Re: ide_cd modifies hard_* members of request

[Jens Axboe <jens.axboe@oracle.com>]

On Tue, Mar 18 2008, Boris Petkov wrote:
> (sorry, forgot to cc all)
> 
> On Tue, Mar 18, 2008 at 9:54 AM, Nikanth Karthikesan <knikanth@suse.de> wrote:
> > Hi
> >
> >  Despite the warning in blkdev.h, that members of struct request starting
> >  with hard_* are block layer internals, and no driver should touch them,
> >  the ide_cd driver seems to fiddle around with it.
> >
> >  Is this reqd, or can this be made to work without the need for this?
> >
> >  If this reqd, why not provide a helper function for that in block layer
> >  itself, may be, with a warning about usage?
> >
> >  Thanks
> >  Nikanth Karthikesan
> 
> Hi,
> 
> the code pieces you're referring to are in there since at least the
> first git import of the kernel source (i.e. 2005) aka long time so
> Bart/Jens should know that...

restore_request(), I'm assuming. It was probably me who modified this
code, I don't think there's any justification for it. I would suggest
killing it. And cdrom_read_from_buffer() as well, it's a silly addition.

-- 
Jens Axboe