Re: TYPE_RBC cache fixes (sbp2.c affected)

[Douglas Gilbert <dougg@torque.net>]

Al Viro wrote:
> 	a) TYPE_SDAD renamed to TYPE_RBC and taken to scsi.h
> 	b) in sbp2.c remapping of TYPE_RPB to TYPE_DISK turned off
> 	c) relevant places in midlayer and sd.c taught to accept TYPE_RBC
> 	d) sd.c::sd_read_cache_type() looks into page 6 when dealing with

Al,
Adding some information:

The name of mode page 6 is "RBC device parameters".
References for RBC:
http://www.t10.org/ftp/t10/drafts/rbc/rbc-r10a.pdf
http://www.t10.org/ftp/t10/drafts/rbc/rbc-a101.pdf

I just added mode page 6 to my sdparm beta.

> TYPE_RBC - these guys have writeback cache flag there and are not guaranteed
> to have page 8 at all.

The Write(back) Cache Disable (WCD) flag in mode page 6
is logically flipped from WCE in DASD (i.e. disk) caching
mode page (page number 8).

> 	e) sd_read_cache_type() got an extra sanity check - it checks that
> it got the page it asked for before using its contents.  And screams if
> mismatch had happened.  Rationale: there are broken devices out there that
> are "helpful" enough to go for "I don't have a page you've asked for, here,
> have another one".  For example, PL3507 had been caught doing just that...
> 	f) sbp2 sets sdev->use_10_for_rw and sdev->use_10_for_ms instead
> of bothering to remap READ6/WRITE6/MOD_SENSE, so most of the conversions
> in there are gone now.

RBC makes MODE_SENSE/SELECT (6) support mandatory but does
not support MODE_SENSE/SELECT(10).

> 	Incidentally, I wonder if USB storage devices that have no
> mode page 8 are simply RBC ones.  I haven't touched that, but it might
> be interesting to check...

I tried on a SanDisk SDDR-31 which doesn't support
any mode pages at all! Same with a SanDisk mini Cruzer.
My USB enclosure with a ATA disk inside is an
abomination wrt MODE SENSE. It doesn't respond to
MODE SENSE 6 but does to a MODE SENSE 10 but with
a mode parameter header for the 6 byte variant.
Any code that silently switches mode sense 6 to
10 (or vice versa) and doesn't fix the response is just
wrong.

Doug Gilbert