Re: Implement mode_select for atapi tape drives

[Marc Balmer <marc%msys.ch@localhost>]

> On Sat, 8 Aug 2009, Marc Balmer wrote:
>
> > > But the calculated value of *_select_len is actually passed to  
> > > routine scsipi_mode_select() a little bit later on, so we still  
> > > need the calculated value.
> >
> > why cant sizeof be used there, too?
>
> It actually controls amount of data that is sent to the device.   
> Most devices have 0 bytes for st->page_0_size but some have 12 bytes.
>
>
> The attached patch revision includes the change to use sizeof() in  
> the calls to memset(), adds a KASSERT when loading the quirk data to  
> ensure that the st->page_0_size is not larger than MAX_PAGE_0_SIZE,  
> and adds the necessary quirk for my tape drive.

I like this.  Maybe KASSERT that MAX_PAG_0_SIZE is <= sizeof(struct  
atapi_select) ?  That could be combined into a single KASSERT, even.

> Additional comments/criticisms welcomed!