]> arthur.barton.de Git - bup.git/blobdiff - lib/bup/_helpers.c
Remove vestigial and inappropriate isnan()/isinf() timestamp tests.
[bup.git] / lib / bup / _helpers.c
index d077cd9710c08bd007f1a8424cd4ea2f326333ff..0fd20cc233db19fc0152f1a98aa692abdffbb6c4 100644 (file)
@@ -1,15 +1,50 @@
+#define _LARGEFILE64_SOURCE 1
+#define PY_SSIZE_T_CLEAN 1
 #undef NDEBUG
-#include "bupsplit.h"
+#include "../../config/config.h"
+
+// According to Python, its header has to go first:
+//   http://docs.python.org/2/c-api/intro.html#include-files
 #include <Python.h>
+
 #include <assert.h>
-#include <stdint.h>
+#include <errno.h>
 #include <fcntl.h>
 #include <arpa/inet.h>
-#include <unistd.h>
+#include <stdint.h>
 #include <stdlib.h>
 #include <stdio.h>
+#include <sys/mman.h>
 
-static int istty = 0;
+#ifdef HAVE_SYS_TYPES_H
+#include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_STAT_H
+#include <sys/stat.h>
+#endif
+#ifdef HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+
+#ifdef HAVE_LINUX_FS_H
+#include <linux/fs.h>
+#endif
+#ifdef HAVE_SYS_IOCTL_H
+#include <sys/ioctl.h>
+#endif
+
+#include "bupsplit.h"
+
+#if defined(FS_IOC_GETFLAGS) && defined(FS_IOC_SETFLAGS)
+#define BUP_HAVE_FILE_ATTRS 1
+#endif
+
+#ifndef FS_NOCOW_FL
+// Of course, this assumes it's a bitfield value.
+#define FS_NOCOW_FL 0
+#endif
+
+static int istty2 = 0;
 
 // Probably we should use autoconf or something and set HAVE_PY_GETARGCARGV...
 #if __WIN32__ || __CYGWIN__
@@ -82,10 +117,12 @@ static PyObject *blobbits(PyObject *self, PyObject *args)
 static PyObject *splitbuf(PyObject *self, PyObject *args)
 {
     unsigned char *buf = NULL;
-    int len = 0, out = 0, bits = -1;
+    Py_ssize_t len = 0;
+    int out = 0, bits = -1;
 
     if (!PyArg_ParseTuple(args, "t#", &buf, &len))
        return NULL;
+    assert(len <= INT_MAX);
     out = bupsplit_find_ofs(buf, len, &bits);
     if (out) assert(bits >= BUP_BLOBBITS);
     return Py_BuildValue("ii", out, bits);
@@ -95,8 +132,9 @@ static PyObject *splitbuf(PyObject *self, PyObject *args)
 static PyObject *bitmatch(PyObject *self, PyObject *args)
 {
     unsigned char *buf1 = NULL, *buf2 = NULL;
-    int len1 = 0, len2 = 0;
-    int byte, bit;
+    Py_ssize_t len1 = 0, len2 = 0;
+    Py_ssize_t byte;
+    int bit;
 
     if (!PyArg_ParseTuple(args, "t#t#", &buf1, &len1, &buf2, &len2))
        return NULL;
@@ -114,6 +152,7 @@ static PyObject *bitmatch(PyObject *self, PyObject *args)
        }
     }
     
+    assert(byte <= (INT_MAX >> 3));
     return Py_BuildValue("i", byte*8 + bit);
 }
 
@@ -121,7 +160,7 @@ static PyObject *bitmatch(PyObject *self, PyObject *args)
 static PyObject *firstword(PyObject *self, PyObject *args)
 {
     unsigned char *buf = NULL;
-    int len = 0;
+    Py_ssize_t len = 0;
     uint32_t v;
 
     if (!PyArg_ParseTuple(args, "t#", &buf, &len))
@@ -137,66 +176,66 @@ static PyObject *firstword(PyObject *self, PyObject *args)
 
 #define BLOOM2_HEADERLEN 16
 
-typedef struct {
-    uint32_t high;
-    unsigned char low;
-} bits40_t;
-
-static void to_bloom_address_bitmask4(const bits40_t *buf,
+static void to_bloom_address_bitmask4(const unsigned char *buf,
        const int nbits, uint64_t *v, unsigned char *bitmask)
 {
     int bit;
+    uint32_t high;
     uint64_t raw, mask;
 
+    memcpy(&high, buf, 4);
     mask = (1<<nbits) - 1;
-    raw = (((uint64_t)ntohl(buf->high)) << 8) | buf->low;
+    raw = (((uint64_t)ntohl(high) << 8) | buf[4]);
     bit = (raw >> (37-nbits)) & 0x7;
     *v = (raw >> (40-nbits)) & mask;
     *bitmask = 1 << bit;
 }
 
-static void to_bloom_address_bitmask5(const uint32_t *buf,
+static void to_bloom_address_bitmask5(const unsigned char *buf,
        const int nbits, uint32_t *v, unsigned char *bitmask)
 {
     int bit;
+    uint32_t high;
     uint32_t raw, mask;
 
+    memcpy(&high, buf, 4);
     mask = (1<<nbits) - 1;
-    raw = ntohl(*buf);
+    raw = ntohl(high);
     bit = (raw >> (29-nbits)) & 0x7;
     *v = (raw >> (32-nbits)) & mask;
     *bitmask = 1 << bit;
 }
 
-#define BLOOM_SET_BIT(name, address, itype, otype) \
-static void name(unsigned char *bloom, const void *buf, const int nbits)\
+#define BLOOM_SET_BIT(name, address, otype) \
+static void name(unsigned char *bloom, const unsigned char *buf, const int nbits)\
 {\
     unsigned char bitmask;\
     otype v;\
-    address((itype *)buf, nbits, &v, &bitmask);\
+    address(buf, nbits, &v, &bitmask);\
     bloom[BLOOM2_HEADERLEN+v] |= bitmask;\
 }
-BLOOM_SET_BIT(bloom_set_bit4, to_bloom_address_bitmask4, bits40_t, uint64_t)
-BLOOM_SET_BIT(bloom_set_bit5, to_bloom_address_bitmask5, uint32_t, uint32_t)
+BLOOM_SET_BIT(bloom_set_bit4, to_bloom_address_bitmask4, uint64_t)
+BLOOM_SET_BIT(bloom_set_bit5, to_bloom_address_bitmask5, uint32_t)
 
 
-#define BLOOM_GET_BIT(name, address, itype, otype) \
-static int name(const unsigned char *bloom, const void *buf, const int nbits)\
+#define BLOOM_GET_BIT(name, address, otype) \
+static int name(const unsigned char *bloom, const unsigned char *buf, const int nbits)\
 {\
     unsigned char bitmask;\
     otype v;\
-    address((itype *)buf, nbits, &v, &bitmask);\
+    address(buf, nbits, &v, &bitmask);\
     return bloom[BLOOM2_HEADERLEN+v] & bitmask;\
 }
-BLOOM_GET_BIT(bloom_get_bit4, to_bloom_address_bitmask4, bits40_t, uint64_t)
-BLOOM_GET_BIT(bloom_get_bit5, to_bloom_address_bitmask5, uint32_t, uint32_t)
+BLOOM_GET_BIT(bloom_get_bit4, to_bloom_address_bitmask4, uint64_t)
+BLOOM_GET_BIT(bloom_get_bit5, to_bloom_address_bitmask5, uint32_t)
 
 
 static PyObject *bloom_add(PyObject *self, PyObject *args)
 {
     unsigned char *sha = NULL, *bloom = NULL;
     unsigned char *end;
-    int len = 0, blen = 0, nbits = 0, k = 0;
+    Py_ssize_t len = 0, blen = 0;
+    int nbits = 0, k = 0;
 
     if (!PyArg_ParseTuple(args, "w#s#ii", &bloom, &blen, &sha, &len, &nbits, &k))
        return NULL;
@@ -222,13 +261,14 @@ static PyObject *bloom_add(PyObject *self, PyObject *args)
        return NULL;
 
 
-    return Py_BuildValue("i", len/20);
+    return Py_BuildValue("n", len/20);
 }
 
 static PyObject *bloom_contains(PyObject *self, PyObject *args)
 {
     unsigned char *sha = NULL, *bloom = NULL;
-    int len = 0, blen = 0, nbits = 0, k = 0;
+    Py_ssize_t len = 0, blen = 0;
+    int nbits = 0, k = 0;
     unsigned char *end;
     int steps;
 
@@ -257,7 +297,7 @@ static PyObject *bloom_contains(PyObject *self, PyObject *args)
     else
        return NULL;
 
-    return Py_BuildValue("Oi", Py_True, k);
+    return Py_BuildValue("ii", 1, k);
 }
 
 
@@ -270,10 +310,13 @@ static uint32_t _extract_bits(unsigned char *buf, int nbits)
     v = (v >> (32-nbits)) & mask;
     return v;
 }
+
+
 static PyObject *extract_bits(PyObject *self, PyObject *args)
 {
     unsigned char *buf = NULL;
-    int len = 0, nbits = 0;
+    Py_ssize_t len = 0;
+    int nbits = 0;
 
     if (!PyArg_ParseTuple(args, "t#i", &buf, &len, &nbits))
        return NULL;
@@ -288,12 +331,14 @@ static PyObject *extract_bits(PyObject *self, PyObject *args)
 struct sha {
     unsigned char bytes[20];
 };
+
+
 struct idx {
     unsigned char *map;
     struct sha *cur;
     struct sha *end;
     uint32_t *cur_name;
-    long bytes;
+    Py_ssize_t bytes;
     int name_base;
 };
 
@@ -360,11 +405,13 @@ static PyObject *merge_into(PyObject *self, PyObject *args)
 {
     PyObject *ilist = NULL;
     unsigned char *fmap = NULL;
-    struct sha *sha_ptr, *sha_start, *last = NULL;
+    struct sha *sha_ptr, *sha_start = NULL;
     uint32_t *table_ptr, *name_ptr, *name_start;
     struct idx **idxs = NULL;
-    int flen = 0, bits = 0, i;
-    uint32_t total, count, prefix;
+    Py_ssize_t flen = 0;
+    int bits = 0, i;
+    unsigned int total;
+    uint32_t count, prefix;
     int num_i;
     int last_i;
 
@@ -400,7 +447,7 @@ static PyObject *merge_into(PyObject *self, PyObject *args)
     {
        struct idx *idx;
        uint32_t new_prefix;
-       if (count % 102424 == 0 && istty)
+       if (count % 102424 == 0 && istty2)
            fprintf(stderr, "midx: writing %.2f%% (%d/%d)\r",
                    count*100.0/total, count, total);
        idx = idxs[last_i];
@@ -409,7 +456,6 @@ static PyObject *merge_into(PyObject *self, PyObject *args)
            table_ptr[prefix++] = htonl(count);
        memcpy(sha_ptr++, idx->cur, sizeof(struct sha));
        *name_ptr++ = htonl(_get_idx_i(idx));
-       last = idx->cur;
        ++idx->cur;
        if (idx->cur_name != NULL)
            ++idx->cur_name;
@@ -439,30 +485,37 @@ static uint64_t htonll(uint64_t value)
     return value; // already in network byte order MSB-LSB
 }
 
-#define PACK_IDX_V2_HEADERLEN 8
 #define FAN_ENTRIES 256
 
 static PyObject *write_idx(PyObject *self, PyObject *args)
 {
-    PyObject *pf = NULL, *idx = NULL;
+    char *filename = NULL;
+    PyObject *idx = NULL;
     PyObject *part;
-    FILE *f;
     unsigned char *fmap = NULL;
-    int flen = 0;
-    uint32_t total = 0;
+    Py_ssize_t flen = 0;
+    unsigned int total = 0;
     uint32_t count;
     int i, j, ofs64_count;
     uint32_t *fan_ptr, *crc_ptr, *ofs_ptr;
+    uint64_t *ofs64_ptr;
     struct sha *sha_ptr;
 
-    if (!PyArg_ParseTuple(args, "Ow#OI", &pf, &fmap, &flen, &idx, &total))
+    if (!PyArg_ParseTuple(args, "sw#OI", &filename, &fmap, &flen, &idx, &total))
        return NULL;
 
-    fan_ptr = (uint32_t *)&fmap[PACK_IDX_V2_HEADERLEN];
+    if (PyList_Size (idx) != FAN_ENTRIES) // Check for list of the right length.
+        return PyErr_Format (PyExc_TypeError, "idx must contain %d entries",
+                             FAN_ENTRIES);
+
+    const char idx_header[] = "\377tOc\0\0\0\002";
+    memcpy (fmap, idx_header, sizeof(idx_header) - 1);
+
+    fan_ptr = (uint32_t *)&fmap[sizeof(idx_header) - 1];
     sha_ptr = (struct sha *)&fan_ptr[FAN_ENTRIES];
     crc_ptr = (uint32_t *)&sha_ptr[total];
     ofs_ptr = (uint32_t *)&crc_ptr[total];
-    f = PyFile_AsFile(pf);
+    ofs64_ptr = (uint64_t *)&ofs_ptr[total];
 
     count = 0;
     ofs64_count = 0;
@@ -477,26 +530,33 @@ static PyObject *write_idx(PyObject *self, PyObject *args)
        for (j = 0; j < plen; ++j)
        {
            unsigned char *sha = NULL;
-           int sha_len = 0;
-           uint32_t crc = 0;
-           uint64_t ofs = 0;
+           Py_ssize_t sha_len = 0;
+           unsigned int crc = 0;
+            unsigned PY_LONG_LONG ofs_py = 0;
+           uint64_t ofs;
            if (!PyArg_ParseTuple(PyList_GET_ITEM(part, j), "t#IK",
-                                 &sha, &sha_len, &crc, &ofs))
+                                 &sha, &sha_len, &crc, &ofs_py))
                return NULL;
+            assert(crc <= UINT32_MAX);
+            assert(ofs_py <= UINT64_MAX);
+           ofs = ofs_py;
            if (sha_len != sizeof(struct sha))
                return NULL;
            memcpy(sha_ptr++, sha, sizeof(struct sha));
            *crc_ptr++ = htonl(crc);
            if (ofs > 0x7fffffff)
            {
-               uint64_t nofs = htonll(ofs);
-               if (fwrite(&nofs, sizeof(uint64_t), 1, f) != 1)
-                   return PyErr_SetFromErrno(PyExc_OSError);
+                *ofs64_ptr++ = htonll(ofs);
                ofs = 0x80000000 | ofs64_count++;
            }
            *ofs_ptr++ = htonl((uint32_t)ofs);
        }
     }
+
+    int rc = msync(fmap, flen, MS_ASYNC);
+    if (rc != 0)
+       return PyErr_SetFromErrnoWithFilename(PyExc_IOError, filename);
+
     return PyLong_FromUnsignedLong(count);
 }
 
@@ -575,13 +635,10 @@ static PyObject *random_sha(PyObject *self, PyObject *args)
 }
 
 
-static PyObject *open_noatime(PyObject *self, PyObject *args)
+static int _open_noatime(const char *filename, int attrs)
 {
-    char *filename = NULL;
-    int attrs, attrs_noatime, fd;
-    if (!PyArg_ParseTuple(args, "s", &filename))
-       return NULL;
-    attrs = O_RDONLY;
+    int attrs_noatime, fd;
+    attrs |= O_RDONLY;
 #ifdef O_NOFOLLOW
     attrs |= O_NOFOLLOW;
 #endif
@@ -602,8 +659,19 @@ static PyObject *open_noatime(PyObject *self, PyObject *args)
        // just harmlessly ignore it, so this branch won't trigger)
        fd = open(filename, attrs);
     }
+    return fd;
+}
+
+
+static PyObject *open_noatime(PyObject *self, PyObject *args)
+{
+    char *filename = NULL;
+    int fd;
+    if (!PyArg_ParseTuple(args, "s", &filename))
+       return NULL;
+    fd = _open_noatime(filename, 0);
     if (fd < 0)
-       return PyErr_SetFromErrnoWithFilename(PyExc_IOError, filename);
+       return PyErr_SetFromErrnoWithFilename(PyExc_OSError, filename);
     return Py_BuildValue("i", fd);
 }
 
@@ -621,7 +689,398 @@ static PyObject *fadvise_done(PyObject *self, PyObject *args)
 }
 
 
-static PyMethodDef faster_methods[] = {
+#ifdef BUP_HAVE_FILE_ATTRS
+static PyObject *bup_get_linux_file_attr(PyObject *self, PyObject *args)
+{
+    int rc;
+    unsigned int attr;
+    char *path;
+    int fd;
+
+    if (!PyArg_ParseTuple(args, "s", &path))
+        return NULL;
+
+    fd = _open_noatime(path, O_NONBLOCK);
+    if (fd == -1)
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
+
+    attr = 0;
+    rc = ioctl(fd, FS_IOC_GETFLAGS, &attr);
+    if (rc == -1)
+    {
+        close(fd);
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
+    }
+
+    close(fd);
+    return Py_BuildValue("I", attr);
+}
+#endif /* def BUP_HAVE_FILE_ATTRS */
+
+
+#ifdef BUP_HAVE_FILE_ATTRS
+static PyObject *bup_set_linux_file_attr(PyObject *self, PyObject *args)
+{
+    int rc;
+    unsigned int orig_attr, attr;
+    char *path;
+    int fd;
+
+    if (!PyArg_ParseTuple(args, "sI", &path, &attr))
+        return NULL;
+
+    fd = open(path, O_RDONLY | O_NONBLOCK | O_LARGEFILE | O_NOFOLLOW);
+    if (fd == -1)
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
+
+    // Restrict attr to modifiable flags acdeijstuADST -- see
+    // chattr(1) and the e2fsprogs source.  Letter to flag mapping is
+    // in pf.c flags_array[].
+    attr &= FS_APPEND_FL | FS_COMPR_FL | FS_NODUMP_FL | FS_EXTENT_FL
+    | FS_IMMUTABLE_FL | FS_JOURNAL_DATA_FL | FS_SECRM_FL | FS_NOTAIL_FL
+    | FS_UNRM_FL | FS_NOATIME_FL | FS_DIRSYNC_FL | FS_SYNC_FL
+    | FS_TOPDIR_FL | FS_NOCOW_FL;
+
+    // The extents flag can't be removed, so don't (see chattr(1) and chattr.c).
+    rc = ioctl(fd, FS_IOC_GETFLAGS, &orig_attr);
+    if (rc == -1)
+    {
+        close(fd);
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
+    }
+    attr |= (orig_attr & FS_EXTENT_FL);
+
+    rc = ioctl(fd, FS_IOC_SETFLAGS, &attr);
+    if (rc == -1)
+    {
+        close(fd);
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
+    }
+
+    close(fd);
+    return Py_BuildValue("O", Py_None);
+}
+#endif /* def BUP_HAVE_FILE_ATTRS */
+
+
+#ifndef HAVE_UTIMENSAT
+#ifndef HAVE_UTIMES
+#error "cannot find utimensat or utimes()"
+#endif
+#ifndef HAVE_LUTIMES
+#error "cannot find utimensat or lutimes()"
+#endif
+#endif
+
+
+#if defined(HAVE_UTIMENSAT) || defined(HAVE_FUTIMES) || defined(HAVE_LUTIMES)
+
+static int bup_parse_xutime_args(char **path,
+                                 long *access,
+                                 long *access_ns,
+                                 long *modification,
+                                 long *modification_ns,
+                                 PyObject *self, PyObject *args)
+{
+    if (!PyArg_ParseTuple(args, "s((ll)(ll))",
+                          path,
+                          access, access_ns,
+                          modification, modification_ns))
+        return 0;
+    return 1;
+}
+
+#endif /* defined(HAVE_UTIMENSAT) || defined(HAVE_FUTIMES)
+          || defined(HAVE_LUTIMES) */
+
+
+#define INTEGRAL_ASSIGNMENT_FITS(dest, src)                             \
+    ({                                                                  \
+        *(dest) = (src);                                                \
+        *(dest) == (src) && (*(dest) < 1) == ((src) < 1);               \
+    })
+
+
+#define ASSIGN_PYLONG_TO_INTEGRAL(dest, pylong, overflow) \
+    ({                                                     \
+        int result = 0;                                                 \
+        *(overflow) = 0;                                                \
+        const long long ltmp = PyLong_AsLongLong(pylong);               \
+        if (ltmp == -1 && PyErr_Occurred())                             \
+        {                                                               \
+            if (PyErr_ExceptionMatches(PyExc_OverflowError))            \
+            {                                                           \
+                const unsigned long ultmp = PyLong_AsUnsignedLongLong(pylong); \
+                if (ultmp == (unsigned long long) -1 && PyErr_Occurred()) \
+                {                                                       \
+                    if (PyErr_ExceptionMatches(PyExc_OverflowError))    \
+                    {                                                   \
+                        PyErr_Clear();                                  \
+                        *(overflow) = 1;                                \
+                    }                                                   \
+                }                                                       \
+                if (INTEGRAL_ASSIGNMENT_FITS((dest), ultmp))            \
+                    result = 1;                                         \
+                else                                                    \
+                    *(overflow) = 1;                                    \
+            }                                                           \
+        }                                                               \
+        else                                                            \
+        {                                                               \
+            if (INTEGRAL_ASSIGNMENT_FITS((dest), ltmp))                 \
+                result = 1;                                             \
+            else                                                        \
+                *(overflow) = 1;                                        \
+        }                                                               \
+        result;                                                         \
+        })
+
+
+#ifdef HAVE_UTIMENSAT
+
+static PyObject *bup_utimensat(PyObject *self, PyObject *args)
+{
+    int rc;
+    int fd, flag;
+    char *path;
+    PyObject *access_py, *modification_py;
+    struct timespec ts[2];
+
+    if (!PyArg_ParseTuple(args, "is((Ol)(Ol))i",
+                          &fd,
+                          &path,
+                          &access_py, &(ts[0].tv_nsec),
+                          &modification_py, &(ts[1].tv_nsec),
+                          &flag))
+        return NULL;
+
+    int overflow;
+    if (!ASSIGN_PYLONG_TO_INTEGRAL(&(ts[0].tv_sec), access_py, &overflow))
+    {
+        if (overflow)
+            PyErr_SetString(PyExc_ValueError,
+                            "unable to convert access time seconds for utimensat");
+        return NULL;
+    }
+    if (!ASSIGN_PYLONG_TO_INTEGRAL(&(ts[1].tv_sec), modification_py, &overflow))
+    {
+        if (overflow)
+            PyErr_SetString(PyExc_ValueError,
+                            "unable to convert modification time seconds for utimensat");
+        return NULL;
+    }
+    rc = utimensat(fd, path, ts, flag);
+    if (rc != 0)
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
+
+    return Py_BuildValue("O", Py_None);
+}
+
+#endif /* def HAVE_UTIMENSAT */
+
+
+#ifdef HAVE_UTIMES
+#define BUP_HAVE_BUP_UTIME_NS 1
+static PyObject *bup_utime_ns(PyObject *self, PyObject *args)
+{
+    int rc;
+    char *path;
+    long access, access_ns, modification, modification_ns;
+    struct timeval tv[2];
+
+    if (!bup_parse_xutime_args(&path, &access, &access_ns,
+                               &modification, &modification_ns,
+                               self, args))
+       return NULL;
+
+    tv[0].tv_sec = access;
+    tv[0].tv_usec = access_ns / 1000;
+    tv[1].tv_sec = modification;
+    tv[1].tv_usec = modification_ns / 1000;
+    rc = utimes(path, tv);
+    if (rc != 0)
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
+
+    return Py_BuildValue("O", Py_None);
+}
+#endif /* def HAVE_UTIMES */
+
+
+#ifdef HAVE_LUTIMES
+#define BUP_HAVE_BUP_LUTIME_NS 1
+static PyObject *bup_lutime_ns(PyObject *self, PyObject *args)
+{
+    int rc;
+    char *path;
+    long access, access_ns, modification, modification_ns;
+    struct timeval tv[2];
+
+    if (!bup_parse_xutime_args(&path, &access, &access_ns,
+                               &modification, &modification_ns,
+                               self, args))
+       return NULL;
+
+    tv[0].tv_sec = access;
+    tv[0].tv_usec = access_ns / 1000;
+    tv[1].tv_sec = modification;
+    tv[1].tv_usec = modification_ns / 1000;
+    rc = lutimes(path, tv);
+    if (rc != 0)
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, path);
+
+    return Py_BuildValue("O", Py_None);
+}
+#endif /* def HAVE_LUTIMES */
+
+
+#ifdef HAVE_STAT_ST_ATIM
+# define BUP_STAT_ATIME_NS(st) (st)->st_atim.tv_nsec
+# define BUP_STAT_MTIME_NS(st) (st)->st_mtim.tv_nsec
+# define BUP_STAT_CTIME_NS(st) (st)->st_ctim.tv_nsec
+#elif defined HAVE_STAT_ST_ATIMENSEC
+# define BUP_STAT_ATIME_NS(st) (st)->st_atimespec.tv_nsec
+# define BUP_STAT_MTIME_NS(st) (st)->st_mtimespec.tv_nsec
+# define BUP_STAT_CTIME_NS(st) (st)->st_ctimespec.tv_nsec
+#else
+# define BUP_STAT_ATIME_NS(st) 0
+# define BUP_STAT_MTIME_NS(st) 0
+# define BUP_STAT_CTIME_NS(st) 0
+#endif
+
+
+static void set_invalid_timespec_msg(const char *field,
+                                     const long long sec,
+                                     const long nsec,
+                                     const char *filename,
+                                     int fd)
+{
+    if (filename != NULL)
+        PyErr_Format(PyExc_ValueError,
+                     "invalid %s timespec (%lld %ld) for file \"%s\"",
+                     field, sec, nsec, filename);
+    else
+        PyErr_Format(PyExc_ValueError,
+                     "invalid %s timespec (%lld %ld) for file descriptor %d",
+                     field, sec, nsec, fd);
+}
+
+
+static int normalize_timespec_values(const char *name,
+                                     long long *sec,
+                                     long *nsec,
+                                     const char *filename,
+                                     int fd)
+{
+    if (*nsec < -999999999 || *nsec > 999999999)
+    {
+        set_invalid_timespec_msg(name, *sec, *nsec, filename, fd);
+        return 0;
+    }
+    if (*nsec < 0)
+    {
+        if (*sec == LONG_MIN)
+        {
+            set_invalid_timespec_msg(name, *sec, *nsec, filename, fd);
+            return 0;
+        }
+        *nsec += 1000000000;
+        *sec -= 1;
+    }
+    return 1;
+}
+
+
+#define INTEGER_TO_PY(x) \
+    (((x) >= 0) ? PyLong_FromUnsignedLongLong(x) : PyLong_FromLongLong(x))
+
+
+static PyObject *stat_struct_to_py(const struct stat *st,
+                                   const char *filename,
+                                   int fd)
+{
+    long long atime = st->st_atime;
+    long long mtime = st->st_mtime;
+    long long ctime = st->st_ctime;
+    long atime_ns = BUP_STAT_ATIME_NS(st);
+    long mtime_ns = BUP_STAT_MTIME_NS(st);
+    long ctime_ns = BUP_STAT_CTIME_NS(st);
+
+    if (!normalize_timespec_values("atime", &atime, &atime_ns, filename, fd))
+        return NULL;
+    if (!normalize_timespec_values("mtime", &mtime, &mtime_ns, filename, fd))
+        return NULL;
+    if (!normalize_timespec_values("ctime", &ctime, &ctime_ns, filename, fd))
+        return NULL;
+
+    // We can check the known (via POSIX) signed and unsigned types at
+    // compile time, but not (easily) the unspecified types, so handle
+    // those via INTEGER_TO_PY().
+    return Py_BuildValue("OKOOOOOL(Ll)(Ll)(Ll)",
+                         INTEGER_TO_PY(st->st_mode),
+                         (unsigned PY_LONG_LONG) st->st_ino,
+                         INTEGER_TO_PY(st->st_dev),
+                         INTEGER_TO_PY(st->st_nlink),
+                         INTEGER_TO_PY(st->st_uid),
+                         INTEGER_TO_PY(st->st_gid),
+                         INTEGER_TO_PY(st->st_rdev),
+                         (PY_LONG_LONG) st->st_size,
+                         (PY_LONG_LONG) atime,
+                         (long) atime_ns,
+                         (PY_LONG_LONG) mtime,
+                         (long) mtime_ns,
+                         (PY_LONG_LONG) ctime,
+                         (long) ctime_ns);
+}
+
+
+static PyObject *bup_stat(PyObject *self, PyObject *args)
+{
+    int rc;
+    char *filename;
+
+    if (!PyArg_ParseTuple(args, "s", &filename))
+        return NULL;
+
+    struct stat st;
+    rc = stat(filename, &st);
+    if (rc != 0)
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, filename);
+    return stat_struct_to_py(&st, filename, 0);
+}
+
+
+static PyObject *bup_lstat(PyObject *self, PyObject *args)
+{
+    int rc;
+    char *filename;
+
+    if (!PyArg_ParseTuple(args, "s", &filename))
+        return NULL;
+
+    struct stat st;
+    rc = lstat(filename, &st);
+    if (rc != 0)
+        return PyErr_SetFromErrnoWithFilename(PyExc_OSError, filename);
+    return stat_struct_to_py(&st, filename, 0);
+}
+
+
+static PyObject *bup_fstat(PyObject *self, PyObject *args)
+{
+    int rc, fd;
+
+    if (!PyArg_ParseTuple(args, "i", &fd))
+        return NULL;
+
+    struct stat st;
+    rc = fstat(fd, &st);
+    if (rc != 0)
+        return PyErr_SetFromErrno(PyExc_OSError);
+    return stat_struct_to_py(&st, NULL, fd);
+}
+
+
+static PyMethodDef helper_methods[] = {
     { "selftest", selftest, METH_VARARGS,
        "Check that the rolling checksum rolls correctly (for unit tests)." },
     { "blobbits", blobbits, METH_VARARGS,
@@ -650,12 +1109,70 @@ static PyMethodDef faster_methods[] = {
        "open() the given filename for read with O_NOATIME if possible" },
     { "fadvise_done", fadvise_done, METH_VARARGS,
        "Inform the kernel that we're finished with earlier parts of a file" },
+#ifdef BUP_HAVE_FILE_ATTRS
+    { "get_linux_file_attr", bup_get_linux_file_attr, METH_VARARGS,
+      "Return the Linux attributes for the given file." },
+#endif
+#ifdef BUP_HAVE_FILE_ATTRS
+    { "set_linux_file_attr", bup_set_linux_file_attr, METH_VARARGS,
+      "Set the Linux attributes for the given file." },
+#endif
+#ifdef HAVE_UTIMENSAT
+    { "bup_utimensat", bup_utimensat, METH_VARARGS,
+      "Change path timestamps with nanosecond precision (POSIX)." },
+#endif
+#ifdef BUP_HAVE_BUP_UTIME_NS
+    { "bup_utime_ns", bup_utime_ns, METH_VARARGS,
+      "Change path timestamps with up to nanosecond precision." },
+#endif
+#ifdef BUP_HAVE_BUP_LUTIME_NS
+    { "bup_lutime_ns", bup_lutime_ns, METH_VARARGS,
+      "Change path timestamps with up to nanosecond precision;"
+      " don't follow symlinks." },
+#endif
+    { "stat", bup_stat, METH_VARARGS,
+      "Extended version of stat." },
+    { "lstat", bup_lstat, METH_VARARGS,
+      "Extended version of lstat." },
+    { "fstat", bup_fstat, METH_VARARGS,
+      "Extended version of fstat." },
     { NULL, NULL, 0, NULL },  // sentinel
 };
 
+
 PyMODINIT_FUNC init_helpers(void)
 {
-    Py_InitModule("_helpers", faster_methods);
-    istty = isatty(2) || getenv("BUP_FORCE_TTY");
+    // FIXME: migrate these tests to configure.  Check against the
+    // type we're going to use when passing to python.  Other stat
+    // types are tested at runtime.
+    assert(sizeof(ino_t) <= sizeof(unsigned PY_LONG_LONG));
+    assert(sizeof(off_t) <= sizeof(PY_LONG_LONG));
+    assert(sizeof(blksize_t) <= sizeof(PY_LONG_LONG));
+    assert(sizeof(blkcnt_t) <= sizeof(PY_LONG_LONG));
+    // Just be sure (relevant when passing timestamps back to Python above).
+    assert(sizeof(PY_LONG_LONG) <= sizeof(long long));
+
+    char *e;
+    PyObject *m = Py_InitModule("_helpers", helper_methods);
+    if (m == NULL)
+        return;
+
+#ifdef HAVE_UTIMENSAT
+    {
+        PyObject *value;
+        value = INTEGER_TO_PY(AT_FDCWD);
+        PyObject_SetAttrString(m, "AT_FDCWD", value);
+        Py_DECREF(value);
+        value = INTEGER_TO_PY(AT_SYMLINK_NOFOLLOW);
+        PyObject_SetAttrString(m, "AT_SYMLINK_NOFOLLOW", value);
+        Py_DECREF(value);
+        value = INTEGER_TO_PY(UTIME_NOW);
+        PyObject_SetAttrString(m, "UTIME_NOW", value);
+        Py_DECREF(value);
+    }
+#endif
+
+    e = getenv("BUP_FORCE_TTY");
+    istty2 = isatty(2) || (atoi(e ? e : "0") & 2);
     unpythonize_argv();
 }