Generate the tables in InstrUtils.c...almost.

This patch adds code to opcode-gen to generate the opcode info tables
currently built at vm start time, but they are left commented out for
now. I wanted to separate the work of getting these tables generated
from the work of getting those new tables hooked up, and this seemed
like a reasonable way to cut through the larger effort.

To be clear, I've already verified that the data in the new
pregenerated tables matches what's in the status quo.

Change-Id: Ie73dc6c6a5721e8f714f845c97b5338141995770

3 files changed, 93 insertions, 21 deletions

diff --git a/opcode-gen/bytecode.txt b/opcode-gen/bytecode.txt
index 4462dac96..b24e6119a 100644
--- a/opcode-gen/bytecode.txt
+++ b/opcode-gen/bytecode.txt
@@ -41,7 +41,7 @@ format 22t
 format 35c 3rc 5rc
 
 # Optimized formats
-format 00x  # not in spec; used for breakpoint
+format 00x  # not in spec; used for undefined opcodes and breakpoint
 format 20bc # not in spec; used for throw-verification-error
 format 22cs
 format 35mi
@@ -57,6 +57,7 @@ format 3rms
 #     y
 #     n
 #   index type; one of:
+#     unknown -- used for undefined opcodes and breakpoint
 #     none
 #     varies
 #     type-ref
@@ -311,7 +312,7 @@ op   e8 +iget-wide-volatile         22c  y field-ref     optimized|continue|thro
 op   e9 +iput-wide-volatile         22c  n field-ref     optimized|continue|throw
 op   ea +sget-wide-volatile         21c  y field-ref     optimized|continue|throw
 op   eb +sput-wide-volatile         21c  n field-ref     optimized|continue|throw
-op   ec ^breakpoint                 00x  n none          optimized|continue|throw
+op   ec ^breakpoint                 00x  n unknown       optimized
 op   ed ^throw-verification-error   20bc n varies        optimized|throw
 op   ee +execute-inline             35mi n inline-method optimized|continue|throw
 op   ef +execute-inline/range       3rmi n inline-method optimized|continue|throw
diff --git a/opcode-gen/opcode-gen b/opcode-gen/opcode-gen
index 976a169ad..17560d107 100755
--- a/opcode-gen/opcode-gen
+++ b/opcode-gen/opcode-gen
@@ -172,8 +172,73 @@ consumeUntil != "" {
     next;
 }
 
+/BEGIN\(libdex-widths\)/ {
+    consumeUntil = "END(libdex-widths)";
+    print;
+
+    col = 1;
+    for (i = 0; i <= MAX_LIBDEX_OPCODE; i++) {
+        value = sprintf("%d,", isUnusedByte(i) ? 0 : width[i]);
+        col = colPrint(value, (i == MAX_LIBDEX_OPCODE), col, 16, 2, "    ");
+    }
+
+    next;
+}
+
+/BEGIN\(libdex-flags\)/ {
+    consumeUntil = "END(libdex-flags)";
+    print;
+
+    for (i = 0; i <= MAX_LIBDEX_OPCODE; i++) {
+        value = flagsToC(isUnusedByte(i) ? 0 : flags[i]);
+        printf("    %s,\n", value);
+    }
+
+    next;
+}
+
+/BEGIN\(libdex-formats\)/ {
+    consumeUntil = "END(libdex-formats)";
+    print;
+
+    col = 1;
+    for (i = 0; i <= MAX_LIBDEX_OPCODE; i++) {
+        value = sprintf("kFmt%s,", isUnusedByte(i) ? "00x" : format[i]);
+        col = colPrint(value, (i == MAX_LIBDEX_OPCODE), col, 7, 9, "    ");
+    }
+
+    next;
+}
+
+/BEGIN\(libdex-index-types\)/ {
+    consumeUntil = "END(libdex-index-types)";
+    print;
+
+    col = 1;
+    for (i = 0; i <= MAX_LIBDEX_OPCODE; i++) {
+        value = isUnusedByte(i) ? "unknown" : indexType[i];
+        value = sprintf("%s,", indexTypeValues[value]);
+        col = colPrint(value, (i == MAX_LIBDEX_OPCODE), col, 3, 19, "    ");
+    }
+
+    next;
+}
+
 { print; }
 
+# Helper to print out an element in a multi-column fashion. It returns
+# the (one-based) column number that the next element will be printed
+# in.
+function colPrint(value, isLast, col, numCols, colWidth, linePrefix) {
+    isLast = (isLast || (col == numCols));
+    printf("%s%-*s%s",
+        (col == 1) ? linePrefix : " ",
+        isLast ? 1 : colWidth, value,
+        isLast ? "\n" : "");
+
+    return (col % numCols) + 1;
+}
+
 # Read the bytecode description file.
 function readBytecodes(i, parts, line, cmd, status, count) {
     # locals: parts, line, cmd, status, count
@@ -230,14 +295,19 @@ function defineOpcode(line, count, parts, idx) {
     flags[idx] = parts[6];
 
     # Calculate derived values.
+
     constName[idx] = toupper(name[idx]);
     gsub("[---/]", "_", constName[idx]); # Dash and slash become underscore.
     gsub("[+^]", "", constName[idx]);    # Plus and caret are removed.
     split(name[idx], parts, "/");
+
     family[idx] = toupper(parts[1]);
     gsub("-", "_", family[idx]);         # Dash becomes underscore.
     gsub("[+^]", "", family[idx]);       # Plus and caret are removed.
 
+    split(format[idx], parts, "");       # Width is the first format char.
+    width[idx] = parts[1];
+
     # This association is used when computing "next" opcodes.
     familyFormat[family[idx],format[idx]] = idx;
 
@@ -398,39 +468,39 @@ function isOptimized(idx, parts, f) {
     return 0;
 }
 
-# Returns true if there is no definition for given opcode (by index).
+# Returns true if there is no definition for the given opcode (by index).
 function isUnused(idx) {
     return (name[idx] == "");
 }
 
+# Returns true if there is no definition for the given opcode (by
+# index), taken as a single-byte opcode. The odd case for this
+# function is 255, which is the first extended (two-byte) opcode. For
+# the purposes of this function, it is considered unused. (This is
+# meant as a stop-gap measure for code that is not yet prepared to
+# deal with extended opcodes.)
+function isUnusedByte(idx) {
+    return (idx == 255) || (name[idx] == "");
+}
+
 # Returns the constant name of the given single-byte opcode (by index)
 # or the string "UNUSED_XX" (where XX is the index in hex) if the
-# opcode is unused. The odd case for this function is 255, which is
-# the first extended (two-byte) opcode. For the purposes of this
-# function, it is considered unused. (This is meant as a stop-gap
-# measure for code that is not yet prepared to deal with extended
-# opcodes.)
-function constNameOrUnusedByte(idx, n) {
-    n = constName[idx];
-    if ((n == "") || (i == 255)) {
+# opcode is unused. See isUnusedByte(), above, for more info.
+function constNameOrUnusedByte(idx) {
+    if (isUnusedByte(idx)) {
        return toupper(sprintf("UNUSED_%02x", idx));
     }
-    return n;
+    return constName[idx];
 }
 
 # Returns the (human-oriented) name of the given single-byte opcode
 # (by index) or the string "unused-xx" (where xx is the index in hex)
-# if the opcode is unused. The odd case for this function is 255,
-# which is the first extended (two-byte) opcode. For the purposes of
-# this function, it is considered unused. (This is meant as a stop-gap
-# measure for code that is not yet prepared to deal with extended
-# opcodes.)
-function nameOrUnusedByte(idx, n) {
-    n = name[idx];
-    if ((n == "") || (i == 255)) {
+# if the opcode is unused. See isUnusedByte(), above, for more info.
+function nameOrUnusedByte(idx) {
+    if (isUnusedByte(idx)) {
        return sprintf("unused-%02x", idx);
     }
-    return n;
+    return name[idx];
 }
 ' "$file" > "$tmpfile"
 
diff --git a/opcode-gen/regen-all b/opcode-gen/regen-all
index 4e5c38229..276893e7e 100755
--- a/opcode-gen/regen-all
+++ b/opcode-gen/regen-all
@@ -37,5 +37,6 @@ cd ".."
 ${progdir}/opcode-gen dx/src/com/android/dx/dex/code/DalvOps.java
 ${progdir}/opcode-gen dx/src/com/android/dx/dex/code/Dops.java
 ${progdir}/opcode-gen dx/src/com/android/dx/dex/code/RopToDop.java
+${progdir}/opcode-gen libdex/InstrUtils.c
 ${progdir}/opcode-gen libdex/OpCode.h
 ${progdir}/opcode-gen libdex/OpCodeNames.c