On Thu, Jan 14, 2021 at 09:28:31AM +0100, Jakub Jelinek via Gcc-patches wrote:
> I'm afraid yes, because it is not a toplevel routine, but something called
> from the c-family pretty-printers, so it can be in the middle of arbitrary
> C/C++ expressions. And printing
> (3 * (access to a memory object of type 'int' at offset 12 bytes from 't') +
> 31) * 42
> would be just weird.
>
> > We could print, instead of *((int *)t + 3), "access to a memory
> > object of type 'int' at offset 12 bytes from 't'", thus explain
> > in plain english.
> >
> > That said, *((int *)t + 3) is exactly what the access is,
>
> *((int *)&t + 3) actually, the code I haven't touched has multiple bugs.
>
> The user generally doesn't know the exact layout of the structures,
> and especially with C++ templates it is extremely hard to figure that out,
> so even when we could print verbose text it would be helpful to give a hint
> (in your text something like (which falls into 't.u.b')).
> I don't see how we can print both the MEM_REF type and TBAA type in a way
> that would be understandable to the user.
>
> Could we print
> t.u.b
> if the TBAA type is compatible with the type of the reference and perhaps
> *(int*)&t.u.b
> if it is incompatible?
> >From the aliasing perspective that is still different, but we don't print
> the TBAA type anyway.
There is another option I forgot about, but perhaps it is too verbose.
Print
*(int*)((char*)&t + offsetof (struct T, u.b))
so like
*(int*)((char*)&t + 12)
but print the offset in a more user-friendly way.
Jakub
