From: James Smart <[email protected]> For some nvme command, when issued by the nvme core layer, there is an internal buffer which can cause blk_rq_payload_bytes() to return a non-zero value yet there is no actual/real command payload and sg list. An example is the WRITE ZEROES command.
To address this, when making choices on whether to dma map an sgl, use blk_rq_nr_phys_segments() instead of blk_rq_payload_bytes(). When there is a sgl, blk_rq_payload_bytes() will return the amount of data to be transferred by the sgl. Signed-off-by: Chaitanya Kulkarni <[email protected]> Signed-off-by: James Smart <[email protected]> Signed-off-by: Christoph Hellwig <[email protected]> --- drivers/nvme/host/fc.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-) diff --git a/drivers/nvme/host/fc.c b/drivers/nvme/host/fc.c index b29b12498a1a..ba8f2a9cbdaf 100644 --- a/drivers/nvme/host/fc.c +++ b/drivers/nvme/host/fc.c @@ -2107,7 +2107,7 @@ nvme_fc_map_data(struct nvme_fc_ctrl *ctrl, struct request *rq, freq->sg_cnt = 0; - if (!blk_rq_payload_bytes(rq)) + if (!blk_rq_nr_phys_segments(rq)) return 0; freq->sg_table.sgl = freq->first_sgl; @@ -2304,12 +2304,23 @@ nvme_fc_queue_rq(struct blk_mq_hw_ctx *hctx, if (ret) return ret; - data_len = blk_rq_payload_bytes(rq); - if (data_len) + /* + * nvme core doesn't quite treat the rq opaquely. Commands such + * as WRITE ZEROES will return a non-zero rq payload_bytes yet + * there is no actual payload to be transferred. + * To get it right, key data transmission on there being 1 or + * more physical segments in the sg list. If there is no + * physical segments, there is no payload. + */ + if (blk_rq_nr_phys_segments(rq)) { + data_len = blk_rq_payload_bytes(rq); io_dir = ((rq_data_dir(rq) == WRITE) ? NVMEFC_FCP_WRITE : NVMEFC_FCP_READ); - else + } else { + data_len = 0; io_dir = NVMEFC_FCP_NODATA; + } + return nvme_fc_start_fcp_op(ctrl, queue, op, data_len, io_dir); } -- 2.20.1
