Differences between iscsi and gscsi

[Roman Rakus]

On 02/03/2010 08:27 AM, Roman Rakus wrote:
> On 02/03/2010 01:06 AM, Tony Breeds wrote:
>> On Tue, Feb 02, 2010 at 09:30:58PM +0100, Roman Rakus wrote:
>>
>>> Or maybe we can add TOK_GSCSI and corresponding tests.
>> That seesm like a better approach to me.
>>
>> I have to admit that I don't know what gscsi is.  The only reference 
>> I can
>> really find is in the linux IPR driver.  and that's to "Generic Scsi"
>>
>> Yours Tony
> Well, my apologize, now I have noticed that in the defdevice string: 
> /vdevice/gscsi:dev=bootpath,iscsi,rawio,itname=iqn.2008-05.com.elm3c241:01,ciaddr=9.47.69.241,giaddr=9.47.69.1,subnet-mask=255.255.255.0,siaddr=9.47.69.22,iport=3260,iname=iqn.1992-08.com.netapp:sn.84183797,ilun=0,0/disk at 0 
>
> is `iscsi' substring. So the problem is elsewhere. I will try to 
> figure out why yaboot removes ending `:sn.84183797,ilun=0,0/disk at 0' (I 
> guess it causes problems).
>
> RR
> _______________________________________________
> Yaboot-devel mailing list
> Yaboot-devel at lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/yaboot-devel

Looks like the problem is in the prom_get_devtype() function. We find 
there ISCSI token and then we check what type of device is 
"/vdevice/gscsi/disk". From my fast check on one of the systems, 
"/vdevice/gscsi/disk" is block device type, but prom_get_devtype() 
function should returns iscsi device type. Or at least network device type.
I'm not sure whether "/vdevice/gscsi/disk" is always block device. Can 
anyone confirm, disconfirm or clarify it?
Following patch assumes "/vdevice/gscsi/disk" is always block device and 
whether checking device path contains ISCSI token returns ISCSI device type.
I have not done any tests. Can anyone test is please?
RR
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: yaboot-1.3.14-iscsi.patch
URL: <http://lists.ozlabs.org/pipermail/yaboot-devel/attachments/20100205/7d764f23/attachment.asc>