Re: [PATCH 2.6.15-rc6] block: Make CDROMEJECT more robust

[Jens Axboe <axboe@suse.de>]

On Mon, Dec 19 2005, Ben Collins wrote:
> On Mon, 2005-12-19 at 20:35 +0100, Jens Axboe wrote:
> > On Mon, Dec 19 2005, Ben Collins wrote:
> > > This patch fixes the WRITE vs READ issue, and also sends the extra two
> > > commands. Anyone with an iPod connected via USB (not sure about firewire)
> > > should be able to reproduce this issue, and verify the patch.
> > 
> > The bug was in the SCSI layer, and James already has the fix integrated
> > for that. It really should make 2.6.15, James are you sending it upwards
> > for that?
> 
> You mean this patch?
> 
> James Bottomley:
>       [SCSI] Consolidate REQ_BLOCK_PC handling path (fix ipod panic)
> 
> This fixes an oops with data direction because sbp2 was not checking
> enough itself.
> 
> I seriously doubt this will fix the issue being reported. Changing the
> blk request to a READ did not fix the problem. The problem was only
> fixed by sending the extra two commands. The direction was just a side
> issue.

Your report surely made it seem like an issue, hence I pointed you at
the fix(es). What part of the block layer didn't like these commands?

> Is there a problem with sending the commands? If they don't bother
> unaffected devices, but it does fix a large number of other devices,
> what's the problem?

It's not necessarily a problem, especially if other paths end up doing
it in some way or another. Hard to say whether it bothers other device,
but I agree it should not. Please resend after 2.6.15 is released (and
do make those macros functions, thanks).

-- 
Jens Axboe