[VM/Compiler/AOT] Bare instructions 8: Improve AOT code by using pc-r…

…elative calls

This CL improves AOT code for StackOverflowInstr/CheckNullInstr:

  * On ARM we can do a conditional pc-relative calls for the stack overflow
    checks, getting rid of the slow-paths entirely.

  * On ARM64 we can do pc-relative calls on the slow path, avoiding an
    extra load.

Flutter gallery size impact (in bare instructions mode):
  * ARM: -3.7% RX
  * ARM64: -1.4% RX

Issue dart-lang#33274

Change-Id: Ia1acd76ac6efa26642f99e1ce3e417100aa357f3
Reviewed-on: https://dart-review.googlesource.com/c/89620
Commit-Queue: Martin Kustermann <[email protected]>
Reviewed-by: Vyacheslav Egorov <[email protected]>

assembler_arm.cc

@@ -3406,9 +3406,9 @@ void Assembler::TryAllocateArray(intptr_t cid,
   }
 }
 
-void Assembler::GenerateUnRelocatedPcRelativeCall() {
-  // Emit "blr <offset>".
-  EmitType5(AL, 0x686868, /*link=*/true);
+void Assembler::GenerateUnRelocatedPcRelativeCall(Condition cond) {
+  // Emit "blr.cond <offset>".
+  EmitType5(cond, 0x686868, /*link=*/true);
 }
 
 void Assembler::Stop(const char* message) {

assembler_arm.h

@@ -1106,7 +1106,7 @@ class Assembler : public AssemblerBase {
   //   (Code::kPcRelativeCall & pc_offset, <target-code>, <target-function>)
   //
   // will be used during relocation to fix the offset.
-  void GenerateUnRelocatedPcRelativeCall();
+  void GenerateUnRelocatedPcRelativeCall(Condition cond = AL);
 
   // Emit data (e.g encoded instruction or immediate) in instruction stream.
   void Emit(int32_t value);