Rework the bytecode format file and parser.
This change adds a set of "format" commands to the bytecode.txt file
and prefixes each bytecode definition with an "op" keyword. The new
parser script understands both commands (though it mostly ignores the
"format" ones) and is now a little less wedded to the idea that there
can only be 256 opcodes.
This all is in preparation for (a) adding new opcodes, and (b) making
the opcode/instruction fitter a little more efficient.
Change-Id: I1b4e7274da06b197536823148d91f426d049ec19
diff --git a/dx/etc/bytecode.txt b/dx/etc/bytecode.txt
index f1df5bf..28908be 100644
--- a/dx/etc/bytecode.txt
+++ b/dx/etc/bytecode.txt
@@ -12,267 +12,292 @@
# See the License for the specific language governing permissions and
# limitations under the License.
+#
# Bytecode definition file
#
-# Columns are:
+
+# One line per instruction format family. Each line consists of a
+# series of instruction formats that all take (potentially) compatible
+# arguments. The order is the preferred order (most to least
+# preferable) of formats, when more than one opcode could be used for
+# a given instruction's arguments.
+#
+# Note: The family that starts with 12x has a mix of both two- and
+# three- register formats. This is because some of the two-register
+# opcodes effectively take three, with a destination and two sources
+# where the destination and one of the sources have to be the same.
+
+format 10t 20t 30t
+format 10x
+format 11n 21s 21h 31i 51l
+format 11x
+format 12x 22x 23x 32x 33x # See note, above.
+format 21c 31c 41c
+format 21t 31t
+format 22b 22s 32s
+format 22c 52c
+format 22t
+format 35c 3rc 5rc
+
+# One line per opcode. Columns are:
# hex for opcode
# format
# has result register (y/n)
# opcode name
-00 10x n nop
-01 12x y move
-02 22x y move/from16
-03 32x y move/16
-04 12x y move-wide
-05 22x y move-wide/from16
-06 32x y move-wide/16
-07 12x y move-object
-08 22x y move-object/from16
-09 32x y move-object/16
-0a 11x y move-result
-0b 11x y move-result-wide
-0c 11x y move-result-object
-0d 11x y move-exception
-0e 10x n return-void
-0f 11x n return
-10 11x n return-wide
-11 11x n return-object
-12 11n y const/4
-13 21s y const/16
-14 31i y const
-15 21h y const/high16
-16 21s y const-wide/16
-17 31i y const-wide/32
-18 51l y const-wide
-19 21h y const-wide/high16
-1a 21c y const-string
-1b 31c y const-string/jumbo
-1c 21c y const-class
-1d 11x n monitor-enter
-1e 11x n monitor-exit
-1f 21c y check-cast
-20 22c y instance-of
-21 12x y array-length
-22 21c y new-instance
-23 22c y new-array
-24 35c n filled-new-array
-25 3rc n filled-new-array/range
-26 31t n fill-array-data
-27 11x n throw
-28 10t n goto
-29 20t n goto/16
-2a 30t n goto/32
-2b 31t n packed-switch
-2c 31t n sparse-switch
-2d 23x y cmpl-float
-2e 23x y cmpg-float
-2f 23x y cmpl-double
-30 23x y cmpg-double
-31 23x y cmp-long
-32 22t n if-eq
-33 22t n if-ne
-34 22t n if-lt
-35 22t n if-ge
-36 22t n if-gt
-37 22t n if-le
-38 21t n if-eqz
-39 21t n if-nez
-3a 21t n if-ltz
-3b 21t n if-gez
-3c 21t n if-gtz
-3d 21t n if-lez
-3e 10x n unused-3e
-3f 10x n unused-3f
-40 10x n unused-40
-41 10x n unused-41
-42 10x n unused-42
-43 10x n unused-43
-44 23x y aget
-45 23x y aget-wide
-46 23x y aget-object
-47 23x y aget-boolean
-48 23x y aget-byte
-49 23x y aget-char
-4a 23x y aget-short
-4b 23x n aput
-4c 23x n aput-wide
-4d 23x n aput-object
-4e 23x n aput-boolean
-4f 23x n aput-byte
-50 23x n aput-char
-51 23x n aput-short
-52 22c y iget
-53 22c y iget-wide
-54 22c y iget-object
-55 22c y iget-boolean
-56 22c y iget-byte
-57 22c y iget-char
-58 22c y iget-short
-59 22c n iput
-5a 22c n iput-wide
-5b 22c n iput-object
-5c 22c n iput-boolean
-5d 22c n iput-byte
-5e 22c n iput-char
-5f 22c n iput-short
-60 21c y sget
-61 21c y sget-wide
-62 21c y sget-object
-63 21c y sget-boolean
-64 21c y sget-byte
-65 21c y sget-char
-66 21c y sget-short
-67 21c n sput
-68 21c n sput-wide
-69 21c n sput-object
-6a 21c n sput-boolean
-6b 21c n sput-byte
-6c 21c n sput-char
-6d 21c n sput-short
-6e 35c n invoke-virtual
-6f 35c n invoke-super
-70 35c n invoke-direct
-71 35c n invoke-static
-72 35c n invoke-interface
-73 10x n unused-73
-74 3rc n invoke-virtual/range
-75 3rc n invoke-super/range
-76 3rc n invoke-direct/range
-77 3rc n invoke-static/range
-78 3rc n invoke-interface/range
-79 10x n unused-79
-7a 10x n unused-7a
-7b 12x y neg-int
-7c 12x y not-int
-7d 12x y neg-long
-7e 12x y not-long
-7f 12x y neg-float
-80 12x y neg-double
-81 12x y int-to-long
-82 12x y int-to-float
-83 12x y int-to-double
-84 12x y long-to-int
-85 12x y long-to-float
-86 12x y long-to-double
-87 12x y float-to-int
-88 12x y float-to-long
-89 12x y float-to-double
-8a 12x y double-to-int
-8b 12x y double-to-long
-8c 12x y double-to-float
-8d 12x y int-to-byte
-8e 12x y int-to-char
-8f 12x y int-to-short
-90 23x y add-int
-91 23x y sub-int
-92 23x y mul-int
-93 23x y div-int
-94 23x y rem-int
-95 23x y and-int
-96 23x y or-int
-97 23x y xor-int
-98 23x y shl-int
-99 23x y shr-int
-9a 23x y ushr-int
-9b 23x y add-long
-9c 23x y sub-long
-9d 23x y mul-long
-9e 23x y div-long
-9f 23x y rem-long
-a0 23x y and-long
-a1 23x y or-long
-a2 23x y xor-long
-a3 23x y shl-long
-a4 23x y shr-long
-a5 23x y ushr-long
-a6 23x y add-float
-a7 23x y sub-float
-a8 23x y mul-float
-a9 23x y div-float
-aa 23x y rem-float
-ab 23x y add-double
-ac 23x y sub-double
-ad 23x y mul-double
-ae 23x y div-double
-af 23x y rem-double
-b0 12x y add-int/2addr
-b1 12x y sub-int/2addr
-b2 12x y mul-int/2addr
-b3 12x y div-int/2addr
-b4 12x y rem-int/2addr
-b5 12x y and-int/2addr
-b6 12x y or-int/2addr
-b7 12x y xor-int/2addr
-b8 12x y shl-int/2addr
-b9 12x y shr-int/2addr
-ba 12x y ushr-int/2addr
-bb 12x y add-long/2addr
-bc 12x y sub-long/2addr
-bd 12x y mul-long/2addr
-be 12x y div-long/2addr
-bf 12x y rem-long/2addr
-c0 12x y and-long/2addr
-c1 12x y or-long/2addr
-c2 12x y xor-long/2addr
-c3 12x y shl-long/2addr
-c4 12x y shr-long/2addr
-c5 12x y ushr-long/2addr
-c6 12x y add-float/2addr
-c7 12x y sub-float/2addr
-c8 12x y mul-float/2addr
-c9 12x y div-float/2addr
-ca 12x y rem-float/2addr
-cb 12x y add-double/2addr
-cc 12x y sub-double/2addr
-cd 12x y mul-double/2addr
-ce 12x y div-double/2addr
-cf 12x y rem-double/2addr
-d0 22s y add-int/lit16
-d1 22s y rsub-int
-d2 22s y mul-int/lit16
-d3 22s y div-int/lit16
-d4 22s y rem-int/lit16
-d5 22s y and-int/lit16
-d6 22s y or-int/lit16
-d7 22s y xor-int/lit16
-d8 22b y add-int/lit8
-d9 22b y rsub-int/lit8
-da 22b y mul-int/lit8
-db 22b y div-int/lit8
-dc 22b y rem-int/lit8
-dd 22b y and-int/lit8
-de 22b y or-int/lit8
-df 22b y xor-int/lit8
-e0 22b y shl-int/lit8
-e1 22b y shr-int/lit8
-e2 22b y ushr-int/lit8
-e3 10x n unused-e3
-e4 10x n unused-e4
-e5 10x n unused-e5
-e6 10x n unused-e6
-e7 10x n unused-e7
-e8 10x n unused-e8
-e9 10x n unused-e9
-ea 10x n unused-ea
-eb 10x n unused-eb
-ec 10x n unused-ec
-ed 10x n unused-ed
-ee 10x n unused-ee
-ef 10x n unused-ef
-f0 10x n unused-f0
-f1 10x n unused-f1
-f2 10x n unused-f2
-f3 10x n unused-f3
-f4 10x n unused-f4
-f5 10x n unused-f5
-f6 10x n unused-f6
-f7 10x n unused-f7
-f8 10x n unused-f8
-f9 10x n unused-f9
-fa 10x n unused-fa
-fb 10x n unused-fb
-fc 10x n unused-fc
-fd 10x n unused-fd
-fe 10x n unused-fe
-ff 10x n unused-ff
+op 00 10x n nop
+op 01 12x y move
+op 02 22x y move/from16
+op 03 32x y move/16
+op 04 12x y move-wide
+op 05 22x y move-wide/from16
+op 06 32x y move-wide/16
+op 07 12x y move-object
+op 08 22x y move-object/from16
+op 09 32x y move-object/16
+op 0a 11x y move-result
+op 0b 11x y move-result-wide
+op 0c 11x y move-result-object
+op 0d 11x y move-exception
+op 0e 10x n return-void
+op 0f 11x n return
+op 10 11x n return-wide
+op 11 11x n return-object
+op 12 11n y const/4
+op 13 21s y const/16
+op 14 31i y const
+op 15 21h y const/high16
+op 16 21s y const-wide/16
+op 17 31i y const-wide/32
+op 18 51l y const-wide
+op 19 21h y const-wide/high16
+op 1a 21c y const-string
+op 1b 31c y const-string/jumbo
+op 1c 21c y const-class
+op 1d 11x n monitor-enter
+op 1e 11x n monitor-exit
+op 1f 21c y check-cast
+op 20 22c y instance-of
+op 21 12x y array-length
+op 22 21c y new-instance
+op 23 22c y new-array
+op 24 35c n filled-new-array
+op 25 3rc n filled-new-array/range
+op 26 31t n fill-array-data
+op 27 11x n throw
+op 28 10t n goto
+op 29 20t n goto/16
+op 2a 30t n goto/32
+op 2b 31t n packed-switch
+op 2c 31t n sparse-switch
+op 2d 23x y cmpl-float
+op 2e 23x y cmpg-float
+op 2f 23x y cmpl-double
+op 30 23x y cmpg-double
+op 31 23x y cmp-long
+op 32 22t n if-eq
+op 33 22t n if-ne
+op 34 22t n if-lt
+op 35 22t n if-ge
+op 36 22t n if-gt
+op 37 22t n if-le
+op 38 21t n if-eqz
+op 39 21t n if-nez
+op 3a 21t n if-ltz
+op 3b 21t n if-gez
+op 3c 21t n if-gtz
+op 3d 21t n if-lez
+op 3e 10x n unused-3e
+op 3f 10x n unused-3f
+op 40 10x n unused-40
+op 41 10x n unused-41
+op 42 10x n unused-42
+op 43 10x n unused-43
+op 44 23x y aget
+op 45 23x y aget-wide
+op 46 23x y aget-object
+op 47 23x y aget-boolean
+op 48 23x y aget-byte
+op 49 23x y aget-char
+op 4a 23x y aget-short
+op 4b 23x n aput
+op 4c 23x n aput-wide
+op 4d 23x n aput-object
+op 4e 23x n aput-boolean
+op 4f 23x n aput-byte
+op 50 23x n aput-char
+op 51 23x n aput-short
+op 52 22c y iget
+op 53 22c y iget-wide
+op 54 22c y iget-object
+op 55 22c y iget-boolean
+op 56 22c y iget-byte
+op 57 22c y iget-char
+op 58 22c y iget-short
+op 59 22c n iput
+op 5a 22c n iput-wide
+op 5b 22c n iput-object
+op 5c 22c n iput-boolean
+op 5d 22c n iput-byte
+op 5e 22c n iput-char
+op 5f 22c n iput-short
+op 60 21c y sget
+op 61 21c y sget-wide
+op 62 21c y sget-object
+op 63 21c y sget-boolean
+op 64 21c y sget-byte
+op 65 21c y sget-char
+op 66 21c y sget-short
+op 67 21c n sput
+op 68 21c n sput-wide
+op 69 21c n sput-object
+op 6a 21c n sput-boolean
+op 6b 21c n sput-byte
+op 6c 21c n sput-char
+op 6d 21c n sput-short
+op 6e 35c n invoke-virtual
+op 6f 35c n invoke-super
+op 70 35c n invoke-direct
+op 71 35c n invoke-static
+op 72 35c n invoke-interface
+op 73 10x n unused-73
+op 74 3rc n invoke-virtual/range
+op 75 3rc n invoke-super/range
+op 76 3rc n invoke-direct/range
+op 77 3rc n invoke-static/range
+op 78 3rc n invoke-interface/range
+op 79 10x n unused-79
+op 7a 10x n unused-7a
+op 7b 12x y neg-int
+op 7c 12x y not-int
+op 7d 12x y neg-long
+op 7e 12x y not-long
+op 7f 12x y neg-float
+op 80 12x y neg-double
+op 81 12x y int-to-long
+op 82 12x y int-to-float
+op 83 12x y int-to-double
+op 84 12x y long-to-int
+op 85 12x y long-to-float
+op 86 12x y long-to-double
+op 87 12x y float-to-int
+op 88 12x y float-to-long
+op 89 12x y float-to-double
+op 8a 12x y double-to-int
+op 8b 12x y double-to-long
+op 8c 12x y double-to-float
+op 8d 12x y int-to-byte
+op 8e 12x y int-to-char
+op 8f 12x y int-to-short
+op 90 23x y add-int
+op 91 23x y sub-int
+op 92 23x y mul-int
+op 93 23x y div-int
+op 94 23x y rem-int
+op 95 23x y and-int
+op 96 23x y or-int
+op 97 23x y xor-int
+op 98 23x y shl-int
+op 99 23x y shr-int
+op 9a 23x y ushr-int
+op 9b 23x y add-long
+op 9c 23x y sub-long
+op 9d 23x y mul-long
+op 9e 23x y div-long
+op 9f 23x y rem-long
+op a0 23x y and-long
+op a1 23x y or-long
+op a2 23x y xor-long
+op a3 23x y shl-long
+op a4 23x y shr-long
+op a5 23x y ushr-long
+op a6 23x y add-float
+op a7 23x y sub-float
+op a8 23x y mul-float
+op a9 23x y div-float
+op aa 23x y rem-float
+op ab 23x y add-double
+op ac 23x y sub-double
+op ad 23x y mul-double
+op ae 23x y div-double
+op af 23x y rem-double
+op b0 12x y add-int/2addr
+op b1 12x y sub-int/2addr
+op b2 12x y mul-int/2addr
+op b3 12x y div-int/2addr
+op b4 12x y rem-int/2addr
+op b5 12x y and-int/2addr
+op b6 12x y or-int/2addr
+op b7 12x y xor-int/2addr
+op b8 12x y shl-int/2addr
+op b9 12x y shr-int/2addr
+op ba 12x y ushr-int/2addr
+op bb 12x y add-long/2addr
+op bc 12x y sub-long/2addr
+op bd 12x y mul-long/2addr
+op be 12x y div-long/2addr
+op bf 12x y rem-long/2addr
+op c0 12x y and-long/2addr
+op c1 12x y or-long/2addr
+op c2 12x y xor-long/2addr
+op c3 12x y shl-long/2addr
+op c4 12x y shr-long/2addr
+op c5 12x y ushr-long/2addr
+op c6 12x y add-float/2addr
+op c7 12x y sub-float/2addr
+op c8 12x y mul-float/2addr
+op c9 12x y div-float/2addr
+op ca 12x y rem-float/2addr
+op cb 12x y add-double/2addr
+op cc 12x y sub-double/2addr
+op cd 12x y mul-double/2addr
+op ce 12x y div-double/2addr
+op cf 12x y rem-double/2addr
+op d0 22s y add-int/lit16
+op d1 22s y rsub-int
+op d2 22s y mul-int/lit16
+op d3 22s y div-int/lit16
+op d4 22s y rem-int/lit16
+op d5 22s y and-int/lit16
+op d6 22s y or-int/lit16
+op d7 22s y xor-int/lit16
+op d8 22b y add-int/lit8
+op d9 22b y rsub-int/lit8
+op da 22b y mul-int/lit8
+op db 22b y div-int/lit8
+op dc 22b y rem-int/lit8
+op dd 22b y and-int/lit8
+op de 22b y or-int/lit8
+op df 22b y xor-int/lit8
+op e0 22b y shl-int/lit8
+op e1 22b y shr-int/lit8
+op e2 22b y ushr-int/lit8
+op e3 10x n unused-e3
+op e4 10x n unused-e4
+op e5 10x n unused-e5
+op e6 10x n unused-e6
+op e7 10x n unused-e7
+op e8 10x n unused-e8
+op e9 10x n unused-e9
+op ea 10x n unused-ea
+op eb 10x n unused-eb
+op ec 10x n unused-ec
+op ed 10x n unused-ed
+op ee 10x n unused-ee
+op ef 10x n unused-ef
+op f0 10x n unused-f0
+op f1 10x n unused-f1
+op f2 10x n unused-f2
+op f3 10x n unused-f3
+op f4 10x n unused-f4
+op f5 10x n unused-f5
+op f6 10x n unused-f6
+op f7 10x n unused-f7
+op f8 10x n unused-f8
+op f9 10x n unused-f9
+op fa 10x n unused-fa
+op fb 10x n unused-fb
+op fc 10x n unused-fc
+op fd 10x n unused-fd
+op fe 10x n unused-fe
+op ff 10x n unused-ff
diff --git a/dx/etc/opcode-gen b/dx/etc/opcode-gen
index 390a6c3..518be4d 100755
--- a/dx/etc/opcode-gen
+++ b/dx/etc/opcode-gen
@@ -16,8 +16,9 @@
# opcode-gen <file>
#
-# Use the file bytecodes.txt to generate code inside <file>, based on
-# the directives found in that file:
+# This script uses the file bytecodes.txt (in this directory) to
+# generate code inside the given <file>, based on the directives found
+# in that file:
#
# opcodes: static final ints for each opcode
# dops: static final objects for each opcode
@@ -114,27 +115,88 @@
{ print; }
-function readBytecodes(i, parts) {
- for (i = 0; i < 256; i++) {
- $0 = "";
- while (($0 == "") || (index($0, "#") != 0)) {
- if ((getline <bytecodeFile) != 1) {
- print "trouble reading bytecode file";
- exit 1;
- }
+# Read the bytecode description file.
+function readBytecodes(i, parts, line, cmd, status, count) {
+ # locals: parts, line, cmd, status, count
+ for (;;) {
+ # Read a line.
+ status = getline line <bytecodeFile;
+ if (status == 0) break;
+ if (status < 0) {
+ print "trouble reading bytecode file";
+ exit 1;
}
- split($0, parts);
- hex[i] = parts[1];
- format[i] = parts[2];
- hasres[i] = (parts[3] == "n") ? "false" : "true";
- name[i] = parts[4];
- uppername[i] = toupper(parts[4]);
- gsub("[---/]", "_", uppername[i]);
- split(name[i], parts, "/");
- family[i] = toupper(parts[1]);
- gsub("-", "_", family[i]);
+
+ # Clean up the line and extract the command
+ gsub(/ */, " ", line);
+ sub(/ *#.*$/, "", line);
+ sub(/ $/, "", line);
+ sub(/^ /, "", line);
+ count = split(line, parts);
+ if (count == 0) continue; # Blank or comment line.
+ cmd = parts[1];
+ sub(/^[a-z][a-z]* */, "", line); # Remove the command from line.
+
+ if (cmd == "op") {
+ status = defineOpcode(line);
+ } else if (cmd == "format") {
+ status = defineFormat(line);
+ } else {
+ status = -1;
+ }
+
+ if (status != 0) {
+ printf("syntax error on line: %s\n", line);
+ }
}
}
+
+# Define an opcode.
+function defineOpcode(line, count, parts, idx) {
+ # locals: count, parts, idx
+ count = split(line, parts);
+ if (count != 4) return -1;
+ idx = parseHex(parts[1]);
+ if (idx < 0) return -1;
+
+ hex[idx] = parts[1];
+ format[idx] = parts[2];
+ hasres[idx] = (parts[3] == "n") ? "false" : "true";
+ name[idx] = parts[4];
+ uppername[idx] = toupper(parts[4]);
+ gsub("[---/]", "_", uppername[idx]);
+ split(name[idx], parts, "/");
+ family[idx] = toupper(parts[1]);
+ gsub("-", "_", family[idx]);
+
+ return 0;
+}
+
+# Define a format family.
+function defineFormat(line, count, parts) {
+ # locals: count, parts
+ count = split(line, parts);
+ if (count < 1) return -1;
+ formats[parts[1]] = line;
+ return 0;
+}
+
+# Convert a hex value to an int.
+function parseHex(hex, result, chars, count, c, i) {
+ # locals: result, chars, count, c, i
+ hex = tolower(hex);
+ count = split(hex, chars, "");
+ result = 0;
+ for (i = 1; i <= count; i++) {
+ c = index("0123456789abcdef", chars[i]);
+ if (c == 0) {
+ printf("bogus hex value: %s\n", hex);
+ return -1;
+ }
+ result = (result * 16) + c - 1;
+ }
+ return result;
+}
' "$file" > "$tmpfile"
cp "$tmpfile" "$file"
