[AArch64] Exclude instructions after setting the FP from SEH prologues

[mstorsjo]

Summary

After setting up the FP, the rest of the prologue doesn't need to
be replayed for unwinding the stack frame.

This allows reverting the functional parts of
2f7fbf837625267193351cc334e506a3a9161958.

[efriedma]

llvm/lib/Target/AArch64/AArch64InstrInfo.cpp

4350

I don't think you need to revert the part of the change involving the redundant *HasWinCFI = true;?

[mstorsjo]

llvm/lib/Target/AArch64/AArch64InstrInfo.cpp

4350

Oh, right. I noted that after removing the else case, there’s one possible codepath where no SEH opcodes are added, and we perhaps shouldn’t be setting HasWinCFI, but apparently we had that inconsistency before this change. So either we should remove the catch-all setting of it below to remove the inconsistency, or just leave that part as it is.

[efriedma]

llvm/lib/Target/AArch64/AArch64InstrInfo.cpp

4350

Whatever you think makes sense.

[mstorsjo]

Removed the duplicate setting of HasWinCFI.

[efriedma]

LGTM

llvm/lib/Target/AArch64/AArch64FrameLowering.cpp

if (!IsFunclet && HasFP) {
     } else {
       // Issue    sub fp, sp, FPOffset or
       //          mov fp,sp          when FPOffset is zero.
       // Note: All stores of callee-saved registers are marked as "FrameSetup".
       // This code marks the instruction(s) that set the FP also.
       emitFrameOffset(MBB, MBBI, DL, AArch64::FP, AArch64::SP,
                       StackOffset::getFixed(FPOffset), TII,
                       MachineInstr::FrameSetup, false, NeedsWinCFI, &HasWinCFI);
+      if (NeedsWinCFI && HasWinCFI) {
+        BuildMI(MBB, MBBI, DL, TII->get(AArch64::SEH_PrologEnd))
+            .setMIFlag(MachineInstr::FrameSetup);
+        // After setting up the FP, the rest of the prolog doesn't need to be
+        // included in the SEH unwind info.
+        NeedsWinCFI = false;
+      }
     }
     if (EmitCFI) {
       // Define the current CFA rule to use the provided FP.
       const int OffsetToFirstCalleeSaveFromFP =
           AFI->getCalleeSaveBaseToFrameRecordOffset() -
           AFI->getCalleeSavedStackSize();
       Register FramePtr = RegInfo->getFrameRegister(MF);
       unsigned Reg = RegInfo->getDwarfRegNum(FramePtr, true);

llvm/lib/Target/AArch64/AArch64InstrInfo.cpp

if (EmitCFAOffset && DestReg == TmpReg) {
           .setMIFlags(Flag);
     }
 
     if (NeedsWinCFI) {
       assert(Sign == 1 && "SEH directives should always have a positive sign");
       int Imm = (int)(ThisVal << LocalShiftSize);
       if ((DestReg == AArch64::FP && SrcReg == AArch64::SP) ||
           (SrcReg == AArch64::FP && DestReg == AArch64::SP)) {
+        if (HasWinCFI)
+          *HasWinCFI = true;
         if (Imm == 0)
           BuildMI(MBB, MBBI, DL, TII->get(AArch64::SEH_SetFP)).setMIFlag(Flag);
         else
           BuildMI(MBB, MBBI, DL, TII->get(AArch64::SEH_AddFP))
               .addImm(Imm)
               .setMIFlag(Flag);
         assert(Offset == 0 && "Expected remaining offset to be zero to "
                               "emit a single SEH directive");
       } else if (DestReg == AArch64::SP) {
+        if (HasWinCFI)
+          *HasWinCFI = true;
         assert(SrcReg == AArch64::SP && "Unexpected SrcReg for SEH_StackAlloc");
         BuildMI(MBB, MBBI, DL, TII->get(AArch64::SEH_StackAlloc))
             .addImm(Imm)
             .setMIFlag(Flag);
-      } else {
-        BuildMI(MBB, MBBI, DL, TII->get(AArch64::SEH_Nop))
-            .setMIFlag(Flag);
       }
-      if (HasWinCFI)
-        *HasWinCFI = true;
     }
 
     SrcReg = TmpReg;
   } while (Offset);
 }
 
 void llvm::emitFrameOffset(MachineBasicBlock &MBB,
                            MachineBasicBlock::iterator MBBI, const DebugLoc &DL,

llvm/test/CodeGen/AArch64/wineh-align-stack.ll

 
 ; CHECK-LABEL: func:
 ; CHECK-NEXT: .seh_proc func
 ; CHECK-NEXT: // %bb.0:
 ; CHECK-NEXT: stp x29, x30, [sp, #-16]!
 ; CHECK-NEXT: .seh_save_fplr_x 16
 ; CHECK-NEXT: mov x29, sp
 ; CHECK-NEXT: .seh_set_fp
+; CHECK-NEXT: .seh_endprologue
 ; CHECK-NEXT: sub x9, sp, #80
-; CHECK-NEXT: .seh_nop
 ; CHECK-NEXT: and sp, x9, #0xffffffffffffffe0
-; CHECK-NEXT: .seh_stackalloc 80
-; CHECK-NEXT: .seh_endprologue
-
-; FIXME: Ideally, the SEH prologue wouldn't include the stack realigning
-; instructions at all, but it's enough to end the prologue after .seh_set_fp.

llvm/test/CodeGen/AArch64/wineh-frame6.mir

 # RUN: llc -o - %s -mtriple=aarch64-windows -start-before=prologepilog \
 # RUN:   -stop-after=prologepilog | FileCheck %s
 # Test that stack probe results in Nop unwind codes in the prologue.  Test
 # save_fplr, save_reg_x and stack_alloc with multiple updates
 
 # CHECK:      early-clobber $sp = frame-setup STPXpre killed $fp, killed $lr, $sp, -2
 # CHECK-NEXT: frame-setup SEH_SaveFPLR_X -16
 # CHECK-NEXT: $fp = frame-setup ADDXri $sp, 0, 0
 # CHECK-NEXT: frame-setup SEH_SetFP
-# CHECK-NEXT: $sp = frame-setup SUBXri $sp, 32, 0
-# CHECK-NEXT: frame-setup SEH_StackAlloc 32
 # CHECK-NEXT: frame-setup SEH_PrologEnd
+# CHECK-NEXT: $sp = frame-setup SUBXri $sp, 32, 0
 # CHECK:      frame-destroy SEH_EpilogStart
 # CHECK-NEXT: $sp = frame-destroy ADDXri $fp, 0, 0
 # CHECK-NEXT: frame-destroy SEH_SetFP
 # CHECK-NEXT: early-clobber $sp, $fp, $lr = frame-destroy LDPXpost $sp, 2
 # CHECK-NEXT: frame-destroy SEH_SaveFPLR_X -16
 # CHECK-NEXT: frame-destroy SEH_EpilogEnd
 # CHECK-NEXT: RET_ReallyLR implicit killed $w0
 --- |

llvm/test/CodeGen/AArch64/wineh-try-catch.ll

 ; CHECK-NEXT:         .word   1                       // NumTryBlocks
 ; CHECK-NEXT:         .word   ("$tryMap$?func@@YAHXZ")@IMGREL // TryBlockMap
 ; CHECK-NEXT:         .word   4                       // IPMapEntries
 ; CHECK-NEXT:         .word   ("$ip2state$?func@@YAHXZ")@IMGREL // IPToStateXData
 ; CHECK-NEXT:         .word   -16                     // UnwindHelp
 
 ; UNWIND: Function: ?func@@YAHXZ (0x0)
 ; UNWIND: Prologue [
-; UNWIND-NEXT: ; nop
-; UNWIND-NEXT: ; sub sp, #624
 ; UNWIND-NEXT: ; add fp, sp, #32
 ; UNWIND-NEXT: ; stp x29, x30, [sp, #32]
 ; UNWIND-NEXT: ; str x28, [sp, #24]
 ; UNWIND-NEXT: ; str x21, [sp, #16]
 ; UNWIND-NEXT: ; stp x19, x20, [sp, #-64]!
 ; UNWIND-NEXT: ; end
 ; UNWIND: Function: ?catch$2@?0??func@@YAHXZ@4HA
 ; UNWIND: Prologue [