Re: lingering loopback bugs?

[Jens Axboe]

On Sat, Mar 03 2001, Alexander Viro wrote:
> > Look for the patch I posted yesterday (hint: just remove these two
> > lines from loop_end_io_transfer)
> > 
> >                if (atomic_dec_and_test(&lo->lo_pending))
> >                        up(&lo->lo_bh_mutex);
> 
> Uhh... And what will compensate for atomic_inc() in loop_make_request() in
> case of loop over block device?

There is no atomic_inc() in loop_make_request, I moved it to
count pending bh on the queue only. But hmm, we should not
be able to remove it while it has bhs in flight. Ok, I'll
rearrange this a bit again...

-- 
Jens Axboe