xfs
[Top] [All Lists]

Re: [PATCH 4/4] XFS: Unicode case-insensitive lookup implementation

To: Barry Naujok <bnaujok@xxxxxxx>
Subject: Re: [PATCH 4/4] XFS: Unicode case-insensitive lookup implementation
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Mon, 21 Apr 2008 05:15:20 -0400
Cc: xfs@xxxxxxxxxxx, linux-fsdevel@xxxxxxxxxxxxxxx
In-reply-to: <20080421083645.799556386@xxxxxxxxxxxxxxxxxxxxxxx>
References: <20080421083103.433280025@xxxxxxxxxxxxxxxxxxxxxxx> <20080421083645.799556386@xxxxxxxxxxxxxxxxxxxxxxx>
Sender: xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.5.17 (2007-11-01)
The patch looks quite clean, but there still is the fundamental issue
of rejecting perfectly fine UNIX filenames in xfs_unicode_validate,
as discussed in the last round a Unix filename is any valid
NULL-terminated bytestream with only ".", "..", and "/" special-cased
and this patch breaks this assumptions.

In addition to that I think xfs_unicode.[ch] should be split into
a small xfs-specific part for reading the on-disk case-folding table
and implementing the actual nameops, and the generic code for case
folding and compares and case-folded strings should be moved into lib/
as it's quite generic.


<Prev in Thread] Current Thread [Next in Thread>