https://sourceware.org/bugzilla/show_bug.cgi?id=34172
Bug ID: 34172
Summary: [RISC-V ld] intra section boundary misses C.J relax
Product: binutils
Version: unspecified
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: ld
Assignee: unassigned at sourceware dot org
Reporter: l784896635 at gmail dot com
Target Milestone: ---
During RISC-V linker regression/conformance testing, I reduced this
linker-stage issue.
## Actual behavior
GNU ld.bfd leaves the final tail call as a 32-bit jal zero even though the
resolved displacement is 0x7fe, which is inside the c.j range, while ld.lld
emits c.j on the same object.
## Expected behavior
The linker should produce the semantically correct RISC-V output for this
reduced testcase, or issue a controlled diagnostic when the input is outside
the supported semantics. It should not silently emit a wrong output, lose
required metadata/relocations/attributes, miss a required relaxation, report a
false range/non-PIC/non-convergence error, or crash.
## Test input and intent
A same-section tail call sits exactly near the c.j boundary and should use the
16-bit compressed form once the exact post-relax displacement is known.
## Target and route
- Target tracker: GNU binutils / ld
- Target ISA / ABI: `rv64imac` / `lp64`
- First failing stage: link
- Toolchain route used in reduction: gas+ld.bfd with gas+ld.lld control
## Reproduction evidence
3 clean reproductions in packaged run1..run3 evidence show stable ld.bfd output
at 0x10812: jal zero, 0x11010 <c>; the ld.lld control on the same a.o emits c.j
to c instead.
The reproducer bundle contains the reduced testcase plus three local
reproduction runs. The evidence is text-only: linker outputs, disassembly,
relocation dumps, symbol dumps, and short summaries where available.
Reproducer bundle:
https://github.com/anonymous25989/riscv-linker-48-reproducers/raw/main/r48-v2/attachments/gnu_001_intra_section_boundary_misses_cj_relax_reproducer.zip
## Notes
This is reported as a linker-stage behavior defect found during RISC-V linker
regression/conformance testing. The reduced testcase is intentionally small so
that the failure mode is easy to inspect. The attached evidence is already
reproduced; no additional local path or private workspace is required.
--
You are receiving this mail because:
You are on the CC list for the bug.
