Index: llvm/trunk/lib/CodeGen/RegisterCoalescer.cpp =================================================================== --- llvm/trunk/lib/CodeGen/RegisterCoalescer.cpp +++ llvm/trunk/lib/CodeGen/RegisterCoalescer.cpp @@ -1810,6 +1810,11 @@ return false; } + if (!LIS->intervalIsInOneMBB(RHS)) { + DEBUG(dbgs() << "\t\tComplex control flow!\n"); + return false; + } + MachineInstr &DestMI = *MRI->getVRegDef(SrcReg); CopyMI = &*MRI->use_instr_nodbg_begin(SrcReg); SlotIndex CopyRegIdx = LIS->getInstructionIndex(*CopyMI).getRegSlot(); Index: llvm/trunk/test/CodeGen/AArch64/regcoal-physreg.mir =================================================================== --- llvm/trunk/test/CodeGen/AArch64/regcoal-physreg.mir +++ llvm/trunk/test/CodeGen/AArch64/regcoal-physreg.mir @@ -2,40 +2,33 @@ --- | declare void @f2() - define void @func() { ret void } + define void @func0() { ret void } + define void @func1() { ret void } + define void @func2() { ret void } ... --- # Check coalescing of COPYs from reserved physregs. -# CHECK-LABEL: name: func -name: func -registers: - - { id: 0, class: gpr32 } - - { id: 1, class: gpr64 } - - { id: 2, class: gpr64 } - - { id: 3, class: gpr32 } - - { id: 4, class: gpr64 } - - { id: 5, class: gpr32 } - - { id: 6, class: xseqpairsclass } - - { id: 7, class: gpr64 } +# CHECK-LABEL: name: func0 +name: func0 body: | bb.0: ; We usually should not coalesce copies from allocatable physregs. ; CHECK: %0 = COPY %w7 ; CHECK: STRWui %0, %x1, 0 - %0 = COPY %w7 + %0 : gpr32 = COPY %w7 STRWui %0, %x1, 0 ; It is fine to coalesce copies from reserved physregs ; CHECK-NOT: COPY ; CHECK: STRXui %fp, %x1, 0 - %1 = COPY %fp + %1 : gpr64 = COPY %fp STRXui %1, %x1, 0 ; It is not fine to coalesce copies from reserved physregs when they are ; clobbered. ; CHECK: %2 = COPY %fp ; CHECK: STRXui %2, %x1, 0 - %2 = COPY %fp + %2 : gpr64 = COPY %fp %fp = SUBXri %fp, 4, 0 STRXui %2, %x1, 0 @@ -43,7 +36,7 @@ ; clobbered. ; CHECK-NOT: COPY ; CHECK: STRWui %wzr, %x1 - %3 = COPY %wzr + %3 : gpr32 = COPY %wzr dead %wzr = SUBSWri %w1, 0, 0, implicit-def %nzcv STRWui %3, %x1, 0 @@ -51,13 +44,13 @@ ; clobbered. ; CHECK-NOT: COPY ; CHECK: STRXui %xzr, %x1 - %4 = COPY %xzr + %4 : gpr64 = COPY %xzr dead %wzr = SUBSWri %w1, 0, 0, implicit-def %nzcv STRXui %4, %x1, 0 ; Coalescing COPYs into constant physregs. ; CHECK: %wzr = SUBSWri %w1, 0, 0 - %5 = SUBSWri %w1, 0, 0, implicit-def %nzcv + %5 : gpr32 = SUBSWri %w1, 0, 0, implicit-def %nzcv %wzr = COPY %5 ; Only coalesce when the source register is reserved as a whole (this is @@ -65,7 +58,7 @@ ; of the non-reserved part). ; CHECK: %6 = COPY %x28_fp ; CHECK: HINT 0, implicit %6 - %6 = COPY %x28_fp + %6 : xseqpairsclass = COPY %x28_fp HINT 0, implicit %6 ; It is not fine to coalesce copies from reserved physregs when they are @@ -76,7 +69,69 @@ ; Need a def of x18 so that it's not deduced as "constant". %x18 = COPY %xzr - %7 = COPY %x18 + %7 : gpr64 = COPY %x18 BL @f2, csr_aarch64_aapcs, implicit-def dead %lr, implicit %sp, implicit-def %sp STRXui %7, %x1, 0 + + ; This can be coalesced. + ; CHECK: %fp = SUBXri %fp, 4, 0 + %8 : gpr64sp = SUBXri %fp, 4, 0 + %fp = COPY %8 + + ; Cannot coalesce when there are reads of the physreg. + ; CHECK-NOT: %fp = SUBXri %fp, 8, 0 + ; CHECK: %9 = SUBXri %fp, 8, 0 + ; CHECK: STRXui %fp, %fp, 0 + ; CHECK: %fp = COPY %9 + %9 : gpr64sp = SUBXri %fp, 8, 0 + STRXui %fp, %fp, 0 + %fp = COPY %9 +... +--- +# Check coalescing of COPYs from reserved physregs. +# CHECK-LABEL: name: func1 +name: func1 +body: | + bb.0: + successors: %bb.1, %bb.2 + ; Cannot coalesce physreg because we have reads on other CFG paths (we + ; currently abort for any control flow) + ; CHECK-NOT: %fp = SUBXri + ; CHECK: %0 = SUBXri %fp, 12, 0 + ; CHECK: CBZX undef %x0, %bb.1 + ; CHECK: B %bb.2 + %0 : gpr64sp = SUBXri %fp, 12, 0 + CBZX undef %x0, %bb.1 + B %bb.2 + + bb.1: + %fp = COPY %0 + RET_ReallyLR + + bb.2: + STRXui %fp, %fp, 0 + RET_ReallyLR +... +--- +# CHECK-LABEL: name: func2 +name: func2 +body: | + bb.0: + successors: %bb.1, %bb.2 + ; We can coalesce copies from physreg to vreg across multiple blocks. + ; CHECK-NOT: COPY + ; CHECK: CBZX undef %x0, %bb.1 + ; CHECK-NEXT: B %bb.2 + %0 : gpr64sp = COPY %fp + CBZX undef %x0, %bb.1 + B %bb.2 + + bb.1: + ; CHECK: STRXui undef %x0, %fp, 0 + ; CHECK-NEXT: RET_ReallyLR + STRXui undef %x0, %0, 0 + RET_ReallyLR + + bb.2: + RET_ReallyLR ...