Chris Adams <[email protected]> wrote:
> rc.sysinit calls /sbin/fsck, which comes from e2fsprogs.

Yes, it comes from e2fsprogs.  That doesn't mean it has issues with
kicking off other programs.

> It has a compiled-in list of recognized filesystems, and
> ext4/ext4dev is not on the list.  Anything not on the list is skipped.

Yeah, I hit the code after the prior correspondence.  This does, indeed,
seem to be the case in even the latest RHEL5 package release.  From
e2fsprogs-1.39/misc ...

  ...
  static const char *really_wanted[] = {
    "minix",
    "ext2",
    "ext3",
    "jfs",
    "reiserfs",
    "xiafs",
    "xfs",
    NULL
  };
  ...
  /* Do we really really want to check this fs?
  for(ip = really_wanted; *ip; ip++)
    if (strcmp(fs->type, *ip) == 0) {
      wanted = 1;
      break;
    }
  /* See if the <fsck.fs> program is available. */
  if (find_fsck(fs->type) == NULL) {
    if (wanted)
      fprintf(stderr, _("fsck: cannot check %s: fsck.%s not found\n"),
                                fs->device, fs->type);
      return 1;
    }
  ...

> It would be better for /sbin/fsck to try to call "fsck.foo"
> when it encounters filesystem type "foo", rather than have a
> strict list (which can always get out of date).

Actually, it's not a "strict list," but a "wanted" list.  It
still checks for an actual fsck.(fstype).

I was thinking through this and decided to hit the code.  Then
it hit me.  There are filesystems we probably don't want to check
by default, including msdos/vfat, on system init (among others,
over the history of Linux).

> Newer Fedora releases have moved fsck to util-linux-ng, but it
> still has a hard-coded list of "wanted" filesystems.

In any case, this should be filed as a bug.  It is always helpful
to file such things when they are found.  Customers who have higher
levels of SLAs will get priority as well.

Although this is a serious oversight at first glance.  Thank you
for noticing it, and the maintainer may either be able to explain
the current logic (I'm just showing a snippet) or confirm the logic
is incorrect for the Ext4 Tech Preview in 5.4+.


_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list

Reply via email to