Mercurial > hg > mercurial-crew-with-dirclash
annotate mercurial/mpatch.c @ 384:a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
mpatch: attempt to handle unpack alignment issues on Solaris
manifest hash: e185dc380bab61cf11a9973ee3ddd2e904e56299
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)
iD8DBQFCsnOtywK+sNU5EO8RAlzQAJ9YIhbL8BJjT+J/pOiQVES2wsF0igCgnFRl
ok5f8i8GbNk77sRbpsGnUF0=
=m0Zh
-----END PGP SIGNATURE-----
| author | mpm@selenic.com |
|---|---|
| date | Thu, 16 Jun 2005 22:54:37 -0800 |
| parents | 97d83e7fbf2f |
| children | 7c678976df3e 688d03d6997a |
| rev | line source |
|---|---|
| 72 | 1 /* |
| 2 mpatch.c - efficient binary patching for Mercurial | |
| 3 | |
| 4 This implements a patch algorithm that's O(m + nlog n) where m is the | |
| 5 size of the output and n is the number of patches. | |
| 6 | |
| 7 Given a list of binary patches, it unpacks each into a hunk list, | |
| 8 then combines the hunk lists with a treewise recursion to form a | |
| 9 single hunk list. This hunk list is then applied to the original | |
| 10 text. | |
| 11 | |
| 12 The text (or binary) fragments are copied directly from their source | |
| 13 Python objects into a preallocated output string to avoid the | |
| 14 allocation of intermediate Python objects. Working memory is about 2x | |
| 15 the total number of hunks. | |
| 16 | |
| 17 Copyright 2005 Matt Mackall <mpm@selenic.com> | |
| 18 | |
| 19 This software may be used and distributed according to the terms | |
| 20 of the GNU General Public License, incorporated herein by reference. | |
| 21 */ | |
| 22 | |
| 23 #include <Python.h> | |
| 24 #include <stdlib.h> | |
| 25 #include <string.h> | |
| 26 #include <netinet/in.h> | |
| 27 #include <sys/types.h> | |
| 28 | |
| 29 static char mpatch_doc[] = "Efficient binary patching."; | |
| 30 | |
| 31 struct frag { | |
| 32 int start, end, len; | |
| 33 char *data; | |
| 34 }; | |
| 35 | |
| 36 struct flist { | |
| 37 struct frag *base, *head, *tail; | |
| 38 }; | |
| 39 | |
| 40 static struct flist *lalloc(int size) | |
| 41 { | |
| 128 | 42 struct flist *a = NULL; |
| 72 | 43 |
| 44 a = malloc(sizeof(struct flist)); | |
| 128 | 45 if (a) { |
| 46 a->base = malloc(sizeof(struct frag) * size); | |
| 282 | 47 if (!a->base) { |
| 128 | 48 free(a); |
| 282 | 49 a = NULL; |
| 50 } else | |
| 128 | 51 a->head = a->tail = a->base; |
| 52 } | |
| 72 | 53 return a; |
| 54 } | |
| 55 | |
| 56 static void lfree(struct flist *a) | |
| 57 { | |
| 128 | 58 if (a) { |
| 59 free(a->base); | |
| 60 free(a); | |
| 61 } | |
| 72 | 62 } |
| 63 | |
| 64 static int lsize(struct flist *a) | |
| 65 { | |
| 66 return a->tail - a->head; | |
| 67 } | |
| 68 | |
| 69 /* move hunks in source that are less cut to dest, compensating | |
| 70 for changes in offset. the last hunk may be split if necessary. | |
| 71 */ | |
| 72 static int gather(struct flist *dest, struct flist *src, int cut, int offset) | |
| 73 { | |
| 74 struct frag *d = dest->tail, *s = src->head; | |
| 75 int postend, c, l; | |
| 76 | |
| 77 while (s != src->tail) { | |
| 78 if (s->start + offset >= cut) | |
| 82 | 79 break; /* we've gone far enough */ |
| 72 | 80 |
| 81 postend = offset + s->start + s->len; | |
| 82 if (postend <= cut) { | |
| 83 /* save this hunk */ | |
| 84 offset += s->start + s->len - s->end; | |
| 85 *d++ = *s++; | |
| 86 } | |
| 87 else { | |
| 88 /* break up this hunk */ | |
| 89 c = cut - offset; | |
| 90 if (s->end < c) | |
| 91 c = s->end; | |
| 92 l = cut - offset - s->start; | |
| 93 if (s->len < l) | |
| 94 l = s->len; | |
| 95 | |
| 96 offset += s->start + l - c; | |
| 97 | |
| 98 d->start = s->start; | |
| 99 d->end = c; | |
| 100 d->len = l; | |
| 101 d->data = s->data; | |
| 102 d++; | |
| 103 s->start = c; | |
| 104 s->len = s->len - l; | |
| 105 s->data = s->data + l; | |
| 106 | |
| 82 | 107 break; |
| 72 | 108 } |
| 109 } | |
| 110 | |
| 111 dest->tail = d; | |
| 112 src->head = s; | |
| 113 return offset; | |
| 114 } | |
| 115 | |
| 116 /* like gather, but with no output list */ | |
| 117 static int discard(struct flist *src, int cut, int offset) | |
| 118 { | |
| 119 struct frag *s = src->head; | |
| 120 int postend, c, l; | |
| 121 | |
| 122 while (s != src->tail) { | |
| 123 if (s->start + offset >= cut) | |
| 82 | 124 break; |
| 72 | 125 |
| 126 postend = offset + s->start + s->len; | |
| 127 if (postend <= cut) { | |
| 128 offset += s->start + s->len - s->end; | |
| 129 s++; | |
| 130 } | |
| 131 else { | |
| 132 c = cut - offset; | |
| 133 if (s->end < c) | |
| 134 c = s->end; | |
| 135 l = cut - offset - s->start; | |
| 136 if (s->len < l) | |
| 137 l = s->len; | |
| 138 | |
| 139 offset += s->start + l - c; | |
| 140 s->start = c; | |
| 141 s->len = s->len - l; | |
| 142 s->data = s->data + l; | |
| 143 | |
| 82 | 144 break; |
| 72 | 145 } |
| 146 } | |
| 147 | |
| 148 src->head = s; | |
| 149 return offset; | |
| 150 } | |
| 151 | |
| 152 /* combine hunk lists a and b, while adjusting b for offset changes in a/ | |
| 153 this deletes a and b and returns the resultant list. */ | |
| 154 static struct flist *combine(struct flist *a, struct flist *b) | |
| 155 { | |
| 128 | 156 struct flist *c = NULL; |
| 157 struct frag *bh, *ct; | |
| 72 | 158 int offset = 0, post; |
| 159 | |
| 128 | 160 if (a && b) |
| 161 c = lalloc((lsize(a) + lsize(b)) * 2); | |
| 162 | |
| 163 if (c) { | |
| 72 | 164 |
| 128 | 165 for (bh = b->head; bh != b->tail; bh++) { |
| 166 /* save old hunks */ | |
| 167 offset = gather(c, a, bh->start, offset); | |
| 72 | 168 |
| 128 | 169 /* discard replaced hunks */ |
| 170 post = discard(a, bh->end, offset); | |
| 72 | 171 |
| 128 | 172 /* insert new hunk */ |
| 173 ct = c->tail; | |
| 174 ct->start = bh->start - offset; | |
| 175 ct->end = bh->end - post; | |
| 176 ct->len = bh->len; | |
| 177 ct->data = bh->data; | |
| 178 c->tail++; | |
| 179 offset = post; | |
| 180 } | |
| 181 | |
| 182 /* hold on to tail from a */ | |
| 183 memcpy(c->tail, a->head, sizeof(struct frag) * lsize(a)); | |
| 184 c->tail += lsize(a); | |
| 72 | 185 } |
| 186 | |
| 187 lfree(a); | |
| 188 lfree(b); | |
| 189 return c; | |
| 190 } | |
| 191 | |
| 192 /* decode a binary patch into a hunk list */ | |
| 193 static struct flist *decode(char *bin, int len) | |
| 194 { | |
| 195 struct flist *l; | |
| 196 struct frag *lt; | |
| 197 char *end = bin + len; | |
|
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
198 char decode[12]; /* for dealing with alignment issues */ |
| 72 | 199 |
| 200 /* assume worst case size, we won't have many of these lists */ | |
| 201 l = lalloc(len / 12); | |
| 202 lt = l->tail; | |
| 203 | |
| 204 while (bin < end) { | |
|
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
205 memcpy(decode, bin, 12); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
206 lt->start = ntohl(*(uint32_t *)decode); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
207 lt->end = ntohl(*(uint32_t *)(decode + 4)); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
208 lt->len = ntohl(*(uint32_t *)(decode + 8)); |
| 72 | 209 lt->data = bin + 12; |
| 210 bin += 12 + lt->len; | |
| 211 lt++; | |
| 212 } | |
| 213 | |
| 214 l->tail = lt; | |
| 215 return l; | |
| 216 } | |
| 217 | |
| 218 /* calculate the size of resultant text */ | |
| 219 static int calcsize(int len, struct flist *l) | |
| 220 { | |
| 221 int outlen = 0, last = 0; | |
| 222 struct frag *f = l->head; | |
| 223 | |
| 224 while (f != l->tail) { | |
| 225 outlen += f->start - last; | |
| 226 last = f->end; | |
| 227 outlen += f->len; | |
| 228 f++; | |
| 229 } | |
| 230 | |
| 231 outlen += len - last; | |
| 232 return outlen; | |
| 233 } | |
| 234 | |
| 235 static void apply(char *buf, char *orig, int len, struct flist *l) | |
| 236 { | |
| 237 struct frag *f = l->head; | |
| 238 int last = 0; | |
| 239 char *p = buf; | |
| 240 | |
| 241 while (f != l->tail) { | |
| 242 memcpy(p, orig + last, f->start - last); | |
| 243 p += f->start - last; | |
| 244 memcpy(p, f->data, f->len); | |
| 245 last = f->end; | |
| 246 p += f->len; | |
| 247 f++; | |
| 248 } | |
| 249 memcpy(p, orig + last, len - last); | |
| 250 } | |
| 251 | |
| 252 /* recursively generate a patch of all bins between start and end */ | |
| 253 static struct flist *fold(PyObject *bins, int start, int end) | |
| 254 { | |
| 255 int len; | |
| 256 | |
| 257 if (start + 1 == end) { | |
| 258 /* trivial case, output a decoded list */ | |
| 259 PyObject *tmp = PyList_GetItem(bins, start); | |
| 128 | 260 if (!tmp) |
| 261 return NULL; | |
| 72 | 262 return decode(PyString_AsString(tmp), PyString_Size(tmp)); |
| 263 } | |
| 264 | |
| 265 /* divide and conquer, memory management is elsewhere */ | |
| 266 len = (end - start) / 2; | |
| 267 return combine(fold(bins, start, start + len), | |
| 268 fold(bins, start + len, end)); | |
| 269 } | |
| 270 | |
| 271 static PyObject * | |
| 272 patches(PyObject *self, PyObject *args) | |
| 273 { | |
| 274 PyObject *text, *bins, *result; | |
| 275 struct flist *patch; | |
| 276 char *in, *out; | |
| 277 int len, outlen; | |
| 278 | |
| 128 | 279 if (!PyArg_ParseTuple(args, "SO:mpatch", &text, &bins)) |
| 72 | 280 return NULL; |
| 281 | |
| 282 len = PyList_Size(bins); | |
| 283 if (!len) { | |
| 284 /* nothing to do */ | |
| 285 Py_INCREF(text); | |
| 286 return text; | |
| 287 } | |
| 288 | |
| 289 patch = fold(bins, 0, len); | |
| 128 | 290 if (!patch) |
| 291 return PyErr_NoMemory(); | |
| 292 | |
| 72 | 293 outlen = calcsize(PyString_Size(text), patch); |
| 294 result = PyString_FromStringAndSize(NULL, outlen); | |
| 128 | 295 if (result) { |
| 296 in = PyString_AsString(text); | |
| 297 out = PyString_AsString(result); | |
| 298 apply(out, in, PyString_Size(text), patch); | |
| 299 } | |
| 300 | |
| 72 | 301 lfree(patch); |
| 302 return result; | |
| 303 } | |
| 304 | |
| 305 static PyMethodDef methods[] = { | |
| 306 {"patches", patches, METH_VARARGS, "apply a series of patches\n"}, | |
| 307 {NULL, NULL} | |
| 308 }; | |
| 309 | |
| 310 PyMODINIT_FUNC | |
| 311 initmpatch(void) | |
| 312 { | |
| 313 Py_InitModule3("mpatch", methods, mpatch_doc); | |
| 314 } | |
| 315 |