[PATCH] D44559: Wrong width of result of mul operation for x86

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri added a comment.

In https://reviews.llvm.org/D44559#1040319, @craig.topper wrote:

> Why should the behavior be X86 specific?
>
> There also seems to be something weirder going on with gcc. They also warn on these.
>
> void foo(char c) {
>
>   c = c + c;
>
> }
>
> void bar(short c) {
>
>   c = c + c;
>
> }


Just want to point the obvious, https://godbolt.org/g/1SQSda
So as per AST (and language rules), it's actually

  char foo(char c) {
    int d = (int)c + (int)c;
    return (char)d;
  }


https://reviews.llvm.org/D44559