On 06/13/2017 01:24 PM, Andreas Dilger wrote: > On Jun 13, 2017, at 11:15 AM, Jens Axboe <axboe@xxxxxxxxx> wrote: >> >> No functional changes in this patch, just in preparation for >> allowing applications to pass in hints about data life times >> for writes. >> >> Signed-off-by: Jens Axboe <axboe@xxxxxxxxx> > > This is fine from the POV of passing the stream ID unchanged down the stack. > If we really wanted to limit the ID to a 16-bit value, the i_stream could fit > into a few 16-bit holes in struct inode, to avoid increasing its size. Thanks, I'll change it to a short and pack the hole instead. Even with everything else in there, having 64k streams is more than we need, might as well do that. -- Jens Axboe
