On Wed, Jun 11, 2025 at 03:38:22PM -0600, Keith Busch wrote:
> On Tue, Jun 10, 2025 at 07:06:41AM +0200, Christoph Hellwig wrote:
> > +static inline bool nvme_pci_metadata_use_sgls(struct request *req)
> > {
> > - if (!nvme_ctrl_meta_sgl_supported(&dev->ctrl))
> > - return false;
> > return req->nr_integrity_segments > 1 ||
> > nvme_req(req)->flags & NVME_REQ_USERCMD;
> > }
>
> ...
>
>
> > @@ -981,7 +979,7 @@ static blk_status_t nvme_map_metadata(struct nvme_dev *dev, struct request *req)
> > struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
> >
> > if ((iod->cmd.common.flags & NVME_CMD_SGL_METABUF) &&
> > - nvme_pci_metadata_use_sgls(dev, req))
> > + nvme_pci_metadata_use_sgls(req))
> > return nvme_pci_setup_meta_sgls(dev, req);
> > return nvme_pci_setup_meta_mptr(dev, req);
> > }
>
> Am I missing something here? This looks like it forces user commands to
> use metadata SGLs even if the controller doesn't support it.
Yeah, looks like I messed up something here.
