[dm-devel] dm writecache: reject asynchronous pmem.

[Michal Suchánek]

On Tue, Jun 30, 2020 at 09:36:33AM -0400, Mike Snitzer wrote:
> On Tue, Jun 30 2020 at 10:10am -0400,
> Michal Suchánek <msuchanek at suse.de> wrote:
> 
> > On Tue, Jun 30, 2020 at 09:32:01AM -0400, Mikulas Patocka wrote:
> > > 
> > > 
> > > On Tue, 30 Jun 2020, Michal Suchanek wrote:
> > > 
> > > > The writecache driver does not handle asynchronous pmem. Reject it when
> > > > supplied as cache.
> > > > 
> > > > Link: https://lore.kernel.org/linux-nvdimm/87lfk5hahc.fsf@linux.ibm.com/
> > > > Fixes: 6e84200c0a29 ("virtio-pmem: Add virtio pmem driver")
> > > > 
> > > > Signed-off-by: Michal Suchanek <msuchanek at suse.de>
> > > > ---
> > > >  drivers/md/dm-writecache.c | 6 ++++++
> > > >  1 file changed, 6 insertions(+)
> > > > 
> > > > diff --git a/drivers/md/dm-writecache.c b/drivers/md/dm-writecache.c
> > > > index 30505d70f423..57b0a972f6fd 100644
> > > > --- a/drivers/md/dm-writecache.c
> > > > +++ b/drivers/md/dm-writecache.c
> > > > @@ -2277,6 +2277,12 @@ static int writecache_ctr(struct dm_target *ti, unsigned argc, char **argv)
> > > >  
> > > >  		wc->memory_map_size -= (uint64_t)wc->start_sector << SECTOR_SHIFT;
> > > >  
> > > > +		if (!dax_synchronous(wc->ssd_dev->dax_dev)) {
> > > > +			r = -EOPNOTSUPP;
> > > > +			ti->error = "Asynchronous persistent memory not supported as pmem cache";
> > > > +			goto bad;
> > > > +		}
> > > > +
> > > >  		bio_list_init(&wc->flush_list);
> > > >  		wc->flush_thread = kthread_create(writecache_flush_thread, wc, "dm_writecache_flush");
> > > >  		if (IS_ERR(wc->flush_thread)) {
> > > > -- 
> > > 
> > > Hi
> > > 
> > > Shouldn't this be in the "if (WC_MODE_PMEM(wc))" block?
> > That should be always the case at this point.
> > > 
> > > WC_MODE_PMEM(wc) retrurns true if we are using persistent memory as a 
> > > cache device, otherwise we are using generic block device as a cache 
> > > device.
> >
> > This is to prevent the situation where we have WC_MODE_PMEM(wc) but
> > cannot guarantee consistency because the async flush is not handled.
> 
> The writecache operates in 2 modes.  SSD or PMEM.  Mikulas is saying
> your dax_synchronous() check should go within a WC_MODE_PMEM(wc) block
> because it doesn't make sense to do the check when in SSD mode.

Indeed, it's in the wrong if/else branch.

Thanks

Michal