kern/164516: [ext2fs] unable to mount EXT2 filesystem

[Kostik Belousov]

On Sun, Jan 29, 2012 at 12:56:52AM +0100, Uffe Jakobsen wrote:
> 
> 
> On 2012-01-28 23:31, Chris Rees wrote:
> >
> >On 28 Jan 2012 22:14, "Uffe Jakobsen" <uffe at uffe.org
> ><mailto:uffe at uffe.org>> wrote:
> > >
> > >
> > >
> > > On 2012-01-28 01:47, Jeremy Chadwick wrote:
> > >>
> > >> On Fri, Jan 27, 2012 at 11:10:11PM +0000, Uffe Jakobsen wrote:
> > >>>
> > >>> The following reply was made to PR kern/164516; it has been noted
> >by GNATS.
> > >>>
> > >>> From: Uffe Jakobsen<uffe at uffe.org <mailto:uffe at uffe.org>>
> > >>> To: bug-followup at FreeBSD.org
> > >>> Cc:
> > >>> Subject: Re: kern/164516: [ext2fs] unable to mount EXT2 filesystem
> > >>> Date: Fri, 27 Jan 2012 20:49:05 +0100
> > >>>
> > >>> >
> > >>> >  Not a bug. The file system type name is "ext2fs" not "ext2".
> > >>> >  Use "mount -t ext2fs /dev/md0 /mnt/tmp0" instead.
> > >>> >
> > >>>
> > >>>  Instead of reporting "Operation not supported by device"
> > >>>  Shouldn't it report unknown filesystem type ?
> > >>
> > >>
> > >> Please look through /usr/include/errno.h and let us know what error 
> > code
> > >> would represent "unknown filesystem type".  :-)
> > >>
> > >
> > > There are plenty of other places in the mount src where we exit
> >without a specific error code. Question is if there exists a scenario
> >where the requested filesystem type would/could not be found by
> >getvfsbyname() ?
> > >
> > > I've met this problem myself a number of times - and even in this
> >case I did not spot the spelling error right away ('ext2fs' and not 
> >'ext2').
> > >
> > > The returned error 'Operation not supported by device' - atleast to
> >me - indicates that the mount process got much further before running
> >into some kind of problem. The times when I've met this error I've begun
> >inspecting mount options etc before realizing that the fstype had a typo.
> > >
> >
> >Normally when an error comes up I check the manpage.
> 
> Manpage for mount does not at all mention any error conditions.
> 
> You'll have to know the inner workings of mount in order to know that in 
> certain cases everything is thrown into nmount() and any error that 
> comes up would come directly from nmount()
> 
> Anyway - would you accept a patch that checked the fstype using 
> getvfsbyname() before calling nmount() ?

Kernel tries to load the named module if it is not registered in the
list of know VFSes. Your proposal would break this functionality.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-fs/attachments/20120129/0f758d76/attachment.pgp