TODO

   1 login-utils:
   2 -----------
   3
   4  - clean up error codes
   5  - use err() and warn() macros rather than fprintf(stderr, ...)
   6
   7
   8 losetup
   9 -------
  10
  11  - don't use ioclts if sysfs supports loop attributes (since 2.6.37,
  12    block/loopN/loop/* files)
  13
  14     - don't open loopdev in looplist_next(), check if /sys/block/loopN/loop
  15       exists and returns ll.name rather than file descriptor
  16
  17     - add sysfs based version of show_loop() and use it for non-root users
  18
  19 sysfs
  20 -----
  21
  22  - add lib/sysfs.c with functions that read block device attributes from /sys.
  23    For more details see devno.c from libblkid, lsblk.c and lomount.c.
  24
  25
  26 libblkid
  27 --------
  28
  29  - remove strerrr() from debug messages (use %m) to make BLKID_DEBUG= output
  30    thread-safe
  31
  32  - add -<BE|LE> suffix to test images for native-endian filesystems (e.g. swap)
  33    and add support for such functionality to tests/ts/blkid/low-probe
  34
  35  - add FSSIZE value  -- filesystem size (klibc requirement)
  36
  37  - add support for dasd PT (used for example on s390)
  38
  39
  40 blkid(8)
  41 -------
  42
  43   - add something like "blkid --list-known" to list all supported filesystems/raids
  44
  45   - add a new option (-r) that allows to print removable block devices that
  46     are missing in /proc/partitions, see blkid_probe_all_removable().
  47
  48 agetty:
  49 ------
  50
  51   - review code and make it more simple for normal console (see mingetty)
  52   - merge functionality from unmaintained mingetty package to agetty
  53
  54 wipefs
  55 ------
  56
  57  - some filesystem (namely FAT) contains more magic strings, so if you erase
  58    one magic string the FS is still detectable by libblkid. We have to inform
  59    users that there is more valid magic string for the same FS.
  60
  61 fdisk(s)
  62 --------
  63
  64  * sfdisk rounds to cylinders is -uM (megabyte units) is specified, this is
  65    pretty stupid feature. It has to round to sectors if -uS or -uM is specified.
  66
  67  * Sun label support is completely useless for large disks, it uses number of
  68    cylinders from on-disk-label where the geometry is stored by int16 values.
  69    It seems better to completely ignore this stuff from the label and always
  70    use geometry + BLKGETSIZE64 from kernel.
  71
  72  * use off_t instead "long long"
  73
  74  * catch SIGINT (Ctrl-C) and return to main menu.
  75    From Red Hat bugzilla #545488:
  76
  77    While using fdisk normally, if you accidentally pressed the wrong button (to
  78    start a sequence of questions for some operation, e.g. 'c' to create
  79    partition).  The tool tries too hard to keep asking you for valid input.  You
  80    can't provide a blank or invalid input to get it to break out of the current
  81    dialog sequence and get back to the main menu.
  82
  83  * fdisk/* refactoring
  84
  85  * add GPT support (probably implement libfdisk)
  86
  87 misc
  88 ----
  89
  90  * switch_root:
  91      - move all mountpoints to the newroot (there are hardcoded /proc /sys and /dev paths now)
  92      - add --dont-move[=<list of dirs>] options
  93
  94  * use ngettext() for strings with plurals, for example
  95
  96     /* include/nls.h */
  97     #define P_(id, id_plural, n)        ngettext(id, id_plural, n)
  98
  99     printf(P_("%d used sector",
 100               "%d used sectors", sectors),
 101               sectors);
 102
 103  * use something better than gtk-doc (doxyden?)
 104
 105  * use TZ=UTC and LANG=en_US.UTF-8 for tests
 106
 107  * add NLS and err.h stuff to schedutils (chrt.c, taskset.c)
 108
 109  * add mllockall() and SCHED_FIFO to hwclock,
 110    see http://lkml.org/lkml/2008/10/12/132
 111
 112  * use rpmatch() for all Y/N questions
 113
 114  * make chrt, taskset, ... threads aware. Currently these utils work with
 115    group leader (master thread) and don't propagate requested changes to
 116    the whole group of threads.  It meas add a new option (e.g. -t):
 117
 118         chrt <prio> <pid>     # master thread only
 119         chrt -t <prio> <pid>  # whole group of threads
 120
 121   Note that we need to scan /proc and call ched_set{scheduler,affinity}
 122   syscalls for all individual threads.
 123
 124  * umount by label:
 125         # mount LABEL=mylabel
 126         # umount LABEL=mylabel
 127
 128  * mount -a -- reorder fstab entries by paths before mount (just idea only)
 129
 130  * mount -a  (just idea only)
 131
 132         Date:   Sun, 3 Jun 2007 18:04:24 +0300 (MET DST)
 133         From: Szabolcs Szakacsits <szaka@sienet.hu>
 134
 135         In the past the right record order could be figured out easily by just
 136         checking out fstab (if one knew what to look for) but considering the
 137         fastly increasing number of user space file systems and their usage, with
 138         their path, library, etc dependencies, it's getting trickier and is a black
 139         magic for most users because they simply expect drives to be mounted
 140         independently of their order in fstab.
 141
 142         One typical, wrongly edited fstab example is:
 143
 144           /dev/hda2 /             ext3      defaults  1 1
 145           /dev/hda1 /mnt/windows  ntfs-3g   defaults  0 0
 146           /dev/hda3 /usr          ext3      defaults  0 0
 147
 148         The events:
 149
 150           mount  ->  /sbin/mount.ntfs-3g  ->
 151                  ->  resolves to <path1>/ntfs-3g via a symlink ->
 152                  ->  ntfs-3g requires at least <path2>/libfuse*
 153
 154         There are many potential solutions. For example installing everything on
 155         the root file system which may be needed for successful mount. But this
 156         is not always feasible or practical since we could end up putting almost
 157         everything on the root file system in the end.
 158
 159         Another idea is an improved mount strategy:
 160
 161           do {
 162                 try to mount all unmounted entries
 163           } while (not all mounted && at least one new was successfully mounted)
 164
 165  * rewrite ipcs to use /proc/sys/kernel rather than unreliable syscalls
 166    (there are problems with 32bit userspace on 64bit kernel)
 167
 168  * minix v3
 169
 170         From: Matthias Koenig <mkoenig@suse.de>
 171         Date:   Tue, 25 Sep 2007 12:00:01 +0200
 172
 173         It seems that the kernel has support for minix fs v3
 174         (though I have not tried it, just inspected some code when
 175         trying to find a mkfs.minix issue).
 176         It might be worth a thought implementing v3 support
 177         (though I am not really sure how much people us minix fs ;-)
 178         This might require some major code cleanup in mkfs.minix.
 179
 180  * add SELinux security contexts support to the 'ipcs' utility
 181    http://bugzilla.redhat.com/show_bug.cgi?id=225342
 182
 183    Would be great to list the current system IPC Objects with their respective
 184    security labels (where allowed) with something like 'ipcs -Z' - following the
 185    way other tools reports those.
 186
 187