[PATCH] D124406: [X86] Use indirect addressing for high 2GB of x32 address space

[Harald van Dijk via Phabricator via llvm-commits]

hvdijk added inline comments.


================
Comment at: llvm/lib/Target/X86/X86ISelDAGToDAG.cpp:1696
+    if (Subtarget->isTarget64BitILP32() && !isUInt<31>(Val) &&
+        !AM.hasBaseOrIndexReg())
+      return true;
----------------
efriedma wrote:
> The reasoning here seems strange.  For example, suppose I write `void f(int a) { ((char*)0x80000000)[a] = a; }`.  That has a base register, but sign-extension is wrong.
> 
> I guess you're trying to allow negative pointer offsets here, but I think SelectionDAG is throwing away the distinction you need here.  (At the IR level, it's easy to distinguish between the base of a GEP and the offset.)
That does the right thing, that's okay to allow. That results in `movb %dil, -2147483648(%edi)`, and the fact that `%edi` is part of the address means `%edi - 2147483648` is calculated as a 32-bit value, and then zero-extended, so it calculates the exact same thing as `%edi + 2147483648` would if it were possible to do that directly. x86 is weird.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D124406/new/

https://reviews.llvm.org/D124406