Re: Softupdates a mount option?

From: Ivan Voras (
Date: 05/27/04

  • Next message: Drew Broadley: "ndis: build error on make buildkernel + FIX"
    Date: Thu, 27 May 2004 00:27:38 +0200
    To: Wilko Bulte <>

    Wilko Bulte wrote:

    > On Wed, May 26, 2004 at 09:15:20PM +0200, Ivan Voras wrote:
    >>This has been really nagging me for a long time: Why aren't softupdates
    >>made a mount option (like 'sync' and 'async')? Do I remember correctly
    >>that it is done so in NetBSD (where it's called softdeps), so it's doable?
    > Its been discussed before, so I suggest to search the mailing list
    > archives (I really don't remember all the arguments used btw ;)

    Thanks for all the replies - I read some (most?) of the posts, and would
    like to add my reasons for it (in case they were missed in previous
    discussions :) ):

    - I was creating a md drive with mdmfs, and it felt rather awkward to
    control softupdates via command line parameters (a sidequestion: does it
    make any sense enabling SU on a memory drive by default?). As it seems
    now, every such utility that handles (well, at least creates) a ffs
    filesystem must handle SU-controlling options as command line parameters.

    - It seems to me that SU is just another "mode" of accessing a
    filesystem, just like sync and async, so it belongs with those two.

    If fsck needs to be aware if the filesystem was mounted with SU (why?
    only to cut down on repair time?), maybe the softupdates flag (as used
    by tunefs) could be treated as 'last mounted as' information.

    I am not an expert, these are just my wishes :)

    C isn't that hard: void (*(*f[])())() defines f as an array of
    unspecified size, of pointers to functions that return pointers to
    functions that return void.

  • Next message: Drew Broadley: "ndis: build error on make buildkernel + FIX"