Re: [Re: Linux 2.6.26-rc2] Write protect on on

[Boaz Harrosh <bharrosh@panasas.com>]

Linus Torvalds wrote:
> 
> On Mon, 19 May 2008, Boaz Harrosh wrote:
> 
>> Alan Stern wrote:
>>> Yes, it never worked properly.  But now it fails in a bad way whereas 
>>> before it failed in a benign way.
>> You do realize that, that was pure lock to have a zero'ed buffer.
> 
> Umm. Maybe it SHOULD NOT HAVE BEEN!
> 
> The thing is, if we can get partial results back, we really *should* 
> either error out, or we should have at least cleared the buffer (either 
> beforehand or when seeing the partial result). Returning a buffer with the 
> old random contents is a bug.
> 
> And if clearing the buffer not only avoids any security holes and possible 
> undefined behavior, but _also_ ends up fixing the write protect sense 
> issue, all the better!
> 
> 		Linus
> --

Sure, inspecting other places that emulate MODE_SENSE, (And inspecting the scsi
spec) all zeros is a very good scsi response. Alan do you want to send a fix for all
places that initiate a MODE_SENSE command, specifically at
scsi_scan.c::scsi_unlock_floptical() ? (Some other places do)

Boaz