[PATCH] D79482: Fix stack clash probing on the tail of dynamic allocation

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: llvm/lib/Target/X86/X86FrameLowering.cpp:715
   if (Offset % StackProbeSize) {
-    const unsigned Opc = getSUBriOpcode(Uses64BitFramePtr, Offset);
-    BuildMI(*tailMBB, tailMBB->begin(), DL, TII.get(Opc), StackPtr)
-        .addReg(StackPtr)
-        .addImm(Offset % StackProbeSize)
+    BuildMI(*tailMBB, tailMBB->begin(), DL, TII.get(X86::MOV64rr), StackPtr)
+        .addReg(FinalStackPtr)
----------------
This can't be MOV64rr for 32-bit. Can we use TargetOpcode::COPY?


================
Comment at: llvm/lib/Target/X86/X86ISelLowering.cpp:31580
 
-  const TargetRegisterClass *SizeRegClass = MRI.getRegClass(sizeVReg);
+  unsigned physSPReg = TFI.Uses64BitFramePtr ? X86::RSP : X86::ESP;
 
----------------
use Register instead of unsigned.


================
Comment at: llvm/lib/Target/X86/X86ISelLowering.cpp:31587
 
-  unsigned physSPReg = TFI.Uses64BitFramePtr ? X86::RSP : X86::ESP;
+  BuildMI(*MBB, {MI}, DL, TII->get(X86::MOV64rr), TmpStackPtr)
+      .addReg(physSPReg);
----------------
This shouldn't be MOV64rr if the stack pointer is 32 bits. Can we just use TargetOpcode::COPY here?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D79482/new/

https://reviews.llvm.org/D79482