sh: support reflinking directories

- modify rm_util_link_type() to specifically detect directory pairs
- modify sh output generator to emit cp_reflink for directory pairs
- modify sh template to support reflinking directories and extend
  stampfile code to preserve attributes recursively

Fixes sahib#618.

Author: intelfx

lib/formats/sh.c.in

@@ -106,6 +106,7 @@ static bool rm_sh_emit_handler_clone(RmFmtHandlerShScript *self, char **out, RmF
     case RM_LINK_XDEV:
     case RM_LINK_SYMLINK:
     case RM_LINK_BOTH_EMPTY:
+    case RM_LINK_DIR:
         rm_log_warning_line("Unexpected return code %d from rm_util_link_type()", link_type);
         return FALSE;
     case RM_LINK_HARDLINK:
@@ -144,6 +145,7 @@ static bool rm_sh_emit_handler_reflink(RmFmtHandlerShScript *self, char **out, R
     case RM_LINK_BOTH_EMPTY:
         rm_log_warning_line("Unexpected return code %d from rm_util_link_type()", link_type);
         return FALSE;
+    case RM_LINK_DIR:
     case RM_LINK_HARDLINK:
     case RM_LINK_SYMLINK:
     case RM_LINK_NONE:

lib/formats/sh.sh

@@ -225,25 +225,34 @@ cp_hardlink() {
 }
 
 cp_reflink() {
-    if [ -d "$1" ]; then
-        # for duplicate dir's, can't clone so use symlink
-        cp_symlink "$@"
-        return $?
-    fi
     print_progress_prefix
     # reflink $1 to $2's data, preserving $1's  mtime
     printf "${COL_YELLOW}Reflinking to original: ${COL_RESET}%%s\n" "$1"
     if original_check "$1" "$2"; then
         if [ -z "$DO_DRY_RUN" ]; then
-            if [ -z "$STAMPFILE2" ]; then
+            if [ -d "$1" ]; then
+                local STAMPFILE2="$(mktemp -d "${TMPDIR:-/tmp}/rmlint.XXXXXXXX.stamp.d")"
+            elif [ -z "$STAMPFILE2" ]; then
                 STAMPFILE2=$(mktemp "${TMPDIR:-/tmp}/rmlint.XXXXXXXX.stamp")
             fi
             cp --archive --attributes-only --no-target-directory -- "$1" "$STAMPFILE2"
             if [ -d "$1" ]; then
+                # to reflink a directory, we will have to delete it, thus changing parent mtime
+                # take care of preserving parent mtime if requested
+                if [ -n "$DO_KEEP_DIR_TIMESTAMPS" ]; then
+                    touch -r "$(dirname "$1")" -- "$STAMPFILE"
+                fi
                 rm -rf -- "$1"
             fi
             cp --archive --reflink=always -- "$2" "$1"
             cp --archive --attributes-only --no-target-directory -- "$STAMPFILE2" "$1"
+            if [ -d "$1" ]; then
+                rm -rf -- "$STAMPFILE2"
+                if [ -n "$DO_KEEP_DIR_TIMESTAMPS" ]; then
+                    # restore parent mtime if we saved it
+                    touch -r "$STAMPFILE" -- "$(dirname "$1")"
+                fi
+            fi
         fi
     fi
 }

lib/reflink.c

@@ -468,6 +468,7 @@ int rm_is_reflink_main(int argc, const char **argv) {
         "     %i:  %s\n"
         "     %i:  %s\n"
         "     %i:  %s\n"
+        "     %i:  %s\n"
         "     %i:  %s\n",
         _("Test if two files are reflinks (share same data extents)"),
         _("Returns 0 if the files are reflinks."),
@@ -481,7 +482,8 @@ int rm_is_reflink_main(int argc, const char **argv) {
         RM_LINK_HARDLINK, desc[RM_LINK_HARDLINK],
         RM_LINK_SYMLINK, desc[RM_LINK_SYMLINK],
         RM_LINK_XDEV, desc[RM_LINK_XDEV],
-        RM_LINK_NONE, desc[RM_LINK_NONE]);
+        RM_LINK_NONE, desc[RM_LINK_NONE],
+        RM_LINK_DIR, desc[RM_LINK_DIR]);
 
 
     g_option_context_set_summary(context, summary);

lib/utilities.c

@@ -1384,7 +1384,7 @@ RmLinkType rm_util_link_type(const char *path1, const char *path2, bool use_fiem
         RM_RETURN(RM_LINK_ERROR);
     }
 
-    if(!S_ISREG(stat1.st_mode)) {
+    if(!S_ISREG(stat1.st_mode) && !S_ISDIR(stat1.st_mode)) {
         RM_RETURN(S_ISLNK(stat1.st_mode) ? RM_LINK_SYMLINK : RM_LINK_NOT_FILE);
     }
 
@@ -1409,11 +1409,23 @@ RmLinkType rm_util_link_type(const char *path1, const char *path2, bool use_fiem
         RM_RETURN(RM_LINK_ERROR);
     }
 
-    if(!S_ISREG(stat2.st_mode)) {
+    if(!S_ISREG(stat2.st_mode) && !S_ISDIR(stat2.st_mode)) {
         RM_RETURN(S_ISLNK(stat2.st_mode) ? RM_LINK_SYMLINK : RM_LINK_NOT_FILE);
     }
 
-    if(stat1.st_size != stat2.st_size) {
+    /* At this point, path1 or path2 may be a regular file or a directory.
+     * Ensure they both have the same type, otherwise fail. */
+    bool is_dir;
+    if(S_ISDIR(stat1.st_mode) && S_ISDIR(stat2.st_mode)) {
+        is_dir = true;
+    } else if (S_ISREG(stat1.st_mode) && S_ISREG(stat2.st_mode)) {
+        is_dir = false;
+    } else {
+        RM_RETURN(RM_LINK_NOT_FILE);
+    }
+
+    if(!is_dir && stat1.st_size != stat2.st_size) {
+        /* st_size is not defined for directories */
 #if _RM_OFFSET_DEBUG
         rm_log_debug_line(
             "rm_util_link_type: Files have different sizes: %" G_GUINT64_FORMAT
@@ -1446,6 +1458,11 @@ RmLinkType rm_util_link_type(const char *path1, const char *path2, bool use_fiem
         }
     }
 
+    if (is_dir) {
+        /* further tests do not make sense for directories */
+        RM_RETURN(RM_LINK_DIR);
+    }
+
     if(use_fiemap) {
         RmLinkType reflink_type = rm_reflink_type_from_fd(fd1, fd2);
         RM_RETURN(reflink_type);
@@ -1468,7 +1485,8 @@ const char **rm_link_type_to_desc() {
                                                  N_("Encountered a symlink"),
                                                  N_("Files are on different devices"),
                                                  N_("Not linked"),
-                                                 N_("Both files are empty")};
+                                                 N_("Both files are empty"),
+                                                 N_("Both files are directories")};
     return RM_LINK_TYPE_TO_DESC;
 }
 

lib/utilities.h

@@ -52,6 +52,7 @@ typedef enum RmLinkType {
     RM_LINK_XDEV            = 10,
     RM_LINK_NONE            = 11,
     RM_LINK_BOTH_EMPTY      = 12,
+    RM_LINK_DIR             = 13,
 } RmLinkType;
 
 #if HAVE_STAT64 && !RM_IS_APPLE