Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 13 Mar 1997 09:48:29 +0100
From:      j@uriah.heep.sax.de (J Wunsch)
To:        nao@sbl.cl.nec.co.jp (Naoki Hamada)
Cc:        hackers@FreeBSD.ORG
Subject:   Re: [2.2-GAMMA] partial write to raw mode disk
Message-ID:  <19970313094829.GV57056@uriah.heep.sax.de>
In-Reply-To: <199703130505.OAA26681@sirius.sbl.cl.nec.co.jp>; from Naoki Hamada on Mar 13, 1997 14:05:18 %2B0900
References:  <199703130505.OAA26681@sirius.sbl.cl.nec.co.jp>

next in thread | previous in thread | raw e-mail | index | archive | help
As Naoki Hamada wrote:

> Writing an incomplete data to a raw mode disk device seems to fail. Is
> this an intended behavior?

Of course!  The raw device is always restricted to be handled in
multiples of its physical block size.  That's why it is `raw'.

> On the other hand, writing the same data to
> the corresponding block device always succeeds.

But you never have any guarantee when the data will actually arrive on
the disk, and you will have poor error reporting due to decoupling the
process.  The only guarantee to get the data onto the drive is to
shutdown the system. :-] (Yes, i have seen systems where this was
true!)  The only other data is to umount the corresponding filesystem,
but since you aren't using one, you cannot do this.  (There's another
bad effect when using block devices: you blow the buffer cache.)

Ergo: do only use block devices in order to mount filesystems over
them.

-- 
cheers, J"org

joerg_wunsch@uriah.heep.sax.de -- http://www.sax.de/~joerg/ -- NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19970313094829.GV57056>