ARM: try harder to detect non-IT eligible instructions
For many Thumb-1 register register instructions, setting the CPSR is not
permitted inside an IT block. We would not correctly flag those instructions.
The previous change to identify this scenario was insufficient as it did not
actually catch all the instances. The current list is formed by manual
inspection of the ARMv6M ARM.
The change to the Thumb2 IT block test is due to the fact that the new more
stringent checking of the MIs results in the If Conversion pass being prevented
from executing (since not all the instructions in the BB are predicable). This
results in code gen changes.
Thanks to Tim Northover for pointing out that the previous patch was
insufficient and hinting that the use of the v6M ARM would be much easier to use
than the v7 or v8!
llvm-svn: 215382
diff --git a/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp b/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp
index bee5bfd..37290ad 100644
--- a/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp
+++ b/llvm/lib/Target/ARM/ARMBaseInstrInfo.cpp
@@ -520,11 +520,40 @@
static bool isCPSRDefined(const MachineInstr *MI) {
for (const auto &MO : MI->operands())
- if (MO.isReg() && MO.getReg() == ARM::CPSR && (MO.isDef() || !MO.isDead()))
+ if (MO.isReg() && MO.getReg() == ARM::CPSR && MO.isDef())
return true;
return false;
}
+static bool isEligibleForITBlock(const MachineInstr *MI) {
+ switch (MI->getOpcode()) {
+ default: return true;
+ case ARM::tADC: // ADC (register) T1
+ case ARM::tADDi3: // ADD (immediate) T1
+ case ARM::tADDi8: // ADD (immediate) T2
+ case ARM::tADDrr: // ADD (register) T1
+ case ARM::tAND: // AND (register) T1
+ case ARM::tASRri: // ASR (immediate) T1
+ case ARM::tASRrr: // ASR (register) T1
+ case ARM::tBIC: // BIC (register) T1
+ case ARM::tEOR: // EOR (register) T1
+ case ARM::tLSLri: // LSL (immediate) T1
+ case ARM::tLSLrr: // LSL (register) T1
+ case ARM::tLSRri: // LSR (immediate) T1
+ case ARM::tLSRrr: // LSR (register) T1
+ case ARM::tMUL: // MUL T1
+ case ARM::tMVN: // MVN (register) T1
+ case ARM::tORR: // ORR (register) T1
+ case ARM::tROR: // ROR (register) T1
+ case ARM::tRSB: // RSB (immediate) T1
+ case ARM::tSBC: // SBC (register) T1
+ case ARM::tSUBi3: // SUB (immediate) T1
+ case ARM::tSUBi8: // SUB (immediate) T2
+ case ARM::tSUBrr: // SUB (register) T1
+ return !isCPSRDefined(MI);
+ }
+}
+
/// isPredicable - Return true if the specified instruction can be predicated.
/// By default, this returns true for every instruction with a
/// PredicateOperand.
@@ -532,12 +561,8 @@
if (!MI->isPredicable())
return false;
- // The ARM Architecture Reference Manual states that the CPSR may only be
- // accessed by MUL in Thumb mode if it is outside an IT block. Thus, if CPSR
- // is defined (or clobbered) by this instruction, it is not predicable.
- if (MI->getOpcode() == ARM::tMUL || MI->getOpcode() == ARM::t2MUL)
- if (isCPSRDefined(MI))
- return false;
+ if (!isEligibleForITBlock(MI))
+ return false;
ARMFunctionInfo *AFI =
MI->getParent()->getParent()->getInfo<ARMFunctionInfo>();
diff --git a/llvm/test/CodeGen/ARM/thumb2-it-block.ll b/llvm/test/CodeGen/ARM/thumb2-it-block.ll
index c5e699c..2675a73 100644
--- a/llvm/test/CodeGen/ARM/thumb2-it-block.ll
+++ b/llvm/test/CodeGen/ARM/thumb2-it-block.ll
@@ -1,15 +1,9 @@
-; RUN: llc -mtriple=thumb-eabi -mcpu=arm1156t2-s -mattr=+thumb2 %s -o - | FileCheck %s
-; RUN: llc -mtriple=thumbv8 %s -o - | FileCheck %s
+; RUN: llc -mtriple=thumb-eabi -mcpu=arm1156t2-s -mattr=+thumb2 %s -o - | FileCheck -check-prefix CHECK-V7 %s
+; RUN: llc -mtriple=thumbv8 %s -o - | FileCheck %s -check-prefix CHECK-V8
; PR11107
define i32 @test(i32 %a, i32 %b) {
entry:
-; CHECK: cmp
-; CHECK-NEXT: it mi
-; CHECK-NEXT: rsb{{s?}}mi
-; CHECK-NEXT: cmp
-; CHECK-NEXT: it mi
-; CHECK-NEXT: rsb{{s?}}mi
%cmp1 = icmp slt i32 %a, 0
%sub1 = sub nsw i32 0, %a
%abs1 = select i1 %cmp1, i32 %sub1, i32 %a
@@ -19,3 +13,18 @@
%add = add nsw i32 %abs1, %abs2
ret i32 %add
}
+
+; CHECK-V7: cmp
+; CHECK-V7-NEXT: it mi
+; CHECK-V7-NEXT: rsbmi
+; CHECK-V7-NEXT: cmp
+; CHECK-V7-NEXT: it mi
+; CHECK-V7-NEXT: rsbmi
+
+; CHECK-V8: cmp
+; CHECK-V8-NEXT: bpl
+; CHECK-V8: rsbs
+; CHECK-V8: cmp
+; CHECK-V8-NEXT: bpl
+; CHECK-V8: rsbs
+