Commit Graph

3 Commits

Author SHA1 Message Date
Nathan Chancellor
f2abc305aa riscv: Define __riscv_copy_{,vec_}{words,bytes}_unaligned() using SYM_TYPED_FUNC_START
After commit 67bdd7b013 ("riscv: Split out measure_cycles() for
reuse") and commit c03ad15f7c ("riscv: Reuse measure_cycles() in
check_vector_unaligned_access()"), there are CFI failure when booting
kernels with CONFIG_CFI=y:

  CFI failure at measure_cycles+0x38/0xe0 (target: __riscv_copy_words_unaligned+0x0/0x50; expected type: ...)
  CFI failure at measure_cycles+0x38/0xe0 (target: __riscv_copy_vec_words_unaligned+0x0/0x24; expected type: ...)

The __riscv_copy_*_unaligned() functions are now called indirectly but
they are not defined with SYM_TYPED_FUNC_START, which is required for
assembly functions called indirectly from C to pass CFI checking. Switch
to SYM_TYPED_FUNC_START to clear up the CFI failures.

Fixes: 67bdd7b013 ("riscv: Split out measure_cycles() for reuse")
Fixes: c03ad15f7c ("riscv: Reuse measure_cycles() in check_vector_unaligned_access()")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Sami Tolvanen <samitolvanen@google.com>
Reviewed-by: Nam Cao <namcao@linutronix.de>
Link: https://patch.msgid.link/20260406-measure_cycles-cfi-failure-v1-1-03e0234ae02f@kernel.org
Signed-off-by: Paul Walmsley <pjw@kernel.org>
2026-04-29 20:38:17 -06:00
Tingbo Liao
475afa39b1
riscv: Fix the __riscv_copy_vec_words_unaligned implementation
Correct the VEC_S macro definition to fix the implementation
of vector words copy in the case of unalignment in RISC-V.

Fixes: e7c9d66e31 ("RISC-V: Report vector unaligned access speed hwprobe")
Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Signed-off-by: Tingbo Liao <tingbo.liao@starfivetech.com>
Link: https://lore.kernel.org/r/20250228090801.8334-1-tingbo.liao@starfivetech.com
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
2025-03-18 13:37:27 +00:00
Jesse Taube
e7c9d66e31
RISC-V: Report vector unaligned access speed hwprobe
Detect if vector misaligned accesses are faster or slower than
equivalent vector byte accesses. This is useful for usermode to know
whether vector byte accesses or vector misaligned accesses have a better
bandwidth for operations like memcpy.

Signed-off-by: Jesse Taube <jesse@rivosinc.com>
Reviewed-by: Charlie Jenkins <charlie@rivosinc.com>
Link: https://lore.kernel.org/r/20241017-jesse_unaligned_vector-v10-5-5b33500160f8@rivosinc.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
2024-10-18 12:38:34 -07:00