1 #define _LARGEFILE64_SOURCE 1
16 #include <sys/ioctl.h>
23 // Probably we should use autoconf or something and set HAVE_PY_GETARGCARGV...
24 #if __WIN32__ || __CYGWIN__
26 // There's no 'ps' on win32 anyway, and Py_GetArgcArgv() isn't available.
27 static void unpythonize_argv(void) { }
29 #else // not __WIN32__
31 // For some reason this isn't declared in Python.h
32 extern void Py_GetArgcArgv(int *argc, char ***argv);
34 static void unpythonize_argv(void)
39 Py_GetArgcArgv(&argc, &argv);
41 for (i = 0; i < argc-1; i++)
43 if (argv[i] + strlen(argv[i]) + 1 != argv[i+1])
45 // The argv block doesn't work the way we expected; it's unsafe
51 arge = argv[argc-1] + strlen(argv[argc-1]) + 1;
53 if (strstr(argv[0], "python") && argv[1] == argv[0] + strlen(argv[0]) + 1)
57 p = strrchr(argv[1], '/');
63 memmove(argv[0], p, len);
64 memset(arge - diff, 0, diff);
65 for (i = 0; i < argc; i++)
66 argv[i] = argv[i+1] ? argv[i+1]-diff : NULL;
71 #endif // not __WIN32__ or __CYGWIN__
74 static PyObject *selftest(PyObject *self, PyObject *args)
76 if (!PyArg_ParseTuple(args, ""))
79 return Py_BuildValue("i", !bupsplit_selftest());
83 static PyObject *blobbits(PyObject *self, PyObject *args)
85 if (!PyArg_ParseTuple(args, ""))
87 return Py_BuildValue("i", BUP_BLOBBITS);
91 static PyObject *splitbuf(PyObject *self, PyObject *args)
93 unsigned char *buf = NULL;
94 int len = 0, out = 0, bits = -1;
96 if (!PyArg_ParseTuple(args, "t#", &buf, &len))
98 out = bupsplit_find_ofs(buf, len, &bits);
99 return Py_BuildValue("ii", out, bits);
103 static PyObject *bitmatch(PyObject *self, PyObject *args)
105 unsigned char *buf1 = NULL, *buf2 = NULL;
106 int len1 = 0, len2 = 0;
109 if (!PyArg_ParseTuple(args, "t#t#", &buf1, &len1, &buf2, &len2))
113 for (byte = 0; byte < len1 && byte < len2; byte++)
115 int b1 = buf1[byte], b2 = buf2[byte];
118 for (bit = 0; bit < 8; bit++)
119 if ( (b1 & (0x80 >> bit)) != (b2 & (0x80 >> bit)) )
125 return Py_BuildValue("i", byte*8 + bit);
129 static PyObject *firstword(PyObject *self, PyObject *args)
131 unsigned char *buf = NULL;
135 if (!PyArg_ParseTuple(args, "t#", &buf, &len))
141 v = ntohl(*(uint32_t *)buf);
142 return PyLong_FromUnsignedLong(v);
146 #define BLOOM2_HEADERLEN 16
153 static void to_bloom_address_bitmask4(const bits40_t *buf,
154 const int nbits, uint64_t *v, unsigned char *bitmask)
159 mask = (1<<nbits) - 1;
160 raw = (((uint64_t)ntohl(buf->high)) << 8) | buf->low;
161 bit = (raw >> (37-nbits)) & 0x7;
162 *v = (raw >> (40-nbits)) & mask;
166 static void to_bloom_address_bitmask5(const uint32_t *buf,
167 const int nbits, uint32_t *v, unsigned char *bitmask)
172 mask = (1<<nbits) - 1;
174 bit = (raw >> (29-nbits)) & 0x7;
175 *v = (raw >> (32-nbits)) & mask;
179 #define BLOOM_SET_BIT(name, address, itype, otype) \
180 static void name(unsigned char *bloom, const void *buf, const int nbits)\
182 unsigned char bitmask;\
184 address((itype *)buf, nbits, &v, &bitmask);\
185 bloom[BLOOM2_HEADERLEN+v] |= bitmask;\
187 BLOOM_SET_BIT(bloom_set_bit4, to_bloom_address_bitmask4, bits40_t, uint64_t)
188 BLOOM_SET_BIT(bloom_set_bit5, to_bloom_address_bitmask5, uint32_t, uint32_t)
191 #define BLOOM_GET_BIT(name, address, itype, otype) \
192 static int name(const unsigned char *bloom, const void *buf, const int nbits)\
194 unsigned char bitmask;\
196 address((itype *)buf, nbits, &v, &bitmask);\
197 return bloom[BLOOM2_HEADERLEN+v] & bitmask;\
199 BLOOM_GET_BIT(bloom_get_bit4, to_bloom_address_bitmask4, bits40_t, uint64_t)
200 BLOOM_GET_BIT(bloom_get_bit5, to_bloom_address_bitmask5, uint32_t, uint32_t)
203 static PyObject *bloom_add(PyObject *self, PyObject *args)
205 unsigned char *sha = NULL, *bloom = NULL;
207 int len = 0, blen = 0, nbits = 0, k = 0;
209 if (!PyArg_ParseTuple(args, "w#s#ii", &bloom, &blen, &sha, &len, &nbits, &k))
212 if (blen < 16+(1<<nbits) || len % 20 != 0)
219 for (end = sha + len; sha < end; sha += 20/k)
220 bloom_set_bit5(bloom, sha, nbits);
226 for (end = sha + len; sha < end; sha += 20/k)
227 bloom_set_bit4(bloom, sha, nbits);
233 return Py_BuildValue("i", len/20);
236 static PyObject *bloom_contains(PyObject *self, PyObject *args)
238 unsigned char *sha = NULL, *bloom = NULL;
239 int len = 0, blen = 0, nbits = 0, k = 0;
243 if (!PyArg_ParseTuple(args, "t#s#ii", &bloom, &blen, &sha, &len, &nbits, &k))
253 for (steps = 1, end = sha + 20; sha < end; sha += 20/k, steps++)
254 if (!bloom_get_bit5(bloom, sha, nbits))
255 return Py_BuildValue("Oi", Py_None, steps);
261 for (steps = 1, end = sha + 20; sha < end; sha += 20/k, steps++)
262 if (!bloom_get_bit4(bloom, sha, nbits))
263 return Py_BuildValue("Oi", Py_None, steps);
268 return Py_BuildValue("Oi", Py_True, k);
272 static uint32_t _extract_bits(unsigned char *buf, int nbits)
276 mask = (1<<nbits) - 1;
277 v = ntohl(*(uint32_t *)buf);
278 v = (v >> (32-nbits)) & mask;
281 static PyObject *extract_bits(PyObject *self, PyObject *args)
283 unsigned char *buf = NULL;
284 int len = 0, nbits = 0;
286 if (!PyArg_ParseTuple(args, "t#i", &buf, &len, &nbits))
292 return PyLong_FromUnsignedLong(_extract_bits(buf, nbits));
297 unsigned char bytes[20];
309 static int _cmp_sha(const struct sha *sha1, const struct sha *sha2)
312 for (i = 0; i < sizeof(struct sha); i++)
313 if (sha1->bytes[i] != sha2->bytes[i])
314 return sha1->bytes[i] - sha2->bytes[i];
319 static void _fix_idx_order(struct idx **idxs, int *last_i)
322 int low, mid, high, c = 0;
325 if (idxs[*last_i]->cur >= idxs[*last_i]->end)
327 idxs[*last_i] = NULL;
340 mid = (low + high) / 2;
341 c = _cmp_sha(idx->cur, idxs[mid]->cur);
353 memmove(&idxs[mid+1], &idxs[mid], (*last_i-mid)*sizeof(struct idx *));
358 static uint32_t _get_idx_i(struct idx *idx)
360 if (idx->cur_name == NULL)
361 return idx->name_base;
362 return ntohl(*idx->cur_name) + idx->name_base;
365 #define MIDX4_HEADERLEN 12
367 static PyObject *merge_into(PyObject *self, PyObject *args)
369 PyObject *ilist = NULL;
370 unsigned char *fmap = NULL;
371 struct sha *sha_ptr, *sha_start, *last = NULL;
372 uint32_t *table_ptr, *name_ptr, *name_start;
373 struct idx **idxs = NULL;
374 int flen = 0, bits = 0, i;
375 uint32_t total, count, prefix;
379 if (!PyArg_ParseTuple(args, "w#iIO", &fmap, &flen, &bits, &total, &ilist))
382 num_i = PyList_Size(ilist);
383 idxs = (struct idx **)PyMem_Malloc(num_i * sizeof(struct idx *));
385 for (i = 0; i < num_i; i++)
387 long len, sha_ofs, name_map_ofs;
388 idxs[i] = (struct idx *)PyMem_Malloc(sizeof(struct idx));
389 PyObject *itup = PyList_GetItem(ilist, i);
390 if (!PyArg_ParseTuple(itup, "t#llli", &idxs[i]->map, &idxs[i]->bytes,
391 &len, &sha_ofs, &name_map_ofs, &idxs[i]->name_base))
393 idxs[i]->cur = (struct sha *)&idxs[i]->map[sha_ofs];
394 idxs[i]->end = &idxs[i]->cur[len];
396 idxs[i]->cur_name = (uint32_t *)&idxs[i]->map[name_map_ofs];
398 idxs[i]->cur_name = NULL;
400 table_ptr = (uint32_t *)&fmap[MIDX4_HEADERLEN];
401 sha_start = sha_ptr = (struct sha *)&table_ptr[1<<bits];
402 name_start = name_ptr = (uint32_t *)&sha_ptr[total];
411 if (count % 102424 == 0 && istty)
412 fprintf(stderr, "midx: writing %.2f%% (%d/%d)\r",
413 count*100.0/total, count, total);
415 new_prefix = _extract_bits((unsigned char *)idx->cur, bits);
416 while (prefix < new_prefix)
417 table_ptr[prefix++] = htonl(count);
418 memcpy(sha_ptr++, idx->cur, sizeof(struct sha));
419 *name_ptr++ = htonl(_get_idx_i(idx));
422 if (idx->cur_name != NULL)
424 _fix_idx_order(idxs, &last_i);
427 while (prefix < (1<<bits))
428 table_ptr[prefix++] = htonl(count);
429 assert(count == total);
430 assert(prefix == (1<<bits));
431 assert(sha_ptr == sha_start+count);
432 assert(name_ptr == name_start+count);
435 return PyLong_FromUnsignedLong(count);
438 // This function should technically be macro'd out if it's going to be used
439 // more than ocasionally. As of this writing, it'll actually never be called
440 // in real world bup scenarios (because our packs are < MAX_INT bytes).
441 static uint64_t htonll(uint64_t value)
443 static const int endian_test = 42;
445 if (*(char *)&endian_test == endian_test) // LSB-MSB
446 return ((uint64_t)htonl(value & 0xFFFFFFFF) << 32) | htonl(value >> 32);
447 return value; // already in network byte order MSB-LSB
450 #define PACK_IDX_V2_HEADERLEN 8
451 #define FAN_ENTRIES 256
453 static PyObject *write_idx(PyObject *self, PyObject *args)
455 PyObject *pf = NULL, *idx = NULL;
458 unsigned char *fmap = NULL;
462 int i, j, ofs64_count;
463 uint32_t *fan_ptr, *crc_ptr, *ofs_ptr;
466 if (!PyArg_ParseTuple(args, "Ow#OI", &pf, &fmap, &flen, &idx, &total))
469 fan_ptr = (uint32_t *)&fmap[PACK_IDX_V2_HEADERLEN];
470 sha_ptr = (struct sha *)&fan_ptr[FAN_ENTRIES];
471 crc_ptr = (uint32_t *)&sha_ptr[total];
472 ofs_ptr = (uint32_t *)&crc_ptr[total];
473 f = PyFile_AsFile(pf);
477 for (i = 0; i < FAN_ENTRIES; ++i)
480 part = PyList_GET_ITEM(idx, i);
482 plen = PyList_GET_SIZE(part);
484 *fan_ptr++ = htonl(count);
485 for (j = 0; j < plen; ++j)
487 unsigned char *sha = NULL;
491 if (!PyArg_ParseTuple(PyList_GET_ITEM(part, j), "t#IK",
492 &sha, &sha_len, &crc, &ofs))
494 if (sha_len != sizeof(struct sha))
496 memcpy(sha_ptr++, sha, sizeof(struct sha));
497 *crc_ptr++ = htonl(crc);
498 if (ofs > 0x7fffffff)
500 uint64_t nofs = htonll(ofs);
501 if (fwrite(&nofs, sizeof(uint64_t), 1, f) != 1)
502 return PyErr_SetFromErrno(PyExc_OSError);
503 ofs = 0x80000000 | ofs64_count++;
505 *ofs_ptr++ = htonl((uint32_t)ofs);
508 return PyLong_FromUnsignedLong(count);
512 // I would have made this a lower-level function that just fills in a buffer
513 // with random values, and then written those values from python. But that's
514 // about 20% slower in my tests, and since we typically generate random
515 // numbers for benchmarking other parts of bup, any slowness in generating
516 // random bytes will make our benchmarks inaccurate. Plus nobody wants
517 // pseudorandom bytes much except for this anyway.
518 static PyObject *write_random(PyObject *self, PyObject *args)
520 uint32_t buf[1024/4];
521 int fd = -1, seed = 0, verbose = 0;
523 long long len = 0, kbytes = 0, written = 0;
525 if (!PyArg_ParseTuple(args, "iLii", &fd, &len, &seed, &verbose))
530 for (kbytes = 0; kbytes < len/1024; kbytes++)
533 for (i = 0; i < sizeof(buf)/sizeof(buf[0]); i++)
535 ret = write(fd, buf, sizeof(buf));
539 if (ret < (int)sizeof(buf))
541 if (verbose && kbytes/1024 > 0 && !(kbytes%1024))
542 fprintf(stderr, "Random: %lld Mbytes\r", kbytes/1024);
545 // handle non-multiples of 1024
549 for (i = 0; i < sizeof(buf)/sizeof(buf[0]); i++)
551 ret = write(fd, buf, len % 1024);
558 fprintf(stderr, "Random: %lld Mbytes, done.\n", kbytes/1024);
559 return Py_BuildValue("L", written);
563 static PyObject *random_sha(PyObject *self, PyObject *args)
565 static int seeded = 0;
566 uint32_t shabuf[20/4];
571 assert(sizeof(shabuf) == 20);
576 if (!PyArg_ParseTuple(args, ""))
579 memset(shabuf, 0, sizeof(shabuf));
580 for (i=0; i < 20/4; i++)
581 shabuf[i] = random();
582 return Py_BuildValue("s#", shabuf, 20);
586 static PyObject *open_noatime(PyObject *self, PyObject *args)
588 char *filename = NULL;
589 int attrs, attrs_noatime, fd;
590 if (!PyArg_ParseTuple(args, "s", &filename))
597 attrs |= O_LARGEFILE;
599 attrs_noatime = attrs;
601 attrs_noatime |= O_NOATIME;
603 fd = open(filename, attrs_noatime);
604 if (fd < 0 && errno == EPERM)
606 // older Linux kernels would return EPERM if you used O_NOATIME
607 // and weren't the file's owner. This pointless restriction was
608 // relaxed eventually, but we have to handle it anyway.
609 // (VERY old kernels didn't recognized O_NOATIME, but they would
610 // just harmlessly ignore it, so this branch won't trigger)
611 fd = open(filename, attrs);
614 return PyErr_SetFromErrnoWithFilename(PyExc_IOError, filename);
615 return Py_BuildValue("i", fd);
619 static PyObject *fadvise_done(PyObject *self, PyObject *args)
623 if (!PyArg_ParseTuple(args, "iL", &fd, &ofs))
625 #ifdef POSIX_FADV_DONTNEED
626 posix_fadvise(fd, 0, ofs, POSIX_FADV_DONTNEED);
628 return Py_BuildValue("");
633 static PyObject *bup_get_linux_file_attr(PyObject *self, PyObject *args)
640 if (!PyArg_ParseTuple(args, "s", &path))
643 fd = open(path, O_RDONLY | O_NONBLOCK | O_LARGEFILE | O_NOFOLLOW);
645 return PyErr_SetFromErrnoWithFilename(PyExc_IOError, path);
648 rc = ioctl(fd, FS_IOC_GETFLAGS, &attr);
652 return PyErr_SetFromErrnoWithFilename(PyExc_IOError, path);
656 return Py_BuildValue("k", attr);
660 static PyObject *bup_set_linux_file_attr(PyObject *self, PyObject *args)
667 if (!PyArg_ParseTuple(args, "sk", &path, &attr))
670 fd = open(path, O_RDONLY | O_NONBLOCK | O_LARGEFILE | O_NOFOLLOW);
672 return PyErr_SetFromErrnoWithFilename(PyExc_IOError, path);
674 rc = ioctl(fd, FS_IOC_SETFLAGS, &attr);
678 return PyErr_SetFromErrnoWithFilename(PyExc_IOError, path);
684 #endif /* def linux */
687 #if defined(_ATFILE_SOURCE) \
688 || _XOPEN_SOURCE >= 700 || _POSIX_C_SOURCE >= 200809L
689 #define HAVE_BUP_UTIMENSAT 1
691 static PyObject *bup_utimensat(PyObject *self, PyObject *args)
693 int rc, dirfd, flags;
695 long access, access_ns, modification, modification_ns;
696 struct timespec ts[2];
698 if (!PyArg_ParseTuple(args, "is((ll)(ll))i",
702 &modification, &modification_ns,
708 PyErr_SetString(PyExc_ValueError, "access time is NaN");
711 else if (isinf(access))
713 PyErr_SetString(PyExc_ValueError, "access time is infinite");
716 else if (isnan(modification))
718 PyErr_SetString(PyExc_ValueError, "modification time is NaN");
721 else if (isinf(modification))
723 PyErr_SetString(PyExc_ValueError, "modification time is infinite");
727 if (isnan(access_ns))
729 PyErr_SetString(PyExc_ValueError, "access time ns is NaN");
732 else if (isinf(access_ns))
734 PyErr_SetString(PyExc_ValueError, "access time ns is infinite");
737 else if (isnan(modification_ns))
739 PyErr_SetString(PyExc_ValueError, "modification time ns is NaN");
742 else if (isinf(modification_ns))
744 PyErr_SetString(PyExc_ValueError, "modification time ns is infinite");
748 ts[0].tv_sec = access;
749 ts[0].tv_nsec = access_ns;
750 ts[1].tv_sec = modification;
751 ts[1].tv_nsec = modification_ns;
753 rc = utimensat(dirfd, path, ts, flags);
755 return PyErr_SetFromErrnoWithFilename(PyExc_IOError, path);
760 #endif /* defined(_ATFILE_SOURCE)
761 || _XOPEN_SOURCE >= 700 || _POSIX_C_SOURCE >= 200809L */
764 #ifdef linux /* and likely others */
766 #define HAVE_BUP_STAT 1
767 static PyObject *bup_stat(PyObject *self, PyObject *args)
772 if (!PyArg_ParseTuple(args, "s", &filename))
776 rc = stat(filename, &st);
778 return PyErr_SetFromErrnoWithFilename(PyExc_IOError, filename);
780 return Py_BuildValue("kkkkkkkk"
784 (unsigned long) st.st_mode,
785 (unsigned long) st.st_ino,
786 (unsigned long) st.st_dev,
787 (unsigned long) st.st_nlink,
788 (unsigned long) st.st_uid,
789 (unsigned long) st.st_gid,
790 (unsigned long) st.st_rdev,
791 (unsigned long) st.st_size,
793 (long) st.st_atim.tv_nsec,
795 (long) st.st_mtim.tv_nsec,
797 (long) st.st_ctim.tv_nsec);
801 #define HAVE_BUP_LSTAT 1
802 static PyObject *bup_lstat(PyObject *self, PyObject *args)
807 if (!PyArg_ParseTuple(args, "s", &filename))
811 rc = lstat(filename, &st);
813 return PyErr_SetFromErrnoWithFilename(PyExc_IOError, filename);
815 return Py_BuildValue("kkkkkkkk"
819 (unsigned long) st.st_mode,
820 (unsigned long) st.st_ino,
821 (unsigned long) st.st_dev,
822 (unsigned long) st.st_nlink,
823 (unsigned long) st.st_uid,
824 (unsigned long) st.st_gid,
825 (unsigned long) st.st_rdev,
826 (unsigned long) st.st_size,
828 (long) st.st_atim.tv_nsec,
830 (long) st.st_mtim.tv_nsec,
832 (long) st.st_ctim.tv_nsec);
836 #define HAVE_BUP_FSTAT 1
837 static PyObject *bup_fstat(PyObject *self, PyObject *args)
841 if (!PyArg_ParseTuple(args, "i", &fd))
847 return PyErr_SetFromErrno(PyExc_IOError);
849 return Py_BuildValue("kkkkkkkk"
853 (unsigned long) st.st_mode,
854 (unsigned long) st.st_ino,
855 (unsigned long) st.st_dev,
856 (unsigned long) st.st_nlink,
857 (unsigned long) st.st_uid,
858 (unsigned long) st.st_gid,
859 (unsigned long) st.st_rdev,
860 (unsigned long) st.st_size,
862 (long) st.st_atim.tv_nsec,
864 (long) st.st_mtim.tv_nsec,
866 (long) st.st_ctim.tv_nsec);
869 #endif /* def linux */
872 static PyMethodDef faster_methods[] = {
873 { "selftest", selftest, METH_VARARGS,
874 "Check that the rolling checksum rolls correctly (for unit tests)." },
875 { "blobbits", blobbits, METH_VARARGS,
876 "Return the number of bits in the rolling checksum." },
877 { "splitbuf", splitbuf, METH_VARARGS,
878 "Split a list of strings based on a rolling checksum." },
879 { "bitmatch", bitmatch, METH_VARARGS,
880 "Count the number of matching prefix bits between two strings." },
881 { "firstword", firstword, METH_VARARGS,
882 "Return an int corresponding to the first 32 bits of buf." },
883 { "bloom_contains", bloom_contains, METH_VARARGS,
884 "Check if a bloom filter of 2^nbits bytes contains an object" },
885 { "bloom_add", bloom_add, METH_VARARGS,
886 "Add an object to a bloom filter of 2^nbits bytes" },
887 { "extract_bits", extract_bits, METH_VARARGS,
888 "Take the first 'nbits' bits from 'buf' and return them as an int." },
889 { "merge_into", merge_into, METH_VARARGS,
890 "Merges a bunch of idx and midx files into a single midx." },
891 { "write_idx", write_idx, METH_VARARGS,
892 "Write a PackIdxV2 file from an idx list of lists of tuples" },
893 { "write_random", write_random, METH_VARARGS,
894 "Write random bytes to the given file descriptor" },
895 { "random_sha", random_sha, METH_VARARGS,
896 "Return a random 20-byte string" },
897 { "open_noatime", open_noatime, METH_VARARGS,
898 "open() the given filename for read with O_NOATIME if possible" },
899 { "fadvise_done", fadvise_done, METH_VARARGS,
900 "Inform the kernel that we're finished with earlier parts of a file" },
902 { "get_linux_file_attr", bup_get_linux_file_attr, METH_VARARGS,
903 "Return the Linux attributes for the given file." },
904 { "set_linux_file_attr", bup_set_linux_file_attr, METH_VARARGS,
905 "Set the Linux attributes for the given file." },
907 #ifdef HAVE_BUP_UTIMENSAT
908 { "utimensat", bup_utimensat, METH_VARARGS,
909 "Change file timestamps with nanosecond precision." },
912 { "stat", bup_stat, METH_VARARGS,
913 "Extended version of stat." },
915 #ifdef HAVE_BUP_LSTAT
916 { "lstat", bup_lstat, METH_VARARGS,
917 "Extended version of lstat." },
919 #ifdef HAVE_BUP_FSTAT
920 { "fstat", bup_fstat, METH_VARARGS,
921 "Extended version of fstat." },
923 { NULL, NULL, 0, NULL }, // sentinel
927 PyMODINIT_FUNC init_helpers(void)
929 PyObject *m = Py_InitModule("_helpers", faster_methods);
932 #ifdef HAVE_BUP_UTIMENSAT
933 PyModule_AddObject(m, "AT_FDCWD", Py_BuildValue("i", AT_FDCWD));
934 PyModule_AddObject(m, "AT_SYMLINK_NOFOLLOW",
935 Py_BuildValue("i", AT_SYMLINK_NOFOLLOW));
939 PyModule_AddObject(m, "_have_ns_fs_timestamps", Py_True);
942 PyModule_AddObject(m, "_have_ns_fs_timestamps", Py_False);
944 istty = isatty(2) || getenv("BUP_FORCE_TTY");