Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 31 Jul 2023 07:55:39 +0200
From:   Thomas =?utf-8?Q?Wei=C3=9Fschuh?= <thomas@t-8ch.de>
To:     tjcw@cantab.net
Cc:     linux-kernel@vger.kernel.org
Subject: Re: vfat file system does not allow a space at the end of a
 directory name
Message-ID: <196f7191-5820-4480-99d3-e7ffc66e9c98@t-8ch.de>
References: <CAC=wTOim6VZM7gKGQ8KuvbVSdmjyg=4G5sFbr+xY89ujPVOq8g@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAC=wTOim6VZM7gKGQ8KuvbVSdmjyg=4G5sFbr+xY89ujPVOq8g@mail.gmail.com>
Precedence: bulk

Hi Chris,

On 2023-07-30 23:33:03+0100, Chris Ward wrote:
> I had a zip file (from downloading pictures from Google Photos) where
> some of the directory names had a space at the end. This unzipped fine
> on an ext4 file system, but gave errors when unpacking onto a USB key
> which had a FAT file system.
> 
> On further investigation, file names have the same problem. On my USB key, I get
> 
> tjcw@tjcw-OptiPlex-7010:/media/tjcw/Photos$ touch 'aab '
> touch: setting times of 'aab ': No such file or directory
> tjcw@tjcw-OptiPlex-7010:/media/tjcw/Photos$ echo ok >'aab '
> bash: aab : Invalid argument
> tjcw@tjcw-OptiPlex-7010:/media/tjcw/Photos$ touch 'aab'
> tjcw@tjcw-OptiPlex-7010:/media/tjcw/Photos$ echo ok >'aab'
> tjcw@tjcw-OptiPlex-7010:/media/tjcw/Photos$
> 
> Is this as expected, or is it a bug in the kernel's handling of FAT
> file systems ? Probably the next step is to check the behaviour under
> Microsoft Windows.

AFAIK this is a limitation in the FAT filesystem itself.

The problem is that FAT stores filenames in fixed width fields.
Unused space in these fields is filled with ' '.
This means that storing "foo", "foo " or "foo ". would all result in the
exact same result and it will be impossible to figure out the correct
one afterwards.

From Wikipedia [0]:

    trailing spaces in either the base name or the extension are
    considered to be padding and not a part of the file name

From fs/fat/namei_vfat.c vfat_is_used_badchars():

    last character cannot be space

[0] https://en.wikipedia.org/wiki/Design_of_the_FAT_file_system