[8618] in linux-scsi channel archive
Re: is (rscsi_disks[i].sector_size != 512) ok ?
daemon@ATHENA.MIT.EDU (Eric Youngdale)
Thu Apr 13 09:45:01 2000
Message-ID: <02db01bfa54d$77a1a3d0$0f17a8c0@eric.home>
From: "Eric Youngdale" <eric@andante.org>
To: <linux-scsi@vger.rutgers.edu>,
"Marc SCHAEFER" <schaefer@alphanet.ch>
Date: Thu, 13 Apr 2000 09:37:52 -0400
MIME-Version: 1.0
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
----- Original Message -----
From: "Marc SCHAEFER" <schaefer@alphanet.ch>
Newsgroups: alphanet.ml.linux.scsi
To: <linux-scsi@vger.rutgers.edu>
Sent: Thursday, April 13, 2000 3:44 AM
Subject: Re: is (rscsi_disks[i].sector_size != 512) ok ?
> Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
> > They do. You need to tell them to use that block size sometimes. Eg
fdisk
> > needs to be told to use 1K block scaling
>
> > > SCSI disk error : host 0 channel 0 id 9 lun 0 return code = 8000002
> > > Info fld=0x10f5a9c, Current sd08:01: sense key Illegal Request
>
> > or you run off the end of the disk 8)
>
> Question: there is really nothing which prevents bad SCSI commands
> to be sent out ? I was thinking that as READ_CAPACITY is used to
> initialize the /dev/sda size, sub-partitions of /dev/sda might never
> be bigger than the whole.
Yes, this is already happening.
> Shouldn't it be a test in the ll_rw_blk() or otherwise that
> would cause the *kernel* to error, and not a SCSI command with
> illegal blocks to be send out ?
There already is such a test in sd.c. The commands in question should
never make it out to the device. This is why I suspect that the data for
READ_CAPACITY is being reported incorrectly.
-Eric
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
