rewrites discard support

index 75982b9..331d437 100644
@@ -181,6 +181,14 @@ static inline u64 nvme_block_nr(struct nvme_ns *ns, sector_t sector)
 	return (sector >> (ns->lba_shift - 9));
 }
 
+static inline unsigned nvme_map_len(struct request *rq)
+{
+	if (rq->cmd_flags & REQ_DISCARD)
+		return sizeof(struct nvme_dsm_range);
+	else
+		return blk_rq_bytes(rq);
+}
+
 static inline void nvme_setup_flush(struct nvme_ns *ns,
 		struct nvme_command *cmnd)
 {

index 4301584..2db6f89 100644
@@ -348,16 +348,11 @@ static __le64 **iod_list(struct request *req)
 	return (__le64 **)(iod->sg + req->nr_phys_segments);
 }
 
-static int nvme_init_iod(struct request *rq, struct nvme_dev *dev)
+static int nvme_init_iod(struct request *rq, unsigned size,
+		struct nvme_dev *dev)
 {
 	struct nvme_iod *iod = blk_mq_rq_to_pdu(rq);
 	int nseg = rq->nr_phys_segments;
-	unsigned size;
-
-	if (rq->cmd_flags & REQ_DISCARD)
-		size = sizeof(struct nvme_dsm_range);
-	else
-		size = blk_rq_bytes(rq);
 
 	if (nseg > NVME_INT_PAGES || size > NVME_INT_BYTES(dev)) {
 		iod->sg = kmalloc(nvme_iod_alloc_size(dev, size, nseg), GFP_ATOMIC);

@@ -651,6 +646,7 @@ static int nvme_queue_rq(struct blk_mq_hw_ctx *hctx,
 	struct nvme_dev *dev = nvmeq->dev;
 	struct request *req = bd->rq;
 	struct nvme_command cmnd;
+	unsigned map_len;
 	int ret = BLK_MQ_RQ_QUEUE_OK;
 
 	/*

@@ -666,7 +662,8 @@ static int nvme_queue_rq(struct blk_mq_hw_ctx *hctx,
 		}
 	}
 
-	ret = nvme_init_iod(req, dev);
+	map_len = nvme_map_len(req);
+	ret = nvme_init_iod(req, map_len, dev);
 	if (ret)
 		return ret;
 

index 47e9f27..f320c38 100644
@@ -1523,6 +1523,7 @@ EXPORT_SYMBOL(blk_put_request);
  * blk_add_request_payload - add a payload to a request
  * @rq: request to update
  * @page: page backing the payload
+ * @offset: offset in page
  * @len: length of the payload.
  *
  * This allows to later add a payload to an already submitted request by

@@ -1533,12 +1534,12 @@ EXPORT_SYMBOL(blk_put_request);
  * discard requests should ever use it.
  */
 void blk_add_request_payload(struct request *rq, struct page *page,
-		unsigned int len)
+		int offset, unsigned int len)
 {
 	struct bio *bio = rq->bio;
 
 	bio->bi_io_vec->bv_page = page;
-	bio->bi_io_vec->bv_offset = 0;
+	bio->bi_io_vec->bv_offset = offset;
 	bio->bi_io_vec->bv_len = len;
 
 	bio->bi_iter.bi_size = len;

index 586f916..9a9ec21 100644
@@ -562,7 +562,7 @@ skd_prep_discard_cdb(struct skd_scsi_request *scsi_req,
 	put_unaligned_be32(count, &buf[16]);
 
 	req = skreq->req;
-	blk_add_request_payload(req, page, len);
+	blk_add_request_payload(req, page, 0, len);
 }
 
 static void skd_request_fn_not_online(struct request_queue *q);

index d749da7..a4b927c2 100644
@@ -779,7 +779,7 @@ static int sd_setup_discard_cmnd(struct scsi_cmnd *cmd)
 	 * discarded on disk. This allows us to report completion on the full
 	 * amount of blocks described by the request.
 	 */
-	blk_add_request_payload(rq, page, len);
+	blk_add_request_payload(rq, page, 0, len);
 	ret = scsi_init_io(cmd);
 	rq->__data_len = nr_bytes;
 

index 413c84f..121eebc 100644
@@ -778,7 +778,7 @@ extern struct request *blk_make_request(struct request_queue *, struct bio *,
 extern void blk_rq_set_block_pc(struct request *);
 extern void blk_requeue_request(struct request_queue *, struct request *);
 extern void blk_add_request_payload(struct request *rq, struct page *page,
-		unsigned int len);
+		int offset, unsigned int len);
 extern int blk_lld_busy(struct request_queue *q);
 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
 			     struct bio_set *bs, gfp_t gfp_mask,

index 2db6f89..f2f8a4d 100644
@@ -377,6 +377,9 @@ static void nvme_free_iod(struct nvme_dev *dev, struct request *req)
 	__le64 **list = iod_list(req);
 	dma_addr_t prp_dma = iod->first_dma;
 
+	if (req->cmd_flags & REQ_DISCARD)
+		kfree(req->completion_data);
+
 	if (iod->npages == 0)
 		dma_pool_free(dev->prp_small_pool, list[0], prp_dma);
 	for (i = 0; i < iod->npages; i++) {

@@ -538,7 +541,7 @@ static bool nvme_setup_prps(struct nvme_dev *dev, struct request *req,
 }
 
 static int nvme_map_data(struct nvme_dev *dev, struct request *req,
-		struct nvme_command *cmnd)
+		unsigned size, struct nvme_command *cmnd)
 {
 	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
 	struct request_queue *q = req->q;

@@ -555,7 +558,7 @@ static int nvme_map_data(struct nvme_dev *dev, struct request *req,
 	if (!dma_map_sg(dev->dev, iod->sg, iod->nents, dma_dir))
 		goto out;
 
-	if (!nvme_setup_prps(dev, req, blk_rq_bytes(req)))
+	if (!nvme_setup_prps(dev, req, size))
 		goto out_unmap;
 
 	ret = BLK_MQ_RQ_QUEUE_ERROR;

@@ -604,35 +607,41 @@ static void nvme_unmap_data(struct nvme_dev *dev, struct request *req)
 	nvme_free_iod(dev, req);
 }
 
-/*
- * We reuse the small pool to allocate the 16-byte range here as it is not
- * worth having a special pool for these or additional cases to handle freeing
- * the iod.
- */
-static int nvme_setup_discard(struct nvme_queue *nvmeq, struct nvme_ns *ns,
-		struct request *req, struct nvme_command *cmnd)
+static inline int nvme_setup_discard(struct nvme_ns *ns, struct request *req,
+		struct nvme_command *cmnd)
 {
-	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
 	struct nvme_dsm_range *range;
+	struct page *page;
+	int offset;
+	unsigned int nr_bytes = blk_rq_bytes(req);
 
-	range = dma_pool_alloc(nvmeq->dev->prp_small_pool, GFP_ATOMIC,
-						&iod->first_dma);
+	range = kmalloc(sizeof(*range), GFP_ATOMIC);
 	if (!range)
 		return BLK_MQ_RQ_QUEUE_BUSY;
-	iod_list(req)[0] = (__le64 *)range;
-	iod->npages = 0;
 
 	range->cattr = cpu_to_le32(0);
-	range->nlb = cpu_to_le32(blk_rq_bytes(req) >> ns->lba_shift);
+	range->nlb = cpu_to_le32(nr_bytes >> ns->lba_shift);
 	range->slba = cpu_to_le64(nvme_block_nr(ns, blk_rq_pos(req)));
 
 	memset(cmnd, 0, sizeof(*cmnd));
 	cmnd->dsm.opcode = nvme_cmd_dsm;
 	cmnd->dsm.nsid = cpu_to_le32(ns->ns_id);
-	cmnd->dsm.prp1 = cpu_to_le64(iod->first_dma);
 	cmnd->dsm.nr = 0;
 	cmnd->dsm.attributes = cpu_to_le32(NVME_DSMGMT_AD);
-	return BLK_MQ_RQ_QUEUE_OK;
+
+	req->completion_data = range;
+	page = virt_to_page(range);
+	offset = offset_in_page(range);
+	blk_add_request_payload(req, page, offset, sizeof(*range));
+
+	/*
+	 * we set __data_len back to the size of the area to be discarded
+	 * on disk. This allows us to report completion on the full amount
+	 * of blocks described by the request.
+	 */
+	req->__data_len = nr_bytes;
+
+	return 0;
 }
 
 /*

@@ -667,19 +676,20 @@ static int nvme_queue_rq(struct blk_mq_hw_ctx *hctx,
 	if (ret)
 		return ret;
 
-	if (req->cmd_flags & REQ_DISCARD) {
-		ret = nvme_setup_discard(nvmeq, ns, req, &cmnd);
-	} else {
-		if (req->cmd_type == REQ_TYPE_DRV_PRIV)
-			memcpy(&cmnd, req->cmd, sizeof(cmnd));
-		else if (req->cmd_flags & REQ_FLUSH)
-			nvme_setup_flush(ns, &cmnd);
-		else
-			nvme_setup_rw(ns, req, &cmnd);
+	if (req->cmd_type == REQ_TYPE_DRV_PRIV)
+		memcpy(&cmnd, req->cmd, sizeof(cmnd));
+	else if (req->cmd_flags & REQ_FLUSH)
+		nvme_setup_flush(ns, &cmnd);
+	else if (req->cmd_flags & REQ_DISCARD)
+		ret = nvme_setup_discard(ns, req, &cmnd);
+	else
+		nvme_setup_rw(ns, req, &cmnd);
 
-		if (req->nr_phys_segments)
-			ret = nvme_map_data(dev, req, &cmnd);
-	}
+	if (ret)
+		goto out;
+
+	if (req->nr_phys_segments)
+		ret = nvme_map_data(dev, req, map_len, &cmnd);
 
 	if (ret)
 		goto out;

index ff9cf2b..dfb5b6e 100644
@@ -127,6 +127,111 @@ struct request *nvme_alloc_request(struct request_queue *q,
 	return req;
 }
 
+static inline void nvme_setup_flush(struct nvme_ns *ns,
+		struct nvme_command *cmnd)
+{
+	memset(cmnd, 0, sizeof(*cmnd));
+	cmnd->common.opcode = nvme_cmd_flush;
+	cmnd->common.nsid = cpu_to_le32(ns->ns_id);
+}
+
+static inline int nvme_setup_discard(struct nvme_ns *ns, struct request *req,
+		struct nvme_command *cmnd)
+{
+	struct nvme_dsm_range *range;
+	struct page *page;
+	int offset;
+	unsigned int nr_bytes = blk_rq_bytes(req);
+
+	range = kmalloc(sizeof(*range), GFP_ATOMIC);
+	if (!range)
+		return BLK_MQ_RQ_QUEUE_BUSY;
+
+	range->cattr = cpu_to_le32(0);
+	range->nlb = cpu_to_le32(nr_bytes >> ns->lba_shift);
+	range->slba = cpu_to_le64(nvme_block_nr(ns, blk_rq_pos(req)));
+
+	memset(cmnd, 0, sizeof(*cmnd));
+	cmnd->dsm.opcode = nvme_cmd_dsm;
+	cmnd->dsm.nsid = cpu_to_le32(ns->ns_id);
+	cmnd->dsm.nr = 0;
+	cmnd->dsm.attributes = cpu_to_le32(NVME_DSMGMT_AD);
+
+	req->completion_data = range;
+	page = virt_to_page(range);
+	offset = offset_in_page(range);
+	blk_add_request_payload(req, page, offset, sizeof(*range));
+
+	/*
+	 * we set __data_len back to the size of the area to be discarded
+	 * on disk. This allows us to report completion on the full amount
+	 * of blocks described by the request.
+	 */
+	req->__data_len = nr_bytes;
+
+	return 0;
+}
+
+static inline void nvme_setup_rw(struct nvme_ns *ns, struct request *req,
+		struct nvme_command *cmnd)
+{
+	u16 control = 0;
+	u32 dsmgmt = 0;
+
+	if (req->cmd_flags & REQ_FUA)
+		control |= NVME_RW_FUA;
+	if (req->cmd_flags & (REQ_FAILFAST_DEV | REQ_RAHEAD))
+		control |= NVME_RW_LR;
+
+	if (req->cmd_flags & REQ_RAHEAD)
+		dsmgmt |= NVME_RW_DSM_FREQ_PREFETCH;
+
+	memset(cmnd, 0, sizeof(*cmnd));
+	cmnd->rw.opcode = (rq_data_dir(req) ? nvme_cmd_write : nvme_cmd_read);
+	cmnd->rw.command_id = req->tag;
+	cmnd->rw.nsid = cpu_to_le32(ns->ns_id);
+	cmnd->rw.slba = cpu_to_le64(nvme_block_nr(ns, blk_rq_pos(req)));
+	cmnd->rw.length = cpu_to_le16((blk_rq_bytes(req) >> ns->lba_shift) - 1);
+
+	if (ns->ms) {
+		switch (ns->pi_type) {
+		case NVME_NS_DPS_PI_TYPE3:
+			control |= NVME_RW_PRINFO_PRCHK_GUARD;
+			break;
+		case NVME_NS_DPS_PI_TYPE1:
+		case NVME_NS_DPS_PI_TYPE2:
+			control |= NVME_RW_PRINFO_PRCHK_GUARD |
+					NVME_RW_PRINFO_PRCHK_REF;
+			cmnd->rw.reftag = cpu_to_le32(
+					nvme_block_nr(ns, blk_rq_pos(req)));
+			break;
+		}
+		if (!blk_integrity_rq(req))
+			control |= NVME_RW_PRINFO_PRACT;
+	}
+
+	cmnd->rw.control = cpu_to_le16(control);
+	cmnd->rw.dsmgmt = cpu_to_le32(dsmgmt);
+}
+
+int nvme_setup_cmd(struct nvme_ns *ns, struct request *req,
+		struct nvme_command *cmd)
+{
+	int ret = 0;
+
+	if (req->cmd_type == REQ_TYPE_DRV_PRIV)
+		memcpy(cmd, req->cmd, sizeof(*cmd));
+	else if (req->cmd_flags & REQ_FLUSH)
+		nvme_setup_flush(ns, cmd);
+	else if (req->cmd_flags & REQ_DISCARD)
+		ret = nvme_setup_discard(ns, req, cmd);
+	else
+		nvme_setup_rw(ns, req, cmd);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(nvme_setup_cmd);
+
 /*
  * Returns 0 on success.  If the result is negative, it's a Linux error code;
  * if the result is positive, it's an NVM Express status code

65 lines not on this page · raw message

59 lines not on this page · raw message