[PATCH] D49720: [ARM] Fix over-alignment in arguments that are HA of 128-bit vectors

[Eli Friedman via Phabricator via llvm-commits]

efriedma added inline comments.


================
Comment at: lib/Target/ARM/ARMCallingConv.h:279
     // we'll be allocating a bunch of i32 slots).
-    Align = Size;
+    Align = std::min(Align, Size);
   }
----------------
Could you hoist the `std::min(Align, Size)` out of the loop so it's clear it isn't changing every iteration?


Repository:
  rL LLVM

https://reviews.llvm.org/D49720