RE: QLA2300, Infortrand and LBD?

[James Bottomley <James.Bottomley@SteelEye.com>]

On Wed, 2004-07-28 at 09:46, Frank Borich wrote:
> Looking at an FC trace on an Infortrend controller I have, shows that
> for a LUN 3586680 MB in size, the LBA is B5D3BFFF in the "RETURNED
> LOGICAL BLOCK ADDRESS FIELD" 
> and with in the "BLOCK LENGTH IN BYTES FIELD" field is 200h , ( B5D3BFFF
> * 200h =  approx. 1.6 TB). Seems that infortrend does NOT respond
> properly to READ_CAPACITY_10, at least in my case. 

Well 3586680 MB == 0x1b5d3bfff * 0x200

so I think you can see where the returned figure is coming from...

Infortrends should be made aware of the problem so they can do a bios
upgrade.  I suspect that even if the array returned the correct capacity
it would be unable to process the READ/WRITE(16) commands necessary to
reach it anyway, so their "fix" may be to tell you that no individual
luns can be configured beyond 2TB.

> James, this may be a dumb question, please excuse my ignorance of the
> internals of the scsi subsystem, 
> but is there a plan to implement LBD support in the 2.4 tree ?

The problem is not just in SCSI, it permeates the block layer.  I don't
think there's much point backporting LBD to 2.4, although many vendor
trees have it.

James