From: Christoph Müllner christoph.muellner@vrull.eu

[ Upstream commit 12c16919652b5873f524c8b361336ecfa5ce5e6b ]

When building the mm tests with a riscv32 compiler, we see a range of shift-count-overflow errors from shifting 1UL by more than 32 bits in do_mmaps(). Since, the relevant code is only called from code that is gated by `__riscv_xlen == 64`, we can just apply the same gating to do_mmaps().

Signed-off-by: Christoph Müllner christoph.muellner@vrull.eu Reviewed-by: Alexandre Ghiti alexghiti@rivosinc.com Reviewed-by: Andrew Jones ajones@ventanamicro.com Link: https://lore.kernel.org/r/20231123185821.2272504-6-christoph.muellner@vrull.... Signed-off-by: Palmer Dabbelt palmer@rivosinc.com Signed-off-by: Sasha Levin sashal@kernel.org --- tools/testing/selftests/riscv/mm/mmap_test.h | 3 +++ 1 file changed, 3 insertions(+)

diff --git a/tools/testing/selftests/riscv/mm/mmap_test.h b/tools/testing/selftests/riscv/mm/mmap_test.h index 9b8434f62f57..2e0db9c5be6c 100644 --- a/tools/testing/selftests/riscv/mm/mmap_test.h +++ b/tools/testing/selftests/riscv/mm/mmap_test.h @@ -18,6 +18,8 @@ struct addresses { int *on_56_addr; };

+// Only works on 64 bit +#if __riscv_xlen == 64 static inline void do_mmaps(struct addresses *mmap_addresses) { /* @@ -50,6 +52,7 @@ static inline void do_mmaps(struct addresses *mmap_addresses) mmap_addresses->on_56_addr = mmap(on_56_bits, 5 * sizeof(int), prot, flags, 0, 0); } +#endif /* __riscv_xlen == 64 */

static inline int memory_layout(void) {