linux-kernel - Re: [PATCH] mass storage : emulation of sat scsi_pass

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Sat, 8 Mar 2008 13:21:48 -0800
From:	Matthew Dharm <mdharm-kernel@...-eyed-alien.net>
To:	matthieu castet <castet.matthieu@...e.fr>
Cc:	linux-usb@...r.kernel.org,
	Linux Kernel list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH]  mass storage : emulation of sat scsi_pass_thru with ATACB

On Sat, Mar 08, 2008 at 09:08:26PM +0100, matthieu castet wrote:
> Hi Matthew,
> 
> thanks for your comments
> 
> Matthew Dharm wrote:
> >Why are you using an initializer instead of a new protocol code?
> Because using a new protocol code means I need to patch all the place 
> where there is a comparison between us->subclass and US_SC_SCSI.
> After all I am US_SC_SCSI with a special case for ATA12 & ATA16 commands.
> I don't translate all scsi to atacb (that's what does US_SC_ISD200).

Yet, you call invoke_transport directly, just like any other protocol
handler.

The proper way to do this is as a separate protocol handler.  If you want
to make it clear that you are only intercepting a couple of command types,
then don't call invoke_transport() directly, call the transparent scsi
protocol handler (which, of course, does the same thing but provides
clearer layering).

Oh, and you should add some "unlikely" tags to these if() conditionals.

> >Actually, why do you even have a separate 'dispatcher' function?  Why not
> >just one protocol handler function which checks the command at the top and
> >calls invoke_transport there?
> What do you means by having a separate 'dispatcher' function?
> You means why I have 2 functions emulate_pass_thru_with_atacb and 
> usb_stor_transparent_scsi_command_atacb ?
> I did 2 functions for having a code more clean.
> 
> You suggest something like
> void usb_stor_transparent_scsi_command_atacb(struct scsi_cmnd *srb,
>                        struct us_data *us)
> {
>     if (srb->cmnd[0] != ATA_16 && srb->cmnd[0] != ATA_12) {
>         usb_stor_invoke_transport(srb, us);
>         return;
>     }
>     copy emulate_pass_thru_with_atacb code here
> }

Yes, modulo my comment above about calling the transparent scsi protocol
handler instead of invoke_transport directly.

> >Also, unless ATACB is a new standard (and I don't think it is, as the
> >Cypress datasheet uses the term 'vendor specific'), then your functions
> >need renaming.  Instead of 'emulate_pass_thru_with_atacb', how about
> >something like 'cypress_atacb' --  since it's already a protocol handler,
> >everyone already knows it's for passing commands.
> But 'emulate_pass_thru_with_atacb' only handle ATA pass_thru scsi 
> commands. It doesn't translate all scsi commands to atacb like 
> 'cypress_atacb' could suggest.
> That's why I put 'usb_stor_transparent_scsi_command_atacb' saying it is 
> transparent_scsi_command + atacb support.

Yes, but your name suggests that ATACB is a new industry standard which is
implemented by more than a few chips from one specific vendor.  That's not
acceptable.

Try 'cypress_atacb_passthrough' instead?

Matt

-- 
Matthew Dharm                              Home: mdharm-usb@...-eyed-alien.net 
Maintainer, Linux USB Mass Storage Driver

P:  Nine more messages in admin.policy.
M: I know, I'm typing as fast as I can!
					-- Pitr and Mike
User Friendly, 11/27/97

Content of type "application/pgp-signature" skipped