All of lore.kernel.org
 help / color / mirror / Atom feed
* jfs, special characters
@ 2005-02-26  4:34 Frederik Eaton
  2005-02-26 16:51 ` Sonny Rao
  2005-02-26 17:58 ` Dave Kleikamp
  0 siblings, 2 replies; 6+ messages in thread
From: Frederik Eaton @ 2005-02-26  4:34 UTC (permalink / raw)
  To: linux-fsdevel

Does JFS have problems with files with special characters in them?

$ ls
ls: 04-The_Dark_Of_The_Matin?e.ogg: No such file or directory
[1]$ rm 04-The_Dark_Of_The_Matin\?e.ogg
rm: cannot lstat `04-The_Dark_Of_The_Matin?e.ogg': No such file or directory

Frederik


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: jfs, special characters
  2005-02-26  4:34 jfs, special characters Frederik Eaton
@ 2005-02-26 16:51 ` Sonny Rao
  2005-02-26 17:58 ` Dave Kleikamp
  1 sibling, 0 replies; 6+ messages in thread
From: Sonny Rao @ 2005-02-26 16:51 UTC (permalink / raw)
  To: frederik; +Cc: linux-fsdevel

On Fri, Feb 25, 2005 at 08:34:39PM -0800, Frederik Eaton wrote:
> Does JFS have problems with files with special characters in them?
> 
> $ ls
> ls: 04-The_Dark_Of_The_Matin?e.ogg: No such file or directory
> [1]$ rm 04-The_Dark_Of_The_Matin\?e.ogg
> rm: cannot lstat `04-The_Dark_Of_The_Matin?e.ogg': No such file or directory
> 

Probably, you should ask this question on the jfs-discussion mailing
list. 

Sonny

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: jfs, special characters
  2005-02-26  4:34 jfs, special characters Frederik Eaton
  2005-02-26 16:51 ` Sonny Rao
@ 2005-02-26 17:58 ` Dave Kleikamp
  2005-02-27 23:00   ` Frederik Eaton
  1 sibling, 1 reply; 6+ messages in thread
From: Dave Kleikamp @ 2005-02-26 17:58 UTC (permalink / raw)
  To: frederik; +Cc: fsdevel

On Fri, 2005-02-25 at 20:34 -0800, Frederik Eaton wrote:
> Does JFS have problems with files with special characters in them?

Sometimes.  jfs attempts to store pathnames in 16-bit unicode, and uses
the mount option iocharset to determine what to convert to/from.

In the 2.4 kernel, the default value of iocharset is determined by the
kernel config option, CONFIG_NLS_DEFAULT.  If an existing pathname
exists that doesn't map to the iocharset, jfs has the problem you
describe.  Mounting with -o iocharset=utf8 should let you access any
existing files.

In the 2.6 kernel, the default was changed to store each byte of the
filename as a 16-bit value in the directory without using any
conversion.  There can still be a problem if an existing pathname has a
value with a non-zero high-order byte.  Again mounting with -o
iocharset=utf8 will let you access all files.

> $ ls
> ls: 04-The_Dark_Of_The_Matin?e.ogg: No such file or directory
> [1]$ rm 04-The_Dark_Of_The_Matin\?e.ogg
> rm: cannot lstat `04-The_Dark_Of_The_Matin?e.ogg': No such file or directory
> 
> Frederik

-- 
David Kleikamp
IBM Linux Technology Center


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: jfs, special characters
  2005-02-26 17:58 ` Dave Kleikamp
@ 2005-02-27 23:00   ` Frederik Eaton
  2005-02-28  4:47     ` Dave Kleikamp
  0 siblings, 1 reply; 6+ messages in thread
From: Frederik Eaton @ 2005-02-27 23:00 UTC (permalink / raw)
  To: Dave Kleikamp; +Cc: fsdevel

On Sat, Feb 26, 2005 at 11:58:34AM -0600, Dave Kleikamp wrote:
> On Fri, 2005-02-25 at 20:34 -0800, Frederik Eaton wrote:
> > Does JFS have problems with files with special characters in them?
> 
> Sometimes.  jfs attempts to store pathnames in 16-bit unicode, and uses
> the mount option iocharset to determine what to convert to/from.
> 
> In the 2.4 kernel, the default value of iocharset is determined by the
> kernel config option, CONFIG_NLS_DEFAULT.  If an existing pathname
> exists that doesn't map to the iocharset, jfs has the problem you
> describe.  Mounting with -o iocharset=utf8 should let you access any
> existing files.
> 
> In the 2.6 kernel, the default was changed to store each byte of the
> filename as a 16-bit value in the directory without using any
> conversion.  There can still be a problem if an existing pathname has a
> value with a non-zero high-order byte.  Again mounting with -o
> iocharset=utf8 will let you access all files.
> 
> > $ ls
> > ls: 04-The_Dark_Of_The_Matin?e.ogg: No such file or directory
> > [1]$ rm 04-The_Dark_Of_The_Matin\?e.ogg
> > rm: cannot lstat `04-The_Dark_Of_The_Matin?e.ogg': No such file or directory

Thanks, you were right that I recently upgraded to 2.6 (actually I
transferred the hard drive between machines), and with a "mount /home -o
remount,iocharset=utf8" at least I can access the files now. But the
names aren't what they should be, for instance I see:

04-The_Dark_Of_The_MatinΘe.ogg

(that's a theta). Were they supposed to be correct? 

As an aside, now, I just tried "mount /home -o
remount,iocharset=iso8859-1", and got same problem

ls: '/home/shared/frederik/Franz_Ferdinand/Franz_Ferdinand/04-The_Dark_Of_The_Matin?e.ogg': No such file or directory

But when I repeated the "mount /home -o remount,iocharset=utf8"
command, the listing didn't show the name with a theta character, but
instead with two question marks:

04-The_Dark_Of_The_Matin??e.ogg

Strange behavior. I.e., upon changing the mount options and then
changing them back I get a different listing.

Frederik
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: jfs, special characters
  2005-02-27 23:00   ` Frederik Eaton
@ 2005-02-28  4:47     ` Dave Kleikamp
  2005-02-28  5:42       ` Frederik Eaton
  0 siblings, 1 reply; 6+ messages in thread
From: Dave Kleikamp @ 2005-02-28  4:47 UTC (permalink / raw)
  To: frederik; +Cc: fsdevel

On Sun, 2005-02-27 at 15:00 -0800, Frederik Eaton wrote:

> Thanks, you were right that I recently upgraded to 2.6 (actually I
> transferred the hard drive between machines), and with a "mount /home -o
> remount,iocharset=utf8" at least I can access the files now. But the
> names aren't what they should be, for instance I see:
> 
> 04-The_Dark_Of_The_MatinΘe.ogg
> 
> (that's a theta). Were they supposed to be correct? 

Not necessarily.  Even if the character returned is correct utf8, the
terminal (xterm?) displaying the name may be assuming a different
character set.  It's also possible that on the older system,
CONFIG_NLS_DEFAULT was not set to the same locale that the system used,
so the character stored in the name was not the right unicode, but it
was accessed consistently on the old system so it worked as long as
nothing changed.

> As an aside, now, I just tried "mount /home -o
> remount,iocharset=iso8859-1", and got same problem

This is no surprise.  iso8859-1 has the same mapping as iocharset=none
(the 2.6 default).  (The "none" keyword is first supported in
2.6.11-rc4-mm1 and 2.4.30-pre2, so you probably won't be able to use
it.)

> ls: '/home/shared/frederik/Franz_Ferdinand/Franz_Ferdinand/04-The_Dark_Of_The_Matin?e.ogg': No such file or directory
> 
> But when I repeated the "mount /home -o remount,iocharset=utf8"
> command, the listing didn't show the name with a theta character, but
> instead with two question marks:
> 
> 04-The_Dark_Of_The_Matin??e.ogg
> 
> Strange behavior. I.e., upon changing the mount options and then
> changing them back I get a different listing.

That I wouldn't expect.  I really have no clue why it isn't consistent.

The easiest way to "fix" these files (assuming there aren't too many) is
to mount with iocharset=utf8, rename the files using only ascii
characters, remount again with iocharset=iso8859-1, and rename the files
to their proper names.  After that you should be able to mount without
any iocharset flag and get sane behavior.

> Frederik
-- 
David Kleikamp
IBM Linux Technology Center

-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: jfs, special characters
  2005-02-28  4:47     ` Dave Kleikamp
@ 2005-02-28  5:42       ` Frederik Eaton
  0 siblings, 0 replies; 6+ messages in thread
From: Frederik Eaton @ 2005-02-28  5:42 UTC (permalink / raw)
  To: Dave Kleikamp; +Cc: fsdevel

> The easiest way to "fix" these files (assuming there aren't too many) is
> to mount with iocharset=utf8, rename the files using only ascii
> characters, remount again with iocharset=iso8859-1, and rename the files
> to their proper names.  After that you should be able to mount without
> any iocharset flag and get sane behavior.

OK, thanks.

Frederik


^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2005-02-28  5:42 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2005-02-26  4:34 jfs, special characters Frederik Eaton
2005-02-26 16:51 ` Sonny Rao
2005-02-26 17:58 ` Dave Kleikamp
2005-02-27 23:00   ` Frederik Eaton
2005-02-28  4:47     ` Dave Kleikamp
2005-02-28  5:42       ` Frederik Eaton

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.