| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 1 | #!/usr/bin/env python |
| 2 | # Merge or print the coverage data collected by asan's coverage. |
| 3 | # Input files are sequences of 4-byte integers. |
| 4 | # We need to merge these integers into a set and then |
| 5 | # either print them (as hex) or dump them into another file. |
| 6 | import array |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 7 | import struct |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 8 | import sys |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 9 | import bisect |
| 10 | import os.path |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 11 | |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 12 | prog_name = "" |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 13 | |
| 14 | def Usage(): |
| 15 | print >> sys.stderr, "Usage: \n" + \ |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 16 | " " + prog_name + " [32|64] merge file1 [file2 ...] > output\n" \ |
| 17 | " " + prog_name + " [32|64] print file1 [file2 ...]\n" \ |
| 18 | " " + prog_name + " [32|64] unpack file1 [file2 ...]\n" \ |
| 19 | " " + prog_name + " [32|64] rawunpack file1 [file2 ...]\n" |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 20 | exit(1) |
| 21 | |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 22 | def CheckBits(bits): |
| 23 | if bits != 32 and bits != 64: |
| 24 | raise Exception("Wrong bitness: %d" % bits) |
| 25 | |
| 26 | def TypeCodeForBits(bits): |
| 27 | CheckBits(bits) |
| 28 | return 'L' if bits == 64 else 'I' |
| 29 | |
| 30 | kMagic32SecondHalf = 0xFFFFFF32; |
| 31 | kMagic64SecondHalf = 0xFFFFFF64; |
| 32 | kMagicFirstHalf = 0xC0BFFFFF; |
| 33 | |
| 34 | def MagicForBits(bits): |
| 35 | CheckBits(bits) |
| 36 | if sys.byteorder == 'little': |
| 37 | return [kMagic64SecondHalf if bits == 64 else kMagic32SecondHalf, kMagicFirstHalf] |
| 38 | else: |
| 39 | return [kMagicFirstHalf, kMagic64SecondHalf if bits == 64 else kMagic32SecondHalf] |
| 40 | |
| 41 | def ReadMagicAndReturnBitness(f, path): |
| 42 | magic_bytes = f.read(8) |
| 43 | magic_words = struct.unpack('II', magic_bytes); |
| 44 | bits = 0 |
| 45 | idx = 1 if sys.byteorder == 'little' else 0 |
| 46 | if magic_words[idx] == kMagicFirstHalf: |
| 47 | if magic_words[1-idx] == kMagic64SecondHalf: |
| 48 | bits = 64 |
| 49 | elif magic_words[1-idx] == kMagic32SecondHalf: |
| 50 | bits = 32 |
| 51 | if bits == 0: |
| 52 | raise Exception('Bad magic word in %s' % path) |
| 53 | return bits |
| 54 | |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 55 | def ReadOneFile(path): |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 56 | with open(path, mode="rb") as f: |
| 57 | f.seek(0, 2) |
| 58 | size = f.tell() |
| 59 | f.seek(0, 0) |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 60 | if size < 8: |
| 61 | raise Exception('File %s is short (< 8 bytes)' % path) |
| 62 | bits = ReadMagicAndReturnBitness(f, path) |
| 63 | size -= 8 |
| 64 | s = array.array(TypeCodeForBits(bits), f.read(size)) |
| 65 | print >>sys.stderr, "%s: read %d %d-bit PCs from %s" % (prog_name, size * 8 / bits, bits, path) |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 66 | return s |
| 67 | |
| 68 | def Merge(files): |
| 69 | s = set() |
| 70 | for f in files: |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 71 | s = s.union(set(ReadOneFile(f))) |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 72 | print >> sys.stderr, "%s: %d files merged; %d PCs total" % \ |
| 73 | (prog_name, len(files), len(s)) |
| 74 | return sorted(s) |
| 75 | |
| 76 | def PrintFiles(files): |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 77 | if len(files) > 1: |
| 78 | s = Merge(files) |
| 79 | else: # If there is just on file, print the PCs in order. |
| 80 | s = ReadOneFile(files[0]) |
| 81 | print >> sys.stderr, "%s: 1 file merged; %d PCs total" % \ |
| 82 | (prog_name, len(s)) |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 83 | for i in s: |
| 84 | print "0x%x" % i |
| 85 | |
| 86 | def MergeAndPrint(files): |
| 87 | if sys.stdout.isatty(): |
| 88 | Usage() |
| 89 | s = Merge(files) |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 90 | bits = 32 |
| 91 | if max(s) > 0xFFFFFFFF: |
| 92 | bits = 64 |
| 93 | array.array('I', MagicForBits(bits)).tofile(sys.stdout) |
| 94 | a = array.array(TypeCodeForBits(bits), s) |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 95 | a.tofile(sys.stdout) |
| 96 | |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 97 | |
| 98 | def UnpackOneFile(path): |
| 99 | with open(path, mode="rb") as f: |
| 100 | print >> sys.stderr, "%s: unpacking %s" % (prog_name, path) |
| 101 | while True: |
| 102 | header = f.read(12) |
| 103 | if not header: return |
| 104 | if len(header) < 12: |
| 105 | break |
| 106 | pid, module_length, blob_size = struct.unpack('iII', header) |
| 107 | module = f.read(module_length) |
| 108 | blob = f.read(blob_size) |
| 109 | assert(len(module) == module_length) |
| 110 | assert(len(blob) == blob_size) |
| 111 | extracted_file = "%s.%d.sancov" % (module, pid) |
| 112 | print >> sys.stderr, "%s: extracting %s" % \ |
| 113 | (prog_name, extracted_file) |
| 114 | # The packed file may contain multiple blobs for the same pid/module |
| 115 | # pair. Append to the end of the file instead of overwriting. |
| 116 | with open(extracted_file, 'ab') as f2: |
| 117 | f2.write(blob) |
| 118 | # fail |
| 119 | raise Exception('Error reading file %s' % path) |
| 120 | |
| 121 | |
| 122 | def Unpack(files): |
| 123 | for f in files: |
| 124 | UnpackOneFile(f) |
| 125 | |
| 126 | def UnpackOneRawFile(path, map_path): |
| 127 | mem_map = [] |
| 128 | with open(map_path, mode="rt") as f_map: |
| 129 | print >> sys.stderr, "%s: reading map %s" % (prog_name, map_path) |
| 130 | bits = int(f_map.readline()) |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 131 | if bits != 32 and bits != 64: |
| 132 | raise Exception('Wrong bits size in the map') |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 133 | for line in f_map: |
| 134 | parts = line.rstrip().split() |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 135 | mem_map.append((int(parts[0], 16), |
| 136 | int(parts[1], 16), |
| 137 | int(parts[2], 16), |
| Stephen Hines | 6a211c5 | 2014-07-21 00:49:56 -0700 | [diff] [blame] | 138 | ' '.join(parts[3:]))) |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 139 | mem_map.sort(key=lambda m : m[0]) |
| 140 | mem_map_keys = [m[0] for m in mem_map] |
| 141 | |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 142 | with open(path, mode="rb") as f: |
| 143 | print >> sys.stderr, "%s: unpacking %s" % (prog_name, path) |
| 144 | |
| 145 | f.seek(0, 2) |
| 146 | size = f.tell() |
| 147 | f.seek(0, 0) |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 148 | pcs = array.array(TypeCodeForBits(bits), f.read(size)) |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 149 | mem_map_pcs = [[] for i in range(0, len(mem_map))] |
| 150 | |
| 151 | for pc in pcs: |
| 152 | if pc == 0: continue |
| 153 | map_idx = bisect.bisect(mem_map_keys, pc) - 1 |
| 154 | (start, end, base, module_path) = mem_map[map_idx] |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 155 | assert pc >= start |
| 156 | if pc >= end: |
| 157 | print >> sys.stderr, "warning: %s: pc %x outside of any known mapping" % (prog_name, pc) |
| 158 | continue |
| 159 | mem_map_pcs[map_idx].append(pc - base) |
| 160 | |
| 161 | for ((start, end, base, module_path), pc_list) in zip(mem_map, mem_map_pcs): |
| 162 | if len(pc_list) == 0: continue |
| 163 | assert path.endswith('.sancov.raw') |
| 164 | dst_path = module_path + '.' + os.path.basename(path)[:-4] |
| Stephen Hines | 86277eb | 2015-03-23 12:06:32 -0700 | [diff] [blame] | 165 | print >> sys.stderr, "%s: writing %d PCs to %s" % (prog_name, len(pc_list), dst_path) |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 166 | arr = array.array(TypeCodeForBits(bits)) |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 167 | arr.fromlist(sorted(pc_list)) |
| 168 | with open(dst_path, 'ab') as f2: |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 169 | array.array('I', MagicForBits(bits)).tofile(f2) |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 170 | arr.tofile(f2) |
| 171 | |
| 172 | def RawUnpack(files): |
| 173 | for f in files: |
| 174 | if not f.endswith('.sancov.raw'): |
| 175 | raise Exception('Unexpected raw file name %s' % f) |
| 176 | f_map = f[:-3] + 'map' |
| 177 | UnpackOneRawFile(f, f_map) |
| 178 | |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 179 | if __name__ == '__main__': |
| 180 | prog_name = sys.argv[0] |
| 181 | if len(sys.argv) <= 2: |
| 182 | Usage(); |
| Pirama Arumuga Nainar | 7c91505 | 2015-04-08 08:58:29 -0700 | [diff] [blame] | 183 | |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 184 | if sys.argv[1] == "print": |
| 185 | PrintFiles(sys.argv[2:]) |
| 186 | elif sys.argv[1] == "merge": |
| 187 | MergeAndPrint(sys.argv[2:]) |
| Stephen Hines | 2d1fdb2 | 2014-05-28 23:58:16 -0700 | [diff] [blame] | 188 | elif sys.argv[1] == "unpack": |
| 189 | Unpack(sys.argv[2:]) |
| 190 | elif sys.argv[1] == "rawunpack": |
| 191 | RawUnpack(sys.argv[2:]) |
| Kostya Serebryany | 106cb08 | 2013-11-15 11:51:08 +0000 | [diff] [blame] | 192 | else: |
| 193 | Usage() |