[Bug target/107238] New: aarch64: Wrong code converting pointer to __int128

[Bug target/107238] New: aarch64: Wrong code converting pointer to __int128

[acoplan at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107238

            Bug ID: 107238
           Summary: aarch64: Wrong code converting pointer to __int128
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: acoplan at gcc dot gnu.org
  Target Milestone: ---

For this testcase:

__int128 f(int *p) {
    return (__int128)p;
}

AArch64 GCC at -O2 generates:

f:
        asr     x1, x0, 63
        ret

i.e. it sign-extends the pointer, but I think it should be a zero extend
instead. The ABI [1] says:

> Code and data pointers are either 64-bit or 32-bit unsigned types.

Indeed LLVM does a zero-extend:

f:                                      // @f
        mov     x1, xzr
        ret

[1] :
https://github.com/ARM-software/abi-aa/blob/main/aapcs64/aapcs64.rst#pointers

[Bug target/107238] aarch64: Wrong code converting pointer to __int128

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107238

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |INVALID
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
https://gcc.gnu.org/onlinedocs/gcc-12.2.0/gcc/Arrays-and-pointers-implementation.html#Arrays-and-pointers-implementation

> A cast from integer to pointer discards most-significant bits if the pointer representation is smaller than the integer type, extends according to the signedness of the integer type if the pointer representation is larger than the integer type, otherwise the bits are unchanged.

[Bug target/107238] aarch64: Wrong code converting pointer to __int128

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107238

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
clang does not even define the behavior here even though it should be as it is
an implementation defined behavior.
File a bug with clang about not documenting their behavior ...

[Bug target/107238] aarch64: Wrong code converting pointer to __int128

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107238

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #2)
> File a bug with clang about not documenting their behavior ...

Oh there is one already:
https://github.com/llvm/llvm-project/issues/11644
Still open after (almost) 10 years.