[PATCH v3 2/2] block-dma: properly take MMIO path

[Leon Romanovsky]

On Wed, Nov 12, 2025 at 10:23:19AM -0500, Keith Busch wrote:
> On Wed, Nov 12, 2025 at 05:16:04PM +0200, Leon Romanovsky wrote:
> > On Wed, Nov 12, 2025 at 09:38:07AM -0500, Keith Busch wrote:
> > > On Mon, Oct 27, 2025 at 09:30:21AM +0200, Leon Romanovsky wrote:
> > > > @@ -732,13 +746,20 @@ static void nvme_unmap_metadata(struct request *req)
> > > >  		return;
> > > >  	}
> > > >  
> > > > -	if (!blk_rq_integrity_dma_unmap(req, dma_dev, &iod->meta_dma_state,
> > > > -					iod->meta_total_len)) {
> > > > +	if (iod->flags & IOD_META_P2P)
> > > > +		map = PCI_P2PDMA_MAP_BUS_ADDR;
> > > > +	if (iod->flags & IOD_META_MMIO) {
> > > > +		map = PCI_P2PDMA_MAP_THRU_HOST_BRIDGE;
> > > > +		attrs |= DMA_ATTR_MMIO;
> > > > +	}
> > > 
> > > This should be an 'else if' no?
> > 
> > We can't have both IOD_META_P2P and IOD_META_MMIO. It can be only one or
> > IOD_META_P2P or IOD_META_MMIO. In this case "else if' is not necessary.
> 
> I get that and I'm not saying this is a bug, but it just looks odd.

Sorry, I didn't get the intent.

Thanks

> You're checking the conditions as if they're independent of each other
> when they're not. If the first 'if' is true, there's no need to test the
> second one as we know it's false, but this code flow will test it
> anyway.