X-Authentication-Warning: delorie.com: mail set sender to djgpp-workers-bounces using -f
X-Recipient: djgpp-workers AT delorie DOT com
Date: Sun, 11 Mar 2012 18:57:40 +0200
From: Eli Zaretskii <eliz AT gnu DOT org>
Subject: Re: Value stored in cache_blksize if _is_remote_drive fails
In-reply-to: <201203111650.53344.juan.guerrero@gmx.de>
X-012-Sender: halo1 AT inter DOT net DOT il
To: djgpp-workers AT delorie DOT com
Message-id: <83haxvxegr.fsf@gnu.org>
References: <201203111650 DOT 53344 DOT juan DOT guerrero AT gmx DOT de>
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

> From: Juan Manuel Guerrero <juan DOT guerrero AT gmx DOT de>
> Date: Sun, 11 Mar 2012 16:50:53 +0100
> 
>   if (!cache_blksize[d])
>   {
>     if (_is_remote_drive(d))  /* A = 0, B = 1, C = 2,  etc. */
>     {
>       /* Default remote drives to 4K block size, to improve performance.
>        *
>        * Also the size returned by statfs() may not be correct. Testing
>        * against files shared by Samba 2.0.10 on Linux kernel 2.2.19
>        * returned a 32K block size, even though the ext2 filesystem
>        * holding the share share had a 4K block size. */
>       cache_blksize[d] = 4096;
>     }
>     else
>     {
>       /* No entry => retrieve cluster size */
>       if (statfs(path, &sbuf) != 0)
>       {
>         /* Failed, pass error through */
>         return -1;
>       }
> 
>       cache_blksize[d] = sbuf.f_bsize;
>     }
>   }
> -- code end --
> 
> 
> If _is_remote_drive returns with a value different from 0 then cache_blksize[d]
> is set to 4096.  The issue is that _is_remote_drive may return with 1 if d is a
> remote drive but also it may return with -1 if the function fails for drive
> number d.  In both cases cache_blksize[d] = 4096.  The question arises if this
> is a bug or a feature?  Drive "d" does not exist but a valid block size is 
> assigned to cache_blksize[d] at the same time the errno (= ENODEV) set by
> _is_remote_drive may get lost in future operations.

There's not enough context in what you show to make up my mind whether
this is a bug or a feature.  Specifically, it's not clear why it would
be better to do something different when _is_remote_drive fails.

Perhaps you could show a couple of use cases where this does some
harm.

Thanks.