > --- a/ld/emulparams/elf32lriscv-defs.sh
> > +++ b/ld/emulparams/elf32lriscv-defs.sh
> > @@ -16,15 +16,8 @@ if test `echo "$host" | sed -e s/64//` = `echo "$target" | sed -e s/64//`; then
> > esac
> > fi
> >
> > -# Enable shared library support for everything except an embedded elf target.
> > -case "$target" in
> > - riscv*-elf)
> > - ;;
> > - *)
> > - GENERATE_SHLIB_SCRIPT=yes
> > - GENERATE_PIE_SCRIPT=yes
> > - ;;
> > -esac
> > +GENERATE_SHLIB_SCRIPT=yes
> > +GENERATE_PIE_SCRIPT=yes
> >
> > IREL_IN_PLT=
> > TEXT_START_ADDR=0x10000
> > --
> > 2.55.0
> >
> >
> > On Tuesday, September 8th, 2026 at 11:25 PM, mintsuki <mintsuki@protonmail.com> wrote:
> >
> > > ld refuses both -shared and -pie for riscv*-elf and loongarch*-elf,
> > > because their emulparams files withhold GENERATE_SHLIB_SCRIPT and
> > > GENERATE_PIE_SCRIPT from those two targets specifically. This is not
> > > the EMBEDDED policy at work -- neither file sets EMBEDDED -- but a pair
> > > of one-off "case $target" checks: be1dce26256a for RISC-V, copied into
> > > the LoongArch port from it. Other bare-metal ELF targets generate the
> > > scripts; aarch64*-elf was given PIE support on purpose in 3af09e8db308,
> > > and f2aaebdb9797, which would have taken it away from EMBEDDED targets
> > > generally, was reverted in 8a46e447a88d:
> > >
> > > https://inbox.sourceware.org/binutils/20191108103411.GU6708@bubble.grove.modra.org/
> > >
> > > What this blocks in practice is freestanding relocatable outputs.
> > > Those need -pie (or, alternatively, -shared -Bsymbolic), as already
> > > supported by x86_64-elf, i686-elf, aarch64-elf, etc...
> > > On riscv64-elf and loongarch64-elf compilers accepts -fPIE and -fPIC,
> > > but the link step with ld.bfd dies with "-pie not supported" or
> > > "-shared not supported", so a bare-metal toolchain that is otherwise
> > > complete cannot build one, and a *-linux (or otherwise other hosted)
> > > ld.bfd or lld has to be used instead for the link step.
> > >
> > > Patch 2 is testsuite preparation. Three LoongArch DT_RELR tests write
> > > out the symbol index of r_info in full, and that index depends on the
> > > default hash style: GNU hash requires undefined symbols to precede
> > > defined ones in .dynsym, so an undefined weak in those tests sorts to
> > > index 2 under the linux default of "both" and to the end of the table
> > > under the bare-metal default of sysv. The relocations themselves are
> > > correct either way. Without that patch, patch 3 turns those three from
> > > unsupported into FAIL.
> > >
> > > Based on 2986ed8ba365 ("Constify targ_desc"). Tested with gcc 16.2.0
> > > riscv64-elf and loongarch64-elf toolchains and dejagnu 1.6.3.
> > > make check-ld, before and after the series:
> > >
> > > riscv64-elf 541 -> 763 passes, 0 unexpected failures either way
> > > loongarch64-elf 540 -> 703 passes, 0 unexpected failures either way
> > >
> > > No test regresses on either target, and there are no unexpected
> > > successes; the increase is the shared and PIE tests that were previously
> > > skipped now running. I have not tested the 32-bit variants, riscv32-elf
> > > and loongarch32-elf, which the same emulparams files cover.
> > >
> > > Mintsuki (3):
> > > RISC-V: Enable shared library and PIE support for riscv*-elf
> > > ld: Pin the hash style in the LoongArch DT_RELR shared tests
> > > LoongArch: Enable shared library and PIE support for loongarch*-elf
> > >
> > > ld/emulparams/elf32loongarch-defs.sh | 11 ++---------
> > > ld/emulparams/elf32lriscv-defs.sh | 11 ++---------
> > > ld/emulparams/elf64loongarch-defs.sh | 11 ++---------
> > > ld/testsuite/ld-loongarch-elf/relr-data-shared.d | 2 +-
> > > ld/testsuite/ld-loongarch-elf/relr-discard-shared.d | 2 +-
> > > ld/testsuite/ld-loongarch-elf/relr-got-shared.d | 2 +-
> > > 6 files changed, 9 insertions(+), 30 deletions(-)
> > >
> > >
More information about the Binutils
mailing list