mount and linux

mount and linux

[Chmouel Boudjnah]

Hi,

Looking at the _mount function i see :

    fss=( minix ext ext2 xiafs hpfs msdos umsdos vfat proc nfs iso9660
          smbfs ncpfs affs ufs romfs sysv adfs autofs coda devpts efs
          hfs ntfs qnx4 smbfs udf ext3 xfs reiserfs )
    [[ -r /proc/filesystems ]] &&
        fss=( ${$(</proc/filesystems)#nodev} )


if i undertstand correctly it purpose fs according of what he found in
/proc/filesystems, 

This is wrong in case of fully modular system (75% of users) since it
only register to /proc/filesystems at module insertion. 

The best way would be to combine that with /etc/filesystems..

RE: mount and linux

[Borsenkow Andrej]

> 
> Looking at the _mount function i see :
> 
>     fss=( minix ext ext2 xiafs hpfs msdos umsdos vfat proc nfs iso9660
>           smbfs ncpfs affs ufs romfs sysv adfs autofs coda devpts efs
>           hfs ntfs qnx4 smbfs udf ext3 xfs reiserfs )
>     [[ -r /proc/filesystems ]] &&
>         fss=( ${$(</proc/filesystems)#nodev} )
> 
> 
> if i undertstand correctly it purpose fs according of what he found in
> /proc/filesystems,
> 
> This is wrong in case of fully modular system (75% of users) since it
> only register to /proc/filesystems at module insertion.
> 
> The best way would be to combine that with /etc/filesystems..

On up-to-date cooker:

[root@cooker root]# cat /etc/filesystems
ext2
nodev proc
nodev devpts
iso9660
reiserfs
vfat

not overly usesful (ext3, xfs, supermount, jfs, nfs ...).

Re: mount and linux

[Oliver Kiddle]

Borsenkow Andrej wrote:
> 
> > This is wrong in case of fully modular system (75% of users) since it
> > only register to /proc/filesystems at module insertion.
> >
> > The best way would be to combine that with /etc/filesystems..
> 
> On up-to-date cooker:
> 
> [root@cooker root]# cat /etc/filesystems
> ext2
> nodev proc
> nodev devpts
> iso9660
> reiserfs
> vfat
> 
> not overly usesful (ext3, xfs, supermount, jfs, nfs ...).

On slightly out-of-date Debian /etc/filesystems doesn't exist at all. It
isn't Mandrake specific is it? What is the purpose of it and where does
it come from? Certainly it is a fair point about /proc/filesystems
limitations though so if /etc/filesystems allows us to correctly add
more supported filesystems, it should be used.

_____________________________________________________________________
This message has been checked for all known viruses by the 
MessageLabs Virus Scanning Service. For further information visit
http://www.messagelabs.com/stats.asp

Re: mount and linux

[Clint Adams]

> On slightly out-of-date Debian /etc/filesystems doesn't exist at all. It
> isn't Mandrake specific is it? What is the purpose of it and where does
> it come from? Certainly it is a fair point about /proc/filesystems
> limitations though so if /etc/filesystems allows us to correctly add
> more supported filesystems, it should be used.

It's used by the mount in util-linux; I'm not aware of who uses
that mount and who doesn't.  It's used in preference to
/proc/filesystems for auto-probing.  As such, I have absolutely
no use for it.

I'm starting to think that this would be better served
by a style.

RE: mount and linux

[Borsenkow Andrej]

> >
> > > This is wrong in case of fully modular system (75% of users) since
it
> > > only register to /proc/filesystems at module insertion.
> > >
> > > The best way would be to combine that with /etc/filesystems..
> >
> > On up-to-date cooker:
> >
> > [root@cooker root]# cat /etc/filesystems
> > ext2
> > nodev proc
> > nodev devpts
> > iso9660
> > reiserfs
> > vfat
> >
> > not overly usesful (ext3, xfs, supermount, jfs, nfs ...).
> 
> On slightly out-of-date Debian /etc/filesystems doesn't exist at all.
It
> isn't Mandrake specific is it? What is the purpose of it and where
does
> it come from? Certainly it is a fair point about /proc/filesystems
> limitations though so if /etc/filesystems allows us to correctly add
> more supported filesystems, it should be used.
> 

man mount:
              The type iso9660 is the default.  If no  -t  option
              is  given,  or  if  the auto type is specified, the
              superblock  is  probed  for  the  filesystem   type
              (minix,  ext, ext2, ext3, xiafs, iso9660, reiserfs,
              romfs, ufs, ntfs,  qnx4,  bfs,  xfs,  cramfs,  hfs,
              hpfs,  adfs,  vxfs  are  supported).  If this probe
              fails,  mount   will   try   to   read   the   file
              /etc/filesystems,  or,  if  that  does  not  exist,
              /proc/filesystems.  All  of  the  filesystem  types
              listed  there  will be tried, except for those that
              are labeled "nodev" (e.g., devpts, proc and nfs).

[root@cooker root]# rpm -q mount
mount-2.11j-2mdk

of course, man page for mount is in different RPM so it may apply to
totally different command altogether :-)

-andrej

Re: mount and linux

[Chmouel Boudjnah]

Borsenkow Andrej <Andrej.Borsenkow@mow.siemens.ru> writes:

> > This is wrong in case of fully modular system (75% of users) since it
> > only register to /proc/filesystems at module insertion.
> > 
> > The best way would be to combine that with /etc/filesystems..
> 
> On up-to-date cooker:
> 
> [root@cooker root]# cat /etc/filesystems
> ext2
> nodev proc
> nodev devpts
> iso9660
> reiserfs
> vfat
> 
> not overly usesful (ext3, xfs, supermount, jfs, nfs ...).

yes but that the only file that give you the ability to force the auto
type, or at least make it that by default is purposing (from
util-linux/mount/mount_guess_fstype.c) :

static char
*magic_known[] = {
	"adfs", "bfs", "cramfs", "ext", "ext2", "ext3",
	"hfs", "hpfs", "iso9660", "minix", "ntfs",
	"qnx4", "reiserfs", "romfs", "swap", "udf", "ufs",
	"vxfs", "xfs", "xiafs"
};

since they are auto detected by magic number....

Re: mount and linux

[Clint Adams]

> static char
> *magic_known[] = {
> 	"adfs", "bfs", "cramfs", "ext", "ext2", "ext3",
> 	"hfs", "hpfs", "iso9660", "minix", "ntfs",
> 	"qnx4", "reiserfs", "romfs", "swap", "udf", "ufs",
> 	"vxfs", "xfs", "xiafs"
> };
> 
> since they are auto detected by magic number....

It might be nicer to use the original array, /proc/filesystems,
and /etc/filesystems all with different tags.

Index: Completion/Unix/Command/_mount
===================================================================
RCS file: /cvsroot/zsh/zsh/Completion/Unix/Command/_mount,v
retrieving revision 1.10
diff -u -r1.10 _mount
--- Completion/Unix/Command/_mount	2001/10/08 15:08:02	1.10
+++ Completion/Unix/Command/_mount	2001/10/10 19:23:23
@@ -559,11 +559,13 @@
       ':dev or dir:->devordir'
       ':mount point:_files -/'
     )
-    fss=( minix ext ext2 xiafs hpfs msdos umsdos vfat proc nfs iso9660
-          smbfs ncpfs affs ufs romfs sysv adfs autofs coda devpts efs
-          hfs ntfs qnx4 smbfs udf ext3 xfs reiserfs )
+    typeset -aU fss
+    fss=( adfs bfs cramfs ext2 ext3 hfs hpfs iso9660 minix ntfs qnx4
+          reiserfs romfs swap udf ufs vxfs xfs xiafs )
     [[ -r /proc/filesystems ]] &&
-        fss=( ${$(</proc/filesystems)#nodev} )
+        fss=( $fss ${$(</proc/filesystems)#nodev} )
+    [[ -r /etc/filesystems ]] &&
+        fss=( $fss ${$(</etc/filesystems)#nodev} )
     ;;
   osf*)
     args=( -s