New option --clamp-mtime

The new `--clamp-mtime` option will change the behavior of `--mtime` to only
use the time specified if the file mtime is newer than the given time.
The `--clamp-mtime` option can only be used together with `--mtime`.

Typical use case is to make builds reproducible: to loose less
information, it's better to keep the original date of an archive, except for
files modified during the build process. In that case, using a reference
(and thus reproducible) timestamps for the latter is good enough. See
<https://wiki.debian.org/ReproducibleBuilds> for more information.

Patch submitted by Jeremy Bobbio and
Daniel Kahn Gillmor <dkg@fifthhorseman.net>

* doc/tar.1: Document --clamp-mtime
* doc/tar.texi: Likewise.

* src/common.h (set_mtime_option_mode): New enum
(set_mtime_option): Change type to enum set_mtime_option_mode.
(NEWER_OPTION_INITIALIZED): Rename to NEWER_OPTION_INITIALIZED.
* src/create.c (start_header): Set mtime depending on set_mtime_option.
* src/tar.c (options,parse_opt): New option --clamp-mtime
(decode_options): Initialize mtime_option

* tests/time02.at: New testcase.
* tests/Makefile.am: Add new testcase
* tests/testsuite.at: Likewise.

src/common.h

@@ -211,13 +211,20 @@ GLOBAL bool multi_volume_option;
    do not get archived (also see after_date_option above).  */
 GLOBAL struct timespec newer_mtime_option;
 
-/* If true, override actual mtime (see below) */
-GLOBAL bool set_mtime_option;
-/* Value to be put in mtime header field instead of the actual mtime */
+enum set_mtime_option_mode
+{
+  USE_FILE_MTIME,
+  FORCE_MTIME,
+  CLAMP_MTIME,
+};
+
+/* Override actual mtime if set to FORCE_MTIME or CLAMP_MTIME */
+GLOBAL enum set_mtime_option_mode set_mtime_option;
+/* Value to use when forcing or clamping the mtime header field. */
 GLOBAL struct timespec mtime_option;
 
-/* Return true if newer_mtime_option is initialized.  */
-#define NEWER_OPTION_INITIALIZED(opt) (0 <= (opt).tv_nsec)
+/* Return true if mtime_option or newer_mtime_option is initialized.  */
+#define TIME_OPTION_INITIALIZED(opt) (0 <= (opt).tv_nsec)
 
 /* Return true if the struct stat ST's M time is less than
    newer_mtime_option.  */

src/create.c

@@ -823,7 +823,24 @@ start_header (struct tar_stat_info *st)
   }
 
   {
-    struct timespec mtime = set_mtime_option ? mtime_option : st->mtime;
+    struct timespec mtime;
+
+    switch (set_mtime_option)
+      {
+      case USE_FILE_MTIME:
+	mtime = st->mtime;
+	break;
+	  
+      case FORCE_MTIME:
+	mtime = mtime_option;
+	break;
+	  
+      case CLAMP_MTIME:
+	mtime = timespec_cmp (st->mtime, mtime_option) > 0
+	           ? mtime_option : st->mtime;
+	break;
+      }
+
     if (archive_format == POSIX_FORMAT)
       {
 	if (MAX_OCTAL_VAL (header->header.mtime) < mtime.tv_sec

src/list.c

@@ -195,7 +195,7 @@ read_and (void (*do_something) (void))
 	  decode_header (current_header, &current_stat_info,
 			 &current_format, 1);
 	  if (! name_match (current_stat_info.file_name)
-	      || (NEWER_OPTION_INITIALIZED (newer_mtime_option)
+	      || (TIME_OPTION_INITIALIZED (newer_mtime_option)
 		  /* FIXME: We get mtime now, and again later; this causes
 		     duplicate diagnostics if header.mtime is bogus.  */
 		  && ((mtime.tv_sec

src/tar.c

@@ -276,6 +276,7 @@ enum
   CHECK_DEVICE_OPTION,
   CHECKPOINT_OPTION,
   CHECKPOINT_ACTION_OPTION,
+  CLAMP_MTIME_OPTION,
   DELAY_DIRECTORY_RESTORE_OPTION,
   HARD_DEREFERENCE_OPTION,
   DELETE_OPTION,
@@ -514,6 +515,8 @@ static struct argp_option options[] = {
    N_("use FILE to map file owner GIDs and names"), GRID+1 },
   {"mtime", MTIME_OPTION, N_("DATE-OR-FILE"), 0,
    N_("set mtime for added files from DATE-OR-FILE"), GRID+1 },
+  {"clamp-mtime", CLAMP_MTIME_OPTION, 0, 0,
+   N_("only set time when the file is more recent than what was given with --mtime"), GRID+1 },
   {"mode", MODE_OPTION, N_("CHANGES"), 0,
    N_("force (symbolic) mode CHANGES for added files"), GRID+1 },
   {"atime-preserve", ATIME_PRESERVE_OPTION,
@@ -1364,6 +1367,10 @@ parse_opt (int key, char *arg, struct argp_state *state)
       set_subcommand_option (CREATE_SUBCOMMAND);
       break;
 
+    case CLAMP_MTIME_OPTION:
+      set_mtime_option = CLAMP_MTIME;
+      break;
+
     case 'd':
       set_subcommand_option (DIFF_SUBCOMMAND);
       break;
@@ -1505,7 +1512,8 @@ parse_opt (int key, char *arg, struct argp_state *state)
 
     case MTIME_OPTION:
       get_date_or_file (args, "--mtime", arg, &mtime_option);
-      set_mtime_option = true;
+      if (set_mtime_option == USE_FILE_MTIME)
+        set_mtime_option = FORCE_MTIME;
       break;
 
     case 'n':
@@ -1521,7 +1529,7 @@ parse_opt (int key, char *arg, struct argp_state *state)
       /* Fall through.  */
 
     case NEWER_MTIME_OPTION:
-      if (NEWER_OPTION_INITIALIZED (newer_mtime_option))
+      if (TIME_OPTION_INITIALIZED (newer_mtime_option))
 	USAGE_ERROR ((0, 0, _("More than one threshold date")));
       get_date_or_file (args,
 			key == NEWER_MTIME_OPTION ? "--newer-mtime"
@@ -2236,6 +2244,8 @@ decode_options (int argc, char **argv)
 
   newer_mtime_option.tv_sec = TYPE_MINIMUM (time_t);
   newer_mtime_option.tv_nsec = -1;
+  mtime_option.tv_sec = TYPE_MINIMUM (time_t);
+  mtime_option.tv_nsec = -1;
   recursion_option = FNM_LEADING_DIR;
   unquote_option = true;
   tar_sparse_major = 1;
@@ -2397,7 +2407,7 @@ decode_options (int argc, char **argv)
 		  _("Multiple archive files require '-M' option")));
 
   if (listed_incremental_option
-      && NEWER_OPTION_INITIALIZED (newer_mtime_option))
+      && TIME_OPTION_INITIALIZED (newer_mtime_option))
     {
       struct option_locus *listed_loc = optloc_lookup (OC_LISTED_INCREMENTAL);
       struct option_locus *newer_loc = optloc_lookup (OC_NEWER);
@@ -2464,6 +2474,13 @@ decode_options (int argc, char **argv)
 	USAGE_ERROR ((0, 0, _("Cannot concatenate compressed archives")));
     }
 
+  if (set_mtime_option == CLAMP_MTIME)
+    {
+      if (!TIME_OPTION_INITIALIZED (mtime_option))
+	USAGE_ERROR ((0, 0,
+		      _("--clamp-mtime needs a date specified using --mtime")));
+    }
+
   /* It is no harm to use --pax-option on non-pax archives in archive
      reading mode. It may even be useful, since it allows to override
      file attributes from tar headers. Therefore I allow such usage.