On Wed, Jan 17, 2024 at 04:25:14PM +0800, Yunhui Cui wrote:
When compiling with -O0, the following error will occur: cbo.c: In function 'cbo_insn': cbo.c:43:9: warning: 'asm' operand 1 probably does not match constraints 43 | asm volatile( | ^~~ cbo.c:43:9: warning: 'asm' operand 2 probably does not match constraints cbo.c:43:9: error: impossible constraint in 'asm'
Add __attribute__((optimize("O"))) to fix.
Fixes: a29e2a48afe3 ("RISC-V: selftests: Add CBO tests") Signed-off-by: Yunhui Cui cuiyunhui@bytedance.com Suggested-by: Zhipeng Xu xuzhipeng.1973@bytedance.com
tools/testing/selftests/riscv/hwprobe/cbo.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/tools/testing/selftests/riscv/hwprobe/cbo.c b/tools/testing/selftests/riscv/hwprobe/cbo.c index 50a2cc8aef38..ff1d8e843d70 100644 --- a/tools/testing/selftests/riscv/hwprobe/cbo.c +++ b/tools/testing/selftests/riscv/hwprobe/cbo.c @@ -36,7 +36,7 @@ static void sigill_handler(int sig, siginfo_t *info, void *context) regs[0] += 4; } -static void cbo_insn(char *base, int fn) +static __always_inline void cbo_insn(char *base, int fn) { uint32_t insn = MK_CBO(fn); @@ -47,10 +47,11 @@ static void cbo_insn(char *base, int fn) : : "r" (base), "i" (fn), "i" (insn) : "a0", "a1", "memory"); } -static void cbo_inval(char *base) { cbo_insn(base, 0); } -static void cbo_clean(char *base) { cbo_insn(base, 1); } -static void cbo_flush(char *base) { cbo_insn(base, 2); } -static void cbo_zero(char *base) { cbo_insn(base, 4); } +#define OPTIMIZE __attribute__((optimize("O"))) +OPTIMIZE static void cbo_inval(char *base) { cbo_insn(base, 0); } +OPTIMIZE static void cbo_clean(char *base) { cbo_insn(base, 1); } +OPTIMIZE static void cbo_flush(char *base) { cbo_insn(base, 2); } +OPTIMIZE static void cbo_zero(char *base) { cbo_insn(base, 4); } static void test_no_zicbom(void *arg) { -- 2.20.1
Hi Yunhui,
Thanks for the bug report, but this isn't the right fix. The real problem is that I didn't ensure operands 1 and 2 match their constraints, just as the warning you discovered says. To do that, I should have made cbo_insn() a macro and not used the local variable, i.e. ensure 'fn' and 'insn' are indeed constants derived from the 0,1,2,4 constants.
I'll send a patch with your reported-by.
Thanks, drew