[JIT] Trace profiling support

In preparation for method compilation, this CL causes all traces to
include two entry points: profiling and non-profiling.  For now, the
profiling entry will only be used if dalvik is run with -Xjitprofile,
and largely works like it did before.  The difference is that profiling
support no longer requires the "assert" build - it's always there now.

This will enable us to do a form of sampling profiling of
traces in order to identify hot methods or hot trace groups,
while keeping the overhead low by only switching profiling on periodically.

To turn the periodic profiling on and off, we simply unchain all existing
translations and set the appropriate global profile state.  The underlying
translation lookup and chaining utilties will examine the profile state to
determine which entry point to use (i.e. - profiling or non-profiling) while
the traces naturally rechain during further execution.

Change-Id: I9ee33e69e33869b9fab3a57e88f9bc524175172b

1 files changed, 52 insertions, 1 deletions

diff --git a/vm/compiler/codegen/arm/Thumb2/Gen.c b/vm/compiler/codegen/arm/Thumb2/Gen.c
index 0891524f7..f5e1096ea 100644
--- a/vm/compiler/codegen/arm/Thumb2/Gen.c
+++ b/vm/compiler/codegen/arm/Thumb2/Gen.c
@@ -15,13 +15,64 @@
  */
 
 /*
- * This file contains codegen for the Thumb ISA and is intended to be
+ * This file contains codegen for the Thumb2 ISA and is intended to be
  * includes by:
  *
  *        Codegen-$(TARGET_ARCH_VARIANT).c
  *
  */
 
+/*
+ * Reserve 6 bytes at the beginning of the trace
+ *        +----------------------------+
+ *        | prof count addr (4 bytes)  |
+ *        +----------------------------+
+ *        | chain cell offset (2 bytes)|
+ *        +----------------------------+
+ *
+ * ...and then code to increment the execution
+ *
+ * For continuous profiling (10 bytes)
+ *       ldr   r0, [pc-8]   @ get prof count addr    [4 bytes]
+ *       ldr   r1, [r0]     @ load counter           [2 bytes]
+ *       add   r1, #1       @ increment              [2 bytes]
+ *       str   r1, [r0]     @ store                  [2 bytes]
+ *
+ * For periodic profiling (4 bytes)
+ *       call  TEMPLATE_PERIODIC_PROFILING
+ *
+ * and return the size (in bytes) of the generated code.
+ */
+
+static int genTraceProfileEntry(CompilationUnit *cUnit)
+{
+    intptr_t addr = (intptr_t)dvmJitNextTraceCounter();
+    assert(__BYTE_ORDER == __LITTLE_ENDIAN);
+    newLIR1(cUnit, kArm16BitData, addr & 0xffff);
+    newLIR1(cUnit, kArm16BitData, (addr >> 16) & 0xffff);
+    cUnit->chainCellOffsetLIR =
+        (LIR *) newLIR1(cUnit, kArm16BitData, CHAIN_CELL_OFFSET_TAG);
+    cUnit->headerSize = 6;
+    if ((gDvmJit.profileMode == kTraceProfilingContinuous) ||
+        (gDvmJit.profileMode == kTraceProfilingDisabled)) {
+        /* Thumb[2] instruction used directly here to ensure correct size */
+        newLIR2(cUnit, kThumb2LdrPcReln12, r0, 8);
+        newLIR3(cUnit, kThumbLdrRRI5, r1, r0, 0);
+        newLIR2(cUnit, kThumbAddRI8, r1, 1);
+        newLIR3(cUnit, kThumbStrRRI5, r1, r0, 0);
+        return 10;
+    } else {
+        int opcode = TEMPLATE_PERIODIC_PROFILING;
+        newLIR2(cUnit, kThumbBlx1,
+            (int) gDvmJit.codeCache + templateEntryOffsets[opcode],
+            (int) gDvmJit.codeCache + templateEntryOffsets[opcode]);
+        newLIR2(cUnit, kThumbBlx2,
+            (int) gDvmJit.codeCache + templateEntryOffsets[opcode],
+            (int) gDvmJit.codeCache + templateEntryOffsets[opcode]);
+        return 4;
+    }
+}
+
 static void genNegFloat(CompilationUnit *cUnit, RegLocation rlDest,
                         RegLocation rlSrc)
 {