- Revamped the hard-link algorithm to save memory.
authorWayne Davison <wayned@samba.org>
Sat, 9 Dec 2006 00:23:55 +0000 (00:23 +0000)
committerWayne Davison <wayned@samba.org>
Sat, 9 Dec 2006 00:23:55 +0000 (00:23 +0000)
- Improved the function names.
- Improved a few variable names.
- Got rid of the "is a hard link" message in favor of the normal status
  messages for files (e.g. --itemize-changes indicates hard-links).
- Fixed a long-standing bug when combining -H with --compare-dest.
- Made sure that code compiles when hard-linking is not available.

hlink.c

diff --git a/hlink.c b/hlink.c
index 23f6a9b..7fc92ac 100644 (file)
--- a/hlink.c
+++ b/hlink.c
@@ -30,17 +30,24 @@ extern int link_dest;
 extern int make_backups;
 extern int remove_source_files;
 extern int stdout_format_has_i;
+extern int maybe_ATTRS_REPORT;
 extern char *basis_dir[];
 extern struct file_list *the_file_list;
 
 #ifdef SUPPORT_HARD_LINKS
 
-#define SKIPPED_LINK (-1)
-#define FINISHED_LINK (-2)
+alloc_pool_t hlink_pool;
 
 #define FPTR(i) (the_file_list->files[i])
 #define LINKED(i1,i2) ((i1)->dev == (i2)->dev && (i1)->ino == (i2)->ino)
 
+void init_hard_links(void)
+{
+       if (!(hlink_pool = pool_create(HLINK_EXTENT, sizeof (struct idev),
+                                      out_of_memory, POOL_INTERN)))
+               out_of_memory("init_hard_links");
+}
+
 static int hlink_compare(int *int1, int *int2)
 {
        struct file_struct *f1 = FPTR(*int1);
@@ -57,111 +64,76 @@ static int hlink_compare(int *int1, int *int2)
        return f_name_cmp(f1, f2);
 }
 
-static int32 *hlink_list;
-static int32 hlink_count;
-
-/* Analyze the data in the hlink_list[], remove items that aren't multiply
- * linked, and replace the dev+inode data with the hlindex+next linked list. */
-static void link_idev_data(void)
+/* Analyze the dev+inode data in the file-list by creating a list of all
+ * the items that have hlink data, sorting them, and matching up identical
+ * values into clusters.  These will be a single linked list from last to
+ * first when we're done. */
+void match_hard_links(void)
 {
-       int32 from, to, start;
+       int32 from, prev, *ndx_list;
        struct file_struct *file, *file_next;
        struct idev *idev, *idev_next;
-       struct hlist *hl;
+       int i, ndx_count = 0;
 
-       alloc_pool_t hlink_pool;
-       alloc_pool_t idev_pool = the_file_list->hlink_pool;
+       if (!(ndx_list = new_array(int32, the_file_list->count)))
+               out_of_memory("match_hard_links");
 
-       hlink_pool = pool_create(128 * 1024, sizeof (struct hlist), out_of_memory, POOL_INTERN);
+       for (i = 0; i < the_file_list->count; i++) {
+               if (F_IS_HLINKED(FPTR(i)))
+                       ndx_list[ndx_count++] = i;
+       }
 
-       for (from = to = 0; from < hlink_count; from++) {
-               start = from;
-               for (file = FPTR(hlink_list[from]), idev = F_HL_IDEV(file);
-                    from < hlink_count-1;
-                    file = file_next, idev = idev_next)
+       if (!ndx_count) {
+               free(ndx_list);
+               return;
+       }
+
+       qsort(ndx_list, ndx_count, sizeof ndx_list[0],
+            (int (*)()) hlink_compare);
+
+       for (from = 0; from < ndx_count; from++) {
+               for (file = FPTR(ndx_list[from]), idev = F_HL_IDEV(file), prev = -1;
+                    from < ndx_count-1;
+                    file = file_next, idev = idev_next, prev = ndx_list[from++])
                {
-                       file_next = FPTR(hlink_list[from+1]);
+                       file_next = FPTR(ndx_list[from+1]);
                        idev_next = F_HL_IDEV(file_next);
                        if (!LINKED(idev, idev_next))
                                break;
-                       pool_free(idev_pool, 0, idev);
-                       hl = pool_talloc(hlink_pool, struct hlist, 1,
-                                        "hlink_list");
-                       hl->hlindex = to;
-                       hl->next = hlink_list[++from];
-                       hl->dest_used = 0;
-                       F_HL_LIST(file) = hl;
+                       pool_free(hlink_pool, 0, idev);
+                       if (prev < 0)
+                               file->flags |= FLAG_HLINK_FIRST;
+                       F_HL_PREV(file) = prev;
                }
-               pool_free(idev_pool, 0, idev);
-               if (from > start) {
-                       int head = hlink_list[start];
-                       hl = pool_talloc(hlink_pool, struct hlist, 1,
-                                        "hlink_list");
-                       FPTR(head)->flags |= FLAG_HLINK_FIRST;
-                       hl->hlindex = to;
-                       hl->next = head;
-                       hl->dest_used = 0;
-                       hlink_list[to++] = head;
-                       file->flags |= FLAG_HLINK_LAST;
-                       F_HL_LIST(file) = hl;
-               } else
+               pool_free(hlink_pool, 0, idev);
+               if (prev < 0)
                        file->flags &= ~FLAG_HLINKED;
+               else {
+                       file->flags |= FLAG_HLINK_LAST;
+                       F_HL_PREV(file) = prev;
+               }
        }
 
-       if (!to) {
-               free(hlink_list);
-               hlink_list = NULL;
-               pool_destroy(hlink_pool);
-               hlink_pool = NULL;
-       } else {
-               hlink_count = to;
-               hlink_list = realloc_array(hlink_list, int32, hlink_count);
-               if (!hlink_list)
-                       out_of_memory("init_hard_links");
-       }
-       the_file_list->hlink_pool = hlink_pool;
-       pool_destroy(idev_pool);
-}
-
-void init_hard_links(void)
-{
-       int i;
-
-       if (hlink_list)
-               free(hlink_list);
-
-       if (!(hlink_list = new_array(int32, the_file_list->count)))
-               out_of_memory("init_hard_links");
-
-       hlink_count = 0;
-       for (i = 0; i < the_file_list->count; i++) {
-               if (F_IS_HLINKED(FPTR(i)))
-                       hlink_list[hlink_count++] = i;
-       }
-
-       qsort(hlink_list, hlink_count,
-           sizeof hlink_list[0], (int (*)()) hlink_compare);
-
-       if (!hlink_count) {
-               free(hlink_list);
-               hlink_list = NULL;
-       } else
-               link_idev_data();
+       pool_destroy(hlink_pool);
+       free(ndx_list);
 }
 
 static int maybe_hard_link(struct file_struct *file, int ndx,
                           const char *fname, int statret, STRUCT_STAT *stp,
-                          const char *toname, STRUCT_STAT *to_stp,
-                          int itemizing, enum logcode code)
+                          const char *oldname, STRUCT_STAT *old_stp,
+                          const char *realname, int itemizing, enum logcode code)
 {
        if (statret == 0) {
-               if (stp->st_dev == to_stp->st_dev
-                && stp->st_ino == to_stp->st_ino) {
+               if (stp->st_dev == old_stp->st_dev
+                && stp->st_ino == old_stp->st_ino) {
                        if (itemizing) {
                                itemize(file, ndx, statret, stp,
                                        ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
                                        0, "");
                        }
+                       if (verbose > 1 && maybe_ATTRS_REPORT)
+                               rprintf(FCLIENT, "%s is uptodate\n", fname);
+                       file->flags |= FLAG_HLINK_DONE;
                        return 0;
                }
                if (make_backups) {
@@ -173,86 +145,118 @@ static int maybe_hard_link(struct file_struct *file, int ndx,
                        return -1;
                }
        }
-       return hard_link_one(file, ndx, fname, statret, stp, toname,
-                            0, itemizing, code);
+
+       if (hard_link_one(file, fname, oldname, 0)) {
+               if (itemizing) {
+                       itemize(file, ndx, statret, stp,
+                               ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
+                               realname);
+               }
+               if (code != FNONE && verbose)
+                       rprintf(code, "%s => %s\n", fname, realname);
+               return 0;
+       }
+       return -1;
 }
 
+/* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
+ * 0 = process the file, 1 = skip the file, -1 = error occurred. */
 int hard_link_check(struct file_struct *file, int ndx, const char *fname,
                    int statret, STRUCT_STAT *stp, int itemizing,
-                   enum logcode code, int skip)
+                   enum logcode code)
 {
-       int head;
-       struct hlist *hl = F_HL_LIST(file);
-
-       if (skip && !(file->flags & FLAG_HLINK_LAST))
-               head = hlink_list[hl->hlindex] = hl->next;
-       else
-               head = hlink_list[hl->hlindex];
-       if (ndx != head) {
-               struct file_struct *head_file = FPTR(head);
-               struct hlist *hf_hl = F_HL_LIST(head_file);
-               if (!stdout_format_has_i && verbose > 1) {
-                       rprintf(FINFO, "\"%s\" is a hard link\n",
-                               f_name(file, NULL));
+       STRUCT_STAT prev_st;
+       char prev_name[MAXPATHLEN], altbuf[MAXPATHLEN], *realname;
+       int alt_dest, prev_ndx = F_HL_PREV(file);
+       struct file_struct *prev_file = FPTR(prev_ndx);
+
+       /* Is the previous link is not complete yet? */
+       if (!(prev_file->flags & FLAG_HLINK_DONE)) {
+               /* Is the previous link being transferred? */
+               if (prev_file->flags & FLAG_SENT) {
+                       /* Add ourselves to the list of files that will be
+                        * updated when the transfer completes, and mark
+                        * ourself as waiting for the transfer. */
+                       F_HL_PREV(file) = F_HL_PREV(prev_file);
+                       F_HL_PREV(prev_file) = ndx;
+                       file->flags |= FLAG_SENT;
+                       return 1;
                }
-               if (hf_hl->hlindex == FINISHED_LINK) {
-                       STRUCT_STAT st2, st3;
-                       char toname[MAXPATHLEN];
-                       int ldu = hf_hl->dest_used;
-                       if (ldu) {
-                               pathjoin(toname, MAXPATHLEN, basis_dir[ldu-1],
-                                        f_name(head_file, NULL));
-                       } else
-                               f_name(head_file, toname);
-                       if (link_stat(toname, &st2, 0) < 0) {
-                               rsyserr(FERROR, errno, "stat %s failed",
-                                       full_fname(toname));
-                               return -1;
-                       }
-                       if (statret < 0 && basis_dir[0] != NULL) {
-                               char cmpbuf[MAXPATHLEN];
-                               int j = 0;
-                               do {
-                                       pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
-                                       if (link_stat(cmpbuf, &st3, 0) < 0)
-                                               continue;
-                                       if (link_dest) {
-                                               if (st2.st_dev != st3.st_dev
-                                                || st2.st_ino != st3.st_ino)
-                                                       continue;
-                                               statret = 1;
-                                               stp = &st3;
-                                               if (verbose < 2 || !stdout_format_has_i) {
-                                                       itemizing = 0;
-                                                       code = FNONE;
-                                               }
-                                               break;
-                                       }
-                                       if (!unchanged_file(cmpbuf, file, &st3))
-                                               continue;
-                                       statret = 1;
-                                       stp = &st3;
-                                       if (unchanged_attrs(file, &st3))
-                                               break;
-                               } while (basis_dir[++j] != NULL);
+               return 0;
+       }
+
+       /* There is a finished file to link with! */
+       if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
+               /* The previous previous will be marked with FIRST. */
+               prev_ndx = F_HL_PREV(prev_file);
+               prev_file = FPTR(prev_ndx);
+               /* Update our previous pointer to point to the first. */
+               F_HL_PREV(file) = prev_ndx;
+       }
+       alt_dest = F_HL_PREV(prev_file); /* alternate value when DONE && FIRST */
+       if (alt_dest >= 0 && dry_run) {
+               pathjoin(prev_name, MAXPATHLEN, basis_dir[alt_dest],
+                        f_name(prev_file, NULL));
+               f_name(prev_file, altbuf);
+               realname = altbuf;
+       } else {
+               f_name(prev_file, prev_name);
+               realname = prev_name;
+       }
+
+       if (link_stat(prev_name, &prev_st, 0) < 0) {
+               rsyserr(FERROR, errno, "stat %s failed",
+                       full_fname(prev_name));
+               return -1;
+       }
+
+       if (statret < 0 && basis_dir[0] != NULL) {
+               /* If we match an alt-dest item, we don't output this as a change. */
+               char cmpbuf[MAXPATHLEN];
+               STRUCT_STAT alt_st;
+               int j = 0;
+               do {
+                       pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
+                       if (link_stat(cmpbuf, &alt_st, 0) < 0)
+                               continue;
+                       if (link_dest) {
+                               if (prev_st.st_dev != alt_st.st_dev
+                                || prev_st.st_ino != alt_st.st_ino)
+                                       continue;
+                               statret = 1;
+                               *stp = alt_st;
+                               if (verbose < 2 || !stdout_format_has_i) {
+                                       itemizing = 0;
+                                       code = FNONE;
+                                       if (verbose > 1 && maybe_ATTRS_REPORT)
+                                               rprintf(FCLIENT, "%s is uptodate\n", fname);
+                               }
+                               break;
                        }
-                       maybe_hard_link(file, ndx, fname, statret, stp,
-                                       toname, &st2, itemizing, code);
-                       if (remove_source_files == 1 && do_xfers)
-                               send_msg_int(MSG_SUCCESS, ndx);
-                       hl->hlindex = FINISHED_LINK;
-               } else
-                       hl->hlindex = SKIPPED_LINK;
-               return 1;
+                       if (!unchanged_file(cmpbuf, file, &alt_st))
+                               continue;
+                       statret = 1;
+                       *stp = alt_st;
+                       if (unchanged_attrs(file, &alt_st))
+                               break;
+               } while (basis_dir[++j] != NULL);
        }
-       return 0;
+
+       if (maybe_hard_link(file, ndx, fname, statret, stp, prev_name, &prev_st,
+                           realname, itemizing, code) < 0)
+               return -1;
+
+       if (remove_source_files == 1 && do_xfers)
+               send_msg_int(MSG_SUCCESS, ndx);
+
+       return 1;
 }
 
-int hard_link_one(struct file_struct *file, int ndx, const char *fname,
-                 int statret, STRUCT_STAT *stp, const char *toname, int terse,
-                 int itemizing, enum logcode code)
+int hard_link_one(struct file_struct *file, const char *fname,
+                 const char *oldname, int terse)
 {
-       if (do_link(toname, fname)) {
+       if (do_link(oldname, fname) < 0) {
+               enum logcode code;
                if (terse) {
                        if (!verbose)
                                return -1;
@@ -260,54 +264,54 @@ int hard_link_one(struct file_struct *file, int ndx, const char *fname,
                } else
                        code = FERROR;
                rsyserr(code, errno, "link %s => %s failed",
-                       full_fname(fname), toname);
-               return -1;
+                       full_fname(fname), oldname);
+               return 0;
        }
 
-       if (itemizing) {
-               itemize(file, ndx, statret, stp,
-                       ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
-                       terse ? "" : toname);
-       }
-       if (code != FNONE && verbose && !terse)
-               rprintf(code, "%s => %s\n", fname, toname);
-       return 0;
+       file->flags |= FLAG_HLINK_DONE;
+
+       return 1;
 }
 
-void hard_link_cluster(struct file_struct *file, int master, int itemizing,
-                      enum logcode code, int dest_used)
+void finish_hard_link(struct file_struct *file, const char *fname,
+                     STRUCT_STAT *stp, int itemizing, enum logcode code,
+                     int alt_dest)
 {
-       char hlink1[MAXPATHLEN];
-       char *hlink2;
-       STRUCT_STAT st1, st2;
-       int statret, ndx = master;
-       struct hlist *hl = F_HL_LIST(file);
-
-       hl->hlindex = FINISHED_LINK;
-       if (dry_run)
-               hl->dest_used = dest_used + 1;
-       if (link_stat(f_name(file, hlink1), &st1, 0) < 0)
-               return;
-       if (!(file->flags & FLAG_HLINK_FIRST)) {
-               while (!(file->flags & FLAG_HLINK_LAST)) {
-                       ndx = hl->next;
-                       file = FPTR(ndx);
-                       hl = F_HL_LIST(file);
+       STRUCT_STAT st, prev_st;
+       char alt_name[MAXPATHLEN], *prev_name;
+       const char *our_name;
+       int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
+
+       if (stp == NULL && prev_ndx >= 0) {
+               if (link_stat(fname, &st, 0) < 0) {
+                       rsyserr(FERROR, errno, "stat %s failed",
+                               full_fname(fname));
+                       return;
                }
+               stp = &st;
        }
-       do {
-               ndx = hl->next;
+
+       /* FIRST combined with DONE means we were the first to get done. */
+       file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
+       F_HL_PREV(file) = alt_dest;
+       if (alt_dest >= 0 && dry_run) {
+               pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
+                        f_name(file, NULL));
+               our_name = alt_name;
+       } else
+               our_name = fname;
+
+       while ((ndx = prev_ndx) >= 0) {
                file = FPTR(ndx);
-               hl = F_HL_LIST(file);
-               if (hl->hlindex != SKIPPED_LINK)
+               file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
+               prev_ndx = F_HL_PREV(file);
+               prev_name = f_name(file, NULL);
+               prev_statret = link_stat(prev_name, &prev_st, 0);
+               if (maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_st,
+                                   our_name, stp, fname, itemizing, code) < 0)
                        continue;
-               hlink2 = f_name(file, NULL);
-               statret = link_stat(hlink2, &st2, 0);
-               maybe_hard_link(file, ndx, hlink2, statret, &st2,
-                               hlink1, &st1, itemizing, code);
                if (remove_source_files == 1 && do_xfers)
                        send_msg_int(MSG_SUCCESS, ndx);
-               hl->hlindex = FINISHED_LINK;
-       } while (!(file->flags & FLAG_HLINK_LAST));
+       }
 }
 #endif