On Fri, 2 Aug 2013, Oliver Neukum wrote: > On Fri, 2013-08-02 at 09:54 -0400, Alan Stern wrote: > > On Fri, 2 Aug 2013, Oliver Neukum wrote: > > > > > On Thu, 2013-08-01 at 11:53 -0400, Alan Stern wrote: > > > > On Thu, 1 Aug 2013, Oliver Neukum wrote: > > > > > > > > > On Wed, 2013-07-31 at 11:13 -0400, Alan Stern wrote: > > > > > > > > > > > More importantly, if we already know that the medium is not present or > > > > > > has been changed since it was last used, then there's no reason to call > > > > > > sd_sync_cache() at all. > > > > > > > > > > Like this? > > > > > > > > Yes, I like this a lot better, except I would put the test for > > > > !sdkp->media_present in sd_suspend_common() -- no need to print the > > > > > > But your observation that it makes no sense while no medium is present > > > is valid whatever be the reason for wanting to flush. > > > > So do the test in both places. > > Code duplication for what reasons? Like I said before, to avoid printing a misleading line in the kernel log. If you prefer, just get rid of the log message. Or make it KERN_DEBUG instead of KERN_NOTICE, along with the "Stopping disk" message. Those two things might get pretty annoying when people start using block-layer runtime PM. Alan Stern
