From 615399896ca3787728c56c499c99be79e40ac125 Mon Sep 17 00:00:00 2001 From: James Smart Date: Tue, 31 Mar 2020 09:49:46 -0700 Subject: nvme-fc: Sync header to FC-NVME-2 rev 1.08 A couple of minor changes occurred between 1.06 and 1.08: - Addition of NVME_SR_RSP opcode - change of SR_RSP status code 1 to Reserved Signed-off-by: James Smart Reviewed-by: Sagi Grimberg Reviewed-by: Himanshu Madhani Reviewed-by: Hannes Reinecke Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- include/linux/nvme-fc.h | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) (limited to 'include') diff --git a/include/linux/nvme-fc.h b/include/linux/nvme-fc.h index e8c30b39bb27..51fe44e0328b 100644 --- a/include/linux/nvme-fc.h +++ b/include/linux/nvme-fc.h @@ -4,8 +4,8 @@ */ /* - * This file contains definitions relative to FC-NVME-2 r1.06 - * (T11-2019-00210-v001). + * This file contains definitions relative to FC-NVME-2 r1.08 + * (T11-2019-00210-v004). */ #ifndef _NVME_FC_H @@ -81,7 +81,8 @@ struct nvme_fc_ersp_iu { }; -#define FCNVME_NVME_SR_OPCODE 0x01 +#define FCNVME_NVME_SR_OPCODE 0x01 +#define FCNVME_NVME_SR_RSP_OPCODE 0x02 struct nvme_fc_nvme_sr_iu { __u8 fc_id; @@ -94,7 +95,7 @@ struct nvme_fc_nvme_sr_iu { enum { FCNVME_SRSTAT_ACC = 0x0, - FCNVME_SRSTAT_INV_FCID = 0x1, + /* reserved 0x1 */ /* reserved 0x2 */ FCNVME_SRSTAT_LOGICAL_ERR = 0x3, FCNVME_SRSTAT_INV_QUALIF = 0x4, @@ -397,7 +398,7 @@ struct fcnvme_ls_disconnect_conn_rqst { struct fcnvme_ls_rqst_w0 w0; __be32 desc_list_len; struct fcnvme_lsdesc_assoc_id associd; - struct fcnvme_lsdesc_disconn_cmd connectid; + struct fcnvme_lsdesc_conn_id connectid; }; struct fcnvme_ls_disconnect_conn_acc { -- cgit v1.2.3 From 72e6329f86c714785ac195d293cb19dd24507880 Mon Sep 17 00:00:00 2001 From: James Smart Date: Tue, 31 Mar 2020 09:49:47 -0700 Subject: nvme-fc and nvmet-fc: revise LLDD api for LS reception and LS request The current LLDD api has: nvme-fc: contains api for transport to do LS requests (and aborts of them). However, there is no interface for reception of LS's and sending responses for them. nvmet-fc: contains api for transport to do reception of LS's and sending of responses for them. However, there is no interface for doing LS requests. Revise the api's so that both nvme-fc and nvmet-fc can send LS's, as well as receiving LS's and sending their responses. Change name of the rcv_ls_req struct to better reflect generic use as a context to used to send an ls rsp. Specifically: nvmefc_tgt_ls_req -> nvmefc_ls_rsp nvmefc_tgt_ls_req.nvmet_fc_private -> nvmefc_ls_rsp.nvme_fc_private Change nvmet_fc_rcv_ls_req() calling sequence to provide handle that can be used by transport in later LS request sequences for an association. nvme-fc nvmet_fc nvme_fcloop: Revise to adapt to changed names in api header. Change calling sequence to nvmet_fc_rcv_ls_req() for hosthandle. Add stubs for new interfaces: host/fc.c: nvme_fc_rcv_ls_req() target/fc.c: nvmet_fc_invalidate_host() lpfc: Revise to adapt code to changed names in api header. Change calling sequence to nvmet_fc_rcv_ls_req() for hosthandle. Signed-off-by: James Smart Reviewed-by: Sagi Grimberg Reviewed-by: Himanshu Madhani Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- drivers/nvme/host/fc.c | 35 ++++ drivers/nvme/target/fc.c | 77 ++++++--- drivers/nvme/target/fcloop.c | 20 +-- drivers/scsi/lpfc/lpfc_nvmet.c | 10 +- drivers/scsi/lpfc/lpfc_nvmet.h | 2 +- include/linux/nvme-fc-driver.h | 368 ++++++++++++++++++++++++++++++----------- 6 files changed, 378 insertions(+), 134 deletions(-) (limited to 'include') diff --git a/drivers/nvme/host/fc.c b/drivers/nvme/host/fc.c index 7dfc4a2ecf1e..8012099fc3ee 100644 --- a/drivers/nvme/host/fc.c +++ b/drivers/nvme/host/fc.c @@ -1464,6 +1464,41 @@ nvme_fc_xmt_disconnect_assoc(struct nvme_fc_ctrl *ctrl) kfree(lsop); } +/** + * nvme_fc_rcv_ls_req - transport entry point called by an LLDD + * upon the reception of a NVME LS request. + * + * The nvme-fc layer will copy payload to an internal structure for + * processing. As such, upon completion of the routine, the LLDD may + * immediately free/reuse the LS request buffer passed in the call. + * + * If this routine returns error, the LLDD should abort the exchange. + * + * @remoteport: pointer to the (registered) remote port that the LS + * was received from. The remoteport is associated with + * a specific localport. + * @lsrsp: pointer to a nvmefc_ls_rsp response structure to be + * used to reference the exchange corresponding to the LS + * when issuing an ls response. + * @lsreqbuf: pointer to the buffer containing the LS Request + * @lsreqbuf_len: length, in bytes, of the received LS request + */ +int +nvme_fc_rcv_ls_req(struct nvme_fc_remote_port *portptr, + struct nvmefc_ls_rsp *lsrsp, + void *lsreqbuf, u32 lsreqbuf_len) +{ + struct nvme_fc_rport *rport = remoteport_to_rport(portptr); + struct nvme_fc_lport *lport = rport->lport; + + /* validate there's a routine to transmit a response */ + if (!lport->ops->xmt_ls_rsp) + return(-EINVAL); + + return 0; +} +EXPORT_SYMBOL_GPL(nvme_fc_rcv_ls_req); + /* *********************** NVME Ctrl Routines **************************** */ diff --git a/drivers/nvme/target/fc.c b/drivers/nvme/target/fc.c index a8ceb7721640..aac7869a70bb 100644 --- a/drivers/nvme/target/fc.c +++ b/drivers/nvme/target/fc.c @@ -28,7 +28,7 @@ struct nvmet_fc_tgtport; struct nvmet_fc_tgt_assoc; struct nvmet_fc_ls_iod { - struct nvmefc_tgt_ls_req *lsreq; + struct nvmefc_ls_rsp *lsrsp; struct nvmefc_tgt_fcp_req *fcpreq; /* only if RS */ struct list_head ls_list; /* tgtport->ls_list */ @@ -1146,6 +1146,42 @@ __nvmet_fc_free_assocs(struct nvmet_fc_tgtport *tgtport) spin_unlock_irqrestore(&tgtport->lock, flags); } +/** + * nvmet_fc_invalidate_host - transport entry point called by an LLDD + * to remove references to a hosthandle for LS's. + * + * The nvmet-fc layer ensures that any references to the hosthandle + * on the targetport are forgotten (set to NULL). The LLDD will + * typically call this when a login with a remote host port has been + * lost, thus LS's for the remote host port are no longer possible. + * + * If an LS request is outstanding to the targetport/hosthandle (or + * issued concurrently with the call to invalidate the host), the + * LLDD is responsible for terminating/aborting the LS and completing + * the LS request. It is recommended that these terminations/aborts + * occur after calling to invalidate the host handle to avoid additional + * retries by the nvmet-fc transport. The nvmet-fc transport may + * continue to reference host handle while it cleans up outstanding + * NVME associations. The nvmet-fc transport will call the + * ops->host_release() callback to notify the LLDD that all references + * are complete and the related host handle can be recovered. + * Note: if there are no references, the callback may be called before + * the invalidate host call returns. + * + * @target_port: pointer to the (registered) target port that a prior + * LS was received on and which supplied the transport the + * hosthandle. + * @hosthandle: the handle (pointer) that represents the host port + * that no longer has connectivity and that LS's should + * no longer be directed to. + */ +void +nvmet_fc_invalidate_host(struct nvmet_fc_target_port *target_port, + void *hosthandle) +{ +} +EXPORT_SYMBOL_GPL(nvmet_fc_invalidate_host); + /* * nvmet layer has called to terminate an association */ @@ -1371,7 +1407,7 @@ nvmet_fc_ls_create_association(struct nvmet_fc_tgtport *tgtport, dev_err(tgtport->dev, "Create Association LS failed: %s\n", validation_errors[ret]); - iod->lsreq->rsplen = nvmet_fc_format_rjt(acc, + iod->lsrsp->rsplen = nvmet_fc_format_rjt(acc, NVME_FC_MAX_LS_BUFFER_SIZE, rqst->w0.ls_cmd, FCNVME_RJT_RC_LOGIC, FCNVME_RJT_EXP_NONE, 0); @@ -1384,7 +1420,7 @@ nvmet_fc_ls_create_association(struct nvmet_fc_tgtport *tgtport, /* format a response */ - iod->lsreq->rsplen = sizeof(*acc); + iod->lsrsp->rsplen = sizeof(*acc); nvmet_fc_format_rsp_hdr(acc, FCNVME_LS_ACC, fcnvme_lsdesc_len( @@ -1462,7 +1498,7 @@ nvmet_fc_ls_create_connection(struct nvmet_fc_tgtport *tgtport, dev_err(tgtport->dev, "Create Connection LS failed: %s\n", validation_errors[ret]); - iod->lsreq->rsplen = nvmet_fc_format_rjt(acc, + iod->lsrsp->rsplen = nvmet_fc_format_rjt(acc, NVME_FC_MAX_LS_BUFFER_SIZE, rqst->w0.ls_cmd, (ret == VERR_NO_ASSOC) ? FCNVME_RJT_RC_INV_ASSOC : @@ -1477,7 +1513,7 @@ nvmet_fc_ls_create_connection(struct nvmet_fc_tgtport *tgtport, /* format a response */ - iod->lsreq->rsplen = sizeof(*acc); + iod->lsrsp->rsplen = sizeof(*acc); nvmet_fc_format_rsp_hdr(acc, FCNVME_LS_ACC, fcnvme_lsdesc_len(sizeof(struct fcnvme_ls_cr_conn_acc)), @@ -1542,7 +1578,7 @@ nvmet_fc_ls_disconnect(struct nvmet_fc_tgtport *tgtport, dev_err(tgtport->dev, "Disconnect LS failed: %s\n", validation_errors[ret]); - iod->lsreq->rsplen = nvmet_fc_format_rjt(acc, + iod->lsrsp->rsplen = nvmet_fc_format_rjt(acc, NVME_FC_MAX_LS_BUFFER_SIZE, rqst->w0.ls_cmd, (ret == VERR_NO_ASSOC) ? FCNVME_RJT_RC_INV_ASSOC : @@ -1555,7 +1591,7 @@ nvmet_fc_ls_disconnect(struct nvmet_fc_tgtport *tgtport, /* format a response */ - iod->lsreq->rsplen = sizeof(*acc); + iod->lsrsp->rsplen = sizeof(*acc); nvmet_fc_format_rsp_hdr(acc, FCNVME_LS_ACC, fcnvme_lsdesc_len( @@ -1577,9 +1613,9 @@ static void nvmet_fc_fcp_nvme_cmd_done(struct nvmet_req *nvme_req); static const struct nvmet_fabrics_ops nvmet_fc_tgt_fcp_ops; static void -nvmet_fc_xmt_ls_rsp_done(struct nvmefc_tgt_ls_req *lsreq) +nvmet_fc_xmt_ls_rsp_done(struct nvmefc_ls_rsp *lsrsp) { - struct nvmet_fc_ls_iod *iod = lsreq->nvmet_fc_private; + struct nvmet_fc_ls_iod *iod = lsrsp->nvme_fc_private; struct nvmet_fc_tgtport *tgtport = iod->tgtport; fc_dma_sync_single_for_cpu(tgtport->dev, iod->rspdma, @@ -1597,9 +1633,9 @@ nvmet_fc_xmt_ls_rsp(struct nvmet_fc_tgtport *tgtport, fc_dma_sync_single_for_device(tgtport->dev, iod->rspdma, NVME_FC_MAX_LS_BUFFER_SIZE, DMA_TO_DEVICE); - ret = tgtport->ops->xmt_ls_rsp(&tgtport->fc_target_port, iod->lsreq); + ret = tgtport->ops->xmt_ls_rsp(&tgtport->fc_target_port, iod->lsrsp); if (ret) - nvmet_fc_xmt_ls_rsp_done(iod->lsreq); + nvmet_fc_xmt_ls_rsp_done(iod->lsrsp); } /* @@ -1612,12 +1648,12 @@ nvmet_fc_handle_ls_rqst(struct nvmet_fc_tgtport *tgtport, struct fcnvme_ls_rqst_w0 *w0 = (struct fcnvme_ls_rqst_w0 *)iod->rqstbuf; - iod->lsreq->nvmet_fc_private = iod; - iod->lsreq->rspbuf = iod->rspbuf; - iod->lsreq->rspdma = iod->rspdma; - iod->lsreq->done = nvmet_fc_xmt_ls_rsp_done; + iod->lsrsp->nvme_fc_private = iod; + iod->lsrsp->rspbuf = iod->rspbuf; + iod->lsrsp->rspdma = iod->rspdma; + iod->lsrsp->done = nvmet_fc_xmt_ls_rsp_done; /* Be preventative. handlers will later set to valid length */ - iod->lsreq->rsplen = 0; + iod->lsrsp->rsplen = 0; iod->assoc = NULL; @@ -1640,7 +1676,7 @@ nvmet_fc_handle_ls_rqst(struct nvmet_fc_tgtport *tgtport, nvmet_fc_ls_disconnect(tgtport, iod); break; default: - iod->lsreq->rsplen = nvmet_fc_format_rjt(iod->rspbuf, + iod->lsrsp->rsplen = nvmet_fc_format_rjt(iod->rspbuf, NVME_FC_MAX_LS_BUFFER_SIZE, w0->ls_cmd, FCNVME_RJT_RC_INVAL, FCNVME_RJT_EXP_NONE, 0); } @@ -1674,14 +1710,15 @@ nvmet_fc_handle_ls_rqst_work(struct work_struct *work) * * @target_port: pointer to the (registered) target port the LS was * received on. - * @lsreq: pointer to a lsreq request structure to be used to reference + * @lsrsp: pointer to a lsrsp structure to be used to reference * the exchange corresponding to the LS. * @lsreqbuf: pointer to the buffer containing the LS Request * @lsreqbuf_len: length, in bytes, of the received LS request */ int nvmet_fc_rcv_ls_req(struct nvmet_fc_target_port *target_port, - struct nvmefc_tgt_ls_req *lsreq, + void *hosthandle, + struct nvmefc_ls_rsp *lsrsp, void *lsreqbuf, u32 lsreqbuf_len) { struct nvmet_fc_tgtport *tgtport = targetport_to_tgtport(target_port); @@ -1699,7 +1736,7 @@ nvmet_fc_rcv_ls_req(struct nvmet_fc_target_port *target_port, return -ENOENT; } - iod->lsreq = lsreq; + iod->lsrsp = lsrsp; iod->fcpreq = NULL; memcpy(iod->rqstbuf, lsreqbuf, lsreqbuf_len); iod->rqstdatalen = lsreqbuf_len; diff --git a/drivers/nvme/target/fcloop.c b/drivers/nvme/target/fcloop.c index f69ce66e2d44..c11805a155e8 100644 --- a/drivers/nvme/target/fcloop.c +++ b/drivers/nvme/target/fcloop.c @@ -228,7 +228,7 @@ struct fcloop_nport { struct fcloop_lsreq { struct nvmefc_ls_req *lsreq; - struct nvmefc_tgt_ls_req tgt_ls_req; + struct nvmefc_ls_rsp ls_rsp; int status; struct list_head ls_list; /* fcloop_rport->ls_list */ }; @@ -267,9 +267,9 @@ struct fcloop_ini_fcpreq { }; static inline struct fcloop_lsreq * -tgt_ls_req_to_lsreq(struct nvmefc_tgt_ls_req *tgt_lsreq) +ls_rsp_to_lsreq(struct nvmefc_ls_rsp *lsrsp) { - return container_of(tgt_lsreq, struct fcloop_lsreq, tgt_ls_req); + return container_of(lsrsp, struct fcloop_lsreq, ls_rsp); } static inline struct fcloop_fcpreq * @@ -344,7 +344,7 @@ fcloop_ls_req(struct nvme_fc_local_port *localport, } tls_req->status = 0; - ret = nvmet_fc_rcv_ls_req(rport->targetport, &tls_req->tgt_ls_req, + ret = nvmet_fc_rcv_ls_req(rport->targetport, NULL, &tls_req->ls_rsp, lsreq->rqstaddr, lsreq->rqstlen); return ret; @@ -352,19 +352,19 @@ fcloop_ls_req(struct nvme_fc_local_port *localport, static int fcloop_xmt_ls_rsp(struct nvmet_fc_target_port *targetport, - struct nvmefc_tgt_ls_req *tgt_lsreq) + struct nvmefc_ls_rsp *lsrsp) { - struct fcloop_lsreq *tls_req = tgt_ls_req_to_lsreq(tgt_lsreq); + struct fcloop_lsreq *tls_req = ls_rsp_to_lsreq(lsrsp); struct nvmefc_ls_req *lsreq = tls_req->lsreq; struct fcloop_tport *tport = targetport->private; struct nvme_fc_remote_port *remoteport = tport->remoteport; struct fcloop_rport *rport; - memcpy(lsreq->rspaddr, tgt_lsreq->rspbuf, - ((lsreq->rsplen < tgt_lsreq->rsplen) ? - lsreq->rsplen : tgt_lsreq->rsplen)); + memcpy(lsreq->rspaddr, lsrsp->rspbuf, + ((lsreq->rsplen < lsrsp->rsplen) ? + lsreq->rsplen : lsrsp->rsplen)); - tgt_lsreq->done(tgt_lsreq); + lsrsp->done(lsrsp); if (remoteport) { rport = remoteport->private; diff --git a/drivers/scsi/lpfc/lpfc_nvmet.c b/drivers/scsi/lpfc/lpfc_nvmet.c index 565419bf8d74..3b25bcb150ea 100644 --- a/drivers/scsi/lpfc/lpfc_nvmet.c +++ b/drivers/scsi/lpfc/lpfc_nvmet.c @@ -302,7 +302,7 @@ lpfc_nvmet_xmt_ls_rsp_cmp(struct lpfc_hba *phba, struct lpfc_iocbq *cmdwqe, struct lpfc_wcqe_complete *wcqe) { struct lpfc_nvmet_tgtport *tgtp; - struct nvmefc_tgt_ls_req *rsp; + struct nvmefc_ls_rsp *rsp; struct lpfc_nvmet_rcv_ctx *ctxp; uint32_t status, result; @@ -335,7 +335,7 @@ lpfc_nvmet_xmt_ls_rsp_cmp(struct lpfc_hba *phba, struct lpfc_iocbq *cmdwqe, } out: - rsp = &ctxp->ctx.ls_req; + rsp = &ctxp->ctx.ls_rsp; lpfc_nvmeio_data(phba, "NVMET LS CMPL: xri x%x stat x%x result x%x\n", ctxp->oxid, status, result); @@ -828,10 +828,10 @@ lpfc_nvmet_xmt_fcp_op_cmp(struct lpfc_hba *phba, struct lpfc_iocbq *cmdwqe, static int lpfc_nvmet_xmt_ls_rsp(struct nvmet_fc_target_port *tgtport, - struct nvmefc_tgt_ls_req *rsp) + struct nvmefc_ls_rsp *rsp) { struct lpfc_nvmet_rcv_ctx *ctxp = - container_of(rsp, struct lpfc_nvmet_rcv_ctx, ctx.ls_req); + container_of(rsp, struct lpfc_nvmet_rcv_ctx, ctx.ls_rsp); struct lpfc_hba *phba = ctxp->phba; struct hbq_dmabuf *nvmebuf = (struct hbq_dmabuf *)ctxp->rqb_buffer; @@ -1999,7 +1999,7 @@ dropit: * lpfc_nvmet_xmt_ls_rsp_cmp should free the allocated ctxp. */ atomic_inc(&tgtp->rcv_ls_req_in); - rc = nvmet_fc_rcv_ls_req(phba->targetport, &ctxp->ctx.ls_req, + rc = nvmet_fc_rcv_ls_req(phba->targetport, NULL, &ctxp->ctx.ls_rsp, payload, size); lpfc_printf_log(phba, KERN_INFO, LOG_NVME_DISC, diff --git a/drivers/scsi/lpfc/lpfc_nvmet.h b/drivers/scsi/lpfc/lpfc_nvmet.h index b80b1639b9a7..f0196f3ef90d 100644 --- a/drivers/scsi/lpfc/lpfc_nvmet.h +++ b/drivers/scsi/lpfc/lpfc_nvmet.h @@ -105,7 +105,7 @@ struct lpfc_nvmet_ctx_info { struct lpfc_nvmet_rcv_ctx { union { - struct nvmefc_tgt_ls_req ls_req; + struct nvmefc_ls_rsp ls_rsp; struct nvmefc_tgt_fcp_req fcp_req; } ctx; struct list_head list; diff --git a/include/linux/nvme-fc-driver.h b/include/linux/nvme-fc-driver.h index 10f81629b9ce..41e7795a3ee4 100644 --- a/include/linux/nvme-fc-driver.h +++ b/include/linux/nvme-fc-driver.h @@ -10,47 +10,26 @@ /* - * ********************** LLDD FC-NVME Host API ******************** + * ********************** FC-NVME LS API ******************** * - * For FC LLDD's that are the NVME Host role. + * Data structures used by both FC-NVME hosts and FC-NVME + * targets to perform FC-NVME LS requests or transmit + * responses. * - * ****************************************************************** + * *********************************************************** */ - - /** - * struct nvme_fc_port_info - port-specific ids and FC connection-specific - * data element used during NVME Host role - * registrations - * - * Static fields describing the port being registered: - * @node_name: FC WWNN for the port - * @port_name: FC WWPN for the port - * @port_role: What NVME roles are supported (see FC_PORT_ROLE_xxx) - * @dev_loss_tmo: maximum delay for reconnects to an association on - * this device. Used only on a remoteport. + * struct nvmefc_ls_req - Request structure passed from the transport + * to the LLDD to perform a NVME-FC LS request and obtain + * a response. + * Used by nvme-fc transport (host) to send LS's such as + * Create Association, Create Connection and Disconnect + * Association. + * Used by the nvmet-fc transport (controller) to send + * LS's such as Disconnect Association. * - * Initialization values for dynamic port fields: - * @port_id: FC N_Port_ID currently assigned the port. Upper 8 bits must - * be set to 0. - */ -struct nvme_fc_port_info { - u64 node_name; - u64 port_name; - u32 port_role; - u32 port_id; - u32 dev_loss_tmo; -}; - - -/** - * struct nvmefc_ls_req - Request structure passed from NVME-FC transport - * to LLDD in order to perform a NVME FC-4 LS - * request and obtain a response. - * - * Values set by the NVME-FC layer prior to calling the LLDD ls_req - * entrypoint. + * Values set by the requestor prior to calling the LLDD ls_req entrypoint: * @rqstaddr: pointer to request buffer * @rqstdma: PCI DMA address of request buffer * @rqstlen: Length, in bytes, of request buffer @@ -63,8 +42,8 @@ struct nvme_fc_port_info { * @private: pointer to memory allocated alongside the ls request structure * that is specifically for the LLDD to use while processing the * request. The length of the buffer corresponds to the - * lsrqst_priv_sz value specified in the nvme_fc_port_template - * supplied by the LLDD. + * lsrqst_priv_sz value specified in the xxx_template supplied + * by the LLDD. * @done: The callback routine the LLDD is to invoke upon completion of * the LS request. req argument is the pointer to the original LS * request structure. Status argument must be 0 upon success, a @@ -86,6 +65,101 @@ struct nvmefc_ls_req { } __aligned(sizeof(u64)); /* alignment for other things alloc'd with */ +/** + * struct nvmefc_ls_rsp - Structure passed from the transport to the LLDD + * to request the transmit the NVME-FC LS response to a + * NVME-FC LS request. The structure originates in the LLDD + * and is given to the transport via the xxx_rcv_ls_req() + * transport routine. As such, the structure represents the + * FC exchange context for the NVME-FC LS request that was + * received and which the response is to be sent for. + * Used by the LLDD to pass the nvmet-fc transport (controller) + * received LS's such as Create Association, Create Connection + * and Disconnect Association. + * Used by the LLDD to pass the nvme-fc transport (host) + * received LS's such as Disconnect Association or Disconnect + * Connection. + * + * The structure is allocated by the LLDD whenever a LS Request is received + * from the FC link. The address of the structure is passed to the nvmet-fc + * or nvme-fc layer via the xxx_rcv_ls_req() transport routines. + * + * The address of the structure is to be passed back to the LLDD + * when the response is to be transmit. The LLDD will use the address to + * map back to the LLDD exchange structure which maintains information such + * the remote N_Port that sent the LS as well as any FC exchange context. + * Upon completion of the LS response transmit, the LLDD will pass the + * address of the structure back to the transport LS rsp done() routine, + * allowing the transport release dma resources. Upon completion of + * the done() routine, no further access to the structure will be made by + * the transport and the LLDD can de-allocate the structure. + * + * Field initialization: + * At the time of the xxx_rcv_ls_req() call, there is no content that + * is valid in the structure. + * + * When the structure is used for the LLDD->xmt_ls_rsp() call, the + * transport layer will fully set the fields in order to specify the + * response payload buffer and its length as well as the done routine + * to be called upon completion of the transmit. The transport layer + * will also set a private pointer for its own use in the done routine. + * + * Values set by the transport layer prior to calling the LLDD xmt_ls_rsp + * entrypoint: + * @rspbuf: pointer to the LS response buffer + * @rspdma: PCI DMA address of the LS response buffer + * @rsplen: Length, in bytes, of the LS response buffer + * @done: The callback routine the LLDD is to invoke upon completion of + * transmitting the LS response. req argument is the pointer to + * the original ls request. + * @nvme_fc_private: pointer to an internal transport-specific structure + * used as part of the transport done() processing. The LLDD is + * not to access this pointer. + */ +struct nvmefc_ls_rsp { + void *rspbuf; + dma_addr_t rspdma; + u16 rsplen; + + void (*done)(struct nvmefc_ls_rsp *rsp); + void *nvme_fc_private; /* LLDD is not to access !! */ +}; + + + +/* + * ********************** LLDD FC-NVME Host API ******************** + * + * For FC LLDD's that are the NVME Host role. + * + * ****************************************************************** + */ + + +/** + * struct nvme_fc_port_info - port-specific ids and FC connection-specific + * data element used during NVME Host role + * registrations + * + * Static fields describing the port being registered: + * @node_name: FC WWNN for the port + * @port_name: FC WWPN for the port + * @port_role: What NVME roles are supported (see FC_PORT_ROLE_xxx) + * @dev_loss_tmo: maximum delay for reconnects to an association on + * this device. Used only on a remoteport. + * + * Initialization values for dynamic port fields: + * @port_id: FC N_Port_ID currently assigned the port. Upper 8 bits must + * be set to 0. + */ +struct nvme_fc_port_info { + u64 node_name; + u64 port_name; + u32 port_role; + u32 port_id; + u32 dev_loss_tmo; +}; + enum nvmefc_fcp_datadir { NVMEFC_FCP_NODATA, /* payload_length and sg_cnt will be zero */ NVMEFC_FCP_WRITE, @@ -337,6 +411,21 @@ struct nvme_fc_remote_port { * indicating an FC transport Aborted status. * Entrypoint is Mandatory. * + * @xmt_ls_rsp: Called to transmit the response to a FC-NVME FC-4 LS service. + * The nvmefc_ls_rsp structure is the same LLDD-supplied exchange + * structure specified in the nvme_fc_rcv_ls_req() call made when + * the LS request was received. The structure will fully describe + * the buffers for the response payload and the dma address of the + * payload. The LLDD is to transmit the response (or return a + * non-zero errno status), and upon completion of the transmit, call + * the "done" routine specified in the nvmefc_ls_rsp structure + * (argument to done is the address of the nvmefc_ls_rsp structure + * itself). Upon the completion of the done routine, the LLDD shall + * consider the LS handling complete and the nvmefc_ls_rsp structure + * may be freed/released. + * Entrypoint is mandatory if the LLDD calls the nvme_fc_rcv_ls_req() + * entrypoint. + * * @max_hw_queues: indicates the maximum number of hw queues the LLDD * supports for cpu affinitization. * Value is Mandatory. Must be at least 1. @@ -371,7 +460,7 @@ struct nvme_fc_remote_port { * @lsrqst_priv_sz: The LLDD sets this field to the amount of additional * memory that it would like fc nvme layer to allocate on the LLDD's * behalf whenever a ls request structure is allocated. The additional - * memory area solely for the of the LLDD and its location is + * memory area is solely for use by the LLDD and its location is * specified by the ls_request->private pointer. * Value is Mandatory. Allowed to be zero. * @@ -405,6 +494,9 @@ struct nvme_fc_port_template { struct nvme_fc_remote_port *, void *hw_queue_handle, struct nvmefc_fcp_req *); + int (*xmt_ls_rsp)(struct nvme_fc_local_port *localport, + struct nvme_fc_remote_port *rport, + struct nvmefc_ls_rsp *ls_rsp); u32 max_hw_queues; u16 max_sgl_segments; @@ -441,6 +533,34 @@ void nvme_fc_rescan_remoteport(struct nvme_fc_remote_port *remoteport); int nvme_fc_set_remoteport_devloss(struct nvme_fc_remote_port *remoteport, u32 dev_loss_tmo); +/* + * Routine called to pass a NVME-FC LS request, received by the lldd, + * to the nvme-fc transport. + * + * If the return value is zero: the LS was successfully accepted by the + * transport. + * If the return value is non-zero: the transport has not accepted the + * LS. The lldd should ABTS-LS the LS. + * + * Note: if the LLDD receives and ABTS for the LS prior to the transport + * calling the ops->xmt_ls_rsp() routine to transmit a response, the LLDD + * shall mark the LS as aborted, and when the xmt_ls_rsp() is called: the + * response shall not be transmit and the struct nvmefc_ls_rsp() done + * routine shall be called. The LLDD may transmit the ABTS response as + * soon as the LS was marked or can delay until the xmt_ls_rsp() call is + * made. + * Note: if an RCV LS was successfully posted to the transport and the + * remoteport is then unregistered before xmt_ls_rsp() was called for + * the lsrsp structure, the transport will still call xmt_ls_rsp() + * afterward to cleanup the outstanding lsrsp structure. The LLDD should + * noop the transmission of the rsp and call the lsrsp->done() routine + * to allow the lsrsp structure to be released. + */ +int nvme_fc_rcv_ls_req(struct nvme_fc_remote_port *remoteport, + struct nvmefc_ls_rsp *lsrsp, + void *lsreqbuf, u32 lsreqbuf_len); + + /* * *************** LLDD FC-NVME Target/Subsystem API *************** @@ -470,55 +590,6 @@ struct nvmet_fc_port_info { }; -/** - * struct nvmefc_tgt_ls_req - Structure used between LLDD and NVMET-FC - * layer to represent the exchange context for - * a FC-NVME Link Service (LS). - * - * The structure is allocated by the LLDD whenever a LS Request is received - * from the FC link. The address of the structure is passed to the nvmet-fc - * layer via the nvmet_fc_rcv_ls_req() call. The address of the structure - * will be passed back to the LLDD when the response is to be transmit. - * The LLDD is to use the address to map back to the LLDD exchange structure - * which maintains information such as the targetport the LS was received - * on, the remote FC NVME initiator that sent the LS, and any FC exchange - * context. Upon completion of the LS response transmit, the address of the - * structure will be passed back to the LS rsp done() routine, allowing the - * nvmet-fc layer to release dma resources. Upon completion of the done() - * routine, no further access will be made by the nvmet-fc layer and the - * LLDD can de-allocate the structure. - * - * Field initialization: - * At the time of the nvmet_fc_rcv_ls_req() call, there is no content that - * is valid in the structure. - * - * When the structure is used for the LLDD->xmt_ls_rsp() call, the nvmet-fc - * layer will fully set the fields in order to specify the response - * payload buffer and its length as well as the done routine to be called - * upon compeletion of the transmit. The nvmet-fc layer will also set a - * private pointer for its own use in the done routine. - * - * Values set by the NVMET-FC layer prior to calling the LLDD xmt_ls_rsp - * entrypoint. - * @rspbuf: pointer to the LS response buffer - * @rspdma: PCI DMA address of the LS response buffer - * @rsplen: Length, in bytes, of the LS response buffer - * @done: The callback routine the LLDD is to invoke upon completion of - * transmitting the LS response. req argument is the pointer to - * the original ls request. - * @nvmet_fc_private: pointer to an internal NVMET-FC layer structure used - * as part of the NVMET-FC processing. The LLDD is not to access - * this pointer. - */ -struct nvmefc_tgt_ls_req { - void *rspbuf; - dma_addr_t rspdma; - u16 rsplen; - - void (*done)(struct nvmefc_tgt_ls_req *req); - void *nvmet_fc_private; /* LLDD is not to access !! */ -}; - /* Operations that NVME-FC layer may request the LLDD to perform for FCP */ enum { NVMET_FCOP_READDATA = 1, /* xmt data to initiator */ @@ -693,17 +764,19 @@ struct nvmet_fc_target_port { * Entrypoint is Mandatory. * * @xmt_ls_rsp: Called to transmit the response to a FC-NVME FC-4 LS service. - * The nvmefc_tgt_ls_req structure is the same LLDD-supplied exchange + * The nvmefc_ls_rsp structure is the same LLDD-supplied exchange * structure specified in the nvmet_fc_rcv_ls_req() call made when - * the LS request was received. The structure will fully describe + * the LS request was received. The structure will fully describe * the buffers for the response payload and the dma address of the - * payload. The LLDD is to transmit the response (or return a non-zero - * errno status), and upon completion of the transmit, call the - * "done" routine specified in the nvmefc_tgt_ls_req structure - * (argument to done is the ls reqwuest structure itself). - * After calling the done routine, the LLDD shall consider the - * LS handling complete and the nvmefc_tgt_ls_req structure may - * be freed/released. + * payload. The LLDD is to transmit the response (or return a + * non-zero errno status), and upon completion of the transmit, call + * the "done" routine specified in the nvmefc_ls_rsp structure + * (argument to done is the address of the nvmefc_ls_rsp structure + * itself). Upon the completion of the done() routine, the LLDD shall + * consider the LS handling complete and the nvmefc_ls_rsp structure + * may be freed/released. + * The transport will always call the xmt_ls_rsp() routine for any + * LS received. * Entrypoint is Mandatory. * * @fcp_op: Called to perform a data transfer or transmit a response. @@ -798,6 +871,39 @@ struct nvmet_fc_target_port { * should cause the initiator to rescan the discovery controller * on the targetport. * + * @ls_req: Called to issue a FC-NVME FC-4 LS service request. + * The nvme_fc_ls_req structure will fully describe the buffers for + * the request payload and where to place the response payload. + * The targetport that is to issue the LS request is identified by + * the targetport argument. The remote port that is to receive the + * LS request is identified by the hosthandle argument. The nvmet-fc + * transport is only allowed to issue FC-NVME LS's on behalf of an + * association that was created prior by a Create Association LS. + * The hosthandle will originate from the LLDD in the struct + * nvmefc_ls_rsp structure for the Create Association LS that + * was delivered to the transport. The transport will save the + * hosthandle as an attribute of the association. If the LLDD + * loses connectivity with the remote port, it must call the + * nvmet_fc_invalidate_host() routine to remove any references to + * the remote port in the transport. + * The LLDD is to allocate an exchange, issue the LS request, obtain + * the LS response, and call the "done" routine specified in the + * request structure (argument to done is the ls request structure + * itself). + * Entrypoint is Optional - but highly recommended. + * + * @ls_abort: called to request the LLDD to abort the indicated ls request. + * The call may return before the abort has completed. After aborting + * the request, the LLDD must still call the ls request done routine + * indicating an FC transport Aborted status. + * Entrypoint is Mandatory if the ls_req entry point is specified. + * + * @host_release: called to inform the LLDD that the request to invalidate + * the host port indicated by the hosthandle has been fully completed. + * No associations exist with the host port and there will be no + * further references to hosthandle. + * Entrypoint is Mandatory if the lldd calls nvmet_fc_invalidate_host(). + * * @max_hw_queues: indicates the maximum number of hw queues the LLDD * supports for cpu affinitization. * Value is Mandatory. Must be at least 1. @@ -826,11 +932,19 @@ struct nvmet_fc_target_port { * area solely for the of the LLDD and its location is specified by * the targetport->private pointer. * Value is Mandatory. Allowed to be zero. + * + * @lsrqst_priv_sz: The LLDD sets this field to the amount of additional + * memory that it would like nvmet-fc layer to allocate on the LLDD's + * behalf whenever a ls request structure is allocated. The additional + * memory area is solely for use by the LLDD and its location is + * specified by the ls_request->private pointer. + * Value is Mandatory. Allowed to be zero. + * */ struct nvmet_fc_target_template { void (*targetport_delete)(struct nvmet_fc_target_port *tgtport); int (*xmt_ls_rsp)(struct nvmet_fc_target_port *tgtport, - struct nvmefc_tgt_ls_req *tls_req); + struct nvmefc_ls_rsp *ls_rsp); int (*fcp_op)(struct nvmet_fc_target_port *tgtport, struct nvmefc_tgt_fcp_req *fcpreq); void (*fcp_abort)(struct nvmet_fc_target_port *tgtport, @@ -840,6 +954,11 @@ struct nvmet_fc_target_template { void (*defer_rcv)(struct nvmet_fc_target_port *tgtport, struct nvmefc_tgt_fcp_req *fcpreq); void (*discovery_event)(struct nvmet_fc_target_port *tgtport); + int (*ls_req)(struct nvmet_fc_target_port *targetport, + void *hosthandle, struct nvmefc_ls_req *lsreq); + void (*ls_abort)(struct nvmet_fc_target_port *targetport, + void *hosthandle, struct nvmefc_ls_req *lsreq); + void (*host_release)(void *hosthandle); u32 max_hw_queues; u16 max_sgl_segments; @@ -848,7 +967,9 @@ struct nvmet_fc_target_template { u32 target_features; + /* sizes of additional private data for data structures */ u32 target_priv_sz; + u32 lsrqst_priv_sz; }; @@ -859,10 +980,61 @@ int nvmet_fc_register_targetport(struct nvmet_fc_port_info *portinfo, int nvmet_fc_unregister_targetport(struct nvmet_fc_target_port *tgtport); +/* + * Routine called to pass a NVME-FC LS request, received by the lldd, + * to the nvmet-fc transport. + * + * If the return value is zero: the LS was successfully accepted by the + * transport. + * If the return value is non-zero: the transport has not accepted the + * LS. The lldd should ABTS-LS the LS. + * + * Note: if the LLDD receives and ABTS for the LS prior to the transport + * calling the ops->xmt_ls_rsp() routine to transmit a response, the LLDD + * shall mark the LS as aborted, and when the xmt_ls_rsp() is called: the + * response shall not be transmit and the struct nvmefc_ls_rsp() done + * routine shall be called. The LLDD may transmit the ABTS response as + * soon as the LS was marked or can delay until the xmt_ls_rsp() call is + * made. + * Note: if an RCV LS was successfully posted to the transport and the + * targetport is then unregistered before xmt_ls_rsp() was called for + * the lsrsp structure, the transport will still call xmt_ls_rsp() + * afterward to cleanup the outstanding lsrsp structure. The LLDD should + * noop the transmission of the rsp and call the lsrsp->done() routine + * to allow the lsrsp structure to be released. + */ int nvmet_fc_rcv_ls_req(struct nvmet_fc_target_port *tgtport, - struct nvmefc_tgt_ls_req *lsreq, + void *hosthandle, + struct nvmefc_ls_rsp *rsp, void *lsreqbuf, u32 lsreqbuf_len); +/* + * Routine called by the LLDD whenever it has a logout or loss of + * connectivity to a NVME-FC host port which there had been active + * NVMe controllers for. The host port is indicated by the + * hosthandle. The hosthandle is given to the nvmet-fc transport + * when a NVME LS was received, typically to create a new association. + * The nvmet-fc transport will cache the hostport value with the + * association for use in LS requests for the association. + * When the LLDD calls this routine, the nvmet-fc transport will + * immediately terminate all associations that were created with + * the hosthandle host port. + * The LLDD, after calling this routine and having control returned, + * must assume the transport may subsequently utilize hosthandle as + * part of sending LS's to terminate the association. The LLDD + * should reject the LS's if they are attempted. + * Once the last association has terminated for the hosthandle host + * port, the nvmet-fc transport will call the ops->host_release() + * callback. As of the callback, the nvmet-fc transport will no + * longer reference hosthandle. + */ +void nvmet_fc_invalidate_host(struct nvmet_fc_target_port *tgtport, + void *hosthandle); + +/* + * If nvmet_fc_rcv_fcp_req returns non-zero, the transport has not accepted + * the FCP cmd. The lldd should ABTS-LS the cmd. + */ int nvmet_fc_rcv_fcp_req(struct nvmet_fc_target_port *tgtport, struct nvmefc_tgt_fcp_req *fcpreq, void *cmdiubuf, u32 cmdiubuf_len); -- cgit v1.2.3 From d02abd198633a4c40411b9a5994111452720470d Mon Sep 17 00:00:00 2001 From: Chaitanya Kulkarni Date: Mon, 4 May 2020 01:56:48 -0700 Subject: nvmet: align addrfam list to spec With reference to the NVMeOF Specification (page 44, Figure 38) discovery log page entry provides address family field. We do set the transport type field but the adrfam field is not set when using loop transport and also it doesn't have support in the nvme-cli. So when reading discovery log page with a loop transport it leads to confusing output. As per the spec for adrfam value 254 is reserved for Intra Host Transport i.e. loopback), we add a required macro in the protocol header file, set default port disc addr entry's adrfam to NVMF_ADDR_FAMILY_MAX, and update nvmet_addr_family configfs array for show/store attribute. Without this patch, setting adrfam to (ipv4/ipv6/ib/fc/loop/" ") we get following output for nvme discover command from nvme-cli which is confusing. trtype: loop adrfam: ipv4 trtype: loop adrfam: ipv6 trtype: loop adrfam: infiniband trtype: loop adrfam: fibre-channel trtype: loop # ${CFGFS_HOME}/nvmet/ports/1/addr_adrfam = loop adrfam: pci # <----- pci for loop trtype: loop # ${CFGFS_HOME}/nvmet/ports/1/addr_adrfam = " " adrfam: pci # <----- pci for unrecognized This patch fixes above output :- trtype: loop adrfam: ipv4 trtype: loop adrfam: ipv6 trtype: loop adrfam: infiniband trtype: loop adrfam: fibre-channel trtype: loop # ${CFGFS_HOME}/nvmet/ports/1/addr_adrfam = loop adrfam: loop # <----- loop for loop trtype: loop # ${CFGFS_HOME}/config/nvmet/ports/adrfam = " " adrfam: unrecognized # <----- unrecognized when invalid value Signed-off-by: Chaitanya Kulkarni Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- drivers/nvme/target/configfs.c | 14 ++++++++------ include/linux/nvme.h | 2 ++ 2 files changed, 10 insertions(+), 6 deletions(-) (limited to 'include') diff --git a/drivers/nvme/target/configfs.c b/drivers/nvme/target/configfs.c index 20779587eefe..ae8fb4489a10 100644 --- a/drivers/nvme/target/configfs.c +++ b/drivers/nvme/target/configfs.c @@ -33,11 +33,12 @@ static struct nvmet_type_name_map nvmet_transport[] = { }; static const struct nvmet_type_name_map nvmet_addr_family[] = { - { NVMF_ADDR_FAMILY_PCI, "pcie" }, - { NVMF_ADDR_FAMILY_IP4, "ipv4" }, - { NVMF_ADDR_FAMILY_IP6, "ipv6" }, - { NVMF_ADDR_FAMILY_IB, "ib" }, - { NVMF_ADDR_FAMILY_FC, "fc" }, + { NVMF_ADDR_FAMILY_PCI, "pcie" }, + { NVMF_ADDR_FAMILY_IP4, "ipv4" }, + { NVMF_ADDR_FAMILY_IP6, "ipv6" }, + { NVMF_ADDR_FAMILY_IB, "ib" }, + { NVMF_ADDR_FAMILY_FC, "fc" }, + { NVMF_ADDR_FAMILY_LOOP, "loop" }, }; static bool nvmet_is_port_enabled(struct nvmet_port *p, const char *caller) @@ -83,7 +84,7 @@ static ssize_t nvmet_addr_adrfam_store(struct config_item *item, return -EINVAL; found: - port->disc_addr.adrfam = i; + port->disc_addr.adrfam = nvmet_addr_family[i].type; return count; } @@ -1338,6 +1339,7 @@ static struct config_group *nvmet_ports_make(struct config_group *group, port->inline_data_size = -1; /* < 0 == let the transport choose */ port->disc_addr.portid = cpu_to_le16(portid); + port->disc_addr.adrfam = NVMF_ADDR_FAMILY_MAX; port->disc_addr.treq = NVMF_TREQ_DISABLE_SQFLOW; config_group_init_type_name(&port->group, name, &nvmet_port_type); diff --git a/include/linux/nvme.h b/include/linux/nvme.h index 3d5189f46cb1..2d978d0cbde6 100644 --- a/include/linux/nvme.h +++ b/include/linux/nvme.h @@ -38,6 +38,8 @@ enum { NVMF_ADDR_FAMILY_IP6 = 2, /* IP6 */ NVMF_ADDR_FAMILY_IB = 3, /* InfiniBand */ NVMF_ADDR_FAMILY_FC = 4, /* Fibre Channel */ + NVMF_ADDR_FAMILY_LOOP = 254, /* Reserved for host usage */ + NVMF_ADDR_FAMILY_MAX, }; /* Transport Type codes for Discovery Log Page entry TRTYPE field */ -- cgit v1.2.3 From 92decf118f1da4c866515f80387f9cf4d48611d6 Mon Sep 17 00:00:00 2001 From: Keith Busch Date: Fri, 3 Apr 2020 10:53:46 -0700 Subject: nvme: define constants for identification values Improve code readability by defining the specification's constants that the driver is using when decoding identification payloads. Signed-off-by: Keith Busch Reviewed-by: Bart van Assche Reviewed-by: Chaitanya Kulkarni Acked-by: Sagi Grimberg Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- drivers/nvme/host/core.c | 10 +++++----- drivers/nvme/host/multipath.c | 5 +++-- include/linux/nvme.h | 6 ++++++ 3 files changed, 14 insertions(+), 7 deletions(-) (limited to 'include') diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c index 4a124a28118f..805d289e6cd9 100644 --- a/drivers/nvme/host/core.c +++ b/drivers/nvme/host/core.c @@ -1859,13 +1859,13 @@ static void nvme_update_disk_info(struct gendisk *disk, * and whether it should be used instead of AWUPF. If NAWUPF == * 0 then AWUPF must be used instead. */ - if (id->nsfeat & (1 << 1) && id->nawupf) + if (id->nsfeat & NVME_NS_FEAT_ATOMICS && id->nawupf) atomic_bs = (1 + le16_to_cpu(id->nawupf)) * bs; else atomic_bs = (1 + ns->ctrl->subsys->awupf) * bs; } - if (id->nsfeat & (1 << 4)) { + if (id->nsfeat & NVME_NS_FEAT_IO_OPT) { /* NPWG = Namespace Preferred Write Granularity */ phys_bs = bs * (1 + le16_to_cpu(id->npwg)); /* NOWS = Namespace Optimal Write Size */ @@ -1894,7 +1894,7 @@ static void nvme_update_disk_info(struct gendisk *disk, nvme_config_discard(disk, ns); nvme_config_write_zeroes(disk, ns); - if (id->nsattr & (1 << 0)) + if (id->nsattr & NVME_NS_ATTR_RO) set_disk_ro(disk, true); else set_disk_ro(disk, false); @@ -2685,7 +2685,7 @@ static bool nvme_validate_cntlid(struct nvme_subsystem *subsys, return false; } - if ((id->cmic & (1 << 1)) || + if ((id->cmic & NVME_CTRL_CMIC_MULTI_CTRL) || (ctrl->opts && ctrl->opts->discovery_nqn)) continue; @@ -3497,7 +3497,7 @@ static int nvme_init_ns_head(struct nvme_ns *ns, unsigned nsid, struct nvme_id_ns *id) { struct nvme_ctrl *ctrl = ns->ctrl; - bool is_shared = id->nmic & (1 << 0); + bool is_shared = id->nmic & NVME_NS_NMIC_SHARED; struct nvme_ns_head *head = NULL; struct nvme_ns_ids ids; int ret = 0; diff --git a/drivers/nvme/host/multipath.c b/drivers/nvme/host/multipath.c index 5f5537d0a024..da78e499947a 100644 --- a/drivers/nvme/host/multipath.c +++ b/drivers/nvme/host/multipath.c @@ -372,7 +372,7 @@ int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl, struct nvme_ns_head *head) * We also do this for private namespaces as the namespace sharing data could * change after a rescan. */ - if (!(ctrl->subsys->cmic & (1 << 1)) || !multipath) + if (!(ctrl->subsys->cmic & NVME_CTRL_CMIC_MULTI_CTRL) || !multipath) return 0; q = blk_alloc_queue(nvme_ns_head_make_request, ctrl->numa_node); @@ -694,7 +694,8 @@ int nvme_mpath_init(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id) int error; /* check if multipath is enabled and we have the capability */ - if (!multipath || !ctrl->subsys || !(ctrl->subsys->cmic & (1 << 3))) + if (!multipath || !ctrl->subsys || + !(ctrl->subsys->cmic & NVME_CTRL_CMIC_ANA)) return 0; ctrl->anacap = id->anacap; diff --git a/include/linux/nvme.h b/include/linux/nvme.h index 2d978d0cbde6..b235a48eac8c 100644 --- a/include/linux/nvme.h +++ b/include/linux/nvme.h @@ -301,6 +301,8 @@ struct nvme_id_ctrl { }; enum { + NVME_CTRL_CMIC_MULTI_CTRL = 1 << 1, + NVME_CTRL_CMIC_ANA = 1 << 3, NVME_CTRL_ONCS_COMPARE = 1 << 0, NVME_CTRL_ONCS_WRITE_UNCORRECTABLE = 1 << 1, NVME_CTRL_ONCS_DSM = 1 << 2, @@ -396,8 +398,12 @@ enum { enum { NVME_NS_FEAT_THIN = 1 << 0, + NVME_NS_FEAT_ATOMICS = 1 << 1, + NVME_NS_FEAT_IO_OPT = 1 << 4, + NVME_NS_ATTR_RO = 1 << 0, NVME_NS_FLBAS_LBA_MASK = 0xf, NVME_NS_FLBAS_META_EXT = 0x10, + NVME_NS_NMIC_SHARED = 1 << 0, NVME_LBAF_RP_BEST = 0, NVME_LBAF_RP_BETTER = 1, NVME_LBAF_RP_GOOD = 2, -- cgit v1.2.3 From 7d33850abdb9048c6aa421440a64905eb4ad07a2 Mon Sep 17 00:00:00 2001 From: Willy Tarreau Date: Tue, 31 Mar 2020 11:40:33 +0200 Subject: floppy: add references to 82077's extra registers This controller provides extra status registers SRA and SRB as well as a tape drive register (TDR) and a data rate select register (DSR), which are referenced in the sparc port, so let's have their symbolic definitions centralized. Link: https://lore.kernel.org/r/20200331094054.24441-3-w@1wt.eu Signed-off-by: Willy Tarreau Signed-off-by: Denis Efremov --- include/uapi/linux/fdreg.h | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) (limited to 'include') diff --git a/include/uapi/linux/fdreg.h b/include/uapi/linux/fdreg.h index 1318881954e1..10d33632939d 100644 --- a/include/uapi/linux/fdreg.h +++ b/include/uapi/linux/fdreg.h @@ -7,13 +7,23 @@ * Handbook", Sanches and Canton. */ -/* Fd controller regs. S&C, about page 340 */ -#define FD_STATUS 4 -#define FD_DATA 5 +/* 82077's auxiliary status registers A & B (R) */ +#define FD_SRA 0 +#define FD_SRB 1 /* Digital Output Register */ #define FD_DOR 2 +/* 82077's tape drive register (R/W) */ +#define FD_TDR 3 + +/* 82077's data rate select register (W) */ +#define FD_DSR 4 + +/* Fd controller regs. S&C, about page 340 */ +#define FD_STATUS 4 +#define FD_DATA 5 + /* Digital Input Register (read) */ #define FD_DIR 7 -- cgit v1.2.3 From 9c4c5a24c85585fb8904bd2872501cd8181b3854 Mon Sep 17 00:00:00 2001 From: Denis Efremov Date: Fri, 1 May 2020 16:44:14 +0300 Subject: floppy: add FD_AUTODETECT_SIZE define for struct floppy_drive_params Use FD_AUTODETECT_SIZE for autodetect buffer size in struct floppy_drive_params instead of a magic number. Link: https://lore.kernel.org/r/20200501134416.72248-3-efremov@linux.com Reviewed-by: Christoph Hellwig Signed-off-by: Denis Efremov --- drivers/block/floppy.c | 9 +++++---- include/uapi/linux/fd.h | 5 ++++- 2 files changed, 9 insertions(+), 5 deletions(-) (limited to 'include') diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c index 3ab6e804b5ec..b82b3d38b834 100644 --- a/drivers/block/floppy.c +++ b/drivers/block/floppy.c @@ -2073,7 +2073,8 @@ static int next_valid_format(int drive) probed_format = drive_state[drive].probed_format; while (1) { - if (probed_format >= 8 || !drive_params[drive].autodetect[probed_format]) { + if (probed_format >= FD_AUTODETECT_SIZE || + !drive_params[drive].autodetect[probed_format]) { drive_state[drive].probed_format = 0; return 1; } @@ -3442,13 +3443,13 @@ static int fd_getgeo(struct block_device *bdev, struct hd_geometry *geo) return 0; } -static bool valid_floppy_drive_params(const short autodetect[8], +static bool valid_floppy_drive_params(const short autodetect[FD_AUTODETECT_SIZE], int native_format) { size_t floppy_type_size = ARRAY_SIZE(floppy_type); size_t i = 0; - for (i = 0; i < 8; ++i) { + for (i = 0; i < FD_AUTODETECT_SIZE; ++i) { if (autodetect[i] < 0 || autodetect[i] >= floppy_type_size) return false; @@ -3673,7 +3674,7 @@ struct compat_floppy_drive_params { struct floppy_max_errors max_errors; char flags; char read_track; - short autodetect[8]; + short autodetect[FD_AUTODETECT_SIZE]; compat_int_t checkfreq; compat_int_t native_format; }; diff --git a/include/uapi/linux/fd.h b/include/uapi/linux/fd.h index 90fb94712c41..3f6b7be4c096 100644 --- a/include/uapi/linux/fd.h +++ b/include/uapi/linux/fd.h @@ -172,7 +172,10 @@ struct floppy_drive_params { * used in succession to try to read the disk. If the FDC cannot lock onto * the disk, the next format is tried. This uses the variable 'probing'. */ - short autodetect[8]; /* autodetected formats */ + +#define FD_AUTODETECT_SIZE 8 + + short autodetect[FD_AUTODETECT_SIZE]; /* autodetected formats */ int checkfreq; /* how often should the drive be checked for disk * changes */ -- cgit v1.2.3 From bd10a5f3e21b1cb8e2133c1f08b3e8207cee12dd Mon Sep 17 00:00:00 2001 From: Denis Efremov Date: Fri, 1 May 2020 16:44:15 +0300 Subject: floppy: add defines for sizes of cmd & reply buffers of floppy_raw_cmd Use FD_RAW_CMD_SIZE, FD_RAW_REPLY_SIZE defines instead of magic numbers for cmd & reply buffers of struct floppy_raw_cmd. Remove local to floppy.c MAX_REPLIES define, as it is now FD_RAW_REPLY_SIZE. FD_RAW_CMD_FULLSIZE added as we allow command to also fill reply_count and reply fields. Link: https://lore.kernel.org/r/20200501134416.72248-4-efremov@linux.com Reviewed-by: Christoph Hellwig Signed-off-by: Denis Efremov --- drivers/block/floppy.c | 19 +++++-------------- include/uapi/linux/fd.h | 14 ++++++++++++-- 2 files changed, 17 insertions(+), 16 deletions(-) (limited to 'include') diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c index b82b3d38b834..9e098d53b046 100644 --- a/drivers/block/floppy.c +++ b/drivers/block/floppy.c @@ -337,8 +337,7 @@ static bool initialized; /* * globals used by 'result()' */ -#define MAX_REPLIES 16 -static unsigned char reply_buffer[MAX_REPLIES]; +static unsigned char reply_buffer[FD_RAW_REPLY_SIZE]; static int inr; /* size of reply buffer, when called from interrupt */ #define ST0 0 #define ST1 1 @@ -1162,7 +1161,7 @@ static int result(int fdc) int i; int status = 0; - for (i = 0; i < MAX_REPLIES; i++) { + for (i = 0; i < FD_RAW_REPLY_SIZE; i++) { status = wait_til_ready(fdc); if (status < 0) break; @@ -3079,7 +3078,7 @@ static void raw_cmd_done(int flag) raw_cmd->flags |= FD_RAW_HARDFAILURE; } else { raw_cmd->reply_count = inr; - if (raw_cmd->reply_count > MAX_REPLIES) + if (raw_cmd->reply_count > FD_RAW_REPLY_SIZE) raw_cmd->reply_count = 0; for (i = 0; i < raw_cmd->reply_count; i++) raw_cmd->reply[i] = reply_buffer[i]; @@ -3190,18 +3189,10 @@ loop: if (ret) return -EFAULT; param += sizeof(struct floppy_raw_cmd); - if (ptr->cmd_count > 33) - /* the command may now also take up the space - * initially intended for the reply & the - * reply count. Needed for long 82078 commands - * such as RESTORE, which takes ... 17 command - * bytes. Murphy's law #137: When you reserve - * 16 bytes for a structure, you'll one day - * discover that you really need 17... - */ + if (ptr->cmd_count > FD_RAW_CMD_FULLSIZE) return -EINVAL; - for (i = 0; i < 16; i++) + for (i = 0; i < FD_RAW_REPLY_SIZE; i++) ptr->reply[i] = 0; ptr->resultcode = 0; diff --git a/include/uapi/linux/fd.h b/include/uapi/linux/fd.h index 3f6b7be4c096..2e9c2c1c18e6 100644 --- a/include/uapi/linux/fd.h +++ b/include/uapi/linux/fd.h @@ -360,10 +360,20 @@ struct floppy_raw_cmd { int buffer_length; /* length of allocated buffer */ unsigned char rate; + +#define FD_RAW_CMD_SIZE 16 +#define FD_RAW_REPLY_SIZE 16 +#define FD_RAW_CMD_FULLSIZE (FD_RAW_CMD_SIZE + 1 + FD_RAW_REPLY_SIZE) + + /* The command may take up the space initially intended for the reply + * and the reply count. Needed for long 82078 commands such as RESTORE, + * which takes 17 command bytes. + */ + unsigned char cmd_count; - unsigned char cmd[16]; + unsigned char cmd[FD_RAW_CMD_SIZE]; unsigned char reply_count; - unsigned char reply[16]; + unsigned char reply[FD_RAW_REPLY_SIZE]; int track; int resultcode; -- cgit v1.2.3 From 0836275df4db20daf040fff5d9a1da89c4c08a85 Mon Sep 17 00:00:00 2001 From: Denis Efremov Date: Fri, 1 May 2020 16:44:16 +0300 Subject: floppy: suppress UBSAN warning in setup_rw_floppy() UBSAN: array-index-out-of-bounds in drivers/block/floppy.c:1521:45 index 16 is out of range for type 'unsigned char [16]' Call Trace: ... setup_rw_floppy+0x5c3/0x7f0 floppy_ready+0x2be/0x13b0 process_one_work+0x2c1/0x5d0 worker_thread+0x56/0x5e0 kthread+0x122/0x170 ret_from_fork+0x35/0x40 From include/uapi/linux/fd.h: struct floppy_raw_cmd { ... unsigned char cmd_count; unsigned char cmd[16]; unsigned char reply_count; unsigned char reply[16]; ... } This out-of-bounds access is intentional. The command in struct floppy_raw_cmd may take up the space initially intended for the reply and the reply count. It is needed for long 82078 commands such as RESTORE, which takes 17 command bytes. Initial cmd size is not enough and since struct setup_rw_floppy is a part of uapi we check that cmd_count is in [0:16+1+16] in raw_cmd_copyin(). The patch adds union with original cmd,reply_count,reply fields and fullcmd field of equivalent size. The cmd accesses are turned to fullcmd where appropriate to suppress UBSAN warning. Link: https://lore.kernel.org/r/20200501134416.72248-5-efremov@linux.com Reviewed-by: Christoph Hellwig Signed-off-by: Denis Efremov --- drivers/block/floppy.c | 4 ++-- include/uapi/linux/fd.h | 11 ++++++++--- 2 files changed, 10 insertions(+), 5 deletions(-) (limited to 'include') diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c index 9e098d53b046..064c1acb9f00 100644 --- a/drivers/block/floppy.c +++ b/drivers/block/floppy.c @@ -1070,7 +1070,7 @@ static void setup_DMA(void) if (raw_cmd->length == 0) { print_hex_dump(KERN_INFO, "zero dma transfer size: ", DUMP_PREFIX_NONE, 16, 1, - raw_cmd->cmd, raw_cmd->cmd_count, false); + raw_cmd->fullcmd, raw_cmd->cmd_count, false); cont->done(0); fdc_state[current_fdc].reset = 1; return; @@ -1515,7 +1515,7 @@ static void setup_rw_floppy(void) r = 0; for (i = 0; i < raw_cmd->cmd_count; i++) - r |= output_byte(current_fdc, raw_cmd->cmd[i]); + r |= output_byte(current_fdc, raw_cmd->fullcmd[i]); debugt(__func__, "rw_command"); diff --git a/include/uapi/linux/fd.h b/include/uapi/linux/fd.h index 2e9c2c1c18e6..8b80c63b971c 100644 --- a/include/uapi/linux/fd.h +++ b/include/uapi/linux/fd.h @@ -371,9 +371,14 @@ struct floppy_raw_cmd { */ unsigned char cmd_count; - unsigned char cmd[FD_RAW_CMD_SIZE]; - unsigned char reply_count; - unsigned char reply[FD_RAW_REPLY_SIZE]; + union { + struct { + unsigned char cmd[FD_RAW_CMD_SIZE]; + unsigned char reply_count; + unsigned char reply[FD_RAW_REPLY_SIZE]; + }; + unsigned char fullcmd[FD_RAW_CMD_FULLSIZE]; + }; int track; int resultcode; -- cgit v1.2.3 From faf1d25440d6ad06d509dada4b6fe62fea844370 Mon Sep 17 00:00:00 2001 From: Martijn Coenen Date: Wed, 13 May 2020 15:38:44 +0200 Subject: loop: Clean up LOOP_SET_STATUS lo_flags handling LOOP_SET_STATUS(64) will actually allow some lo_flags to be modified; in particular, LO_FLAGS_AUTOCLEAR can be set and cleared, whereas LO_FLAGS_PARTSCAN can be set to request a partition scan. Make this explicit by updating the UAPI to include the flags that can be set/cleared using this ioctl. The implementation can then blindly take over the passed in flags, and use the previous flags for those flags that can't be set / cleared using LOOP_SET_STATUS. Signed-off-by: Martijn Coenen Reviewed-by: Christoph Hellwig Signed-off-by: Jens Axboe --- drivers/block/loop.c | 19 +++++++++++++------ include/uapi/linux/loop.h | 10 ++++++++-- 2 files changed, 21 insertions(+), 8 deletions(-) (limited to 'include') diff --git a/drivers/block/loop.c b/drivers/block/loop.c index 31f10da4945e..13518ba191f5 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -1036,9 +1036,7 @@ loop_set_status_from_info(struct loop_device *lo, lo->transfer = xfer->transfer; lo->ioctl = xfer->ioctl; - if ((lo->lo_flags & LO_FLAGS_AUTOCLEAR) != - (info->lo_flags & LO_FLAGS_AUTOCLEAR)) - lo->lo_flags ^= LO_FLAGS_AUTOCLEAR; + lo->lo_flags = info->lo_flags; lo->lo_encrypt_key_size = info->lo_encrypt_key_size; lo->lo_init[0] = info->lo_init[0]; @@ -1323,6 +1321,7 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info) int err; struct block_device *bdev; kuid_t uid = current_uid(); + int prev_lo_flags; bool partscan = false; bool size_changed = false; @@ -1359,10 +1358,19 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info) goto out_unfreeze; } + prev_lo_flags = lo->lo_flags; + err = loop_set_status_from_info(lo, info); if (err) goto out_unfreeze; + /* Mask out flags that can't be set using LOOP_SET_STATUS. */ + lo->lo_flags &= ~LOOP_SET_STATUS_SETTABLE_FLAGS; + /* For those flags, use the previous values instead */ + lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_SETTABLE_FLAGS; + /* For flags that can't be cleared, use previous values too */ + lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_CLEARABLE_FLAGS; + if (size_changed) { loff_t new_size = get_size(lo->lo_offset, lo->lo_sizelimit, lo->lo_backing_file); @@ -1377,9 +1385,8 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info) out_unfreeze: blk_mq_unfreeze_queue(lo->lo_queue); - if (!err && (info->lo_flags & LO_FLAGS_PARTSCAN) && - !(lo->lo_flags & LO_FLAGS_PARTSCAN)) { - lo->lo_flags |= LO_FLAGS_PARTSCAN; + if (!err && (lo->lo_flags & LO_FLAGS_PARTSCAN) && + !(prev_lo_flags & LO_FLAGS_PARTSCAN)) { lo->lo_disk->flags &= ~GENHD_FL_NO_PART_SCAN; bdev = lo->lo_device; partscan = true; diff --git a/include/uapi/linux/loop.h b/include/uapi/linux/loop.h index 080a8df134ef..6b32fee80ce0 100644 --- a/include/uapi/linux/loop.h +++ b/include/uapi/linux/loop.h @@ -25,6 +25,12 @@ enum { LO_FLAGS_DIRECT_IO = 16, }; +/* LO_FLAGS that can be set using LOOP_SET_STATUS(64) */ +#define LOOP_SET_STATUS_SETTABLE_FLAGS (LO_FLAGS_AUTOCLEAR | LO_FLAGS_PARTSCAN) + +/* LO_FLAGS that can be cleared using LOOP_SET_STATUS(64) */ +#define LOOP_SET_STATUS_CLEARABLE_FLAGS (LO_FLAGS_AUTOCLEAR) + #include /* for __kernel_old_dev_t */ #include /* for __u64 */ @@ -37,7 +43,7 @@ struct loop_info { int lo_offset; int lo_encrypt_type; int lo_encrypt_key_size; /* ioctl w/o */ - int lo_flags; /* ioctl r/o */ + int lo_flags; char lo_name[LO_NAME_SIZE]; unsigned char lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */ unsigned long lo_init[2]; @@ -53,7 +59,7 @@ struct loop_info64 { __u32 lo_number; /* ioctl r/o */ __u32 lo_encrypt_type; __u32 lo_encrypt_key_size; /* ioctl w/o */ - __u32 lo_flags; /* ioctl r/o */ + __u32 lo_flags; __u8 lo_file_name[LO_NAME_SIZE]; __u8 lo_crypt_name[LO_NAME_SIZE]; __u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */ -- cgit v1.2.3 From 3448914e8cc550ba792d4ccc74471d1ca4293aae Mon Sep 17 00:00:00 2001 From: Martijn Coenen Date: Wed, 13 May 2020 15:38:45 +0200 Subject: loop: Add LOOP_CONFIGURE ioctl This allows userspace to completely setup a loop device with a single ioctl, removing the in-between state where the device can be partially configured - eg the loop device has a backing file associated with it, but is reading from the wrong offset. Besides removing the intermediate state, another big benefit of this ioctl is that LOOP_SET_STATUS can be slow; the main reason for this slowness is that LOOP_SET_STATUS(64) calls blk_mq_freeze_queue() to freeze the associated queue; this requires waiting for RCU synchronization, which I've measured can take about 15-20ms on this device on average. In addition to doing what LOOP_SET_STATUS can do, LOOP_CONFIGURE can also be used to: - Set the correct block size immediately by setting loop_config.block_size (avoids LOOP_SET_BLOCK_SIZE) - Explicitly request direct I/O mode by setting LO_FLAGS_DIRECT_IO in loop_config.info.lo_flags (avoids LOOP_SET_DIRECT_IO) - Explicitly request read-only mode by setting LO_FLAGS_READ_ONLY in loop_config.info.lo_flags Here's setting up ~70 regular loop devices with an offset on an x86 Android device, using LOOP_SET_FD and LOOP_SET_STATUS: vsoc_x86:/system/apex # time for i in `seq 30 100`; do losetup -r -o 4096 /dev/block/loop$i com.android.adbd.apex; done 0m03.40s real 0m00.02s user 0m00.03s system Here's configuring ~70 devices in the same way, but using a modified losetup that uses the new LOOP_CONFIGURE ioctl: vsoc_x86:/system/apex # time for i in `seq 30 100`; do losetup -r -o 4096 /dev/block/loop$i com.android.adbd.apex; done 0m01.94s real 0m00.01s user 0m00.01s system Signed-off-by: Martijn Coenen Reviewed-by: Christoph Hellwig Signed-off-by: Jens Axboe --- drivers/block/loop.c | 104 +++++++++++++++++++++++++++++++++------------- include/uapi/linux/loop.h | 21 ++++++++++ 2 files changed, 97 insertions(+), 28 deletions(-) (limited to 'include') diff --git a/drivers/block/loop.c b/drivers/block/loop.c index 13518ba191f5..a565c5aafa52 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -228,6 +228,19 @@ static void __loop_update_dio(struct loop_device *lo, bool dio) blk_mq_unfreeze_queue(lo->lo_queue); } +/** + * loop_validate_block_size() - validates the passed in block size + * @bsize: size to validate + */ +static int +loop_validate_block_size(unsigned short bsize) +{ + if (bsize < 512 || bsize > PAGE_SIZE || !is_power_of_2(bsize)) + return -EINVAL; + + return 0; +} + /** * loop_set_size() - sets device size and notifies userspace * @lo: struct loop_device to set the size for @@ -1050,23 +1063,24 @@ loop_set_status_from_info(struct loop_device *lo, return 0; } -static int loop_set_fd(struct loop_device *lo, fmode_t mode, - struct block_device *bdev, unsigned int arg) +static int loop_configure(struct loop_device *lo, fmode_t mode, + struct block_device *bdev, + const struct loop_config *config) { struct file *file; struct inode *inode; struct address_space *mapping; struct block_device *claimed_bdev = NULL; - int lo_flags = 0; int error; loff_t size; bool partscan; + unsigned short bsize; /* This is safe, since we have a reference from open(). */ __module_get(THIS_MODULE); error = -EBADF; - file = fget(arg); + file = fget(config->fd); if (!file) goto out; @@ -1075,7 +1089,7 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, * here to avoid changing device under exclusive owner. */ if (!(mode & FMODE_EXCL)) { - claimed_bdev = bd_start_claiming(bdev, loop_set_fd); + claimed_bdev = bd_start_claiming(bdev, loop_configure); if (IS_ERR(claimed_bdev)) { error = PTR_ERR(claimed_bdev); goto out_putf; @@ -1097,11 +1111,26 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, mapping = file->f_mapping; inode = mapping->host; + size = get_loop_size(lo, file); + + if ((config->info.lo_flags & ~LOOP_CONFIGURE_SETTABLE_FLAGS) != 0) { + error = -EINVAL; + goto out_unlock; + } + + if (config->block_size) { + error = loop_validate_block_size(config->block_size); + if (error) + goto out_unlock; + } + + error = loop_set_status_from_info(lo, &config->info); + if (error) + goto out_unlock; + if (!(file->f_mode & FMODE_WRITE) || !(mode & FMODE_WRITE) || !file->f_op->write_iter) - lo_flags |= LO_FLAGS_READ_ONLY; - - size = get_loop_size(lo, file); + lo->lo_flags |= LO_FLAGS_READ_ONLY; error = loop_prepare_queue(lo); if (error) @@ -1109,30 +1138,28 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, error = 0; - set_device_ro(bdev, (lo_flags & LO_FLAGS_READ_ONLY) != 0); + set_device_ro(bdev, (lo->lo_flags & LO_FLAGS_READ_ONLY) != 0); - lo->use_dio = false; + lo->use_dio = lo->lo_flags & LO_FLAGS_DIRECT_IO; lo->lo_device = bdev; - lo->lo_flags = lo_flags; lo->lo_backing_file = file; - lo->transfer = NULL; - lo->ioctl = NULL; - lo->lo_sizelimit = 0; lo->old_gfp_mask = mapping_gfp_mask(mapping); mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS)); - if (!(lo_flags & LO_FLAGS_READ_ONLY) && file->f_op->fsync) + if (!(lo->lo_flags & LO_FLAGS_READ_ONLY) && file->f_op->fsync) blk_queue_write_cache(lo->lo_queue, true, false); - if (io_is_direct(lo->lo_backing_file) && inode->i_sb->s_bdev) { + if (config->block_size) + bsize = config->block_size; + else if (io_is_direct(lo->lo_backing_file) && inode->i_sb->s_bdev) /* In case of direct I/O, match underlying block size */ - unsigned short bsize = bdev_logical_block_size( - inode->i_sb->s_bdev); + bsize = bdev_logical_block_size(inode->i_sb->s_bdev); + else + bsize = 512; - blk_queue_logical_block_size(lo->lo_queue, bsize); - blk_queue_physical_block_size(lo->lo_queue, bsize); - blk_queue_io_min(lo->lo_queue, bsize); - } + blk_queue_logical_block_size(lo->lo_queue, bsize); + blk_queue_physical_block_size(lo->lo_queue, bsize); + blk_queue_io_min(lo->lo_queue, bsize); loop_update_rotational(lo); loop_update_dio(lo); @@ -1155,14 +1182,14 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode, if (partscan) loop_reread_partitions(lo, bdev); if (claimed_bdev) - bd_abort_claiming(bdev, claimed_bdev, loop_set_fd); + bd_abort_claiming(bdev, claimed_bdev, loop_configure); return 0; out_unlock: mutex_unlock(&loop_ctl_mutex); out_bdev: if (claimed_bdev) - bd_abort_claiming(bdev, claimed_bdev, loop_set_fd); + bd_abort_claiming(bdev, claimed_bdev, loop_configure); out_putf: fput(file); out: @@ -1582,8 +1609,9 @@ static int loop_set_block_size(struct loop_device *lo, unsigned long arg) if (lo->lo_state != Lo_bound) return -ENXIO; - if (arg < 512 || arg > PAGE_SIZE || !is_power_of_2(arg)) - return -EINVAL; + err = loop_validate_block_size(arg); + if (err) + return err; if (lo->lo_queue->limits.logical_block_size == arg) return 0; @@ -1645,8 +1673,27 @@ static int lo_ioctl(struct block_device *bdev, fmode_t mode, int err; switch (cmd) { - case LOOP_SET_FD: - return loop_set_fd(lo, mode, bdev, arg); + case LOOP_SET_FD: { + /* + * Legacy case - pass in a zeroed out struct loop_config with + * only the file descriptor set , which corresponds with the + * default parameters we'd have used otherwise. + */ + struct loop_config config; + + memset(&config, 0, sizeof(config)); + config.fd = arg; + + return loop_configure(lo, mode, bdev, &config); + } + case LOOP_CONFIGURE: { + struct loop_config config; + + if (copy_from_user(&config, argp, sizeof(config))) + return -EFAULT; + + return loop_configure(lo, mode, bdev, &config); + } case LOOP_CHANGE_FD: return loop_change_fd(lo, bdev, arg); case LOOP_CLR_FD: @@ -1818,6 +1865,7 @@ static int lo_compat_ioctl(struct block_device *bdev, fmode_t mode, case LOOP_CLR_FD: case LOOP_GET_STATUS64: case LOOP_SET_STATUS64: + case LOOP_CONFIGURE: arg = (unsigned long) compat_ptr(arg); /* fall through */ case LOOP_SET_FD: diff --git a/include/uapi/linux/loop.h b/include/uapi/linux/loop.h index 6b32fee80ce0..24a1c45bd1ae 100644 --- a/include/uapi/linux/loop.h +++ b/include/uapi/linux/loop.h @@ -31,6 +31,10 @@ enum { /* LO_FLAGS that can be cleared using LOOP_SET_STATUS(64) */ #define LOOP_SET_STATUS_CLEARABLE_FLAGS (LO_FLAGS_AUTOCLEAR) +/* LO_FLAGS that can be set using LOOP_CONFIGURE */ +#define LOOP_CONFIGURE_SETTABLE_FLAGS (LO_FLAGS_READ_ONLY | LO_FLAGS_AUTOCLEAR \ + | LO_FLAGS_PARTSCAN | LO_FLAGS_DIRECT_IO) + #include /* for __kernel_old_dev_t */ #include /* for __u64 */ @@ -66,6 +70,22 @@ struct loop_info64 { __u64 lo_init[2]; }; +/** + * struct loop_config - Complete configuration for a loop device. + * @fd: fd of the file to be used as a backing file for the loop device. + * @block_size: block size to use; ignored if 0. + * @info: struct loop_info64 to configure the loop device with. + * + * This structure is used with the LOOP_CONFIGURE ioctl, and can be used to + * atomically setup and configure all loop device parameters at once. + */ +struct loop_config { + __u32 fd; + __u32 block_size; + struct loop_info64 info; + __u64 __reserved[8]; +}; + /* * Loop filter types */ @@ -96,6 +116,7 @@ struct loop_info64 { #define LOOP_SET_CAPACITY 0x4C07 #define LOOP_SET_DIRECT_IO 0x4C08 #define LOOP_SET_BLOCK_SIZE 0x4C09 +#define LOOP_CONFIGURE 0x4C0A /* /dev/loop-control interface */ #define LOOP_CTL_ADD 0x4C80 -- cgit v1.2.3 From 26d7e28e38206b1b3207af1409eee2269ab36f82 Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Tue, 19 May 2020 16:22:59 +0200 Subject: s390/dasd: remove ioctl_by_bdev calls The IBM partition parser requires device type specific information only available to the DASD driver to correctly register partitions. The current approach of using ioctl_by_bdev with a fake user space pointer is discouraged. Fix this by replacing IOCTL calls with direct in-kernel function calls. Suggested-by: Christoph Hellwig Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Reviewed-by: Peter Oberparleiter Reviewed-by: Christoph Hellwig Signed-off-by: Jens Axboe --- MAINTAINERS | 1 + block/partitions/ibm.c | 24 ++++++++++++++++++------ drivers/s390/block/dasd_ioctl.c | 34 ++++++++++++++++++++++++++++++++++ include/linux/dasd_mod.h | 9 +++++++++ 4 files changed, 62 insertions(+), 6 deletions(-) create mode 100644 include/linux/dasd_mod.h (limited to 'include') diff --git a/MAINTAINERS b/MAINTAINERS index b816a453b10e..927fd9845d2f 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -14628,6 +14628,7 @@ S: Supported W: http://www.ibm.com/developerworks/linux/linux390/ F: block/partitions/ibm.c F: drivers/s390/block/dasd* +F: include/linux/dasd_mod.h S390 IOMMU (PCI) M: Gerald Schaefer diff --git a/block/partitions/ibm.c b/block/partitions/ibm.c index 073faa6a69b8..d6e18df9c53c 100644 --- a/block/partitions/ibm.c +++ b/block/partitions/ibm.c @@ -13,10 +13,11 @@ #include #include #include +#include +#include #include "check.h" - union label_t { struct vtoc_volume_label_cdl vol; struct vtoc_volume_label_ldl lnx; @@ -288,7 +289,9 @@ static int find_cms1_partitions(struct parsed_partitions *state, */ int ibm_partition(struct parsed_partitions *state) { + int (*fn)(struct gendisk *disk, dasd_information2_t *info); struct block_device *bdev = state->bdev; + struct gendisk *disk = bdev->bd_disk; int blocksize, res; loff_t i_size, offset, size; dasd_information2_t *info; @@ -299,24 +302,31 @@ int ibm_partition(struct parsed_partitions *state) union label_t *label; res = 0; + if (!disk->fops->getgeo) + goto out_exit; + fn = symbol_get(dasd_biodasdinfo); + if (!fn) + goto out_exit; blocksize = bdev_logical_block_size(bdev); if (blocksize <= 0) - goto out_exit; + goto out_symbol; i_size = i_size_read(bdev->bd_inode); if (i_size == 0) - goto out_exit; + goto out_symbol; info = kmalloc(sizeof(dasd_information2_t), GFP_KERNEL); if (info == NULL) - goto out_exit; + goto out_symbol; geo = kmalloc(sizeof(struct hd_geometry), GFP_KERNEL); if (geo == NULL) goto out_nogeo; label = kmalloc(sizeof(union label_t), GFP_KERNEL); if (label == NULL) goto out_nolab; - if (ioctl_by_bdev(bdev, HDIO_GETGEO, (unsigned long)geo) != 0) + /* set start if not filled by getgeo function e.g. virtblk */ + geo->start = get_start_sect(bdev); + if (disk->fops->getgeo(bdev, geo)) goto out_freeall; - if (ioctl_by_bdev(bdev, BIODASDINFO2, (unsigned long)info) != 0) { + if (fn(disk, info)) { kfree(info); info = NULL; } @@ -359,6 +369,8 @@ out_nolab: kfree(geo); out_nogeo: kfree(info); +out_symbol: + symbol_put(dasd_biodasdinfo); out_exit: return res; } diff --git a/drivers/s390/block/dasd_ioctl.c b/drivers/s390/block/dasd_ioctl.c index 9b7782395c37..777734d1b4e5 100644 --- a/drivers/s390/block/dasd_ioctl.c +++ b/drivers/s390/block/dasd_ioctl.c @@ -22,6 +22,7 @@ #include #include #include +#include /* This is ugly... */ #define PRINTK_HEADER "dasd_ioctl:" @@ -664,3 +665,36 @@ int dasd_ioctl(struct block_device *bdev, fmode_t mode, dasd_put_device(base); return rc; } + + +/** + * dasd_biodasdinfo() - fill out the dasd information structure + * @disk [in]: pointer to gendisk structure that references a DASD + * @info [out]: pointer to the dasd_information2_t structure + * + * Provide access to DASD specific information. + * The gendisk structure is checked if it belongs to the DASD driver by + * comparing the gendisk->fops pointer. + * If it does not belong to the DASD driver -EINVAL is returned. + * Otherwise the provided dasd_information2_t structure is filled out. + * + * Returns: + * %0 on success and a negative error value on failure. + */ +int dasd_biodasdinfo(struct gendisk *disk, struct dasd_information2_t *info) +{ + struct dasd_device *base; + int error; + + if (disk->fops != &dasd_device_operations) + return -EINVAL; + + base = dasd_device_from_gendisk(disk); + if (!base) + return -ENODEV; + error = __dasd_ioctl_information(base->block, info); + dasd_put_device(base); + return error; +} +/* export that symbol_get in partition detection is possible */ +EXPORT_SYMBOL_GPL(dasd_biodasdinfo); diff --git a/include/linux/dasd_mod.h b/include/linux/dasd_mod.h new file mode 100644 index 000000000000..d39abad2ff6e --- /dev/null +++ b/include/linux/dasd_mod.h @@ -0,0 +1,9 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef DASD_MOD_H +#define DASD_MOD_H + +#include + +extern int dasd_biodasdinfo(struct gendisk *disk, dasd_information2_t *info); + +#endif -- cgit v1.2.3 From 3783daeb1d24696ff00125050353cfce4f5b6239 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 19 May 2020 16:33:21 +0200 Subject: block: remove ioctl_by_bdev No callers left. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- fs/block_dev.c | 12 ------------ include/linux/fs.h | 1 - 2 files changed, 13 deletions(-) (limited to 'include') diff --git a/fs/block_dev.c b/fs/block_dev.c index 7cbb7b79935e..3003831c771d 100644 --- a/fs/block_dev.c +++ b/fs/block_dev.c @@ -2166,18 +2166,6 @@ const struct file_operations def_blk_fops = { .fallocate = blkdev_fallocate, }; -int ioctl_by_bdev(struct block_device *bdev, unsigned cmd, unsigned long arg) -{ - int res; - mm_segment_t old_fs = get_fs(); - set_fs(KERNEL_DS); - res = blkdev_ioctl(bdev, 0, cmd, arg); - set_fs(old_fs); - return res; -} - -EXPORT_SYMBOL(ioctl_by_bdev); - /** * lookup_bdev - lookup a struct block_device by name * @pathname: special file representing the block device diff --git a/include/linux/fs.h b/include/linux/fs.h index 1a95e5158811..861ca61d728b 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -2636,7 +2636,6 @@ extern int sync_filesystem(struct super_block *); extern const struct file_operations def_blk_fops; extern const struct file_operations def_chr_fops; #ifdef CONFIG_BLOCK -extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long); extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder); -- cgit v1.2.3 From f1e71d75f04792721d411170168e68019ccb7de3 Mon Sep 17 00:00:00 2001 From: "Gustavo A. R. Silva" Date: Thu, 7 May 2020 14:04:52 -0500 Subject: nvme: replace zero-length array with flexible-array The current codebase makes use of the zero-length array language extension to the C90 standard, but the preferred mechanism to declare variable-length types such as these ones is a flexible array member[1][2], introduced in C99: struct foo { int stuff; struct boo array[]; }; By making use of the mechanism above, we will get a compiler warning in case the flexible array does not occur last in the structure, which will help us prevent some kind of undefined behavior bugs from being inadvertently introduced[3] to the codebase from now on. Also, notice that, dynamic memory allocations won't be affected by this change: "Flexible array members have incomplete type, and so the sizeof operator may not be applied. As a quirk of the original implementation of zero-length arrays, sizeof evaluates to zero."[1] sizeof(flexible-array-member) triggers a warning because flexible array members have incomplete type[1]. There are some instances of code in which the sizeof operator is being incorrectly/erroneously applied to zero-length arrays and the result is zero. Such instances may be hiding some bugs. So, this work (flexible-array member conversions) will also help to get completely rid of those sorts of issues. This issue was found with the help of Coccinelle. [1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html [2] https://github.com/KSPP/linux/issues/21 [3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour") Signed-off-by: Gustavo A. R. Silva Signed-off-by: Christoph Hellwig --- drivers/nvme/host/fc.c | 2 +- drivers/nvme/host/lightnvm.c | 2 +- include/linux/nvme.h | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) (limited to 'include') diff --git a/drivers/nvme/host/fc.c b/drivers/nvme/host/fc.c index 8aabc056c754..cb0007592c12 100644 --- a/drivers/nvme/host/fc.c +++ b/drivers/nvme/host/fc.c @@ -108,7 +108,7 @@ struct nvme_fc_fcp_op { struct nvme_fcp_op_w_sgl { struct nvme_fc_fcp_op op; struct scatterlist sgl[NVME_INLINE_SG_CNT]; - uint8_t priv[0]; + uint8_t priv[]; }; struct nvme_fc_lport { diff --git a/drivers/nvme/host/lightnvm.c b/drivers/nvme/host/lightnvm.c index ec46693f6b64..3002bf972c6b 100644 --- a/drivers/nvme/host/lightnvm.c +++ b/drivers/nvme/host/lightnvm.c @@ -171,7 +171,7 @@ struct nvme_nvm_bb_tbl { __le32 tdresv; __le32 thresv; __le32 rsvd2[8]; - __u8 blk[0]; + __u8 blk[]; }; struct nvme_nvm_id20_addrf { diff --git a/include/linux/nvme.h b/include/linux/nvme.h index b235a48eac8c..e2993e6a9d7c 100644 --- a/include/linux/nvme.h +++ b/include/linux/nvme.h @@ -1185,7 +1185,7 @@ struct nvmf_disc_rsp_page_hdr { __le64 numrec; __le16 recfmt; __u8 resv14[1006]; - struct nvmf_disc_rsp_page_entry entries[0]; + struct nvmf_disc_rsp_page_entry entries[]; }; enum { -- cgit v1.2.3 From c295ee4742fda49de598a304ef9a95cf8da6b1f5 Mon Sep 17 00:00:00 2001 From: Max Gurtovoy Date: Tue, 19 May 2020 17:05:48 +0300 Subject: block: always define struct blk_integrity in genhd.h This will reduce the amount of ifdefs inside the source code for various drivers and also will reduce the amount of stub functions that were created for the !CONFIG_BLK_DEV_INTEGRITY case. Suggested-by: Christoph Hellwig Signed-off-by: Max Gurtovoy Reviewed-by: Israel Rukshin Reviewed-by: Martin K. Petersen Signed-off-by: Christoph Hellwig --- include/linux/genhd.h | 4 ---- 1 file changed, 4 deletions(-) (limited to 'include') diff --git a/include/linux/genhd.h b/include/linux/genhd.h index f9c226f9546a..2590bed6e6b3 100644 --- a/include/linux/genhd.h +++ b/include/linux/genhd.h @@ -169,8 +169,6 @@ struct disk_part_tbl { struct disk_events; struct badblocks; -#if defined(CONFIG_BLK_DEV_INTEGRITY) - struct blk_integrity { const struct blk_integrity_profile *profile; unsigned char flags; @@ -179,8 +177,6 @@ struct blk_integrity { unsigned char tag_size; }; -#endif /* CONFIG_BLK_DEV_INTEGRITY */ - struct gendisk { /* major, first_minor and minors are input parameters only, * don't use directly. Use disk_devt() and disk_max_parts(). -- cgit v1.2.3 From 39481fbd14ee272edd419d73a98bc637e2a3fd35 Mon Sep 17 00:00:00 2001 From: Israel Rukshin Date: Tue, 19 May 2020 17:06:00 +0300 Subject: nvme: add Metadata Capabilities enumerations The enumerations will be used to expose the namespace metadata format by the target. Suggested-by: Christoph Hellwig Signed-off-by: Israel Rukshin Signed-off-by: Max Gurtovoy Reviewed-by: James Smart Reviewed-by: Martin K. Petersen Signed-off-by: Christoph Hellwig --- include/linux/nvme.h | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'include') diff --git a/include/linux/nvme.h b/include/linux/nvme.h index e2993e6a9d7c..5ce51ab4c50e 100644 --- a/include/linux/nvme.h +++ b/include/linux/nvme.h @@ -420,6 +420,12 @@ enum { NVME_NS_DPS_PI_TYPE3 = 3, }; +/* Identify Namespace Metadata Capabilities (MC): */ +enum { + NVME_MC_EXTENDED_LBA = (1 << 0), + NVME_MC_METADATA_PTR = (1 << 1), +}; + struct nvme_ns_id_desc { __u8 nidt; __u8 nidl; -- cgit v1.2.3