AArch64: estimate inline asm length during branch relaxation

To make sure branches are in range, we need to do a better job of estimating
the length of an inline assembly block than "it's probably 1 instruction, who'd
write asm with more than that?".

Fortunately there's already a (highly suspect, see how many ways you can think
of to break it!) callback for this purpose, which is used by the other targets.

rdar://problem/17277590

llvm-svn: 211095

diff --git a/llvm/lib/Target/AArch64/AArch64InstrInfo.cpp b/llvm/lib/Target/AArch64/AArch64InstrInfo.cpp
index 8cec4a1..cabf9ea 100644 (file)
--- a/llvm/lib/Target/AArch64/AArch64InstrInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64InstrInfo.cpp
@@ -35,8 +35,14 @@ AArch64InstrInfo::AArch64InstrInfo(const AArch64Subtarget &STI)
 /// GetInstSize - Return the number of bytes of code the specified
 /// instruction may be.  This returns the maximum number of bytes.
 unsigned AArch64InstrInfo::GetInstSizeInBytes(const MachineInstr *MI) const {
-  const MCInstrDesc &Desc = MI->getDesc();
+  const MachineBasicBlock &MBB = *MI->getParent();
+  const MachineFunction *MF = MBB.getParent();
+  const MCAsmInfo *MAI = MF->getTarget().getMCAsmInfo();
+
+  if (MI->getOpcode() == AArch64::INLINEASM)
+    return getInlineAsmLength(MI->getOperand(0).getSymbolName(), *MAI);
 
+  const MCInstrDesc &Desc = MI->getDesc();
   switch (Desc.getOpcode()) {
   default:
     // Anything not explicitly designated otherwise is a nomal 4-byte insn.

diff --git a/llvm/test/CodeGen/AArch64/branch-relax-asm.ll b/llvm/test/CodeGen/AArch64/branch-relax-asm.ll
new file mode 100644 (file)
index 0000000..7409c84
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/branch-relax-asm.ll
@@ -0,0 +1,35 @@
+; RUN: llc -mtriple=aarch64-apple-ios7.0 -disable-block-placement -aarch64-tbz-offset-bits=4 -o - %s | FileCheck %s
+define i32 @test_asm_length(i32 %in) {
+; CHECK-LABEL: test_asm_length:
+
+  ; It would be more natural to use just one "tbnz %false" here, but if the
+  ; number of instructions in the asm is counted reasonably, that block is out
+  ; of the limited range we gave tbz. So branch relaxation has to invert the
+  ; condition.
+; CHECK:     tbz w0, #0, [[TRUE:LBB[0-9]+_[0-9]+]]
+; CHECK:     b [[FALSE:LBB[0-9]+_[0-9]+]]
+
+; CHECK: [[TRUE]]:
+; CHECK:     orr w0, wzr, #0x4
+; CHECK:     nop
+; CHECK:     nop
+; CHECK:     nop
+; CHECK:     nop
+; CHECK:     nop
+; CHECK:     nop
+; CHECK:     ret
+
+; CHECK: [[FALSE]]:
+; CHECK:     ret
+
+  %val = and i32 %in, 1
+  %tst = icmp eq i32 %val, 0
+  br i1 %tst, label %true, label %false
+
+true:
+  call void asm sideeffect "nop\0A\09nop\0A\09nop\0A\09nop\0A\09nop\0A\09nop", ""()
+  ret i32 4
+
+false:
+  ret i32 0
+}