[PATCH] D37850: AMDGPU: Split MUBUF offset into aligned components

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp:1236
   if (Imm >= 4096) {
-    if (Imm <= 4095 + 64) {
-      // Use an SOffset inline constant for 1..64
-      Overflow = Imm - 4095;
-      Imm = 4095;
+    if (Imm <= 4092 + 64) {
+      // Use an SOffset inline constant for 4..64
----------------
Can you keep using the bit ranges, and then use the various alignTo functions instead of changing the hardcoded values


https://reviews.llvm.org/D37850