[llvm] [LLVM] Make use of s_flbit_i32_b64 and s_ff1_i32_b64 (PR #75158)

[Jay Foad via llvm-commits]

================
@@ -607,13 +607,10 @@ define amdgpu_kernel void @s_ctlz_i64(ptr addrspace(1) noalias %out, [8 x i32],
 ; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    s_flbit_i32_b32 s4, s4
-; SI-NEXT:    s_flbit_i32_b32 s5, s5
-; SI-NEXT:    s_min_u32 s4, s4, 0xffffffdf
-; SI-NEXT:    v_mov_b32_e32 v0, s5
-; SI-NEXT:    s_add_i32 s4, s4, 32
-; SI-NEXT:    v_min3_u32 v0, s4, v0, 64
+; SI-NEXT:    s_flbit_i32_b64 s4, s[4:5]
+; SI-NEXT:    s_min_u32 s4, s4, 32
----------------
jayfoad wrote:

32 is wrong here and many other places. Should be 64.

https://github.com/llvm/llvm-project/pull/75158