[llvm-bugs] [Bug 48257] New: vget_lane_p64 triggers -Wvector-conversion diagnostic

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=48257

            Bug ID: 48257
           Summary: vget_lane_p64 triggers -Wvector-conversion diagnostic
           Product: new-bugs
           Version: 11.0
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: normal
          Priority: P
         Component: new bugs
          Assignee: unassignedbugs at nondot.org
          Reporter: clang at evan.coeusgroup.com
                CC: htmldeveloper at gmail.com, llvm-bugs at lists.llvm.org

vget_lane_p64 is defined in arm_neon.h as:

  #define vget_lane_p64(__p0, __p1) __extension__ ({ \
    poly64x1_t __s0 = __p0; \
    poly64_t __ret; \
    __ret = (poly64_t) __builtin_neon_vget_lane_i64((poly64x1_t)__s0, __p1); \
    __ret; \
  })

Note the cast on the first argument to __builtin_neon_vget_lane_i64;
vget_lane_i64 takes an int64x1_t, not a poly64x1_t, so code which calls this
macro ends up emitting a -Wvector-conversion diagnostic:

  error: incompatible vector types passing 'poly64x1_t' (vector of 1 'poly64_t'
value) to parameter of type '__attribute__((__vector_size__(1 * sizeof(long))))
long' (vector of 1 'long' value) [-Werror,-Wvector-conversion]

For an example, you can use:

  #include <arm_neon.h>

  poly64_t foo (poly64x1_t v);

  poly64_t
  foo (poly64x1_t v) {
    return vget_lane_p64(v, 0);
  }

Or, on Compiler Explorer: https://godbolt.org/z/GWf6fq

It looks like this is a problem on all the vget*_lane_p64 functions.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20201122/bbb4f643/attachment.html>