diff options
Diffstat (limited to 'drivers/scsi')
26 files changed, 5505 insertions, 1120 deletions
diff --git a/drivers/scsi/Kconfig b/drivers/scsi/Kconfig index 1918f5483b23..7d1b4317eccc 100644 --- a/drivers/scsi/Kconfig +++ b/drivers/scsi/Kconfig @@ -838,6 +838,23 @@ config SCSI_IBMVSCSI To compile this driver as a module, choose M here: the module will be called ibmvscsi. +config SCSI_IBMVSCSIS + tristate "IBM Virtual SCSI Server support" + depends on PPC_PSERIES && TARGET_CORE && SCSI && PCI + help + This is the IBM POWER Virtual SCSI Target Server + This driver uses the SRP protocol for communication betwen servers + guest and/or the host that run on the same server. + More information on VSCSI protocol can be found at www.power.org + + The userspace configuration needed to initialize the driver can be + be found here: + + https://github.com/powervm/ibmvscsis/wiki/Configuration + + To compile this driver as a module, choose M here: the + module will be called ibmvscsis. + config SCSI_IBMVFC tristate "IBM Virtual FC support" depends on PPC_PSERIES && SCSI diff --git a/drivers/scsi/Makefile b/drivers/scsi/Makefile index 862ab4efad61..d5397987e731 100644 --- a/drivers/scsi/Makefile +++ b/drivers/scsi/Makefile @@ -128,6 +128,7 @@ obj-$(CONFIG_SCSI_SNI_53C710) += 53c700.o sni_53c710.o obj-$(CONFIG_SCSI_NSP32) += nsp32.o obj-$(CONFIG_SCSI_IPR) += ipr.o obj-$(CONFIG_SCSI_IBMVSCSI) += ibmvscsi/ +obj-$(CONFIG_SCSI_IBMVSCSIS) += ibmvscsi_tgt/ obj-$(CONFIG_SCSI_IBMVFC) += ibmvscsi/ obj-$(CONFIG_SCSI_HPTIOP) += hptiop.o obj-$(CONFIG_SCSI_STEX) += stex.o diff --git a/drivers/scsi/cxgbi/Makefile b/drivers/scsi/cxgbi/Makefile index 86007e344955..a73781ac1800 100644 --- a/drivers/scsi/cxgbi/Makefile +++ b/drivers/scsi/cxgbi/Makefile @@ -1,2 +1,4 @@ +ccflags-y += -Idrivers/net/ethernet/chelsio/libcxgb + obj-$(CONFIG_SCSI_CXGB3_ISCSI) += libcxgbi.o cxgb3i/ obj-$(CONFIG_SCSI_CXGB4_ISCSI) += libcxgbi.o cxgb4i/ diff --git a/drivers/scsi/cxgbi/cxgb3i/Kbuild b/drivers/scsi/cxgbi/cxgb3i/Kbuild index 961a12f6d318..663c52e05d81 100644 --- a/drivers/scsi/cxgbi/cxgb3i/Kbuild +++ b/drivers/scsi/cxgbi/cxgb3i/Kbuild @@ -1,3 +1,4 @@ ccflags-y += -I$(srctree)/drivers/net/ethernet/chelsio/cxgb3 +ccflags-y += -I$(srctree)/drivers/net/ethernet/chelsio/libcxgb obj-$(CONFIG_SCSI_CXGB3_ISCSI) += cxgb3i.o diff --git a/drivers/scsi/cxgbi/cxgb3i/Kconfig b/drivers/scsi/cxgbi/cxgb3i/Kconfig index e4603985dce3..f68c871b16ca 100644 --- a/drivers/scsi/cxgbi/cxgb3i/Kconfig +++ b/drivers/scsi/cxgbi/cxgb3i/Kconfig @@ -5,6 +5,7 @@ config SCSI_CXGB3_ISCSI select ETHERNET select NET_VENDOR_CHELSIO select CHELSIO_T3 + select CHELSIO_LIB select SCSI_ISCSI_ATTRS ---help--- This driver supports iSCSI offload for the Chelsio T3 devices. diff --git a/drivers/scsi/cxgbi/cxgb3i/cxgb3i.c b/drivers/scsi/cxgbi/cxgb3i/cxgb3i.c index e22a268fd311..33e83464e091 100644 --- a/drivers/scsi/cxgbi/cxgb3i/cxgb3i.c +++ b/drivers/scsi/cxgbi/cxgb3i/cxgb3i.c @@ -1028,7 +1028,7 @@ cxgb3_cpl_handler_func cxgb3i_cpl_handlers[NUM_CPL_CMDS] = { * cxgb3i_ofld_init - allocate and initialize resources for each adapter found * @cdev: cxgbi adapter */ -int cxgb3i_ofld_init(struct cxgbi_device *cdev) +static int cxgb3i_ofld_init(struct cxgbi_device *cdev) { struct t3cdev *t3dev = (struct t3cdev *)cdev->lldev; struct adap_ports port; @@ -1076,64 +1076,69 @@ static inline void ulp_mem_io_set_hdr(struct sk_buff *skb, unsigned int addr) req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_BYPASS)); req->cmd_lock_addr = htonl(V_ULP_MEMIO_ADDR(addr >> 5) | V_ULPTX_CMD(ULP_MEM_WRITE)); - req->len = htonl(V_ULP_MEMIO_DATA_LEN(PPOD_SIZE >> 5) | - V_ULPTX_NFLITS((PPOD_SIZE >> 3) + 1)); + req->len = htonl(V_ULP_MEMIO_DATA_LEN(IPPOD_SIZE >> 5) | + V_ULPTX_NFLITS((IPPOD_SIZE >> 3) + 1)); } -static int ddp_set_map(struct cxgbi_sock *csk, struct cxgbi_pagepod_hdr *hdr, - unsigned int idx, unsigned int npods, - struct cxgbi_gather_list *gl) +static struct cxgbi_ppm *cdev2ppm(struct cxgbi_device *cdev) { - struct cxgbi_device *cdev = csk->cdev; - struct cxgbi_ddp_info *ddp = cdev->ddp; - unsigned int pm_addr = (idx << PPOD_SIZE_SHIFT) + ddp->llimit; - int i; + return ((struct t3cdev *)cdev->lldev)->ulp_iscsi; +} - log_debug(1 << CXGBI_DBG_DDP, - "csk 0x%p, idx %u, npods %u, gl 0x%p.\n", - csk, idx, npods, gl); +static int ddp_set_map(struct cxgbi_ppm *ppm, struct cxgbi_sock *csk, + struct cxgbi_task_tag_info *ttinfo) +{ + unsigned int idx = ttinfo->idx; + unsigned int npods = ttinfo->npods; + struct scatterlist *sg = ttinfo->sgl; + struct cxgbi_pagepod *ppod; + struct ulp_mem_io *req; + unsigned int sg_off; + unsigned int pm_addr = (idx << PPOD_SIZE_SHIFT) + ppm->llimit; + int i; - for (i = 0; i < npods; i++, idx++, pm_addr += PPOD_SIZE) { + for (i = 0; i < npods; i++, idx++, pm_addr += IPPOD_SIZE) { struct sk_buff *skb = alloc_wr(sizeof(struct ulp_mem_io) + - PPOD_SIZE, 0, GFP_ATOMIC); + IPPOD_SIZE, 0, GFP_ATOMIC); if (!skb) return -ENOMEM; - ulp_mem_io_set_hdr(skb, pm_addr); - cxgbi_ddp_ppod_set((struct cxgbi_pagepod *)(skb->head + - sizeof(struct ulp_mem_io)), - hdr, gl, i * PPOD_PAGES_MAX); + req = (struct ulp_mem_io *)skb->head; + ppod = (struct cxgbi_pagepod *)(req + 1); + sg_off = i * PPOD_PAGES_MAX; + cxgbi_ddp_set_one_ppod(ppod, ttinfo, &sg, + &sg_off); skb->priority = CPL_PRIORITY_CONTROL; - cxgb3_ofld_send(cdev->lldev, skb); + cxgb3_ofld_send(ppm->lldev, skb); } return 0; } -static void ddp_clear_map(struct cxgbi_hba *chba, unsigned int tag, - unsigned int idx, unsigned int npods) +static void ddp_clear_map(struct cxgbi_device *cdev, struct cxgbi_ppm *ppm, + struct cxgbi_task_tag_info *ttinfo) { - struct cxgbi_device *cdev = chba->cdev; - struct cxgbi_ddp_info *ddp = cdev->ddp; - unsigned int pm_addr = (idx << PPOD_SIZE_SHIFT) + ddp->llimit; + unsigned int idx = ttinfo->idx; + unsigned int pm_addr = (idx << PPOD_SIZE_SHIFT) + ppm->llimit; + unsigned int npods = ttinfo->npods; int i; log_debug(1 << CXGBI_DBG_DDP, - "cdev 0x%p, idx %u, npods %u, tag 0x%x.\n", - cdev, idx, npods, tag); + "cdev 0x%p, clear idx %u, npods %u.\n", + cdev, idx, npods); - for (i = 0; i < npods; i++, idx++, pm_addr += PPOD_SIZE) { + for (i = 0; i < npods; i++, idx++, pm_addr += IPPOD_SIZE) { struct sk_buff *skb = alloc_wr(sizeof(struct ulp_mem_io) + - PPOD_SIZE, 0, GFP_ATOMIC); + IPPOD_SIZE, 0, GFP_ATOMIC); if (!skb) { - pr_err("tag 0x%x, 0x%x, %d/%u, skb OOM.\n", - tag, idx, i, npods); + pr_err("cdev 0x%p, clear ddp, %u,%d/%u, skb OOM.\n", + cdev, idx, i, npods); continue; } ulp_mem_io_set_hdr(skb, pm_addr); skb->priority = CPL_PRIORITY_CONTROL; - cxgb3_ofld_send(cdev->lldev, skb); + cxgb3_ofld_send(ppm->lldev, skb); } } @@ -1203,82 +1208,68 @@ static int ddp_setup_conn_digest(struct cxgbi_sock *csk, unsigned int tid, } /** - * t3_ddp_cleanup - release the cxgb3 adapter's ddp resource - * @cdev: cxgb3i adapter - * release all the resource held by the ddp pagepod manager for a given - * adapter if needed - */ - -static void t3_ddp_cleanup(struct cxgbi_device *cdev) -{ - struct t3cdev *tdev = (struct t3cdev *)cdev->lldev; - - if (cxgbi_ddp_cleanup(cdev)) { - pr_info("t3dev 0x%p, ulp_iscsi no more user.\n", tdev); - tdev->ulp_iscsi = NULL; - } -} - -/** - * ddp_init - initialize the cxgb3 adapter's ddp resource + * cxgb3i_ddp_init - initialize the cxgb3 adapter's ddp resource * @cdev: cxgb3i adapter * initialize the ddp pagepod manager for a given adapter */ static int cxgb3i_ddp_init(struct cxgbi_device *cdev) { struct t3cdev *tdev = (struct t3cdev *)cdev->lldev; - struct cxgbi_ddp_info *ddp = tdev->ulp_iscsi; + struct net_device *ndev = cdev->ports[0]; + struct cxgbi_tag_format tformat; + unsigned int ppmax, tagmask = 0; struct ulp_iscsi_info uinfo; - unsigned int pgsz_factor[4]; int i, err; - if (ddp) { - kref_get(&ddp->refcnt); - pr_warn("t3dev 0x%p, ddp 0x%p already set up.\n", - tdev, tdev->ulp_iscsi); - cdev->ddp = ddp; - return -EALREADY; - } - err = tdev->ctl(tdev, ULP_ISCSI_GET_PARAMS, &uinfo); if (err < 0) { - pr_err("%s, failed to get iscsi param err=%d.\n", - tdev->name, err); + pr_err("%s, failed to get iscsi param %d.\n", + ndev->name, err); return err; } + if (uinfo.llimit >= uinfo.ulimit) { + pr_warn("T3 %s, iscsi NOT enabled %u ~ %u!\n", + ndev->name, uinfo.llimit, uinfo.ulimit); + return -EACCES; + } - err = cxgbi_ddp_init(cdev, uinfo.llimit, uinfo.ulimit, - uinfo.max_txsz, uinfo.max_rxsz); - if (err < 0) - return err; + ppmax = (uinfo.ulimit - uinfo.llimit + 1) >> PPOD_SIZE_SHIFT; + tagmask = cxgbi_tagmask_set(ppmax); - ddp = cdev->ddp; + pr_info("T3 %s: 0x%x~0x%x, 0x%x, tagmask 0x%x -> 0x%x.\n", + ndev->name, uinfo.llimit, uinfo.ulimit, ppmax, uinfo.tagmask, + tagmask); - uinfo.tagmask = ddp->idx_mask << PPOD_IDX_SHIFT; - cxgbi_ddp_page_size_factor(pgsz_factor); + memset(&tformat, 0, sizeof(struct cxgbi_tag_format)); for (i = 0; i < 4; i++) - uinfo.pgsz_factor[i] = pgsz_factor[i]; - uinfo.ulimit = uinfo.llimit + (ddp->nppods << PPOD_SIZE_SHIFT); + tformat.pgsz_order[i] = uinfo.pgsz_factor[i]; + cxgbi_tagmask_check(tagmask, &tformat); - err = tdev->ctl(tdev, ULP_ISCSI_SET_PARAMS, &uinfo); - if (err < 0) { - pr_warn("%s unable to set iscsi param err=%d, ddp disabled.\n", - tdev->name, err); - cxgbi_ddp_cleanup(cdev); - return err; + cxgbi_ddp_ppm_setup(&tdev->ulp_iscsi, cdev, &tformat, ppmax, + uinfo.llimit, uinfo.llimit, 0); + if (!(cdev->flags & CXGBI_FLAG_DDP_OFF)) { + uinfo.tagmask = tagmask; + uinfo.ulimit = uinfo.llimit + (ppmax << PPOD_SIZE_SHIFT); + + err = tdev->ctl(tdev, ULP_ISCSI_SET_PARAMS, &uinfo); + if (err < 0) { + pr_err("T3 %s fail to set iscsi param %d.\n", + ndev->name, err); + cdev->flags |= CXGBI_FLAG_DDP_OFF; + } + err = 0; } - tdev->ulp_iscsi = ddp; cdev->csk_ddp_setup_digest = ddp_setup_conn_digest; cdev->csk_ddp_setup_pgidx = ddp_setup_conn_pgidx; - cdev->csk_ddp_set = ddp_set_map; - cdev->csk_ddp_clear = ddp_clear_map; - - pr_info("tdev 0x%p, nppods %u, bits %u, mask 0x%x,0x%x pkt %u/%u, " - "%u/%u.\n", - tdev, ddp->nppods, ddp->idx_bits, ddp->idx_mask, - ddp->rsvd_tag_mask, ddp->max_txsz, uinfo.max_txsz, - ddp->max_rxsz, uinfo.max_rxsz); + cdev->csk_ddp_set_map = ddp_set_map; + cdev->csk_ddp_clear_map = ddp_clear_map; + cdev->cdev2ppm = cdev2ppm; + cdev->tx_max_size = min_t(unsigned int, ULP2_MAX_PDU_PAYLOAD, + uinfo.max_txsz - ISCSI_PDU_NONPAYLOAD_LEN); + cdev->rx_max_size = min_t(unsigned int, ULP2_MAX_PDU_PAYLOAD, + uinfo.max_rxsz - ISCSI_PDU_NONPAYLOAD_LEN); + return 0; } @@ -1325,7 +1316,6 @@ static void cxgb3i_dev_open(struct t3cdev *t3dev) cdev->rx_credit_thres = cxgb3i_rx_credit_thres; cdev->skb_tx_rsvd = CXGB3I_TX_HEADER_LEN; cdev->skb_rx_extra = sizeof(struct cpl_iscsi_hdr_norss); - cdev->dev_ddp_cleanup = t3_ddp_cleanup; cdev->itp = &cxgb3i_iscsi_transport; err = cxgb3i_ddp_init(cdev); diff --git a/drivers/scsi/cxgbi/cxgb4i/Kbuild b/drivers/scsi/cxgbi/cxgb4i/Kbuild index 37458643749b..38e03c280417 100644 --- a/drivers/scsi/cxgbi/cxgb4i/Kbuild +++ b/drivers/scsi/cxgbi/cxgb4i/Kbuild @@ -1,3 +1,4 @@ ccflags-y += -I$(srctree)/drivers/net/ethernet/chelsio/cxgb4 +ccflags-y += -I$(srctree)/drivers/net/ethernet/chelsio/libcxgb obj-$(CONFIG_SCSI_CXGB4_ISCSI) += cxgb4i.o diff --git a/drivers/scsi/cxgbi/cxgb4i/Kconfig b/drivers/scsi/cxgbi/cxgb4i/Kconfig index 8c4e423037b6..594f593c8821 100644 --- a/drivers/scsi/cxgbi/cxgb4i/Kconfig +++ b/drivers/scsi/cxgbi/cxgb4i/Kconfig @@ -5,6 +5,7 @@ config SCSI_CXGB4_ISCSI select ETHERNET select NET_VENDOR_CHELSIO select CHELSIO_T4 + select CHELSIO_LIB select SCSI_ISCSI_ATTRS ---help--- This driver supports iSCSI offload for the Chelsio T4 devices. diff --git a/drivers/scsi/cxgbi/cxgb4i/cxgb4i.c b/drivers/scsi/cxgbi/cxgb4i/cxgb4i.c index 339f6b7f4803..e4ba2d2616cd 100644 --- a/drivers/scsi/cxgbi/cxgb4i/cxgb4i.c +++ b/drivers/scsi/cxgbi/cxgb4i/cxgb4i.c @@ -1503,7 +1503,7 @@ rel_resource_without_clip: return -EINVAL; } -cxgb4i_cplhandler_func cxgb4i_cplhandlers[NUM_CPL_CMDS] = { +static cxgb4i_cplhandler_func cxgb4i_cplhandlers[NUM_CPL_CMDS] = { [CPL_ACT_ESTABLISH] = do_act_establish, [CPL_ACT_OPEN_RPL] = do_act_open_rpl, [CPL_PEER_CLOSE] = do_peer_close, @@ -1519,7 +1519,7 @@ cxgb4i_cplhandler_func cxgb4i_cplhandlers[NUM_CPL_CMDS] = { [CPL_RX_DATA] = do_rx_data, }; -int cxgb4i_ofld_init(struct cxgbi_device *cdev) +static int cxgb4i_ofld_init(struct cxgbi_device *cdev) { int rc; @@ -1543,24 +1543,22 @@ int cxgb4i_ofld_init(struct cxgbi_device *cdev) return 0; } -/* - * functions to program the pagepod in h/w - */ -#define ULPMEM_IDATA_MAX_NPPODS 4 /* 256/PPOD_SIZE */ -static inline void ulp_mem_io_set_hdr(struct cxgb4_lld_info *lldi, - struct ulp_mem_io *req, - unsigned int wr_len, unsigned int dlen, - unsigned int pm_addr) +static inline void +ulp_mem_io_set_hdr(struct cxgbi_device *cdev, + struct ulp_mem_io *req, + unsigned int wr_len, unsigned int dlen, + unsigned int pm_addr, + int tid) { + struct cxgb4_lld_info *lldi = cxgbi_cdev_priv(cdev); struct ulptx_idata *idata = (struct ulptx_idata *)(req + 1); - INIT_ULPTX_WR(req, wr_len, 0, 0); - if (is_t4(lldi->adapter_type)) - req->cmd = htonl(ULPTX_CMD_V(ULP_TX_MEM_WRITE) | - (ULP_MEMIO_ORDER_F)); - else - req->cmd = htonl(ULPTX_CMD_V(ULP_TX_MEM_WRITE) | - (T5_ULP_MEMIO_IMM_F)); + INIT_ULPTX_WR(req, wr_len, 0, tid); + req->wr.wr_hi = htonl(FW_WR_OP_V(FW_ULPTX_WR) | + FW_WR_ATOMIC_V(0)); + req->cmd = htonl(ULPTX_CMD_V(ULP_TX_MEM_WRITE) | + ULP_MEMIO_ORDER_V(is_t4(lldi->adapter_type)) | + T5_ULP_MEMIO_IMM_V(!is_t4(lldi->adapter_type))); req->dlen = htonl(ULP_MEMIO_DATA_LEN_V(dlen >> 5)); req->lock_addr = htonl(ULP_MEMIO_ADDR_V(pm_addr >> 5)); req->len16 = htonl(DIV_ROUND_UP(wr_len - sizeof(req->wr), 16)); @@ -1569,82 +1567,89 @@ static inline void ulp_mem_io_set_hdr(struct cxgb4_lld_info *lldi, idata->len = htonl(dlen); } -static int ddp_ppod_write_idata(struct cxgbi_device *cdev, unsigned int port_id, - struct cxgbi_pagepod_hdr *hdr, unsigned int idx, - unsigned int npods, - struct cxgbi_gather_list *gl, - unsigned int gl_pidx) +static struct sk_buff * +ddp_ppod_init_idata(struct cxgbi_device *cdev, + struct cxgbi_ppm *ppm, + unsigned int idx, unsigned int npods, + unsigned int tid) { - struct cxgbi_ddp_info *ddp = cdev->ddp; - struct cxgb4_lld_info *lldi = cxgbi_cdev_priv(cdev); - struct sk_buff *skb; - struct ulp_mem_io *req; - struct ulptx_idata *idata; - struct cxgbi_pagepod *ppod; - unsigned int pm_addr = idx * PPOD_SIZE + ddp->llimit; - unsigned int dlen = PPOD_SIZE * npods; + unsigned int pm_addr = (idx << PPOD_SIZE_SHIFT) + ppm->llimit; + unsigned int dlen = npods << PPOD_SIZE_SHIFT; unsigned int wr_len = roundup(sizeof(struct ulp_mem_io) + sizeof(struct ulptx_idata) + dlen, 16); - unsigned int i; + struct sk_buff *skb = alloc_wr(wr_len, 0, GFP_ATOMIC); - skb = alloc_wr(wr_len, 0, GFP_ATOMIC); if (!skb) { - pr_err("cdev 0x%p, idx %u, npods %u, OOM.\n", - cdev, idx, npods); - return -ENOMEM; + pr_err("%s: %s idx %u, npods %u, OOM.\n", + __func__, ppm->ndev->name, idx, npods); + return NULL; } - req = (struct ulp_mem_io *)skb->head; - set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0); - ulp_mem_io_set_hdr(lldi, req, wr_len, dlen, pm_addr); + ulp_mem_io_set_hdr(cdev, (struct ulp_mem_io *)skb->head, wr_len, dlen, + pm_addr, tid); + + return skb; +} + +static int ddp_ppod_write_idata(struct cxgbi_ppm *ppm, struct cxgbi_sock *csk, + struct cxgbi_task_tag_info *ttinfo, + unsigned int idx, unsigned int npods, + struct scatterlist **sg_pp, + unsigned int *sg_off) +{ + struct cxgbi_device *cdev = csk->cdev; + struct sk_buff *skb = ddp_ppod_init_idata(cdev, ppm, idx, npods, + csk->tid); + struct ulp_mem_io *req; + struct ulptx_idata *idata; + struct cxgbi_pagepod *ppod; + int i; + + if (!skb) + return -ENOMEM; + + req = (struct ulp_mem_io *)skb->head; idata = (struct ulptx_idata *)(req + 1); ppod = (struct cxgbi_pagepod *)(idata + 1); - for (i = 0; i < npods; i++, ppod++, gl_pidx += PPOD_PAGES_MAX) { - if (!hdr && !gl) - cxgbi_ddp_ppod_clear(ppod); - else - cxgbi_ddp_ppod_set(ppod, hdr, gl, gl_pidx); - } + for (i = 0; i < npods; i++, ppod++) + cxgbi_ddp_set_one_ppod(ppod, ttinfo, sg_pp, sg_off); + + cxgbi_skcb_set_flag(skb, SKCBF_TX_MEM_WRITE); + cxgbi_skcb_set_flag(skb, SKCBF_TX_FLAG_COMPL); + set_wr_txq(skb, CPL_PRIORITY_DATA, csk->port_id); + + spin_lock_bh(&csk->lock); + cxgbi_sock_skb_entail(csk, skb); + spin_unlock_bh(&csk->lock); - cxgb4_ofld_send(cdev->ports[port_id], skb); return 0; } -static int ddp_set_map(struct cxgbi_sock *csk, struct cxgbi_pagepod_hdr *hdr, - unsigned int idx, unsigned int npods, - struct cxgbi_gather_list *gl) +static int ddp_set_map(struct cxgbi_ppm *ppm, struct cxgbi_sock *csk, + struct cxgbi_task_tag_info *ttinfo) { + unsigned int pidx = ttinfo->idx; + unsigned int npods = ttinfo->npods; unsigned int i, cnt; int err = 0; + struct scatterlist *sg = ttinfo->sgl; + unsigned int offset = 0; - for (i = 0; i < npods; i += cnt, idx += cnt) { - cnt = npods - i; - if (cnt > ULPMEM_IDATA_MAX_NPPODS) - cnt = ULPMEM_IDATA_MAX_NPPODS; - err = ddp_ppod_write_idata(csk->cdev, csk->port_id, hdr, - idx, cnt, gl, 4 * i); - if (err < 0) - break; - } - return err; -} - -static void ddp_clear_map(struct cxgbi_hba *chba, unsigned int tag, - unsigned int idx, unsigned int npods) -{ - unsigned int i, cnt; - int err; + ttinfo->cid = csk->port_id; - for (i = 0; i < npods; i += cnt, idx += cnt) { + for (i = 0; i < npods; i += cnt, pidx += cnt) { cnt = npods - i; + if (cnt > ULPMEM_IDATA_MAX_NPPODS) cnt = ULPMEM_IDATA_MAX_NPPODS; - err = ddp_ppod_write_idata(chba->cdev, chba->port_id, NULL, - idx, cnt, NULL, 0); + err = ddp_ppod_write_idata(ppm, csk, ttinfo, pidx, cnt, + &sg, &offset); if (err < 0) break; } + + return err; } static int ddp_setup_conn_pgidx(struct cxgbi_sock *csk, unsigned int tid, @@ -1710,48 +1715,46 @@ static int ddp_setup_conn_digest(struct cxgbi_sock *csk, unsigned int tid, return 0; } +static struct cxgbi_ppm *cdev2ppm(struct cxgbi_device *cdev) +{ + return (struct cxgbi_ppm *)(*((struct cxgb4_lld_info *) + (cxgbi_cdev_priv(cdev)))->iscsi_ppm); +} + static int cxgb4i_ddp_init(struct cxgbi_device *cdev) { struct cxgb4_lld_info *lldi = cxgbi_cdev_priv(cdev); - struct cxgbi_ddp_info *ddp = cdev->ddp; - unsigned int tagmask, pgsz_factor[4]; - int err; - - if (ddp) { - kref_get(&ddp->refcnt); - pr_warn("cdev 0x%p, ddp 0x%p already set up.\n", - cdev, cdev->ddp); - return -EALREADY; + struct net_device *ndev = cdev->ports[0]; + struct cxgbi_tag_format tformat; + unsigned int ppmax; + int i; + + if (!lldi->vr->iscsi.size) { + pr_warn("%s, iscsi NOT enabled, check config!\n", ndev->name); + return -EACCES; } - err = cxgbi_ddp_init(cdev, lldi->vr->iscsi.start, - lldi->vr->iscsi.start + lldi->vr->iscsi.size - 1, - lldi->iscsi_iolen, lldi->iscsi_iolen); - if (err < 0) - return err; + cdev->flags |= CXGBI_FLAG_USE_PPOD_OFLDQ; + ppmax = lldi->vr->iscsi.size >> PPOD_SIZE_SHIFT; - ddp = cdev->ddp; + memset(&tformat, 0, sizeof(struct cxgbi_tag_format)); + for (i = 0; i < 4; i++) + tformat.pgsz_order[i] = (lldi->iscsi_pgsz_order >> (i << 3)) + & 0xF; + cxgbi_tagmask_check(lldi->iscsi_tagmask, &tformat); - tagmask = ddp->idx_mask << PPOD_IDX_SHIFT; - cxgbi_ddp_page_size_factor(pgsz_factor); - cxgb4_iscsi_init(lldi->ports[0], tagmask, pgsz_factor); + cxgbi_ddp_ppm_setup(lldi->iscsi_ppm, cdev, &tformat, ppmax, + lldi->iscsi_llimit, lldi->vr->iscsi.start, 2); cdev->csk_ddp_setup_digest = ddp_setup_conn_digest; cdev->csk_ddp_setup_pgidx = ddp_setup_conn_pgidx; - cdev->csk_ddp_set = ddp_set_map; - cdev->csk_ddp_clear = ddp_clear_map; - - pr_info("cxgb4i 0x%p tag: sw %u, rsvd %u,%u, mask 0x%x.\n", - cdev, cdev->tag_format.sw_bits, cdev->tag_format.rsvd_bits, - cdev->tag_format.rsvd_shift, cdev->tag_format.rsvd_mask); - pr_info("cxgb4i 0x%p, nppods %u, bits %u, mask 0x%x,0x%x pkt %u/%u, " - " %u/%u.\n", - cdev, ddp->nppods, ddp->idx_bits, ddp->idx_mask, - ddp->rsvd_tag_mask, ddp->max_txsz, lldi->iscsi_iolen, - ddp->max_rxsz, lldi->iscsi_iolen); - pr_info("cxgb4i 0x%p max payload size: %u/%u, %u/%u.\n", - cdev, cdev->tx_max_size, ddp->max_txsz, cdev->rx_max_size, - ddp->max_rxsz); + cdev->csk_ddp_set_map = ddp_set_map; + cdev->tx_max_size = min_t(unsigned int, ULP2_MAX_PDU_PAYLOAD, + lldi->iscsi_iolen - ISCSI_PDU_NONPAYLOAD_LEN); + cdev->rx_max_size = min_t(unsigned int, ULP2_MAX_PDU_PAYLOAD, + lldi->iscsi_iolen - ISCSI_PDU_NONPAYLOAD_LEN); + cdev->cdev2ppm = cdev2ppm; + return 0; } diff --git a/drivers/scsi/cxgbi/libcxgbi.c b/drivers/scsi/cxgbi/libcxgbi.c index ead83a24bcd1..d1421139e6ea 100644 --- a/drivers/scsi/cxgbi/libcxgbi.c +++ b/drivers/scsi/cxgbi/libcxgbi.c @@ -64,6 +64,14 @@ static DEFINE_MUTEX(cdev_mutex); static LIST_HEAD(cdev_rcu_list); static DEFINE_SPINLOCK(cdev_rcu_lock); +static inline void cxgbi_decode_sw_tag(u32 sw_tag, int *idx, int *age) +{ + if (age) + *age = sw_tag & 0x7FFF; + if (idx) + *idx = (sw_tag >> 16) & 0x7FFF; +} + int cxgbi_device_portmap_create(struct cxgbi_device *cdev, unsigned int base, unsigned int max_conn) { @@ -113,12 +121,7 @@ static inline void cxgbi_device_destroy(struct cxgbi_device *cdev) "cdev 0x%p, p# %u.\n", cdev, cdev->nports); cxgbi_hbas_remove(cdev); cxgbi_device_portmap_cleanup(cdev); - if (cdev->dev_ddp_cleanup) - cdev->dev_ddp_cleanup(cdev); - else - cxgbi_ddp_cleanup(cdev); - if (cdev->ddp) - cxgbi_ddp_cleanup(cdev); + cxgbi_ppm_release(cdev->cdev2ppm(cdev)); if (cdev->pmap.max_connect) cxgbi_free_big_mem(cdev->pmap.port_csk); kfree(cdev); @@ -1182,525 +1185,245 @@ out_err: goto done; } -/* - * Direct Data Placement - - * Directly place the iSCSI Data-In or Data-Out PDU's payload into pre-posted - * final destination host-memory buffers based on the Initiator Task Tag (ITT) - * in Data-In or Target Task Tag (TTT) in Data-Out PDUs. - * The host memory address is programmed into h/w in the format of pagepod - * entries. - * The location of the pagepod entry is encoded into ddp tag which is used as - * the base for ITT/TTT. - */ - -static unsigned char ddp_page_order[DDP_PGIDX_MAX] = {0, 1, 2, 4}; -static unsigned char ddp_page_shift[DDP_PGIDX_MAX] = {12, 13, 14, 16}; -static unsigned char page_idx = DDP_PGIDX_MAX; - -static unsigned char sw_tag_idx_bits; -static unsigned char sw_tag_age_bits; - -/* - * Direct-Data Placement page size adjustment - */ -static int ddp_adjust_page_table(void) +static inline void +scmd_get_params(struct scsi_cmnd *sc, struct scatterlist **sgl, + unsigned int *sgcnt, unsigned int *dlen, + unsigned int prot) { - int i; - unsigned int base_order, order; - - if (PAGE_SIZE < (1UL << ddp_page_shift[0])) { - pr_info("PAGE_SIZE 0x%lx too small, min 0x%lx\n", - PAGE_SIZE, 1UL << ddp_page_shift[0]); - return -EINVAL; - } - - base_order = get_order(1UL << ddp_page_shift[0]); - order = get_order(1UL << PAGE_SHIFT); + struct scsi_data_buffer *sdb = prot ? scsi_prot(sc) : scsi_out(sc); - for (i = 0; i < DDP_PGIDX_MAX; i++) { - /* first is the kernel page size, then just doubling */ - ddp_page_order[i] = order - base_order + i; - ddp_page_shift[i] = PAGE_SHIFT + i; - } - return 0; + *sgl = sdb->table.sgl; + *sgcnt = sdb->table.nents; + *dlen = sdb->length; + /* Caution: for protection sdb, sdb->length is invalid */ } -static int ddp_find_page_index(unsigned long pgsz) +void cxgbi_ddp_set_one_ppod(struct cxgbi_pagepod *ppod, + struct cxgbi_task_tag_info *ttinfo, + struct scatterlist **sg_pp, unsigned int *sg_off) { + struct scatterlist *sg = sg_pp ? *sg_pp : NULL; + unsigned int offset = sg_off ? *sg_off : 0; + dma_addr_t addr = 0UL; + unsigned int len = 0; int i; - for (i = 0; i < DDP_PGIDX_MAX; i++) { - if (pgsz == (1UL << ddp_page_shift[i])) - return i; - } - pr_info("ddp page size %lu not supported.\n", pgsz); - return DDP_PGIDX_MAX; -} + memcpy(ppod, &ttinfo->hdr, sizeof(struct cxgbi_pagepod_hdr)); -static void ddp_setup_host_page_size(void) -{ - if (page_idx == DDP_PGIDX_MAX) { - page_idx = ddp_find_page_index(PAGE_SIZE); + if (sg) { + addr = sg_dma_address(sg); + len = sg_dma_len(sg); + } - if (page_idx == DDP_PGIDX_MAX) { - pr_info("system PAGE %lu, update hw.\n", PAGE_SIZE); - if (ddp_adjust_page_table() < 0) { - pr_info("PAGE %lu, disable ddp.\n", PAGE_SIZE); - return; + for (i = 0; i < PPOD_PAGES_MAX; i++) { + if (sg) { + ppod->addr[i] = cpu_to_be64(addr + offset); + offset += PAGE_SIZE; + if (offset == (len + sg->offset)) { + offset = 0; + sg = sg_next(sg); + if (sg) { + addr = sg_dma_address(sg); + len = sg_dma_len(sg); + } } - page_idx = ddp_find_page_index(PAGE_SIZE); + } else { + ppod->addr[i] = 0ULL; } - pr_info("system PAGE %lu, ddp idx %u.\n", PAGE_SIZE, page_idx); } -} - -void cxgbi_ddp_page_size_factor(int *pgsz_factor) -{ - int i; - - for (i = 0; i < DDP_PGIDX_MAX; i++) - pgsz_factor[i] = ddp_page_order[i]; -} -EXPORT_SYMBOL_GPL(cxgbi_ddp_page_size_factor); - -/* - * DDP setup & teardown - */ - -void cxgbi_ddp_ppod_set(struct cxgbi_pagepod *ppod, - struct cxgbi_pagepod_hdr *hdr, - struct cxgbi_gather_list *gl, unsigned int gidx) -{ - int i; - - memcpy(ppod, hdr, sizeof(*hdr)); - for (i = 0; i < (PPOD_PAGES_MAX + 1); i++, gidx++) { - ppod->addr[i] = gidx < gl->nelem ? - cpu_to_be64(gl->phys_addr[gidx]) : 0ULL; - } -} -EXPORT_SYMBOL_GPL(cxgbi_ddp_ppod_set); - -void cxgbi_ddp_ppod_clear(struct cxgbi_pagepod *ppod) -{ - memset(ppod, 0, sizeof(*ppod)); -} -EXPORT_SYMBOL_GPL(cxgbi_ddp_ppod_clear); - -static inline int ddp_find_unused_entries(struct cxgbi_ddp_info *ddp, - unsigned int start, unsigned int max, - unsigned int count, - struct cxgbi_gather_list *gl) -{ - unsigned int i, j, k; - /* not enough entries */ - if ((max - start) < count) { - log_debug(1 << CXGBI_DBG_DDP, - "NOT enough entries %u+%u < %u.\n", start, count, max); - return -EBUSY; + /* + * the fifth address needs to be repeated in the next ppod, so do + * not move sg + */ + if (sg_pp) { + *sg_pp = sg; + *sg_off = offset; } - max -= count; - spin_lock(&ddp->map_lock); - for (i = start; i < max;) { - for (j = 0, k = i; j < count; j++, k++) { - if (ddp->gl_map[k]) - break; - } - if (j == count) { - for (j = 0, k = i; j < count; j++, k++) - ddp->gl_map[k] = gl; - spin_unlock(&ddp->map_lock); - return i; + if (offset == len) { + offset = 0; + sg = sg_next(sg); + if (sg) { + addr = sg_dma_address(sg); + len = sg_dma_len(sg); } - i += j + 1; } - spin_unlock(&ddp->map_lock); - log_debug(1 << CXGBI_DBG_DDP, - "NO suitable entries %u available.\n", count); - return -EBUSY; -} - -static inline void ddp_unmark_entries(struct cxgbi_ddp_info *ddp, - int start, int count) -{ - spin_lock(&ddp->map_lock); - memset(&ddp->gl_map[start], 0, - count * sizeof(struct cxgbi_gather_list *)); - spin_unlock(&ddp->map_lock); + ppod->addr[i] = sg ? cpu_to_be64(addr + offset) : 0ULL; } +EXPORT_SYMBOL_GPL(cxgbi_ddp_set_one_ppod); -static inline void ddp_gl_unmap(struct pci_dev *pdev, - struct cxgbi_gather_list *gl) -{ - int i; +/* + * APIs interacting with open-iscsi libraries + */ - for (i = 0; i < gl->nelem; i++) - dma_unmap_page(&pdev->dev, gl->phys_addr[i], PAGE_SIZE, - PCI_DMA_FROMDEVICE); -} +static unsigned char padding[4]; -static inline int ddp_gl_map(struct pci_dev *pdev, - struct cxgbi_gather_list *gl) +void cxgbi_ddp_ppm_setup(void **ppm_pp, struct cxgbi_device *cdev, + struct cxgbi_tag_format *tformat, unsigned int ppmax, + unsigned int llimit, unsigned int start, + unsigned int rsvd_factor) { - int i; + int err = cxgbi_ppm_init(ppm_pp, cdev->ports[0], cdev->pdev, + cdev->lldev, tformat, ppmax, llimit, start, + rsvd_factor); - for (i = 0; i < gl->nelem; i++) { - gl->phys_addr[i] = dma_map_page(&pdev->dev, gl->pages[i], 0, - PAGE_SIZE, - PCI_DMA_FROMDEVICE); - if (unlikely(dma_mapping_error(&pdev->dev, gl->phys_addr[i]))) { - log_debug(1 << CXGBI_DBG_DDP, - "page %d 0x%p, 0x%p dma mapping err.\n", - i, gl->pages[i], pdev); - goto unmap; - } - } - return i; -unmap: - if (i) { - unsigned int nelem = gl->nelem; + if (err >= 0) { + struct cxgbi_ppm *ppm = (struct cxgbi_ppm *)(*ppm_pp); - gl->nelem = i; - ddp_gl_unmap(pdev, gl); - gl->nelem = nelem; + if (ppm->ppmax < 1024 || + ppm->tformat.pgsz_idx_dflt >= DDP_PGIDX_MAX) + cdev->flags |= CXGBI_FLAG_DDP_OFF; + err = 0; + } else { + cdev->flags |= CXGBI_FLAG_DDP_OFF; } - return -EINVAL; -} - -static void ddp_release_gl(struct cxgbi_gather_list *gl, - struct pci_dev *pdev) -{ - ddp_gl_unmap(pdev, gl); - kfree(gl); } +EXPORT_SYMBOL_GPL(cxgbi_ddp_ppm_setup); -static struct cxgbi_gather_list *ddp_make_gl(unsigned int xferlen, - struct scatterlist *sgl, - unsigned int sgcnt, - struct pci_dev *pdev, - gfp_t gfp) +static int cxgbi_ddp_sgl_check(struct scatterlist *sgl, int nents) { - struct cxgbi_gather_list *gl; + int i; + int last_sgidx = nents - 1; struct scatterlist *sg = sgl; - struct page *sgpage = sg_page(sg); - unsigned int sglen = sg->length; - unsigned int sgoffset = sg->offset; - unsigned int npages = (xferlen + sgoffset + PAGE_SIZE - 1) >> - PAGE_SHIFT; - int i = 1, j = 0; - - if (xferlen < DDP_THRESHOLD) { - log_debug(1 << CXGBI_DBG_DDP, - "xfer %u < threshold %u, no ddp.\n", - xferlen, DDP_THRESHOLD); - return NULL; - } - - gl = kzalloc(sizeof(struct cxgbi_gather_list) + - npages * (sizeof(dma_addr_t) + - sizeof(struct page *)), gfp); - if (!gl) { - log_debug(1 << CXGBI_DBG_DDP, - "xfer %u, %u pages, OOM.\n", xferlen, npages); - return NULL; - } - log_debug(1 << CXGBI_DBG_DDP, - "xfer %u, sgl %u, gl max %u.\n", xferlen, sgcnt, npages); - - gl->pages = (struct page **)&gl->phys_addr[npages]; - gl->nelem = npages; - gl->length = xferlen; - gl->offset = sgoffset; - gl->pages[0] = sgpage; - - for (i = 1, sg = sg_next(sgl), j = 0; i < sgcnt; - i++, sg = sg_next(sg)) { - struct page *page = sg_page(sg); - - if (sgpage == page && sg->offset == sgoffset + sglen) - sglen += sg->length; - else { - /* make sure the sgl is fit for ddp: - * each has the same page size, and - * all of the middle pages are used completely - */ - if ((j && sgoffset) || ((i != sgcnt - 1) && - ((sglen + sgoffset) & ~PAGE_MASK))) { - log_debug(1 << CXGBI_DBG_DDP, - "page %d/%u, %u + %u.\n", - i, sgcnt, sgoffset, sglen); - goto error_out; - } + for (i = 0; i < nents; i++, sg = sg_next(sg)) { + unsigned int len = sg->length + sg->offset; - j++; - if (j == gl->nelem || sg->offset) { - log_debug(1 << CXGBI_DBG_DDP, - "page %d/%u, offset %u.\n", - j, gl->nelem, sg->offset); - goto error_out; - } - gl->pages[j] = page; - sglen = sg->length; - sgoffset = sg->offset; - sgpage = page; - } - } - gl->nelem = ++j; - - if (ddp_gl_map(pdev, gl) < 0) - goto error_out; - - return gl; - -error_out: - kfree(gl); - return NULL; -} - -static void ddp_tag_release(struct cxgbi_hba *chba, u32 tag) -{ - struct cxgbi_device *cdev = chba->cdev; - struct cxgbi_ddp_info *ddp = cdev->ddp; - u32 idx; - - idx = (tag >> PPOD_IDX_SHIFT) & ddp->idx_mask; - if (idx < ddp->nppods) { - struct cxgbi_gather_list *gl = ddp->gl_map[idx]; - unsigned int npods; - - if (!gl || !gl->nelem) { - pr_warn("tag 0x%x, idx %u, gl 0x%p, %u.\n", - tag, idx, gl, gl ? gl->nelem : 0); - return; - } - npods = (gl->nelem + PPOD_PAGES_MAX - 1) >> PPOD_PAGES_SHIFT; - log_debug(1 << CXGBI_DBG_DDP, - "tag 0x%x, release idx %u, npods %u.\n", - tag, idx, npods); - cdev->csk_ddp_clear(chba, tag, idx, npods); - ddp_unmark_entries(ddp, idx, npods); - ddp_release_gl(gl, ddp->pdev); - } else - pr_warn("tag 0x%x, idx %u > max %u.\n", tag, idx, ddp->nppods); -} - -static int ddp_tag_reserve(struct cxgbi_sock *csk, unsigned int tid, - u32 sw_tag, u32 *tagp, struct cxgbi_gather_list *gl, - gfp_t gfp) -{ - struct cxgbi_device *cdev = csk->cdev; - struct cxgbi_ddp_info *ddp = cdev->ddp; - struct cxgbi_tag_format *tformat = &cdev->tag_format; - struct cxgbi_pagepod_hdr hdr; - unsigned int npods; - int idx = -1; - int err = -ENOMEM; - u32 tag; - - npods = (gl->nelem + PPOD_PAGES_MAX - 1) >> PPOD_PAGES_SHIFT; - if (ddp->idx_last == ddp->nppods) - idx = ddp_find_unused_entries(ddp, 0, ddp->nppods, - npods, gl); - else { - idx = ddp_find_unused_entries(ddp, ddp->idx_last + 1, - ddp->nppods, npods, - gl); - if (idx < 0 && ddp->idx_last >= npods) { - idx = ddp_find_unused_entries(ddp, 0, - min(ddp->idx_last + npods, ddp->nppods), - npods, gl); + if ((sg->offset & 0x3) || (i && sg->offset) || + ((i != last_sgidx) && len != PAGE_SIZE)) { + log_debug(1 << CXGBI_DBG_DDP, + "sg %u/%u, %u,%u, not aligned.\n", + i, nents, sg->offset, sg->length); + goto err_out; } } - if (idx < 0) { - log_debug(1 << CXGBI_DBG_DDP, - "xferlen %u, gl %u, npods %u NO DDP.\n", - gl->length, gl->nelem, npods); - return idx; - } - - tag = cxgbi_ddp_tag_base(tformat, sw_tag); - tag |= idx << PPOD_IDX_SHIFT; - - hdr.rsvd = 0; - hdr.vld_tid = htonl(PPOD_VALID_FLAG | PPOD_TID(tid)); - hdr.pgsz_tag_clr = htonl(tag & ddp->rsvd_tag_mask); - hdr.max_offset = htonl(gl->length); - hdr.page_offset = htonl(gl->offset); - - err = cdev->csk_ddp_set(csk, &hdr, idx, npods, gl); - if (err < 0) - goto unmark_entries; - - ddp->idx_last = idx; - log_debug(1 << CXGBI_DBG_DDP, - "xfer %u, gl %u,%u, tid 0x%x, tag 0x%x->0x%x(%u,%u).\n", - gl->length, gl->nelem, gl->offset, tid, sw_tag, tag, idx, - npods); - *tagp = tag; return 0; - -unmark_entries: - ddp_unmark_entries(ddp, idx, npods); - return err; +err_out: + return -EINVAL; } -int cxgbi_ddp_reserve(struct cxgbi_sock *csk, unsigned int *tagp, - unsigned int sw_tag, unsigned int xferlen, - struct scatterlist *sgl, unsigned int sgcnt, gfp_t gfp) +static int cxgbi_ddp_reserve(struct cxgbi_conn *cconn, + struct cxgbi_task_data *tdata, u32 sw_tag, + unsigned int xferlen) { + struct cxgbi_sock *csk = cconn->cep->csk; struct cxgbi_device *cdev = csk->cdev; - struct cxgbi_tag_format *tformat = &cdev->tag_format; - struct cxgbi_gather_list *gl; + struct cxgbi_ppm *ppm = cdev->cdev2ppm(cdev); + struct cxgbi_task_tag_info *ttinfo = &tdata->ttinfo; + struct scatterlist *sgl = ttinfo->sgl; + unsigned int sgcnt = ttinfo->nents; + unsigned int sg_offset = sgl->offset; int err; - if (page_idx >= DDP_PGIDX_MAX || !cdev->ddp || - xferlen < DDP_THRESHOLD) { + if (cdev->flags & CXGBI_FLAG_DDP_OFF) { log_debug(1 << CXGBI_DBG_DDP, - "pgidx %u, xfer %u, NO ddp.\n", page_idx, xferlen); + "cdev 0x%p DDP off.\n", cdev); return -EINVAL; } - if (!cxgbi_sw_tag_usable(tformat, sw_tag)) { + if (!ppm || xferlen < DDP_THRESHOLD || !sgcnt || + ppm->tformat.pgsz_idx_dflt >= DDP_PGIDX_MAX) { log_debug(1 << CXGBI_DBG_DDP, - "sw_tag 0x%x NOT usable.\n", sw_tag); + "ppm 0x%p, pgidx %u, xfer %u, sgcnt %u, NO ddp.\n", + ppm, ppm ? ppm->tformat.pgsz_idx_dflt : DDP_PGIDX_MAX, + xferlen, ttinfo->nents); return -EINVAL; } - gl = ddp_make_gl(xferlen, sgl, sgcnt, cdev->pdev, gfp); - if (!gl) - return -ENOMEM; + /* make sure the buffer is suitable for ddp */ + if (cxgbi_ddp_sgl_check(sgl, sgcnt) < 0) + return -EINVAL; - err = ddp_tag_reserve(csk, csk->tid, sw_tag, tagp, gl, gfp); - if (err < 0) - ddp_release_gl(gl, cdev->pdev); + ttinfo->nr_pages = (xferlen + sgl->offset + (1 << PAGE_SHIFT) - 1) >> + PAGE_SHIFT; - return err; -} + /* + * the ddp tag will be used for the itt in the outgoing pdu, + * the itt genrated by libiscsi is saved in the ppm and can be + * retrieved via the ddp tag + */ + err = cxgbi_ppm_ppods_reserve(ppm, ttinfo->nr_pages, 0, &ttinfo->idx, + &ttinfo->tag, (unsigned long)sw_tag); + if (err < 0) { + cconn->ddp_full++; + return err; + } + ttinfo->npods = err; -static void ddp_destroy(struct kref *kref) -{ - struct cxgbi_ddp_info *ddp = container_of(kref, - struct cxgbi_ddp_info, - refcnt); - struct cxgbi_device *cdev = ddp->cdev; - int i = 0; + /* setup dma from scsi command sgl */ + sgl->offset = 0; + err = dma_map_sg(&ppm->pdev->dev, sgl, sgcnt, DMA_FROM_DEVICE); + sgl->offset = sg_offset; + if (err == 0) { + pr_info("%s: 0x%x, xfer %u, sgl %u dma mapping err.\n", + __func__, sw_tag, xferlen, sgcnt); + goto rel_ppods; + } + if (err != ttinfo->nr_pages) { + log_debug(1 << CXGBI_DBG_DDP, + "%s: sw tag 0x%x, xfer %u, sgl %u, dma count %d.\n", + __func__, sw_tag, xferlen, sgcnt, err); + } - pr_info("kref 0, destroy ddp 0x%p, cdev 0x%p.\n", ddp, cdev); + ttinfo->flags |= CXGBI_PPOD_INFO_FLAG_MAPPED; + ttinfo->cid = csk->port_id; - while (i < ddp->nppods) { - struct cxgbi_gather_list *gl = ddp->gl_map[i]; + cxgbi_ppm_make_ppod_hdr(ppm, ttinfo->tag, csk->tid, sgl->offset, + xferlen, &ttinfo->hdr); - if (gl) { - int npods = (gl->nelem + PPOD_PAGES_MAX - 1) - >> PPOD_PAGES_SHIFT; - pr_info("cdev 0x%p, ddp %d + %d.\n", cdev, i, npods); - kfree(gl); - i += npods; - } else - i++; + if (cdev->flags & CXGBI_FLAG_USE_PPOD_OFLDQ) { + /* write ppod from xmit_pdu (of iscsi_scsi_command pdu) */ + ttinfo->flags |= CXGBI_PPOD_INFO_FLAG_VALID; + } else { + /* write ppod from control queue now */ + err = cdev->csk_ddp_set_map(ppm, csk, ttinfo); + if (err < 0) + goto rel_ppods; } - cxgbi_free_big_mem(ddp); -} - -int cxgbi_ddp_cleanup(struct cxgbi_device *cdev) -{ - struct cxgbi_ddp_info *ddp = cdev->ddp; - log_debug(1 << CXGBI_DBG_DDP, - "cdev 0x%p, release ddp 0x%p.\n", cdev, ddp); - cdev->ddp = NULL; - if (ddp) - return kref_put(&ddp->refcnt, ddp_destroy); return 0; -} -EXPORT_SYMBOL_GPL(cxgbi_ddp_cleanup); -int cxgbi_ddp_init(struct cxgbi_device *cdev, - unsigned int llimit, unsigned int ulimit, - unsigned int max_txsz, unsigned int max_rxsz) -{ - struct cxgbi_ddp_info *ddp; - unsigned int ppmax, bits; +rel_ppods: + cxgbi_ppm_ppod_release(ppm, ttinfo->idx); - ppmax = (ulimit - llimit + 1) >> PPOD_SIZE_SHIFT; - bits = __ilog2_u32(ppmax) + 1; - if (bits > PPOD_IDX_MAX_SIZE) - bits = PPOD_IDX_MAX_SIZE; - ppmax = (1 << (bits - 1)) - 1; - - ddp = cxgbi_alloc_big_mem(sizeof(struct cxgbi_ddp_info) + - ppmax * (sizeof(struct cxgbi_gather_list *) + - sizeof(struct sk_buff *)), - GFP_KERNEL); - if (!ddp) { - pr_warn("cdev 0x%p, ddp ppmax %u OOM.\n", cdev, ppmax); - return -ENOMEM; + if (ttinfo->flags & CXGBI_PPOD_INFO_FLAG_MAPPED) { + ttinfo->flags &= ~CXGBI_PPOD_INFO_FLAG_MAPPED; + dma_unmap_sg(&ppm->pdev->dev, sgl, sgcnt, DMA_FROM_DEVICE); } - ddp->gl_map = (struct cxgbi_gather_list **)(ddp + 1); - cdev->ddp = ddp; - - spin_lock_init(&ddp->map_lock); - kref_init(&ddp->refcnt); - - ddp->cdev = cdev; - ddp->pdev = cdev->pdev; - ddp->llimit = llimit; - ddp->ulimit = ulimit; - ddp->max_txsz = min_t(unsigned int, max_txsz, ULP2_MAX_PKT_SIZE); - ddp->max_rxsz = min_t(unsigned int, max_rxsz, ULP2_MAX_PKT_SIZE); - ddp->nppods = ppmax; - ddp->idx_last = ppmax; - ddp->idx_bits = bits; - ddp->idx_mask = (1 << bits) - 1; - ddp->rsvd_tag_mask = (1 << (bits + PPOD_IDX_SHIFT)) - 1; - - cdev->tag_format.sw_bits = sw_tag_idx_bits + sw_tag_age_bits; - cdev->tag_format.rsvd_bits = ddp->idx_bits; - cdev->tag_format.rsvd_shift = PPOD_IDX_SHIFT; - cdev->tag_format.rsvd_mask = (1 << cdev->tag_format.rsvd_bits) - 1; - - pr_info("%s tag format, sw %u, rsvd %u,%u, mask 0x%x.\n", - cdev->ports[0]->name, cdev->tag_format.sw_bits, - cdev->tag_format.rsvd_bits, cdev->tag_format.rsvd_shift, - cdev->tag_format.rsvd_mask); - - cdev->tx_max_size = min_t(unsigned int, ULP2_MAX_PDU_PAYLOAD, - ddp->max_txsz - ISCSI_PDU_NONPAYLOAD_LEN); - cdev->rx_max_size = min_t(unsigned int, ULP2_MAX_PDU_PAYLOAD, - ddp->max_rxsz - ISCSI_PDU_NONPAYLOAD_LEN); - - log_debug(1 << CXGBI_DBG_DDP, - "%s max payload size: %u/%u, %u/%u.\n", - cdev->ports[0]->name, cdev->tx_max_size, ddp->max_txsz, - cdev->rx_max_size, ddp->max_rxsz); - return 0; + return -EINVAL; } -EXPORT_SYMBOL_GPL(cxgbi_ddp_init); - -/* - * APIs interacting with open-iscsi libraries - */ - -static unsigned char padding[4]; static void task_release_itt(struct iscsi_task *task, itt_t hdr_itt) { struct scsi_cmnd *sc = task->sc; struct iscsi_tcp_conn *tcp_conn = task->conn->dd_data; struct cxgbi_conn *cconn = tcp_conn->dd_data; - struct cxgbi_hba *chba = cconn->chba; - struct cxgbi_tag_format *tformat = &chba->cdev->tag_format; + struct cxgbi_device *cdev = cconn->chba->cdev; + struct cxgbi_ppm *ppm = cdev->cdev2ppm(cdev); u32 tag = ntohl((__force u32)hdr_itt); log_debug(1 << CXGBI_DBG_DDP, - "cdev 0x%p, release tag 0x%x.\n", chba->cdev, tag); + "cdev 0x%p, task 0x%p, release tag 0x%x.\n", + cdev, task, tag); if (sc && (scsi_bidi_cmnd(sc) || sc->sc_data_direction == DMA_FROM_DEVICE) && - cxgbi_is_ddp_tag(tformat, tag)) - ddp_tag_release(chba, tag); + cxgbi_ppm_is_ddp_tag(ppm, tag)) { + struct cxgbi_task_data *tdata = iscsi_task_cxgbi_data(task); + struct cxgbi_task_tag_info *ttinfo = &tdata->ttinfo; + + if (!(cdev->flags & CXGBI_FLAG_USE_PPOD_OFLDQ)) + cdev->csk_ddp_clear_map(cdev, ppm, ttinfo); + cxgbi_ppm_ppod_release(ppm, ttinfo->idx); + dma_unmap_sg(&ppm->pdev->dev, ttinfo->sgl, ttinfo->nents, + DMA_FROM_DEVICE); + } +} + +static inline u32 cxgbi_build_sw_tag(u32 idx, u32 age) +{ + /* assume idx and age both are < 0x7FFF (32767) */ + return (idx << 16) | age; } static int task_reserve_itt(struct iscsi_task *task, itt_t *hdr_itt) @@ -1710,34 +1433,41 @@ static int task_reserve_itt(struct iscsi_task *task, itt_t *hdr_itt) struct iscsi_session *sess = conn->session; struct iscsi_tcp_conn *tcp_conn = conn->dd_data; struct cxgbi_conn *cconn = tcp_conn->dd_data; - struct cxgbi_hba *chba = cconn->chba; - struct cxgbi_tag_format *tformat = &chba->cdev->tag_format; - u32 sw_tag = (sess->age << cconn->task_idx_bits) | task->itt; + struct cxgbi_device *cdev = cconn->chba->cdev; + struct cxgbi_ppm *ppm = cdev->cdev2ppm(cdev); + u32 sw_tag = cxgbi_build_sw_tag(task->itt, sess->age); u32 tag = 0; int err = -EINVAL; if (sc && - (scsi_bidi_cmnd(sc) || sc->sc_data_direction == DMA_FROM_DEVICE)) { - err = cxgbi_ddp_reserve(cconn->cep->csk, &tag, sw_tag, - scsi_in(sc)->length, - scsi_in(sc)->table.sgl, - scsi_in(sc)->table.nents, - GFP_ATOMIC); - if (err < 0) - log_debug(1 << CXGBI_DBG_DDP, - "csk 0x%p, R task 0x%p, %u,%u, no ddp.\n", - cconn->cep->csk, task, scsi_in(sc)->length, - scsi_in(sc)->table.nents); + (scsi_bidi_cmnd(sc) || sc->sc_data_direction == DMA_FROM_DEVICE) + ) { + struct cxgbi_task_data *tdata = iscsi_task_cxgbi_data(task); + struct cxgbi_task_tag_info *ttinfo = &tdata->ttinfo; + + scmd_get_params(sc, &ttinfo->sgl, &ttinfo->nents, + &tdata->dlen, 0); + err = cxgbi_ddp_reserve(cconn, tdata, sw_tag, tdata->dlen); + if (!err) + tag = ttinfo->tag; + else + log_debug(1 << CXGBI_DBG_DDP, + "csk 0x%p, R task 0x%p, %u,%u, no ddp.\n", + cconn->cep->csk, task, tdata->dlen, + ttinfo->nents); } - if (err < 0) - tag = cxgbi_set_non_ddp_tag(tformat, sw_tag); + if (err < 0) { + err = cxgbi_ppm_make_non_ddp_tag(ppm, sw_tag, &tag); + if (err < 0) + return err; + } /* the itt need to sent in big-endian order */ *hdr_itt = (__force itt_t)htonl(tag); log_debug(1 << CXGBI_DBG_DDP, - "cdev 0x%p, task 0x%p, 0x%x(0x%x,0x%x)->0x%x/0x%x.\n", - chba->cdev, task, sw_tag, task->itt, sess->age, tag, *hdr_itt); + "cdev 0x%p, task 0x%p, 0x%x(0x%x,0x%x)->0x%x/0x%x.\n", + cdev, task, sw_tag, task->itt, sess->age, tag, *hdr_itt); return 0; } @@ -1746,19 +1476,24 @@ void cxgbi_parse_pdu_itt(struct iscsi_conn *conn, itt_t itt, int *idx, int *age) struct iscsi_tcp_conn *tcp_conn = conn->dd_data; struct cxgbi_conn *cconn = tcp_conn->dd_data; struct cxgbi_device *cdev = cconn->chba->cdev; - u32 tag = ntohl((__force u32) itt); + struct cxgbi_ppm *ppm = cdev->cdev2ppm(cdev); + u32 tag = ntohl((__force u32)itt); u32 sw_bits; - sw_bits = cxgbi_tag_nonrsvd_bits(&cdev->tag_format, tag); - if (idx) - *idx = sw_bits & ((1 << cconn->task_idx_bits) - 1); - if (age) - *age = (sw_bits >> cconn->task_idx_bits) & ISCSI_AGE_MASK; + if (ppm) { + if (cxgbi_ppm_is_ddp_tag(ppm, tag)) + sw_bits = cxgbi_ppm_get_tag_caller_data(ppm, tag); + else + sw_bits = cxgbi_ppm_decode_non_ddp_tag(ppm, tag); + } else { + sw_bits = tag; + } + cxgbi_decode_sw_tag(sw_bits, idx, age); log_debug(1 << CXGBI_DBG_DDP, - "cdev 0x%p, tag 0x%x/0x%x, -> 0x%x(0x%x,0x%x).\n", - cdev, tag, itt, sw_bits, idx ? *idx : 0xFFFFF, - age ? *age : 0xFF); + "cdev 0x%p, tag 0x%x/0x%x, -> 0x%x(0x%x,0x%x).\n", + cdev, tag, itt, sw_bits, idx ? *idx : 0xFFFFF, + age ? *age : 0xFF); } EXPORT_SYMBOL_GPL(cxgbi_parse_pdu_itt); @@ -2260,7 +1995,9 @@ int cxgbi_conn_xmit_pdu(struct iscsi_task *task) struct iscsi_tcp_conn *tcp_conn = task->conn->dd_data; struct cxgbi_conn *cconn = tcp_conn->dd_data; struct cxgbi_task_data *tdata = iscsi_task_cxgbi_data(task); + struct cxgbi_task_tag_info *ttinfo = &tdata->ttinfo; struct sk_buff *skb = tdata->skb; + struct cxgbi_sock *csk = NULL; unsigned int datalen; int err; @@ -2270,8 +2007,28 @@ int cxgbi_conn_xmit_pdu(struct iscsi_task *task) return 0; } + if (cconn && cconn->cep) + csk = cconn->cep->csk; + if (!csk) { + log_debug(1 << CXGBI_DBG_ISCSI | 1 << CXGBI_DBG_PDU_TX, + "task 0x%p, csk gone.\n", task); + return -EPIPE; + } + datalen = skb->data_len; tdata->skb = NULL; + + /* write ppod first if using ofldq to write ppod */ + if (ttinfo->flags & CXGBI_PPOD_INFO_FLAG_VALID) { + struct cxgbi_ppm *ppm = csk->cdev->cdev2ppm(csk->cdev); + + ttinfo->flags &= ~CXGBI_PPOD_INFO_FLAG_VALID; + if (csk->cdev->csk_ddp_set_map(ppm, csk, ttinfo) < 0) + pr_err("task 0x%p, ppod writing using ofldq failed.\n", + task); + /* continue. Let fl get the data */ + } + err = cxgbi_sock_send_pdus(cconn->cep->csk, skb); if (err > 0) { int pdulen = err; @@ -2313,12 +2070,14 @@ EXPORT_SYMBOL_GPL(cxgbi_conn_xmit_pdu); void cxgbi_cleanup_task(struct iscsi_task *task) { + struct iscsi_tcp_task *tcp_task = task->dd_data; struct cxgbi_task_data *tdata = iscsi_task_cxgbi_data(task); log_debug(1 << CXGBI_DBG_ISCSI, "task 0x%p, skb 0x%p, itt 0x%x.\n", task, tdata->skb, task->hdr_itt); + tcp_task->dd_data = NULL; /* never reached the xmit task callout */ if (tdata->skb) __kfree_skb(tdata->skb); @@ -2528,6 +2287,7 @@ int cxgbi_bind_conn(struct iscsi_cls_session *cls_session, struct iscsi_conn *conn = cls_conn->dd_data; struct iscsi_tcp_conn *tcp_conn = conn->dd_data; struct cxgbi_conn *cconn = tcp_conn->dd_data; + struct cxgbi_ppm *ppm; struct iscsi_endpoint *ep; struct cxgbi_endpoint *cep; struct cxgbi_sock *csk; @@ -2540,7 +2300,10 @@ int cxgbi_bind_conn(struct iscsi_cls_session *cls_session, /* setup ddp pagesize */ cep = ep->dd_data; csk = cep->csk; - err = csk->cdev->csk_ddp_setup_pgidx(csk, csk->tid, page_idx, 0); + + ppm = csk->cdev->cdev2ppm(csk->cdev); + err = csk->cdev->csk_ddp_setup_pgidx(csk, csk->tid, + ppm->tformat.pgsz_idx_dflt, 0); if (err < 0) return err; @@ -2915,16 +2678,7 @@ EXPORT_SYMBOL_GPL(cxgbi_attr_is_visible); static int __init libcxgbi_init_module(void) { - sw_tag_idx_bits = (__ilog2_u32(ISCSI_ITT_MASK)) + 1; - sw_tag_age_bits = (__ilog2_u32(ISCSI_AGE_MASK)) + 1; - pr_info("%s", version); - - pr_info("tag itt 0x%x, %u bits, age 0x%x, %u bits.\n", - ISCSI_ITT_MASK, sw_tag_idx_bits, - ISCSI_AGE_MASK, sw_tag_age_bits); - - ddp_setup_host_page_size(); return 0; } diff --git a/drivers/scsi/cxgbi/libcxgbi.h b/drivers/scsi/cxgbi/libcxgbi.h index 9842301f7980..e7802738f5d2 100644 --- a/drivers/scsi/cxgbi/libcxgbi.h +++ b/drivers/scsi/cxgbi/libcxgbi.h @@ -24,9 +24,12 @@ #include <linux/scatterlist.h> #include <linux/skbuff.h> #include <linux/vmalloc.h> +#include <linux/version.h> #include <scsi/scsi_device.h> #include <scsi/libiscsi_tcp.h> +#include <libcxgb_ppm.h> + enum cxgbi_dbg_flag { CXGBI_DBG_ISCSI, CXGBI_DBG_DDP, @@ -84,92 +87,11 @@ static inline unsigned int cxgbi_ulp_extra_len(int submode) return ulp2_extra_len[submode & 3]; } -/* - * struct pagepod_hdr, pagepod - pagepod format - */ - #define CPL_RX_DDP_STATUS_DDP_SHIFT 16 /* ddp'able */ #define CPL_RX_DDP_STATUS_PAD_SHIFT 19 /* pad error */ #define CPL_RX_DDP_STATUS_HCRC_SHIFT 20 /* hcrc error */ #define CPL_RX_DDP_STATUS_DCRC_SHIFT 21 /* dcrc error */ -struct cxgbi_pagepod_hdr { - u32 vld_tid; - u32 pgsz_tag_clr; - u32 max_offset; - u32 page_offset; - u64 rsvd; -}; - -#define PPOD_PAGES_MAX 4 -struct cxgbi_pagepod { - struct cxgbi_pagepod_hdr hdr; - u64 addr[PPOD_PAGES_MAX + 1]; -}; - -struct cxgbi_tag_format { - unsigned char sw_bits; - unsigned char rsvd_bits; - unsigned char rsvd_shift; - unsigned char filler[1]; - u32 rsvd_mask; -}; - -struct cxgbi_gather_list { - unsigned int tag; - unsigned int length; - unsigned int offset; - unsigned int nelem; - struct page **pages; - dma_addr_t phys_addr[0]; -}; - -struct cxgbi_ddp_info { - struct kref refcnt; - struct cxgbi_device *cdev; - struct pci_dev *pdev; - unsigned int max_txsz; - unsigned int max_rxsz; - unsigned int llimit; - unsigned int ulimit; - unsigned int nppods; - unsigned int idx_last; - unsigned char idx_bits; - unsigned char filler[3]; - unsigned int idx_mask; - unsigned int rsvd_tag_mask; - spinlock_t map_lock; - struct cxgbi_gather_list **gl_map; -}; - -#define DDP_PGIDX_MAX 4 -#define DDP_THRESHOLD 2048 - -#define PPOD_PAGES_SHIFT 2 /* 4 pages per pod */ - -#define PPOD_SIZE sizeof(struct cxgbi_pagepod) /* 64 */ -#define PPOD_SIZE_SHIFT 6 - -#define ULPMEM_DSGL_MAX_NPPODS 16 /* 1024/PPOD_SIZE */ -#define ULPMEM_IDATA_MAX_NPPODS 4 /* 256/PPOD_SIZE */ -#define PCIE_MEMWIN_MAX_NPPODS 16 /* 1024/PPOD_SIZE */ - -#define PPOD_COLOR_SHIFT 0 -#define PPOD_COLOR(x) ((x) << PPOD_COLOR_SHIFT) - -#define PPOD_IDX_SHIFT 6 -#define PPOD_IDX_MAX_SIZE 24 - -#define PPOD_TID_SHIFT 0 -#define PPOD_TID(x) ((x) << PPOD_TID_SHIFT) - -#define PPOD_TAG_SHIFT 6 -#define PPOD_TAG(x) ((x) << PPOD_TAG_SHIFT) - -#define PPOD_VALID_SHIFT 24 -#define PPOD_VALID(x) ((x) << PPOD_VALID_SHIFT) -#define PPOD_VALID_FLAG PPOD_VALID(1U) - /* * sge_opaque_hdr - * Opaque version of structure the SGE stores at skb->head of TX_DATA packets @@ -279,6 +201,8 @@ struct cxgbi_skb_tx_cb { enum cxgbi_skcb_flags { SKCBF_TX_NEED_HDR, /* packet needs a header */ + SKCBF_TX_MEM_WRITE, /* memory write */ + SKCBF_TX_FLAG_COMPL, /* wr completion flag */ SKCBF_RX_COALESCED, /* received whole pdu */ SKCBF_RX_HDR, /* received pdu header */ SKCBF_RX_DATA, /* received pdu payload */ @@ -527,6 +451,9 @@ struct cxgbi_ports_map { #define CXGBI_FLAG_DEV_T4 0x2 #define CXGBI_FLAG_ADAPTER_RESET 0x4 #define CXGBI_FLAG_IPV4_SET 0x10 +#define CXGBI_FLAG_USE_PPOD_OFLDQ 0x40 +#define CXGBI_FLAG_DDP_OFF 0x100 + struct cxgbi_device { struct list_head list_head; struct list_head rcu_node; @@ -548,15 +475,14 @@ struct cxgbi_device { unsigned int tx_max_size; unsigned int rx_max_size; struct cxgbi_ports_map pmap; - struct cxgbi_tag_format tag_format; - struct cxgbi_ddp_info *ddp; void (*dev_ddp_cleanup)(struct cxgbi_device *); - int (*csk_ddp_set)(struct cxgbi_sock *, struct cxgbi_pagepod_hdr *, - unsigned int, unsigned int, - struct cxgbi_gather_list *); - void (*csk_ddp_clear)(struct cxgbi_hba *, - unsigned int, unsigned int, unsigned int); + struct cxgbi_ppm* (*cdev2ppm)(struct cxgbi_device *); + int (*csk_ddp_set_map)(struct cxgbi_ppm *, struct cxgbi_sock *, + struct cxgbi_task_tag_info *); + void (*csk_ddp_clear_map)(struct cxgbi_device *cdev, + struct cxgbi_ppm *, + struct cxgbi_task_tag_info *); int (*csk_ddp_setup_digest)(struct cxgbi_sock *, unsigned int, int, int, int); int (*csk_ddp_setup_pgidx)(struct cxgbi_sock *, @@ -580,6 +506,8 @@ struct cxgbi_conn { struct iscsi_conn *iconn; struct cxgbi_hba *chba; u32 task_idx_bits; + unsigned int ddp_full; + unsigned int ddp_tag_full; }; struct cxgbi_endpoint { @@ -593,85 +521,15 @@ struct cxgbi_task_data { unsigned short nr_frags; struct page_frag frags[MAX_PDU_FRAGS]; struct sk_buff *skb; + unsigned int dlen; unsigned int offset; unsigned int count; unsigned int sgoffset; + struct cxgbi_task_tag_info ttinfo; }; #define iscsi_task_cxgbi_data(task) \ ((task)->dd_data + sizeof(struct iscsi_tcp_task)) -static inline int cxgbi_is_ddp_tag(struct cxgbi_tag_format *tformat, u32 tag) -{ - return !(tag & (1 << (tformat->rsvd_bits + tformat->rsvd_shift - 1))); -} - -static inline int cxgbi_sw_tag_usable(struct cxgbi_tag_format *tformat, - u32 sw_tag) -{ - sw_tag >>= (32 - tformat->rsvd_bits); - return !sw_tag; -} - -static inline u32 cxgbi_set_non_ddp_tag(struct cxgbi_tag_format *tformat, - u32 sw_tag) -{ - unsigned char shift = tformat->rsvd_bits + tformat->rsvd_shift - 1; - u32 mask = (1 << shift) - 1; - - if (sw_tag && (sw_tag & ~mask)) { - u32 v1 = sw_tag & ((1 << shift) - 1); - u32 v2 = (sw_tag >> (shift - 1)) << shift; - - return v2 | v1 | 1 << shift; - } - - return sw_tag | 1 << shift; -} - -static inline u32 cxgbi_ddp_tag_base(struct cxgbi_tag_format *tformat, - u32 sw_tag) -{ - u32 mask = (1 << tformat->rsvd_shift) - 1; - - if (sw_tag && (sw_tag & ~mask)) { - u32 v1 = sw_tag & mask; - u32 v2 = sw_tag >> tformat->rsvd_shift; - - v2 <<= tformat->rsvd_bits + tformat->rsvd_shift; - - return v2 | v1; - } - - return sw_tag; -} - -static inline u32 cxgbi_tag_rsvd_bits(struct cxgbi_tag_format *tformat, - u32 tag) -{ - if (cxgbi_is_ddp_tag(tformat, tag)) - return (tag >> tformat->rsvd_shift) & tformat->rsvd_mask; - - return 0; -} - -static inline u32 cxgbi_tag_nonrsvd_bits(struct cxgbi_tag_format *tformat, - u32 tag) -{ - unsigned char shift = tformat->rsvd_bits + tformat->rsvd_shift - 1; - u32 v1, v2; - - if (cxgbi_is_ddp_tag(tformat, tag)) { - v1 = tag & ((1 << tformat->rsvd_shift) - 1); - v2 = (tag >> (shift + 1)) << tformat->rsvd_shift; - } else { - u32 mask = (1 << shift) - 1; - tag &= ~(1 << shift); - v1 = tag & mask; - v2 = (tag >> 1) & ~mask; - } - return v1 | v2; -} - static inline void *cxgbi_alloc_big_mem(unsigned int size, gfp_t gfp) { @@ -749,7 +607,11 @@ int cxgbi_ddp_init(struct cxgbi_device *, unsigned int, unsigned int, unsigned int, unsigned int); int cxgbi_ddp_cleanup(struct cxgbi_device *); void cxgbi_ddp_page_size_factor(int *); -void cxgbi_ddp_ppod_clear(struct cxgbi_pagepod *); -void cxgbi_ddp_ppod_set(struct cxgbi_pagepod *, struct cxgbi_pagepod_hdr *, - struct cxgbi_gather_list *, unsigned int); +void cxgbi_ddp_set_one_ppod(struct cxgbi_pagepod *, + struct cxgbi_task_tag_info *, + struct scatterlist **sg_pp, unsigned int *sg_off); +void cxgbi_ddp_ppm_setup(void **ppm_pp, struct cxgbi_device *, + struct cxgbi_tag_format *, unsigned int ppmax, + unsigned int llimit, unsigned int start, + unsigned int rsvd_factor); #endif /*__LIBCXGBI_H__*/ diff --git a/drivers/scsi/cxlflash/main.h b/drivers/scsi/cxlflash/main.h index f54bbd5a6062..e43545c86bcf 100644 --- a/drivers/scsi/cxlflash/main.h +++ b/drivers/scsi/cxlflash/main.h @@ -102,8 +102,4 @@ struct asyc_intr_info { #define SCAN_HOST 0x04 }; -#ifndef CONFIG_CXL_EEH -#define cxl_perst_reloads_same_image(_a, _b) do { } while (0) -#endif - #endif /* _CXLFLASH_MAIN_H */ diff --git a/drivers/scsi/ibmvscsi/ibmvfc.h b/drivers/scsi/ibmvscsi/ibmvfc.h index 8fae03215a85..5c70a52ad346 100644 --- a/drivers/scsi/ibmvscsi/ibmvfc.h +++ b/drivers/scsi/ibmvscsi/ibmvfc.h @@ -26,7 +26,7 @@ #include <linux/list.h> #include <linux/types.h> -#include "viosrp.h" +#include <scsi/viosrp.h> #define IBMVFC_NAME "ibmvfc" #define IBMVFC_DRIVER_VERSION "1.0.11" diff --git a/drivers/scsi/ibmvscsi/ibmvscsi.h b/drivers/scsi/ibmvscsi/ibmvscsi.h index 1067367395cd..e0f6c3aeb4ee 100644 --- a/drivers/scsi/ibmvscsi/ibmvscsi.h +++ b/drivers/scsi/ibmvscsi/ibmvscsi.h @@ -33,7 +33,7 @@ #include <linux/list.h> #include <linux/completion.h> #include <linux/interrupt.h> -#include "viosrp.h" +#include <scsi/viosrp.h> struct scsi_cmnd; struct Scsi_Host; diff --git a/drivers/scsi/ibmvscsi/viosrp.h b/drivers/scsi/ibmvscsi/viosrp.h deleted file mode 100644 index c1ab8a4c3161..000000000000 --- a/drivers/scsi/ibmvscsi/viosrp.h +++ /dev/null @@ -1,225 +0,0 @@ -/*****************************************************************************/ -/* srp.h -- SCSI RDMA Protocol definitions */ -/* */ -/* Written By: Colin Devilbis, IBM Corporation */ -/* */ -/* Copyright (C) 2003 IBM Corporation */ -/* */ -/* This program is free software; you can redistribute it and/or modify */ -/* it under the terms of the GNU General Public License as published by */ -/* the Free Software Foundation; either version 2 of the License, or */ -/* (at your option) any later version. */ -/* */ -/* This program is distributed in the hope that it will be useful, */ -/* but WITHOUT ANY WARRANTY; without even the implied warranty of */ -/* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the */ -/* GNU General Public License for more details. */ -/* */ -/* You should have received a copy of the GNU General Public License */ -/* along with this program; if not, write to the Free Software */ -/* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ -/* */ -/* */ -/* This file contains structures and definitions for IBM RPA (RS/6000 */ -/* platform architecture) implementation of the SRP (SCSI RDMA Protocol) */ -/* standard. SRP is used on IBM iSeries and pSeries platforms to send SCSI */ -/* commands between logical partitions. */ -/* */ -/* SRP Information Units (IUs) are sent on a "Command/Response Queue" (CRQ) */ -/* between partitions. The definitions in this file are architected, */ -/* and cannot be changed without breaking compatibility with other versions */ -/* of Linux and other operating systems (AIX, OS/400) that talk this protocol*/ -/* between logical partitions */ -/*****************************************************************************/ -#ifndef VIOSRP_H -#define VIOSRP_H -#include <scsi/srp.h> - -#define SRP_VERSION "16.a" -#define SRP_MAX_IU_LEN 256 -#define SRP_MAX_LOC_LEN 32 - -union srp_iu { - struct srp_login_req login_req; - struct srp_login_rsp login_rsp; - struct srp_login_rej login_rej; - struct srp_i_logout i_logout; - struct srp_t_logout t_logout; - struct srp_tsk_mgmt tsk_mgmt; - struct srp_cmd cmd; - struct srp_rsp rsp; - u8 reserved[SRP_MAX_IU_LEN]; -}; - -enum viosrp_crq_headers { - VIOSRP_CRQ_FREE = 0x00, - VIOSRP_CRQ_CMD_RSP = 0x80, - VIOSRP_CRQ_INIT_RSP = 0xC0, - VIOSRP_CRQ_XPORT_EVENT = 0xFF -}; - -enum viosrp_crq_init_formats { - VIOSRP_CRQ_INIT = 0x01, - VIOSRP_CRQ_INIT_COMPLETE = 0x02 -}; - -enum viosrp_crq_formats { - VIOSRP_SRP_FORMAT = 0x01, - VIOSRP_MAD_FORMAT = 0x02, - VIOSRP_OS400_FORMAT = 0x03, - VIOSRP_AIX_FORMAT = 0x04, - VIOSRP_LINUX_FORMAT = 0x05, - VIOSRP_INLINE_FORMAT = 0x06 -}; - -enum viosrp_crq_status { - VIOSRP_OK = 0x0, - VIOSRP_NONRECOVERABLE_ERR = 0x1, - VIOSRP_VIOLATES_MAX_XFER = 0x2, - VIOSRP_PARTNER_PANIC = 0x3, - VIOSRP_DEVICE_BUSY = 0x8, - VIOSRP_ADAPTER_FAIL = 0x10, - VIOSRP_OK2 = 0x99, -}; - -struct viosrp_crq { - u8 valid; /* used by RPA */ - u8 format; /* SCSI vs out-of-band */ - u8 reserved; - u8 status; /* non-scsi failure? (e.g. DMA failure) */ - __be16 timeout; /* in seconds */ - __be16 IU_length; /* in bytes */ - __be64 IU_data_ptr; /* the TCE for transferring data */ -}; - -/* MADs are Management requests above and beyond the IUs defined in the SRP - * standard. - */ -enum viosrp_mad_types { - VIOSRP_EMPTY_IU_TYPE = 0x01, - VIOSRP_ERROR_LOG_TYPE = 0x02, - VIOSRP_ADAPTER_INFO_TYPE = 0x03, - VIOSRP_CAPABILITIES_TYPE = 0x05, - VIOSRP_ENABLE_FAST_FAIL = 0x08, -}; - -enum viosrp_mad_status { - VIOSRP_MAD_SUCCESS = 0x00, - VIOSRP_MAD_NOT_SUPPORTED = 0xF1, - VIOSRP_MAD_FAILED = 0xF7, -}; - -enum viosrp_capability_type { - MIGRATION_CAPABILITIES = 0x01, - RESERVATION_CAPABILITIES = 0x02, -}; - -enum viosrp_capability_support { - SERVER_DOES_NOT_SUPPORTS_CAP = 0x0, - SERVER_SUPPORTS_CAP = 0x01, - SERVER_CAP_DATA = 0x02, -}; - -enum viosrp_reserve_type { - CLIENT_RESERVE_SCSI_2 = 0x01, -}; - -enum viosrp_capability_flag { - CLIENT_MIGRATED = 0x01, - CLIENT_RECONNECT = 0x02, - CAP_LIST_SUPPORTED = 0x04, - CAP_LIST_DATA = 0x08, -}; - -/* - * Common MAD header - */ -struct mad_common { - __be32 type; - __be16 status; - __be16 length; - __be64 tag; -}; - -/* - * All SRP (and MAD) requests normally flow from the - * client to the server. There is no way for the server to send - * an asynchronous message back to the client. The Empty IU is used - * to hang out a meaningless request to the server so that it can respond - * asynchrouously with something like a SCSI AER - */ -struct viosrp_empty_iu { - struct mad_common common; - __be64 buffer; - __be32 port; -}; - -struct viosrp_error_log { - struct mad_common common; - __be64 buffer; -}; - -struct viosrp_adapter_info { - struct mad_common common; - __be64 buffer; -}; - -struct viosrp_fast_fail { - struct mad_common common; -}; - -struct viosrp_capabilities { - struct mad_common common; - __be64 buffer; -}; - -struct mad_capability_common { - __be32 cap_type; - __be16 length; - __be16 server_support; -}; - -struct mad_reserve_cap { - struct mad_capability_common common; - __be32 type; -}; - -struct mad_migration_cap { - struct mad_capability_common common; - __be32 ecl; -}; - -struct capabilities{ - __be32 flags; - char name[SRP_MAX_LOC_LEN]; - char loc[SRP_MAX_LOC_LEN]; - struct mad_migration_cap migration; - struct mad_reserve_cap reserve; -}; - -union mad_iu { - struct viosrp_empty_iu empty_iu; - struct viosrp_error_log error_log; - struct viosrp_adapter_info adapter_info; - struct viosrp_fast_fail fast_fail; - struct viosrp_capabilities capabilities; -}; - -union viosrp_iu { - union srp_iu srp; - union mad_iu mad; -}; - -struct mad_adapter_info_data { - char srp_version[8]; - char partition_name[96]; - __be32 partition_number; -#define SRP_MAD_VERSION_1 1 - __be32 mad_version; -#define SRP_MAD_OS_LINUX 2 -#define SRP_MAD_OS_AIX 3 - __be32 os_type; - __be32 port_max_txu[8]; /* per-port maximum transfer */ -}; - -#endif diff --git a/drivers/scsi/ibmvscsi_tgt/Makefile b/drivers/scsi/ibmvscsi_tgt/Makefile new file mode 100644 index 000000000000..0c060ce64cb0 --- /dev/null +++ b/drivers/scsi/ibmvscsi_tgt/Makefile @@ -0,0 +1,3 @@ +obj-$(CONFIG_SCSI_IBMVSCSIS) += ibmvscsis.o + +ibmvscsis-y := libsrp.o ibmvscsi_tgt.o diff --git a/drivers/scsi/ibmvscsi_tgt/ibmvscsi_tgt.c b/drivers/scsi/ibmvscsi_tgt/ibmvscsi_tgt.c new file mode 100644 index 000000000000..b29fef9d0f27 --- /dev/null +++ b/drivers/scsi/ibmvscsi_tgt/ibmvscsi_tgt.c @@ -0,0 +1,4087 @@ +/******************************************************************************* + * IBM Virtual SCSI Target Driver + * Copyright (C) 2003-2005 Dave Boutcher (boutcher@us.ibm.com) IBM Corp. + * Santiago Leon (santil@us.ibm.com) IBM Corp. + * Linda Xie (lxie@us.ibm.com) IBM Corp. + * + * Copyright (C) 2005-2011 FUJITA Tomonori <tomof@acm.org> + * Copyright (C) 2010 Nicholas A. Bellinger <nab@kernel.org> + * + * Authors: Bryant G. Ly <bryantly@linux.vnet.ibm.com> + * Authors: Michael Cyr <mikecyr@linux.vnet.ibm.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + ****************************************************************************/ + +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + +#include <linux/module.h> +#include <linux/kernel.h> +#include <linux/slab.h> +#include <linux/types.h> +#include <linux/list.h> +#include <linux/string.h> + +#include <target/target_core_base.h> +#include <target/target_core_fabric.h> + +#include <asm/hvcall.h> +#include <asm/vio.h> + +#include <scsi/viosrp.h> + +#include "ibmvscsi_tgt.h" + +#define IBMVSCSIS_VERSION "v0.2" + +#define INITIAL_SRP_LIMIT 800 +#define DEFAULT_MAX_SECTORS 256 + +static uint max_vdma_size = MAX_H_COPY_RDMA; + +static char system_id[SYS_ID_NAME_LEN] = ""; +static char partition_name[PARTITION_NAMELEN] = "UNKNOWN"; +static uint partition_number = -1; + +/* Adapter list and lock to control it */ +static DEFINE_SPINLOCK(ibmvscsis_dev_lock); +static LIST_HEAD(ibmvscsis_dev_list); + +static long ibmvscsis_parse_command(struct scsi_info *vscsi, + struct viosrp_crq *crq); + +static void ibmvscsis_adapter_idle(struct scsi_info *vscsi); + +static void ibmvscsis_determine_resid(struct se_cmd *se_cmd, + struct srp_rsp *rsp) +{ + u32 residual_count = se_cmd->residual_count; + + if (!residual_count) + return; + + if (se_cmd->se_cmd_flags & SCF_UNDERFLOW_BIT) { + if (se_cmd->data_direction == DMA_TO_DEVICE) { + /* residual data from an underflow write */ + rsp->flags = SRP_RSP_FLAG_DOUNDER; + rsp->data_out_res_cnt = cpu_to_be32(residual_count); + } else if (se_cmd->data_direction == DMA_FROM_DEVICE) { + /* residual data from an underflow read */ + rsp->flags = SRP_RSP_FLAG_DIUNDER; + rsp->data_in_res_cnt = cpu_to_be32(residual_count); + } + } else if (se_cmd->se_cmd_flags & SCF_OVERFLOW_BIT) { + if (se_cmd->data_direction == DMA_TO_DEVICE) { + /* residual data from an overflow write */ + rsp->flags = SRP_RSP_FLAG_DOOVER; + rsp->data_out_res_cnt = cpu_to_be32(residual_count); + } else if (se_cmd->data_direction == DMA_FROM_DEVICE) { + /* residual data from an overflow read */ + rsp->flags = SRP_RSP_FLAG_DIOVER; + rsp->data_in_res_cnt = cpu_to_be32(residual_count); + } + } +} + +/** + * connection_broken() - Determine if the connection to the client is good + * @vscsi: Pointer to our adapter structure + * + * This function attempts to send a ping MAD to the client. If the call to + * queue the request returns H_CLOSED then the connection has been broken + * and the function returns TRUE. + * + * EXECUTION ENVIRONMENT: + * Interrupt or Process environment + */ +static bool connection_broken(struct scsi_info *vscsi) +{ + struct viosrp_crq *crq; + u64 buffer[2] = { 0, 0 }; + long h_return_code; + bool rc = false; + + /* create a PING crq */ + crq = (struct viosrp_crq *)&buffer; + crq->valid = VALID_CMD_RESP_EL; + crq->format = MESSAGE_IN_CRQ; + crq->status = PING; + + h_return_code = h_send_crq(vscsi->dds.unit_id, + cpu_to_be64(buffer[MSG_HI]), + cpu_to_be64(buffer[MSG_LOW])); + + pr_debug("connection_broken: rc %ld\n", h_return_code); + + if (h_return_code == H_CLOSED) + rc = true; + + return rc; +} + +/** + * ibmvscsis_unregister_command_q() - Helper Function-Unregister Command Queue + * @vscsi: Pointer to our adapter structure + * + * This function calls h_free_q then frees the interrupt bit etc. + * It must release the lock before doing so because of the time it can take + * for h_free_crq in PHYP + * NOTE: the caller must make sure that state and or flags will prevent + * interrupt handler from scheduling work. + * NOTE: anyone calling this function may need to set the CRQ_CLOSED flag + * we can't do it here, because we don't have the lock + * + * EXECUTION ENVIRONMENT: + * Process level + */ +static long ibmvscsis_unregister_command_q(struct scsi_info *vscsi) +{ + long qrc; + long rc = ADAPT_SUCCESS; + int ticks = 0; + + do { + qrc = h_free_crq(vscsi->dds.unit_id); + switch (qrc) { + case H_SUCCESS: + break; + + case H_HARDWARE: + case H_PARAMETER: + dev_err(&vscsi->dev, "unregister_command_q: error from h_free_crq %ld\n", + qrc); + rc = ERROR; + break; + + case H_BUSY: + case H_LONG_BUSY_ORDER_1_MSEC: + /* msleep not good for small values */ + usleep_range(1000, 2000); + ticks += 1; + break; + case H_LONG_BUSY_ORDER_10_MSEC: + usleep_range(10000, 20000); + ticks += 10; + break; + case H_LONG_BUSY_ORDER_100_MSEC: + msleep(100); + ticks += 100; + break; + case H_LONG_BUSY_ORDER_1_SEC: + ssleep(1); + ticks += 1000; + break; + case H_LONG_BUSY_ORDER_10_SEC: + ssleep(10); + ticks += 10000; + break; + case H_LONG_BUSY_ORDER_100_SEC: + ssleep(100); + ticks += 100000; + break; + default: + dev_err(&vscsi->dev, "unregister_command_q: unknown error %ld from h_free_crq\n", + qrc); + rc = ERROR; + break; + } + + /* + * dont wait more then 300 seconds + * ticks are in milliseconds more or less + */ + if (ticks > 300000 && qrc != H_SUCCESS) { + rc = ERROR; + dev_err(&vscsi->dev, "Excessive wait for h_free_crq\n"); + } + } while (qrc != H_SUCCESS && rc == ADAPT_SUCCESS); + + pr_debug("Freeing CRQ: phyp rc %ld, rc %ld\n", qrc, rc); + + return rc; +} + +/** + * ibmvscsis_delete_client_info() - Helper function to Delete Client Info + * @vscsi: Pointer to our adapter structure + * @client_closed: True if client closed its queue + * + * Deletes information specific to the client when the client goes away + * + * EXECUTION ENVIRONMENT: + * Interrupt or Process + */ +static void ibmvscsis_delete_client_info(struct scsi_info *vscsi, + bool client_closed) +{ + vscsi->client_cap = 0; + + /* + * Some things we don't want to clear if we're closing the queue, + * because some clients don't resend the host handshake when they + * get a transport event. + */ + if (client_closed) + vscsi->client_data.os_type = 0; +} + +/** + * ibmvscsis_free_command_q() - Free Command Queue + * @vscsi: Pointer to our adapter structure + * + * This function calls unregister_command_q, then clears interrupts and + * any pending interrupt acknowledgments associated with the command q. + * It also clears memory if there is no error. + * + * PHYP did not meet the PAPR architecture so that we must give up the + * lock. This causes a timing hole regarding state change. To close the + * hole this routine does accounting on any change that occurred during + * the time the lock is not held. + * NOTE: must give up and then acquire the interrupt lock, the caller must + * make sure that state and or flags will prevent interrupt handler from + * scheduling work. + * + * EXECUTION ENVIRONMENT: + * Process level, interrupt lock is held + */ +static long ibmvscsis_free_command_q(struct scsi_info *vscsi) +{ + int bytes; + u32 flags_under_lock; + u16 state_under_lock; + long rc = ADAPT_SUCCESS; + + if (!(vscsi->flags & CRQ_CLOSED)) { + vio_disable_interrupts(vscsi->dma_dev); + + state_under_lock = vscsi->new_state; + flags_under_lock = vscsi->flags; + vscsi->phyp_acr_state = 0; + vscsi->phyp_acr_flags = 0; + + spin_unlock_bh(&vscsi->intr_lock); + rc = ibmvscsis_unregister_command_q(vscsi); + spin_lock_bh(&vscsi->intr_lock); + + if (state_under_lock != vscsi->new_state) + vscsi->phyp_acr_state = vscsi->new_state; + + vscsi->phyp_acr_flags = ((~flags_under_lock) & vscsi->flags); + + if (rc == ADAPT_SUCCESS) { + bytes = vscsi->cmd_q.size * PAGE_SIZE; + memset(vscsi->cmd_q.base_addr, 0, bytes); + vscsi->cmd_q.index = 0; + vscsi->flags |= CRQ_CLOSED; + + ibmvscsis_delete_client_info(vscsi, false); + } + + pr_debug("free_command_q: flags 0x%x, state 0x%hx, acr_flags 0x%x, acr_state 0x%hx\n", + vscsi->flags, vscsi->state, vscsi->phyp_acr_flags, + vscsi->phyp_acr_state); + } + return rc; +} + +/** + * ibmvscsis_cmd_q_dequeue() - Get valid Command element + * @mask: Mask to use in case index wraps + * @current_index: Current index into command queue + * @base_addr: Pointer to start of command queue + * + * Returns a pointer to a valid command element or NULL, if the command + * queue is empty + * + * EXECUTION ENVIRONMENT: + * Interrupt environment, interrupt lock held + */ +static struct viosrp_crq *ibmvscsis_cmd_q_dequeue(uint mask, + uint *current_index, + struct viosrp_crq *base_addr) +{ + struct viosrp_crq *ptr; + + ptr = base_addr + *current_index; + + if (ptr->valid) { + *current_index = (*current_index + 1) & mask; + dma_rmb(); + } else { + ptr = NULL; + } + + return ptr; +} + +/** + * ibmvscsis_send_init_message() - send initialize message to the client + * @vscsi: Pointer to our adapter structure + * @format: Which Init Message format to send + * + * EXECUTION ENVIRONMENT: + * Interrupt environment interrupt lock held + */ +static long ibmvscsis_send_init_message(struct scsi_info *vscsi, u8 format) +{ + struct viosrp_crq *crq; + u64 buffer[2] = { 0, 0 }; + long rc; + + crq = (struct viosrp_crq *)&buffer; + crq->valid = VALID_INIT_MSG; + crq->format = format; + rc = h_send_crq(vscsi->dds.unit_id, cpu_to_be64(buffer[MSG_HI]), + cpu_to_be64(buffer[MSG_LOW])); + + return rc; +} + +/** + * ibmvscsis_check_init_msg() - Check init message valid + * @vscsi: Pointer to our adapter structure + * @format: Pointer to return format of Init Message, if any. + * Set to UNUSED_FORMAT if no Init Message in queue. + * + * Checks if an initialize message was queued by the initiatior + * after the queue was created and before the interrupt was enabled. + * + * EXECUTION ENVIRONMENT: + * Process level only, interrupt lock held + */ +static long ibmvscsis_check_init_msg(struct scsi_info *vscsi, uint *format) +{ + struct viosrp_crq *crq; + long rc = ADAPT_SUCCESS; + + crq = ibmvscsis_cmd_q_dequeue(vscsi->cmd_q.mask, &vscsi->cmd_q.index, + vscsi->cmd_q.base_addr); + if (!crq) { + *format = (uint)UNUSED_FORMAT; + } else if (crq->valid == VALID_INIT_MSG && crq->format == INIT_MSG) { + *format = (uint)INIT_MSG; + crq->valid = INVALIDATE_CMD_RESP_EL; + dma_rmb(); + + /* + * the caller has ensured no initialize message was + * sent after the queue was + * created so there should be no other message on the queue. + */ + crq = ibmvscsis_cmd_q_dequeue(vscsi->cmd_q.mask, + &vscsi->cmd_q.index, + vscsi->cmd_q.base_addr); + if (crq) { + *format = (uint)(crq->format); + rc = ERROR; + crq->valid = INVALIDATE_CMD_RESP_EL; + dma_rmb(); + } + } else { + *format = (uint)(crq->format); + rc = ERROR; + crq->valid = INVALIDATE_CMD_RESP_EL; + dma_rmb(); + } + + return rc; +} + +/** + * ibmvscsis_establish_new_q() - Establish new CRQ queue + * @vscsi: Pointer to our adapter structure + * @new_state: New state being established after resetting the queue + * + * Must be called with interrupt lock held. + */ +static long ibmvscsis_establish_new_q(struct scsi_info *vscsi, uint new_state) +{ + long rc = ADAPT_SUCCESS; + uint format; + + vscsi->flags &= PRESERVE_FLAG_FIELDS; + vscsi->rsp_q_timer.timer_pops = 0; + vscsi->debit = 0; + vscsi->credit = 0; + + rc = vio_enable_interrupts(vscsi->dma_dev); + if (rc) { + pr_warn("reset_queue: failed to enable interrupts, rc %ld\n", + rc); + return rc; + } + + rc = ibmvscsis_check_init_msg(vscsi, &format); + if (rc) { + dev_err(&vscsi->dev, "reset_queue: check_init_msg failed, rc %ld\n", + rc); + return rc; + } + + if (format == UNUSED_FORMAT && new_state == WAIT_CONNECTION) { + rc = ibmvscsis_send_init_message(vscsi, INIT_MSG); + switch (rc) { + case H_SUCCESS: + case H_DROPPED: + case H_CLOSED: + rc = ADAPT_SUCCESS; + break; + + case H_PARAMETER: + case H_HARDWARE: + break; + + default: + vscsi->state = UNDEFINED; + rc = H_HARDWARE; + break; + } + } + + return rc; +} + +/** + * ibmvscsis_reset_queue() - Reset CRQ Queue + * @vscsi: Pointer to our adapter structure + * @new_state: New state to establish after resetting the queue + * + * This function calls h_free_q and then calls h_reg_q and does all + * of the bookkeeping to get us back to where we can communicate. + * + * Actually, we don't always call h_free_crq. A problem was discovered + * where one partition would close and reopen his queue, which would + * cause his partner to get a transport event, which would cause him to + * close and reopen his queue, which would cause the original partition + * to get a transport event, etc., etc. To prevent this, we don't + * actually close our queue if the client initiated the reset, (i.e. + * either we got a transport event or we have detected that the client's + * queue is gone) + * + * EXECUTION ENVIRONMENT: + * Process environment, called with interrupt lock held + */ +static void ibmvscsis_reset_queue(struct scsi_info *vscsi, uint new_state) +{ + int bytes; + long rc = ADAPT_SUCCESS; + + pr_debug("reset_queue: flags 0x%x\n", vscsi->flags); + + /* don't reset, the client did it for us */ + if (vscsi->flags & (CLIENT_FAILED | TRANS_EVENT)) { + vscsi->flags &= PRESERVE_FLAG_FIELDS; + vscsi->rsp_q_timer.timer_pops = 0; + vscsi->debit = 0; + vscsi->credit = 0; + vscsi->state = new_state; + vio_enable_interrupts(vscsi->dma_dev); + } else { + rc = ibmvscsis_free_command_q(vscsi); + if (rc == ADAPT_SUCCESS) { + vscsi->state = new_state; + + bytes = vscsi->cmd_q.size * PAGE_SIZE; + rc = h_reg_crq(vscsi->dds.unit_id, + vscsi->cmd_q.crq_token, bytes); + if (rc == H_CLOSED || rc == H_SUCCESS) { + rc = ibmvscsis_establish_new_q(vscsi, + new_state); + } + + if (rc != ADAPT_SUCCESS) { + pr_debug("reset_queue: reg_crq rc %ld\n", rc); + + vscsi->state = ERR_DISCONNECTED; + vscsi->flags |= RESPONSE_Q_DOWN; + ibmvscsis_free_command_q(vscsi); + } + } else { + vscsi->state = ERR_DISCONNECTED; + vscsi->flags |= RESPONSE_Q_DOWN; + } + } +} + +/** + * ibmvscsis_free_cmd_resources() - Free command resources + * @vscsi: Pointer to our adapter structure + * @cmd: Command which is not longer in use + * + * Must be called with interrupt lock held. + */ +static void ibmvscsis_free_cmd_resources(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd) +{ + struct iu_entry *iue = cmd->iue; + + switch (cmd->type) { + case TASK_MANAGEMENT: + case SCSI_CDB: + /* + * When the queue goes down this value is cleared, so it + * cannot be cleared in this general purpose function. + */ + if (vscsi->debit) + vscsi->debit -= 1; + break; + case ADAPTER_MAD: + vscsi->flags &= ~PROCESSING_MAD; + break; + case UNSET_TYPE: + break; + default: + dev_err(&vscsi->dev, "free_cmd_resources unknown type %d\n", + cmd->type); + break; + } + + cmd->iue = NULL; + list_add_tail(&cmd->list, &vscsi->free_cmd); + srp_iu_put(iue); + + if (list_empty(&vscsi->active_q) && list_empty(&vscsi->schedule_q) && + list_empty(&vscsi->waiting_rsp) && (vscsi->flags & WAIT_FOR_IDLE)) { + vscsi->flags &= ~WAIT_FOR_IDLE; + complete(&vscsi->wait_idle); + } +} + +/** + * ibmvscsis_disconnect() - Helper function to disconnect + * @work: Pointer to work_struct, gives access to our adapter structure + * + * An error has occurred or the driver received a Transport event, + * and the driver is requesting that the command queue be de-registered + * in a safe manner. If there is no outstanding I/O then we can stop the + * queue. If we are restarting the queue it will be reflected in the + * the state of the adapter. + * + * EXECUTION ENVIRONMENT: + * Process environment + */ +static void ibmvscsis_disconnect(struct work_struct *work) +{ + struct scsi_info *vscsi = container_of(work, struct scsi_info, + proc_work); + u16 new_state; + bool wait_idle = false; + long rc = ADAPT_SUCCESS; + + spin_lock_bh(&vscsi->intr_lock); + new_state = vscsi->new_state; + vscsi->new_state = 0; + + pr_debug("disconnect: flags 0x%x, state 0x%hx\n", vscsi->flags, + vscsi->state); + + /* + * check which state we are in and see if we + * should transitition to the new state + */ + switch (vscsi->state) { + /* Should never be called while in this state. */ + case NO_QUEUE: + /* + * Can never transition from this state; + * igonore errors and logout. + */ + case UNCONFIGURING: + break; + + /* can transition from this state to UNCONFIGURING */ + case ERR_DISCONNECT: + if (new_state == UNCONFIGURING) + vscsi->state = new_state; + break; + + /* + * Can transition from this state to to unconfiguring + * or err disconnect. + */ + case ERR_DISCONNECT_RECONNECT: + switch (new_state) { + case UNCONFIGURING: + case ERR_DISCONNECT: + vscsi->state = new_state; + break; + + case WAIT_IDLE: + break; + default: + break; + } + break; + + /* can transition from this state to UNCONFIGURING */ + case ERR_DISCONNECTED: + if (new_state == UNCONFIGURING) + vscsi->state = new_state; + break; + + /* + * If this is a transition into an error state. + * a client is attempting to establish a connection + * and has violated the RPA protocol. + * There can be nothing pending on the adapter although + * there can be requests in the command queue. + */ + case WAIT_ENABLED: + case PART_UP_WAIT_ENAB: + switch (new_state) { + case ERR_DISCONNECT: + vscsi->flags |= RESPONSE_Q_DOWN; + vscsi->state = new_state; + vscsi->flags &= ~(SCHEDULE_DISCONNECT | + DISCONNECT_SCHEDULED); + ibmvscsis_free_command_q(vscsi); + break; + case ERR_DISCONNECT_RECONNECT: + ibmvscsis_reset_queue(vscsi, WAIT_ENABLED); + break; + + /* should never happen */ + case WAIT_IDLE: + rc = ERROR; + dev_err(&vscsi->dev, "disconnect: invalid state %d for WAIT_IDLE\n", + vscsi->state); + break; + } + break; + + case WAIT_IDLE: + switch (new_state) { + case ERR_DISCONNECT: + case ERR_DISCONNECT_RECONNECT: + vscsi->state = new_state; + break; + } + break; + + /* + * Initiator has not done a successful srp login + * or has done a successful srp logout ( adapter was not + * busy). In the first case there can be responses queued + * waiting for space on the initiators response queue (MAD) + * The second case the adapter is idle. Assume the worse case, + * i.e. the second case. + */ + case WAIT_CONNECTION: + case CONNECTED: + case SRP_PROCESSING: + wait_idle = true; + vscsi->state = new_state; + break; + + /* can transition from this state to UNCONFIGURING */ + case UNDEFINED: + if (new_state == UNCONFIGURING) + vscsi->state = new_state; + break; + default: + break; + } + + if (wait_idle) { + pr_debug("disconnect start wait, active %d, sched %d\n", + (int)list_empty(&vscsi->active_q), + (int)list_empty(&vscsi->schedule_q)); + if (!list_empty(&vscsi->active_q) || + !list_empty(&vscsi->schedule_q)) { + vscsi->flags |= WAIT_FOR_IDLE; + pr_debug("disconnect flags 0x%x\n", vscsi->flags); + /* + * This routine is can not be called with the interrupt + * lock held. + */ + spin_unlock_bh(&vscsi->intr_lock); + wait_for_completion(&vscsi->wait_idle); + spin_lock_bh(&vscsi->intr_lock); + } + pr_debug("disconnect stop wait\n"); + + ibmvscsis_adapter_idle(vscsi); + } + + spin_unlock_bh(&vscsi->intr_lock); +} + +/** + * ibmvscsis_post_disconnect() - Schedule the disconnect + * @vscsi: Pointer to our adapter structure + * @new_state: State to move to after disconnecting + * @flag_bits: Flags to turn on in adapter structure + * + * If it's already been scheduled, then see if we need to "upgrade" + * the new state (if the one passed in is more "severe" than the + * previous one). + * + * PRECONDITION: + * interrupt lock is held + */ +static void ibmvscsis_post_disconnect(struct scsi_info *vscsi, uint new_state, + uint flag_bits) +{ + uint state; + + /* check the validity of the new state */ + switch (new_state) { + case UNCONFIGURING: + case ERR_DISCONNECT: + case ERR_DISCONNECT_RECONNECT: + case WAIT_IDLE: + break; + + default: + dev_err(&vscsi->dev, "post_disconnect: Invalid new state %d\n", + new_state); + return; + } + + vscsi->flags |= flag_bits; + + pr_debug("post_disconnect: new_state 0x%x, flag_bits 0x%x, vscsi->flags 0x%x, state %hx\n", + new_state, flag_bits, vscsi->flags, vscsi->state); + + if (!(vscsi->flags & (DISCONNECT_SCHEDULED | SCHEDULE_DISCONNECT))) { + vscsi->flags |= SCHEDULE_DISCONNECT; + vscsi->new_state = new_state; + + INIT_WORK(&vscsi->proc_work, ibmvscsis_disconnect); + (void)queue_work(vscsi->work_q, &vscsi->proc_work); + } else { + if (vscsi->new_state) + state = vscsi->new_state; + else + state = vscsi->state; + + switch (state) { + case NO_QUEUE: + case UNCONFIGURING: + break; + + case ERR_DISCONNECTED: + case ERR_DISCONNECT: + case UNDEFINED: + if (new_state == UNCONFIGURING) + vscsi->new_state = new_state; + break; + + case ERR_DISCONNECT_RECONNECT: + switch (new_state) { + case UNCONFIGURING: + case ERR_DISCONNECT: + vscsi->new_state = new_state; + break; + default: + break; + } + break; + + case WAIT_ENABLED: + case PART_UP_WAIT_ENAB: + case WAIT_IDLE: + case WAIT_CONNECTION: + case CONNECTED: + case SRP_PROCESSING: + vscsi->new_state = new_state; + break; + + default: + break; + } + } + + pr_debug("Leaving post_disconnect: flags 0x%x, new_state 0x%x\n", + vscsi->flags, vscsi->new_state); +} + +/** + * ibmvscsis_trans_event() - Handle a Transport Event + * @vscsi: Pointer to our adapter structure + * @crq: Pointer to CRQ entry containing the Transport Event + * + * Do the logic to close the I_T nexus. This function may not + * behave to specification. + * + * EXECUTION ENVIRONMENT: + * Interrupt, interrupt lock held + */ +static long ibmvscsis_trans_event(struct scsi_info *vscsi, + struct viosrp_crq *crq) +{ + long rc = ADAPT_SUCCESS; + + pr_debug("trans_event: format %d, flags 0x%x, state 0x%hx\n", + (int)crq->format, vscsi->flags, vscsi->state); + + switch (crq->format) { + case MIGRATED: + case PARTNER_FAILED: + case PARTNER_DEREGISTER: + ibmvscsis_delete_client_info(vscsi, true); + break; + + default: + rc = ERROR; + dev_err(&vscsi->dev, "trans_event: invalid format %d\n", + (uint)crq->format); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT, + RESPONSE_Q_DOWN); + break; + } + + if (rc == ADAPT_SUCCESS) { + switch (vscsi->state) { + case NO_QUEUE: + case ERR_DISCONNECTED: + case UNDEFINED: + break; + + case UNCONFIGURING: + vscsi->flags |= (RESPONSE_Q_DOWN | TRANS_EVENT); + break; + + case WAIT_ENABLED: + break; + + case WAIT_CONNECTION: + break; + + case CONNECTED: + ibmvscsis_post_disconnect(vscsi, WAIT_IDLE, + (RESPONSE_Q_DOWN | + TRANS_EVENT)); + break; + + case PART_UP_WAIT_ENAB: + vscsi->state = WAIT_ENABLED; + break; + + case SRP_PROCESSING: + if ((vscsi->debit > 0) || + !list_empty(&vscsi->schedule_q) || + !list_empty(&vscsi->waiting_rsp) || + !list_empty(&vscsi->active_q)) { + pr_debug("debit %d, sched %d, wait %d, active %d\n", + vscsi->debit, + (int)list_empty(&vscsi->schedule_q), + (int)list_empty(&vscsi->waiting_rsp), + (int)list_empty(&vscsi->active_q)); + pr_warn("connection lost with outstanding work\n"); + } else { + pr_debug("trans_event: SRP Processing, but no outstanding work\n"); + } + + ibmvscsis_post_disconnect(vscsi, WAIT_IDLE, + (RESPONSE_Q_DOWN | + TRANS_EVENT)); + break; + + case ERR_DISCONNECT: + case ERR_DISCONNECT_RECONNECT: + case WAIT_IDLE: + vscsi->flags |= (RESPONSE_Q_DOWN | TRANS_EVENT); + break; + } + } + + rc = vscsi->flags & SCHEDULE_DISCONNECT; + + pr_debug("Leaving trans_event: flags 0x%x, state 0x%hx, rc %ld\n", + vscsi->flags, vscsi->state, rc); + + return rc; +} + +/** + * ibmvscsis_poll_cmd_q() - Poll Command Queue + * @vscsi: Pointer to our adapter structure + * + * Called to handle command elements that may have arrived while + * interrupts were disabled. + * + * EXECUTION ENVIRONMENT: + * intr_lock must be held + */ +static void ibmvscsis_poll_cmd_q(struct scsi_info *vscsi) +{ + struct viosrp_crq *crq; + long rc; + bool ack = true; + volatile u8 valid; + + pr_debug("poll_cmd_q: flags 0x%x, state 0x%hx, q index %ud\n", + vscsi->flags, vscsi->state, vscsi->cmd_q.index); + + rc = vscsi->flags & SCHEDULE_DISCONNECT; + crq = vscsi->cmd_q.base_addr + vscsi->cmd_q.index; + valid = crq->valid; + dma_rmb(); + + while (valid) { +poll_work: + vscsi->cmd_q.index = + (vscsi->cmd_q.index + 1) & vscsi->cmd_q.mask; + + if (!rc) { + rc = ibmvscsis_parse_command(vscsi, crq); + } else { + if ((uint)crq->valid == VALID_TRANS_EVENT) { + /* + * must service the transport layer events even + * in an error state, dont break out until all + * the consecutive transport events have been + * processed + */ + rc = ibmvscsis_trans_event(vscsi, crq); + } else if (vscsi->flags & TRANS_EVENT) { + /* + * if a tranport event has occurred leave + * everything but transport events on the queue + */ + pr_debug("poll_cmd_q, ignoring\n"); + + /* + * need to decrement the queue index so we can + * look at the elment again + */ + if (vscsi->cmd_q.index) + vscsi->cmd_q.index -= 1; + else + /* + * index is at 0 it just wrapped. + * have it index last element in q + */ + vscsi->cmd_q.index = vscsi->cmd_q.mask; + break; + } + } + + crq->valid = INVALIDATE_CMD_RESP_EL; + + crq = vscsi->cmd_q.base_addr + vscsi->cmd_q.index; + valid = crq->valid; + dma_rmb(); + } + + if (!rc) { + if (ack) { + vio_enable_interrupts(vscsi->dma_dev); + ack = false; + pr_debug("poll_cmd_q, reenabling interrupts\n"); + } + valid = crq->valid; + dma_rmb(); + if (valid) + goto poll_work; + } + + pr_debug("Leaving poll_cmd_q: rc %ld\n", rc); +} + +/** + * ibmvscsis_free_cmd_qs() - Free elements in queue + * @vscsi: Pointer to our adapter structure + * + * Free all of the elements on all queues that are waiting for + * whatever reason. + * + * PRECONDITION: + * Called with interrupt lock held + */ +static void ibmvscsis_free_cmd_qs(struct scsi_info *vscsi) +{ + struct ibmvscsis_cmd *cmd, *nxt; + + pr_debug("free_cmd_qs: waiting_rsp empty %d, timer starter %d\n", + (int)list_empty(&vscsi->waiting_rsp), + vscsi->rsp_q_timer.started); + + list_for_each_entry_safe(cmd, nxt, &vscsi->waiting_rsp, list) { + list_del(&cmd->list); + ibmvscsis_free_cmd_resources(vscsi, cmd); + } +} + +/** + * ibmvscsis_get_free_cmd() - Get free command from list + * @vscsi: Pointer to our adapter structure + * + * Must be called with interrupt lock held. + */ +static struct ibmvscsis_cmd *ibmvscsis_get_free_cmd(struct scsi_info *vscsi) +{ + struct ibmvscsis_cmd *cmd = NULL; + struct iu_entry *iue; + + iue = srp_iu_get(&vscsi->target); + if (iue) { + cmd = list_first_entry_or_null(&vscsi->free_cmd, + struct ibmvscsis_cmd, list); + if (cmd) { + list_del(&cmd->list); + cmd->iue = iue; + cmd->type = UNSET_TYPE; + memset(&cmd->se_cmd, 0, sizeof(cmd->se_cmd)); + } else { + srp_iu_put(iue); + } + } + + return cmd; +} + +/** + * ibmvscsis_adapter_idle() - Helper function to handle idle adapter + * @vscsi: Pointer to our adapter structure + * + * This function is called when the adapter is idle when the driver + * is attempting to clear an error condition. + * The adapter is considered busy if any of its cmd queues + * are non-empty. This function can be invoked + * from the off level disconnect function. + * + * EXECUTION ENVIRONMENT: + * Process environment called with interrupt lock held + */ +static void ibmvscsis_adapter_idle(struct scsi_info *vscsi) +{ + int free_qs = false; + + pr_debug("adapter_idle: flags 0x%x, state 0x%hx\n", vscsi->flags, + vscsi->state); + + /* Only need to free qs if we're disconnecting from client */ + if (vscsi->state != WAIT_CONNECTION || vscsi->flags & TRANS_EVENT) + free_qs = true; + + switch (vscsi->state) { + case ERR_DISCONNECT_RECONNECT: + ibmvscsis_reset_queue(vscsi, WAIT_CONNECTION); + pr_debug("adapter_idle, disc_rec: flags 0x%x\n", vscsi->flags); + break; + + case ERR_DISCONNECT: + ibmvscsis_free_command_q(vscsi); + vscsi->flags &= ~DISCONNECT_SCHEDULED; + vscsi->flags |= RESPONSE_Q_DOWN; + vscsi->state = ERR_DISCONNECTED; + pr_debug("adapter_idle, disc: flags 0x%x, state 0x%hx\n", + vscsi->flags, vscsi->state); + break; + + case WAIT_IDLE: + vscsi->rsp_q_timer.timer_pops = 0; + vscsi->debit = 0; + vscsi->credit = 0; + if (vscsi->flags & TRANS_EVENT) { + vscsi->state = WAIT_CONNECTION; + vscsi->flags &= PRESERVE_FLAG_FIELDS; + } else { + vscsi->state = CONNECTED; + vscsi->flags &= ~DISCONNECT_SCHEDULED; + } + + pr_debug("adapter_idle, wait: flags 0x%x, state 0x%hx\n", + vscsi->flags, vscsi->state); + ibmvscsis_poll_cmd_q(vscsi); + break; + + case ERR_DISCONNECTED: + vscsi->flags &= ~DISCONNECT_SCHEDULED; + pr_debug("adapter_idle, disconnected: flags 0x%x, state 0x%hx\n", + vscsi->flags, vscsi->state); + break; + + default: + dev_err(&vscsi->dev, "adapter_idle: in invalid state %d\n", + vscsi->state); + break; + } + + if (free_qs) + ibmvscsis_free_cmd_qs(vscsi); + + /* + * There is a timing window where we could lose a disconnect request. + * The known path to this window occurs during the DISCONNECT_RECONNECT + * case above: reset_queue calls free_command_q, which will release the + * interrupt lock. During that time, a new post_disconnect call can be + * made with a "more severe" state (DISCONNECT or UNCONFIGURING). + * Because the DISCONNECT_SCHEDULED flag is already set, post_disconnect + * will only set the new_state. Now free_command_q reacquires the intr + * lock and clears the DISCONNECT_SCHEDULED flag (using PRESERVE_FLAG_ + * FIELDS), and the disconnect is lost. This is particularly bad when + * the new disconnect was for UNCONFIGURING, since the unconfigure hangs + * forever. + * Fix is that free command queue sets acr state and acr flags if there + * is a change under the lock + * note free command queue writes to this state it clears it + * before releasing the lock, different drivers call the free command + * queue different times so dont initialize above + */ + if (vscsi->phyp_acr_state != 0) { + /* + * set any bits in flags that may have been cleared by + * a call to free command queue in switch statement + * or reset queue + */ + vscsi->flags |= vscsi->phyp_acr_flags; + ibmvscsis_post_disconnect(vscsi, vscsi->phyp_acr_state, 0); + vscsi->phyp_acr_state = 0; + vscsi->phyp_acr_flags = 0; + + pr_debug("adapter_idle: flags 0x%x, state 0x%hx, acr_flags 0x%x, acr_state 0x%hx\n", + vscsi->flags, vscsi->state, vscsi->phyp_acr_flags, + vscsi->phyp_acr_state); + } + + pr_debug("Leaving adapter_idle: flags 0x%x, state 0x%hx, new_state 0x%x\n", + vscsi->flags, vscsi->state, vscsi->new_state); +} + +/** + * ibmvscsis_copy_crq_packet() - Copy CRQ Packet + * @vscsi: Pointer to our adapter structure + * @cmd: Pointer to command element to use to process the request + * @crq: Pointer to CRQ entry containing the request + * + * Copy the srp information unit from the hosted + * partition using remote dma + * + * EXECUTION ENVIRONMENT: + * Interrupt, interrupt lock held + */ +static long ibmvscsis_copy_crq_packet(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd, + struct viosrp_crq *crq) +{ + struct iu_entry *iue = cmd->iue; + long rc = 0; + u16 len; + + len = be16_to_cpu(crq->IU_length); + if ((len > SRP_MAX_IU_LEN) || (len == 0)) { + dev_err(&vscsi->dev, "copy_crq: Invalid len %d passed", len); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + return SRP_VIOLATION; + } + + rc = h_copy_rdma(len, vscsi->dds.window[REMOTE].liobn, + be64_to_cpu(crq->IU_data_ptr), + vscsi->dds.window[LOCAL].liobn, iue->sbuf->dma); + + switch (rc) { + case H_SUCCESS: + cmd->init_time = mftb(); + iue->remote_token = crq->IU_data_ptr; + iue->iu_len = len; + pr_debug("copy_crq: ioba 0x%llx, init_time 0x%llx\n", + be64_to_cpu(crq->IU_data_ptr), cmd->init_time); + break; + case H_PERMISSION: + if (connection_broken(vscsi)) + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, + (RESPONSE_Q_DOWN | + CLIENT_FAILED)); + else + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, 0); + + dev_err(&vscsi->dev, "copy_crq: h_copy_rdma failed, rc %ld\n", + rc); + break; + case H_DEST_PARM: + case H_SOURCE_PARM: + default: + dev_err(&vscsi->dev, "copy_crq: h_copy_rdma failed, rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + break; + } + + return rc; +} + +/** + * ibmvscsis_adapter_info - Service an Adapter Info MAnagement Data gram + * @vscsi: Pointer to our adapter structure + * @iue: Information Unit containing the Adapter Info MAD request + * + * EXECUTION ENVIRONMENT: + * Interrupt adpater lock is held + */ +static long ibmvscsis_adapter_info(struct scsi_info *vscsi, + struct iu_entry *iue) +{ + struct viosrp_adapter_info *mad = &vio_iu(iue)->mad.adapter_info; + struct mad_adapter_info_data *info; + uint flag_bits = 0; + dma_addr_t token; + long rc; + + mad->common.status = cpu_to_be16(VIOSRP_MAD_SUCCESS); + + if (be16_to_cpu(mad->common.length) > sizeof(*info)) { + mad->common.status = cpu_to_be16(VIOSRP_MAD_FAILED); + return 0; + } + + info = dma_alloc_coherent(&vscsi->dma_dev->dev, sizeof(*info), &token, + GFP_KERNEL); + if (!info) { + dev_err(&vscsi->dev, "bad dma_alloc_coherent %p\n", + iue->target); + mad->common.status = cpu_to_be16(VIOSRP_MAD_FAILED); + return 0; + } + + /* Get remote info */ + rc = h_copy_rdma(be16_to_cpu(mad->common.length), + vscsi->dds.window[REMOTE].liobn, + be64_to_cpu(mad->buffer), + vscsi->dds.window[LOCAL].liobn, token); + + if (rc != H_SUCCESS) { + if (rc == H_PERMISSION) { + if (connection_broken(vscsi)) + flag_bits = (RESPONSE_Q_DOWN | CLIENT_FAILED); + } + pr_warn("adapter_info: h_copy_rdma from client failed, rc %ld\n", + rc); + pr_debug("adapter_info: ioba 0x%llx, flags 0x%x, flag_bits 0x%x\n", + be64_to_cpu(mad->buffer), vscsi->flags, flag_bits); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, + flag_bits); + goto free_dma; + } + + /* + * Copy client info, but ignore partition number, which we + * already got from phyp - unless we failed to get it from + * phyp (e.g. if we're running on a p5 system). + */ + if (vscsi->client_data.partition_number == 0) + vscsi->client_data.partition_number = + be32_to_cpu(info->partition_number); + strncpy(vscsi->client_data.srp_version, info->srp_version, + sizeof(vscsi->client_data.srp_version)); + strncpy(vscsi->client_data.partition_name, info->partition_name, + sizeof(vscsi->client_data.partition_name)); + vscsi->client_data.mad_version = be32_to_cpu(info->mad_version); + vscsi->client_data.os_type = be32_to_cpu(info->os_type); + + /* Copy our info */ + strncpy(info->srp_version, SRP_VERSION, + sizeof(info->srp_version)); + strncpy(info->partition_name, vscsi->dds.partition_name, + sizeof(info->partition_name)); + info->partition_number = cpu_to_be32(vscsi->dds.partition_num); + info->mad_version = cpu_to_be32(MAD_VERSION_1); + info->os_type = cpu_to_be32(LINUX); + memset(&info->port_max_txu[0], 0, sizeof(info->port_max_txu)); + info->port_max_txu[0] = cpu_to_be32(128 * PAGE_SIZE); + + dma_wmb(); + rc = h_copy_rdma(sizeof(*info), vscsi->dds.window[LOCAL].liobn, + token, vscsi->dds.window[REMOTE].liobn, + be64_to_cpu(mad->buffer)); + switch (rc) { + case H_SUCCESS: + break; + + case H_SOURCE_PARM: + case H_DEST_PARM: + case H_PERMISSION: + if (connection_broken(vscsi)) + flag_bits = (RESPONSE_Q_DOWN | CLIENT_FAILED); + default: + dev_err(&vscsi->dev, "adapter_info: h_copy_rdma to client failed, rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, + flag_bits); + break; + } + +free_dma: + dma_free_coherent(&vscsi->dma_dev->dev, sizeof(*info), info, token); + pr_debug("Leaving adapter_info, rc %ld\n", rc); + + return rc; +} + +/** + * ibmvscsis_cap_mad() - Service a Capabilities MAnagement Data gram + * @vscsi: Pointer to our adapter structure + * @iue: Information Unit containing the Capabilities MAD request + * + * NOTE: if you return an error from this routine you must be + * disconnecting or you will cause a hang + * + * EXECUTION ENVIRONMENT: + * Interrupt called with adapter lock held + */ +static int ibmvscsis_cap_mad(struct scsi_info *vscsi, struct iu_entry *iue) +{ + struct viosrp_capabilities *mad = &vio_iu(iue)->mad.capabilities; + struct capabilities *cap; + struct mad_capability_common *common; + dma_addr_t token; + u16 olen, len, status, min_len, cap_len; + u32 flag; + uint flag_bits = 0; + long rc = 0; + + olen = be16_to_cpu(mad->common.length); + /* + * struct capabilities hardcodes a couple capabilities after the + * header, but the capabilities can actually be in any order. + */ + min_len = offsetof(struct capabilities, migration); + if ((olen < min_len) || (olen > PAGE_SIZE)) { + pr_warn("cap_mad: invalid len %d\n", olen); + mad->common.status = cpu_to_be16(VIOSRP_MAD_FAILED); + return 0; + } + + cap = dma_alloc_coherent(&vscsi->dma_dev->dev, olen, &token, + GFP_KERNEL); + if (!cap) { + dev_err(&vscsi->dev, "bad dma_alloc_coherent %p\n", + iue->target); + mad->common.status = cpu_to_be16(VIOSRP_MAD_FAILED); + return 0; + } + rc = h_copy_rdma(olen, vscsi->dds.window[REMOTE].liobn, + be64_to_cpu(mad->buffer), + vscsi->dds.window[LOCAL].liobn, token); + if (rc == H_SUCCESS) { + strncpy(cap->name, dev_name(&vscsi->dma_dev->dev), + SRP_MAX_LOC_LEN); + + len = olen - min_len; + status = VIOSRP_MAD_SUCCESS; + common = (struct mad_capability_common *)&cap->migration; + + while ((len > 0) && (status == VIOSRP_MAD_SUCCESS) && !rc) { + pr_debug("cap_mad: len left %hd, cap type %d, cap len %hd\n", + len, be32_to_cpu(common->cap_type), + be16_to_cpu(common->length)); + + cap_len = be16_to_cpu(common->length); + if (cap_len > len) { + dev_err(&vscsi->dev, "cap_mad: cap len mismatch with total len\n"); + status = VIOSRP_MAD_FAILED; + break; + } + + if (cap_len == 0) { + dev_err(&vscsi->dev, "cap_mad: cap len is 0\n"); + status = VIOSRP_MAD_FAILED; + break; + } + + switch (common->cap_type) { + default: + pr_debug("cap_mad: unsupported capability\n"); + common->server_support = 0; + flag = cpu_to_be32((u32)CAP_LIST_SUPPORTED); + cap->flags &= ~flag; + break; + } + + len = len - cap_len; + common = (struct mad_capability_common *) + ((char *)common + cap_len); + } + + mad->common.status = cpu_to_be16(status); + + dma_wmb(); + rc = h_copy_rdma(olen, vscsi->dds.window[LOCAL].liobn, token, + vscsi->dds.window[REMOTE].liobn, + be64_to_cpu(mad->buffer)); + + if (rc != H_SUCCESS) { + pr_debug("cap_mad: failed to copy to client, rc %ld\n", + rc); + + if (rc == H_PERMISSION) { + if (connection_broken(vscsi)) + flag_bits = (RESPONSE_Q_DOWN | + CLIENT_FAILED); + } + + pr_warn("cap_mad: error copying data to client, rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, + flag_bits); + } + } + + dma_free_coherent(&vscsi->dma_dev->dev, olen, cap, token); + + pr_debug("Leaving cap_mad, rc %ld, client_cap 0x%x\n", + rc, vscsi->client_cap); + + return rc; +} + +/** + * ibmvscsis_process_mad() - Service a MAnagement Data gram + * @vscsi: Pointer to our adapter structure + * @iue: Information Unit containing the MAD request + * + * Must be called with interrupt lock held. + */ +static long ibmvscsis_process_mad(struct scsi_info *vscsi, struct iu_entry *iue) +{ + struct mad_common *mad = (struct mad_common *)&vio_iu(iue)->mad; + struct viosrp_empty_iu *empty; + long rc = ADAPT_SUCCESS; + + switch (be32_to_cpu(mad->type)) { + case VIOSRP_EMPTY_IU_TYPE: + empty = &vio_iu(iue)->mad.empty_iu; + vscsi->empty_iu_id = be64_to_cpu(empty->buffer); + vscsi->empty_iu_tag = be64_to_cpu(empty->common.tag); + mad->status = cpu_to_be16(VIOSRP_MAD_SUCCESS); + break; + case VIOSRP_ADAPTER_INFO_TYPE: + rc = ibmvscsis_adapter_info(vscsi, iue); + break; + case VIOSRP_CAPABILITIES_TYPE: + rc = ibmvscsis_cap_mad(vscsi, iue); + break; + case VIOSRP_ENABLE_FAST_FAIL: + if (vscsi->state == CONNECTED) { + vscsi->fast_fail = true; + mad->status = cpu_to_be16(VIOSRP_MAD_SUCCESS); + } else { + pr_warn("fast fail mad sent after login\n"); + mad->status = cpu_to_be16(VIOSRP_MAD_FAILED); + } + break; + default: + mad->status = cpu_to_be16(VIOSRP_MAD_NOT_SUPPORTED); + break; + } + + return rc; +} + +/** + * srp_snd_msg_failed() - Handle an error when sending a response + * @vscsi: Pointer to our adapter structure + * @rc: The return code from the h_send_crq command + * + * Must be called with interrupt lock held. + */ +static void srp_snd_msg_failed(struct scsi_info *vscsi, long rc) +{ + ktime_t kt; + + if (rc != H_DROPPED) { + ibmvscsis_free_cmd_qs(vscsi); + + if (rc == H_CLOSED) + vscsi->flags |= CLIENT_FAILED; + + /* don't flag the same problem multiple times */ + if (!(vscsi->flags & RESPONSE_Q_DOWN)) { + vscsi->flags |= RESPONSE_Q_DOWN; + if (!(vscsi->state & (ERR_DISCONNECT | + ERR_DISCONNECT_RECONNECT | + ERR_DISCONNECTED | UNDEFINED))) { + dev_err(&vscsi->dev, "snd_msg_failed: setting RESPONSE_Q_DOWN, state 0x%hx, flags 0x%x, rc %ld\n", + vscsi->state, vscsi->flags, rc); + } + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, 0); + } + return; + } + + /* + * The response queue is full. + * If the server is processing SRP requests, i.e. + * the client has successfully done an + * SRP_LOGIN, then it will wait forever for room in + * the queue. However if the system admin + * is attempting to unconfigure the server then one + * or more children will be in a state where + * they are being removed. So if there is even one + * child being removed then the driver assumes + * the system admin is attempting to break the + * connection with the client and MAX_TIMER_POPS + * is honored. + */ + if ((vscsi->rsp_q_timer.timer_pops < MAX_TIMER_POPS) || + (vscsi->state == SRP_PROCESSING)) { + pr_debug("snd_msg_failed: response queue full, flags 0x%x, timer started %d, pops %d\n", + vscsi->flags, (int)vscsi->rsp_q_timer.started, + vscsi->rsp_q_timer.timer_pops); + + /* + * Check if the timer is running; if it + * is not then start it up. + */ + if (!vscsi->rsp_q_timer.started) { + if (vscsi->rsp_q_timer.timer_pops < + MAX_TIMER_POPS) { + kt = ktime_set(0, WAIT_NANO_SECONDS); + } else { + /* + * slide the timeslice if the maximum + * timer pops have already happened + */ + kt = ktime_set(WAIT_SECONDS, 0); + } + + vscsi->rsp_q_timer.started = true; + hrtimer_start(&vscsi->rsp_q_timer.timer, kt, + HRTIMER_MODE_REL); + } + } else { + /* + * TBD: Do we need to worry about this? Need to get + * remove working. + */ + /* + * waited a long time and it appears the system admin + * is bring this driver down + */ + vscsi->flags |= RESPONSE_Q_DOWN; + ibmvscsis_free_cmd_qs(vscsi); + /* + * if the driver is already attempting to disconnect + * from the client and has already logged an error + * trace this event but don't put it in the error log + */ + if (!(vscsi->state & (ERR_DISCONNECT | + ERR_DISCONNECT_RECONNECT | + ERR_DISCONNECTED | UNDEFINED))) { + dev_err(&vscsi->dev, "client crq full too long\n"); + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, + 0); + } + } +} + +/** + * ibmvscsis_send_messages() - Send a Response + * @vscsi: Pointer to our adapter structure + * + * Send a response, first checking the waiting queue. Responses are + * sent in order they are received. If the response cannot be sent, + * because the client queue is full, it stays on the waiting queue. + * + * PRECONDITION: + * Called with interrupt lock held + */ +static void ibmvscsis_send_messages(struct scsi_info *vscsi) +{ + u64 msg_hi = 0; + /* note do not attmempt to access the IU_data_ptr with this pointer + * it is not valid + */ + struct viosrp_crq *crq = (struct viosrp_crq *)&msg_hi; + struct ibmvscsis_cmd *cmd, *nxt; + struct iu_entry *iue; + long rc = ADAPT_SUCCESS; + + if (!(vscsi->flags & RESPONSE_Q_DOWN)) { + list_for_each_entry_safe(cmd, nxt, &vscsi->waiting_rsp, list) { + pr_debug("send_messages cmd %p\n", cmd); + + iue = cmd->iue; + + crq->valid = VALID_CMD_RESP_EL; + crq->format = cmd->rsp.format; + + if (cmd->flags & CMD_FAST_FAIL) + crq->status = VIOSRP_ADAPTER_FAIL; + + crq->IU_length = cpu_to_be16(cmd->rsp.len); + + rc = h_send_crq(vscsi->dma_dev->unit_address, + be64_to_cpu(msg_hi), + be64_to_cpu(cmd->rsp.tag)); + + pr_debug("send_messages: tag 0x%llx, rc %ld\n", + be64_to_cpu(cmd->rsp.tag), rc); + + /* if all ok free up the command element resources */ + if (rc == H_SUCCESS) { + /* some movement has occurred */ + vscsi->rsp_q_timer.timer_pops = 0; + list_del(&cmd->list); + + ibmvscsis_free_cmd_resources(vscsi, cmd); + } else { + srp_snd_msg_failed(vscsi, rc); + break; + } + } + + if (!rc) { + /* + * The timer could pop with the queue empty. If + * this happens, rc will always indicate a + * success; clear the pop count. + */ + vscsi->rsp_q_timer.timer_pops = 0; + } + } else { + ibmvscsis_free_cmd_qs(vscsi); + } +} + +/* Called with intr lock held */ +static void ibmvscsis_send_mad_resp(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd, + struct viosrp_crq *crq) +{ + struct iu_entry *iue = cmd->iue; + struct mad_common *mad = (struct mad_common *)&vio_iu(iue)->mad; + uint flag_bits = 0; + long rc; + + dma_wmb(); + rc = h_copy_rdma(sizeof(struct mad_common), + vscsi->dds.window[LOCAL].liobn, iue->sbuf->dma, + vscsi->dds.window[REMOTE].liobn, + be64_to_cpu(crq->IU_data_ptr)); + if (!rc) { + cmd->rsp.format = VIOSRP_MAD_FORMAT; + cmd->rsp.len = sizeof(struct mad_common); + cmd->rsp.tag = mad->tag; + list_add_tail(&cmd->list, &vscsi->waiting_rsp); + ibmvscsis_send_messages(vscsi); + } else { + pr_debug("Error sending mad response, rc %ld\n", rc); + if (rc == H_PERMISSION) { + if (connection_broken(vscsi)) + flag_bits = (RESPONSE_Q_DOWN | CLIENT_FAILED); + } + dev_err(&vscsi->dev, "mad: failed to copy to client, rc %ld\n", + rc); + + ibmvscsis_free_cmd_resources(vscsi, cmd); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, + flag_bits); + } +} + +/** + * ibmvscsis_mad() - Service a MAnagement Data gram. + * @vscsi: Pointer to our adapter structure + * @crq: Pointer to the CRQ entry containing the MAD request + * + * EXECUTION ENVIRONMENT: + * Interrupt called with adapter lock held + */ +static long ibmvscsis_mad(struct scsi_info *vscsi, struct viosrp_crq *crq) +{ + struct iu_entry *iue; + struct ibmvscsis_cmd *cmd; + struct mad_common *mad; + long rc = ADAPT_SUCCESS; + + switch (vscsi->state) { + /* + * We have not exchanged Init Msgs yet, so this MAD was sent + * before the last Transport Event; client will not be + * expecting a response. + */ + case WAIT_CONNECTION: + pr_debug("mad: in Wait Connection state, ignoring MAD, flags %d\n", + vscsi->flags); + return ADAPT_SUCCESS; + + case SRP_PROCESSING: + case CONNECTED: + break; + + /* + * We should never get here while we're in these states. + * Just log an error and get out. + */ + case UNCONFIGURING: + case WAIT_IDLE: + case ERR_DISCONNECT: + case ERR_DISCONNECT_RECONNECT: + default: + dev_err(&vscsi->dev, "mad: invalid adapter state %d for mad\n", + vscsi->state); + return ADAPT_SUCCESS; + } + + cmd = ibmvscsis_get_free_cmd(vscsi); + if (!cmd) { + dev_err(&vscsi->dev, "mad: failed to get cmd, debit %d\n", + vscsi->debit); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + return ERROR; + } + iue = cmd->iue; + cmd->type = ADAPTER_MAD; + + rc = ibmvscsis_copy_crq_packet(vscsi, cmd, crq); + if (!rc) { + mad = (struct mad_common *)&vio_iu(iue)->mad; + + pr_debug("mad: type %d\n", be32_to_cpu(mad->type)); + + if (be16_to_cpu(mad->length) < 0) { + dev_err(&vscsi->dev, "mad: length is < 0\n"); + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, 0); + rc = SRP_VIOLATION; + } else { + rc = ibmvscsis_process_mad(vscsi, iue); + } + + pr_debug("mad: status %hd, rc %ld\n", be16_to_cpu(mad->status), + rc); + + if (!rc) + ibmvscsis_send_mad_resp(vscsi, cmd, crq); + } else { + ibmvscsis_free_cmd_resources(vscsi, cmd); + } + + pr_debug("Leaving mad, rc %ld\n", rc); + return rc; +} + +/** + * ibmvscsis_login_rsp() - Create/copy a login response notice to the client + * @vscsi: Pointer to our adapter structure + * @cmd: Pointer to the command for the SRP Login request + * + * EXECUTION ENVIRONMENT: + * Interrupt, interrupt lock held + */ +static long ibmvscsis_login_rsp(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd) +{ + struct iu_entry *iue = cmd->iue; + struct srp_login_rsp *rsp = &vio_iu(iue)->srp.login_rsp; + struct format_code *fmt; + uint flag_bits = 0; + long rc = ADAPT_SUCCESS; + + memset(rsp, 0, sizeof(struct srp_login_rsp)); + + rsp->opcode = SRP_LOGIN_RSP; + rsp->req_lim_delta = cpu_to_be32(vscsi->request_limit); + rsp->tag = cmd->rsp.tag; + rsp->max_it_iu_len = cpu_to_be32(SRP_MAX_IU_LEN); + rsp->max_ti_iu_len = cpu_to_be32(SRP_MAX_IU_LEN); + fmt = (struct format_code *)&rsp->buf_fmt; + fmt->buffers = SUPPORTED_FORMATS; + vscsi->credit = 0; + + cmd->rsp.len = sizeof(struct srp_login_rsp); + + dma_wmb(); + rc = h_copy_rdma(cmd->rsp.len, vscsi->dds.window[LOCAL].liobn, + iue->sbuf->dma, vscsi->dds.window[REMOTE].liobn, + be64_to_cpu(iue->remote_token)); + + switch (rc) { + case H_SUCCESS: + break; + + case H_PERMISSION: + if (connection_broken(vscsi)) + flag_bits = RESPONSE_Q_DOWN | CLIENT_FAILED; + dev_err(&vscsi->dev, "login_rsp: error copying to client, rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, + flag_bits); + break; + case H_SOURCE_PARM: + case H_DEST_PARM: + default: + dev_err(&vscsi->dev, "login_rsp: error copying to client, rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + break; + } + + return rc; +} + +/** + * ibmvscsis_srp_login_rej() - Create/copy a login rejection notice to client + * @vscsi: Pointer to our adapter structure + * @cmd: Pointer to the command for the SRP Login request + * @reason: The reason the SRP Login is being rejected, per SRP protocol + * + * EXECUTION ENVIRONMENT: + * Interrupt, interrupt lock held + */ +static long ibmvscsis_srp_login_rej(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd, u32 reason) +{ + struct iu_entry *iue = cmd->iue; + struct srp_login_rej *rej = &vio_iu(iue)->srp.login_rej; + struct format_code *fmt; + uint flag_bits = 0; + long rc = ADAPT_SUCCESS; + + memset(rej, 0, sizeof(*rej)); + + rej->opcode = SRP_LOGIN_REJ; + rej->reason = cpu_to_be32(reason); + rej->tag = cmd->rsp.tag; + fmt = (struct format_code *)&rej->buf_fmt; + fmt->buffers = SUPPORTED_FORMATS; + + cmd->rsp.len = sizeof(*rej); + + dma_wmb(); + rc = h_copy_rdma(cmd->rsp.len, vscsi->dds.window[LOCAL].liobn, + iue->sbuf->dma, vscsi->dds.window[REMOTE].liobn, + be64_to_cpu(iue->remote_token)); + + switch (rc) { + case H_SUCCESS: + break; + case H_PERMISSION: + if (connection_broken(vscsi)) + flag_bits = RESPONSE_Q_DOWN | CLIENT_FAILED; + dev_err(&vscsi->dev, "login_rej: error copying to client, rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, + flag_bits); + break; + case H_SOURCE_PARM: + case H_DEST_PARM: + default: + dev_err(&vscsi->dev, "login_rej: error copying to client, rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + break; + } + + return rc; +} + +static int ibmvscsis_make_nexus(struct ibmvscsis_tport *tport) +{ + char *name = tport->tport_name; + struct ibmvscsis_nexus *nexus; + int rc; + + if (tport->ibmv_nexus) { + pr_debug("tport->ibmv_nexus already exists\n"); + return 0; + } + + nexus = kzalloc(sizeof(*nexus), GFP_KERNEL); + if (!nexus) { + pr_err("Unable to allocate struct ibmvscsis_nexus\n"); + return -ENOMEM; + } + + nexus->se_sess = target_alloc_session(&tport->se_tpg, 0, 0, + TARGET_PROT_NORMAL, name, nexus, + NULL); + if (IS_ERR(nexus->se_sess)) { + rc = PTR_ERR(nexus->se_sess); + goto transport_init_fail; + } + + tport->ibmv_nexus = nexus; + + return 0; + +transport_init_fail: + kfree(nexus); + return rc; +} + +static int ibmvscsis_drop_nexus(struct ibmvscsis_tport *tport) +{ + struct se_session *se_sess; + struct ibmvscsis_nexus *nexus; + + nexus = tport->ibmv_nexus; + if (!nexus) + return -ENODEV; + + se_sess = nexus->se_sess; + if (!se_sess) + return -ENODEV; + + /* + * Release the SCSI I_T Nexus to the emulated ibmvscsis Target Port + */ + transport_deregister_session(se_sess); + tport->ibmv_nexus = NULL; + kfree(nexus); + + return 0; +} + +/** + * ibmvscsis_srp_login() - Process an SRP Login Request + * @vscsi: Pointer to our adapter structure + * @cmd: Command element to use to process the SRP Login request + * @crq: Pointer to CRQ entry containing the SRP Login request + * + * EXECUTION ENVIRONMENT: + * Interrupt, called with interrupt lock held + */ +static long ibmvscsis_srp_login(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd, + struct viosrp_crq *crq) +{ + struct iu_entry *iue = cmd->iue; + struct srp_login_req *req = &vio_iu(iue)->srp.login_req; + struct port_id { + __be64 id_extension; + __be64 io_guid; + } *iport, *tport; + struct format_code *fmt; + u32 reason = 0x0; + long rc = ADAPT_SUCCESS; + + iport = (struct port_id *)req->initiator_port_id; + tport = (struct port_id *)req->target_port_id; + fmt = (struct format_code *)&req->req_buf_fmt; + if (be32_to_cpu(req->req_it_iu_len) > SRP_MAX_IU_LEN) + reason = SRP_LOGIN_REJ_REQ_IT_IU_LENGTH_TOO_LARGE; + else if (be32_to_cpu(req->req_it_iu_len) < 64) + reason = SRP_LOGIN_REJ_UNABLE_ESTABLISH_CHANNEL; + else if ((be64_to_cpu(iport->id_extension) > (MAX_NUM_PORTS - 1)) || + (be64_to_cpu(tport->id_extension) > (MAX_NUM_PORTS - 1))) + reason = SRP_LOGIN_REJ_UNABLE_ASSOCIATE_CHANNEL; + else if (req->req_flags & SRP_MULTICHAN_MULTI) + reason = SRP_LOGIN_REJ_MULTI_CHANNEL_UNSUPPORTED; + else if (fmt->buffers & (~SUPPORTED_FORMATS)) + reason = SRP_LOGIN_REJ_UNSUPPORTED_DESCRIPTOR_FMT; + else if ((fmt->buffers | SUPPORTED_FORMATS) == 0) + reason = SRP_LOGIN_REJ_UNSUPPORTED_DESCRIPTOR_FMT; + + if (vscsi->state == SRP_PROCESSING) + reason = SRP_LOGIN_REJ_CHANNEL_LIMIT_REACHED; + + rc = ibmvscsis_make_nexus(&vscsi->tport); + if (rc) + reason = SRP_LOGIN_REJ_UNABLE_ESTABLISH_CHANNEL; + + cmd->rsp.format = VIOSRP_SRP_FORMAT; + cmd->rsp.tag = req->tag; + + pr_debug("srp_login: reason 0x%x\n", reason); + + if (reason) + rc = ibmvscsis_srp_login_rej(vscsi, cmd, reason); + else + rc = ibmvscsis_login_rsp(vscsi, cmd); + + if (!rc) { + if (!reason) + vscsi->state = SRP_PROCESSING; + + list_add_tail(&cmd->list, &vscsi->waiting_rsp); + ibmvscsis_send_messages(vscsi); + } else { + ibmvscsis_free_cmd_resources(vscsi, cmd); + } + + pr_debug("Leaving srp_login, rc %ld\n", rc); + return rc; +} + +/** + * ibmvscsis_srp_i_logout() - Helper Function to close I_T Nexus + * @vscsi: Pointer to our adapter structure + * @cmd: Command element to use to process the Implicit Logout request + * @crq: Pointer to CRQ entry containing the Implicit Logout request + * + * Do the logic to close the I_T nexus. This function may not + * behave to specification. + * + * EXECUTION ENVIRONMENT: + * Interrupt, interrupt lock held + */ +static long ibmvscsis_srp_i_logout(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd, + struct viosrp_crq *crq) +{ + struct iu_entry *iue = cmd->iue; + struct srp_i_logout *log_out = &vio_iu(iue)->srp.i_logout; + long rc = ADAPT_SUCCESS; + + if ((vscsi->debit > 0) || !list_empty(&vscsi->schedule_q) || + !list_empty(&vscsi->waiting_rsp)) { + dev_err(&vscsi->dev, "i_logout: outstanding work\n"); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT, 0); + } else { + cmd->rsp.format = SRP_FORMAT; + cmd->rsp.tag = log_out->tag; + cmd->rsp.len = sizeof(struct mad_common); + list_add_tail(&cmd->list, &vscsi->waiting_rsp); + ibmvscsis_send_messages(vscsi); + + ibmvscsis_post_disconnect(vscsi, WAIT_IDLE, 0); + } + + return rc; +} + +/* Called with intr lock held */ +static void ibmvscsis_srp_cmd(struct scsi_info *vscsi, struct viosrp_crq *crq) +{ + struct ibmvscsis_cmd *cmd; + struct iu_entry *iue; + struct srp_cmd *srp; + struct srp_tsk_mgmt *tsk; + long rc; + + if (vscsi->request_limit - vscsi->debit <= 0) { + /* Client has exceeded request limit */ + dev_err(&vscsi->dev, "Client exceeded the request limit (%d), debit %d\n", + vscsi->request_limit, vscsi->debit); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + return; + } + + cmd = ibmvscsis_get_free_cmd(vscsi); + if (!cmd) { + dev_err(&vscsi->dev, "srp_cmd failed to get cmd, debit %d\n", + vscsi->debit); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + return; + } + iue = cmd->iue; + srp = &vio_iu(iue)->srp.cmd; + + rc = ibmvscsis_copy_crq_packet(vscsi, cmd, crq); + if (rc) { + ibmvscsis_free_cmd_resources(vscsi, cmd); + return; + } + + if (vscsi->state == SRP_PROCESSING) { + switch (srp->opcode) { + case SRP_LOGIN_REQ: + rc = ibmvscsis_srp_login(vscsi, cmd, crq); + break; + + case SRP_TSK_MGMT: + tsk = &vio_iu(iue)->srp.tsk_mgmt; + pr_debug("tsk_mgmt tag: %llu (0x%llx)\n", tsk->tag, + tsk->tag); + cmd->rsp.tag = tsk->tag; + vscsi->debit += 1; + cmd->type = TASK_MANAGEMENT; + list_add_tail(&cmd->list, &vscsi->schedule_q); + queue_work(vscsi->work_q, &cmd->work); + break; + + case SRP_CMD: + pr_debug("srp_cmd tag: %llu (0x%llx)\n", srp->tag, + srp->tag); + cmd->rsp.tag = srp->tag; + vscsi->debit += 1; + cmd->type = SCSI_CDB; + /* + * We want to keep track of work waiting for + * the workqueue. + */ + list_add_tail(&cmd->list, &vscsi->schedule_q); + queue_work(vscsi->work_q, &cmd->work); + break; + + case SRP_I_LOGOUT: + rc = ibmvscsis_srp_i_logout(vscsi, cmd, crq); + break; + + case SRP_CRED_RSP: + case SRP_AER_RSP: + default: + ibmvscsis_free_cmd_resources(vscsi, cmd); + dev_err(&vscsi->dev, "invalid srp cmd, opcode %d\n", + (uint)srp->opcode); + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, 0); + break; + } + } else if (srp->opcode == SRP_LOGIN_REQ && vscsi->state == CONNECTED) { + rc = ibmvscsis_srp_login(vscsi, cmd, crq); + } else { + ibmvscsis_free_cmd_resources(vscsi, cmd); + dev_err(&vscsi->dev, "Invalid state %d to handle srp cmd\n", + vscsi->state); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + } +} + +/** + * ibmvscsis_ping_response() - Respond to a ping request + * @vscsi: Pointer to our adapter structure + * + * Let the client know that the server is alive and waiting on + * its native I/O stack. + * If any type of error occurs from the call to queue a ping + * response then the client is either not accepting or receiving + * interrupts. Disconnect with an error. + * + * EXECUTION ENVIRONMENT: + * Interrupt, interrupt lock held + */ +static long ibmvscsis_ping_response(struct scsi_info *vscsi) +{ + struct viosrp_crq *crq; + u64 buffer[2] = { 0, 0 }; + long rc; + + crq = (struct viosrp_crq *)&buffer; + crq->valid = VALID_CMD_RESP_EL; + crq->format = (u8)MESSAGE_IN_CRQ; + crq->status = PING_RESPONSE; + + rc = h_send_crq(vscsi->dds.unit_id, cpu_to_be64(buffer[MSG_HI]), + cpu_to_be64(buffer[MSG_LOW])); + + switch (rc) { + case H_SUCCESS: + break; + case H_CLOSED: + vscsi->flags |= CLIENT_FAILED; + case H_DROPPED: + vscsi->flags |= RESPONSE_Q_DOWN; + case H_REMOTE_PARM: + dev_err(&vscsi->dev, "ping_response: h_send_crq failed, rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + break; + default: + dev_err(&vscsi->dev, "ping_response: h_send_crq returned unknown rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT, 0); + break; + } + + return rc; +} + +/** + * ibmvscsis_handle_init_compl_msg() - Respond to an Init Complete Message + * @vscsi: Pointer to our adapter structure + * + * Must be called with interrupt lock held. + */ +static long ibmvscsis_handle_init_compl_msg(struct scsi_info *vscsi) +{ + long rc = ADAPT_SUCCESS; + + switch (vscsi->state) { + case NO_QUEUE: + case ERR_DISCONNECT: + case ERR_DISCONNECT_RECONNECT: + case ERR_DISCONNECTED: + case UNCONFIGURING: + case UNDEFINED: + rc = ERROR; + break; + + case WAIT_CONNECTION: + vscsi->state = CONNECTED; + break; + + case WAIT_IDLE: + case SRP_PROCESSING: + case CONNECTED: + case WAIT_ENABLED: + case PART_UP_WAIT_ENAB: + default: + rc = ERROR; + dev_err(&vscsi->dev, "init_msg: invalid state %d to get init compl msg\n", + vscsi->state); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + break; + } + + return rc; +} + +/** + * ibmvscsis_handle_init_msg() - Respond to an Init Message + * @vscsi: Pointer to our adapter structure + * + * Must be called with interrupt lock held. + */ +static long ibmvscsis_handle_init_msg(struct scsi_info *vscsi) +{ + long rc = ADAPT_SUCCESS; + + switch (vscsi->state) { + case WAIT_ENABLED: + vscsi->state = PART_UP_WAIT_ENAB; + break; + + case WAIT_CONNECTION: + rc = ibmvscsis_send_init_message(vscsi, INIT_COMPLETE_MSG); + switch (rc) { + case H_SUCCESS: + vscsi->state = CONNECTED; + break; + + case H_PARAMETER: + dev_err(&vscsi->dev, "init_msg: failed to send, rc %ld\n", + rc); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT, 0); + break; + + case H_DROPPED: + dev_err(&vscsi->dev, "init_msg: failed to send, rc %ld\n", + rc); + rc = ERROR; + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, 0); + break; + + case H_CLOSED: + pr_warn("init_msg: failed to send, rc %ld\n", rc); + rc = 0; + break; + } + break; + + case UNDEFINED: + rc = ERROR; + break; + + case UNCONFIGURING: + break; + + case PART_UP_WAIT_ENAB: + case CONNECTED: + case SRP_PROCESSING: + case WAIT_IDLE: + case NO_QUEUE: + case ERR_DISCONNECT: + case ERR_DISCONNECT_RECONNECT: + case ERR_DISCONNECTED: + default: + rc = ERROR; + dev_err(&vscsi->dev, "init_msg: invalid state %d to get init msg\n", + vscsi->state); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + break; + } + + return rc; +} + +/** + * ibmvscsis_init_msg() - Respond to an init message + * @vscsi: Pointer to our adapter structure + * @crq: Pointer to CRQ element containing the Init Message + * + * EXECUTION ENVIRONMENT: + * Interrupt, interrupt lock held + */ +static long ibmvscsis_init_msg(struct scsi_info *vscsi, struct viosrp_crq *crq) +{ + long rc = ADAPT_SUCCESS; + + pr_debug("init_msg: state 0x%hx\n", vscsi->state); + + rc = h_vioctl(vscsi->dds.unit_id, H_GET_PARTNER_INFO, + (u64)vscsi->map_ioba | ((u64)PAGE_SIZE << 32), 0, 0, 0, + 0); + if (rc == H_SUCCESS) { + vscsi->client_data.partition_number = + be64_to_cpu(*(u64 *)vscsi->map_buf); + pr_debug("init_msg, part num %d\n", + vscsi->client_data.partition_number); + } else { + pr_debug("init_msg h_vioctl rc %ld\n", rc); + rc = ADAPT_SUCCESS; + } + + if (crq->format == INIT_MSG) { + rc = ibmvscsis_handle_init_msg(vscsi); + } else if (crq->format == INIT_COMPLETE_MSG) { + rc = ibmvscsis_handle_init_compl_msg(vscsi); + } else { + rc = ERROR; + dev_err(&vscsi->dev, "init_msg: invalid format %d\n", + (uint)crq->format); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + } + + return rc; +} + +/** + * ibmvscsis_parse_command() - Parse an element taken from the cmd rsp queue. + * @vscsi: Pointer to our adapter structure + * @crq: Pointer to CRQ element containing the SRP request + * + * This function will return success if the command queue element is valid + * and the srp iu or MAD request it pointed to was also valid. That does + * not mean that an error was not returned to the client. + * + * EXECUTION ENVIRONMENT: + * Interrupt, intr lock held + */ +static long ibmvscsis_parse_command(struct scsi_info *vscsi, + struct viosrp_crq *crq) +{ + long rc = ADAPT_SUCCESS; + + switch (crq->valid) { + case VALID_CMD_RESP_EL: + switch (crq->format) { + case OS400_FORMAT: + case AIX_FORMAT: + case LINUX_FORMAT: + case MAD_FORMAT: + if (vscsi->flags & PROCESSING_MAD) { + rc = ERROR; + dev_err(&vscsi->dev, "parse_command: already processing mad\n"); + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, + 0); + } else { + vscsi->flags |= PROCESSING_MAD; + rc = ibmvscsis_mad(vscsi, crq); + } + break; + + case SRP_FORMAT: + ibmvscsis_srp_cmd(vscsi, crq); + break; + + case MESSAGE_IN_CRQ: + if (crq->status == PING) + ibmvscsis_ping_response(vscsi); + break; + + default: + dev_err(&vscsi->dev, "parse_command: invalid format %d\n", + (uint)crq->format); + ibmvscsis_post_disconnect(vscsi, + ERR_DISCONNECT_RECONNECT, 0); + break; + } + break; + + case VALID_TRANS_EVENT: + rc = ibmvscsis_trans_event(vscsi, crq); + break; + + case VALID_INIT_MSG: + rc = ibmvscsis_init_msg(vscsi, crq); + break; + + default: + dev_err(&vscsi->dev, "parse_command: invalid valid field %d\n", + (uint)crq->valid); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + break; + } + + /* + * Return only what the interrupt handler cares + * about. Most errors we keep right on trucking. + */ + rc = vscsi->flags & SCHEDULE_DISCONNECT; + + return rc; +} + +static int read_dma_window(struct scsi_info *vscsi) +{ + struct vio_dev *vdev = vscsi->dma_dev; + const __be32 *dma_window; + const __be32 *prop; + + /* TODO Using of_parse_dma_window would be better, but it doesn't give + * a way to read multiple windows without already knowing the size of + * a window or the number of windows. + */ + dma_window = (const __be32 *)vio_get_attribute(vdev, + "ibm,my-dma-window", + NULL); + if (!dma_window) { + pr_err("Couldn't find ibm,my-dma-window property\n"); + return -1; + } + + vscsi->dds.window[LOCAL].liobn = be32_to_cpu(*dma_window); + dma_window++; + + prop = (const __be32 *)vio_get_attribute(vdev, "ibm,#dma-address-cells", + NULL); + if (!prop) { + pr_warn("Couldn't find ibm,#dma-address-cells property\n"); + dma_window++; + } else { + dma_window += be32_to_cpu(*prop); + } + + prop = (const __be32 *)vio_get_attribute(vdev, "ibm,#dma-size-cells", + NULL); + if (!prop) { + pr_warn("Couldn't find ibm,#dma-size-cells property\n"); + dma_window++; + } else { + dma_window += be32_to_cpu(*prop); + } + + /* dma_window should point to the second window now */ + vscsi->dds.window[REMOTE].liobn = be32_to_cpu(*dma_window); + + return 0; +} + +static struct ibmvscsis_tport *ibmvscsis_lookup_port(const char *name) +{ + struct ibmvscsis_tport *tport = NULL; + struct vio_dev *vdev; + struct scsi_info *vscsi; + + spin_lock_bh(&ibmvscsis_dev_lock); + list_for_each_entry(vscsi, &ibmvscsis_dev_list, list) { + vdev = vscsi->dma_dev; + if (!strcmp(dev_name(&vdev->dev), name)) { + tport = &vscsi->tport; + break; + } + } + spin_unlock_bh(&ibmvscsis_dev_lock); + + return tport; +} + +/** + * ibmvscsis_parse_cmd() - Parse SRP Command + * @vscsi: Pointer to our adapter structure + * @cmd: Pointer to command element with SRP command + * + * Parse the srp command; if it is valid then submit it to tcm. + * Note: The return code does not reflect the status of the SCSI CDB. + * + * EXECUTION ENVIRONMENT: + * Process level + */ +static void ibmvscsis_parse_cmd(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd) +{ + struct iu_entry *iue = cmd->iue; + struct srp_cmd *srp = (struct srp_cmd *)iue->sbuf->buf; + struct ibmvscsis_nexus *nexus; + u64 data_len = 0; + enum dma_data_direction dir; + int attr = 0; + int rc = 0; + + nexus = vscsi->tport.ibmv_nexus; + /* + * additional length in bytes. Note that the SRP spec says that + * additional length is in 4-byte words, but technically the + * additional length field is only the upper 6 bits of the byte. + * The lower 2 bits are reserved. If the lower 2 bits are 0 (as + * all reserved fields should be), then interpreting the byte as + * an int will yield the length in bytes. + */ + if (srp->add_cdb_len & 0x03) { + dev_err(&vscsi->dev, "parse_cmd: reserved bits set in IU\n"); + spin_lock_bh(&vscsi->intr_lock); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + ibmvscsis_free_cmd_resources(vscsi, cmd); + spin_unlock_bh(&vscsi->intr_lock); + return; + } + + if (srp_get_desc_table(srp, &dir, &data_len)) { + dev_err(&vscsi->dev, "0x%llx: parsing SRP descriptor table failed.\n", + srp->tag); + goto fail; + return; + } + + cmd->rsp.sol_not = srp->sol_not; + + switch (srp->task_attr) { + case SRP_SIMPLE_TASK: + attr = TCM_SIMPLE_TAG; + break; + case SRP_ORDERED_TASK: + attr = TCM_ORDERED_TAG; + break; + case SRP_HEAD_TASK: + attr = TCM_HEAD_TAG; + break; + case SRP_ACA_TASK: + attr = TCM_ACA_TAG; + break; + default: + dev_err(&vscsi->dev, "Invalid task attribute %d\n", + srp->task_attr); + goto fail; + } + + cmd->se_cmd.tag = be64_to_cpu(srp->tag); + + spin_lock_bh(&vscsi->intr_lock); + list_add_tail(&cmd->list, &vscsi->active_q); + spin_unlock_bh(&vscsi->intr_lock); + + srp->lun.scsi_lun[0] &= 0x3f; + + pr_debug("calling submit_cmd, se_cmd %p, lun 0x%llx, cdb 0x%x, attr:%d\n", + &cmd->se_cmd, scsilun_to_int(&srp->lun), (int)srp->cdb[0], + attr); + + rc = target_submit_cmd(&cmd->se_cmd, nexus->se_sess, srp->cdb, + cmd->sense_buf, scsilun_to_int(&srp->lun), + data_len, attr, dir, 0); + if (rc) { + dev_err(&vscsi->dev, "target_submit_cmd failed, rc %d\n", rc); + goto fail; + } + return; + +fail: + spin_lock_bh(&vscsi->intr_lock); + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + spin_unlock_bh(&vscsi->intr_lock); +} + +/** + * ibmvscsis_parse_task() - Parse SRP Task Management Request + * @vscsi: Pointer to our adapter structure + * @cmd: Pointer to command element with SRP task management request + * + * Parse the srp task management request; if it is valid then submit it to tcm. + * Note: The return code does not reflect the status of the task management + * request. + * + * EXECUTION ENVIRONMENT: + * Processor level + */ +static void ibmvscsis_parse_task(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd) +{ + struct iu_entry *iue = cmd->iue; + struct srp_tsk_mgmt *srp_tsk = &vio_iu(iue)->srp.tsk_mgmt; + int tcm_type; + u64 tag_to_abort = 0; + int rc = 0; + struct ibmvscsis_nexus *nexus; + + nexus = vscsi->tport.ibmv_nexus; + + cmd->rsp.sol_not = srp_tsk->sol_not; + + switch (srp_tsk->tsk_mgmt_func) { + case SRP_TSK_ABORT_TASK: + tcm_type = TMR_ABORT_TASK; + tag_to_abort = be64_to_cpu(srp_tsk->task_tag); + break; + case SRP_TSK_ABORT_TASK_SET: + tcm_type = TMR_ABORT_TASK_SET; + break; + case SRP_TSK_CLEAR_TASK_SET: + tcm_type = TMR_CLEAR_TASK_SET; + break; + case SRP_TSK_LUN_RESET: + tcm_type = TMR_LUN_RESET; + break; + case SRP_TSK_CLEAR_ACA: + tcm_type = TMR_CLEAR_ACA; + break; + default: + dev_err(&vscsi->dev, "unknown task mgmt func %d\n", + srp_tsk->tsk_mgmt_func); + cmd->se_cmd.se_tmr_req->response = + TMR_TASK_MGMT_FUNCTION_NOT_SUPPORTED; + rc = -1; + break; + } + + if (!rc) { + cmd->se_cmd.tag = be64_to_cpu(srp_tsk->tag); + + spin_lock_bh(&vscsi->intr_lock); + list_add_tail(&cmd->list, &vscsi->active_q); + spin_unlock_bh(&vscsi->intr_lock); + + srp_tsk->lun.scsi_lun[0] &= 0x3f; + + pr_debug("calling submit_tmr, func %d\n", + srp_tsk->tsk_mgmt_func); + rc = target_submit_tmr(&cmd->se_cmd, nexus->se_sess, NULL, + scsilun_to_int(&srp_tsk->lun), srp_tsk, + tcm_type, GFP_KERNEL, tag_to_abort, 0); + if (rc) { + dev_err(&vscsi->dev, "target_submit_tmr failed, rc %d\n", + rc); + cmd->se_cmd.se_tmr_req->response = + TMR_FUNCTION_REJECTED; + } + } + + if (rc) + transport_send_check_condition_and_sense(&cmd->se_cmd, 0, 0); +} + +static void ibmvscsis_scheduler(struct work_struct *work) +{ + struct ibmvscsis_cmd *cmd = container_of(work, struct ibmvscsis_cmd, + work); + struct scsi_info *vscsi = cmd->adapter; + + spin_lock_bh(&vscsi->intr_lock); + + /* Remove from schedule_q */ + list_del(&cmd->list); + + /* Don't submit cmd if we're disconnecting */ + if (vscsi->flags & (SCHEDULE_DISCONNECT | DISCONNECT_SCHEDULED)) { + ibmvscsis_free_cmd_resources(vscsi, cmd); + + /* ibmvscsis_disconnect might be waiting for us */ + if (list_empty(&vscsi->active_q) && + list_empty(&vscsi->schedule_q) && + (vscsi->flags & WAIT_FOR_IDLE)) { + vscsi->flags &= ~WAIT_FOR_IDLE; + complete(&vscsi->wait_idle); + } + + spin_unlock_bh(&vscsi->intr_lock); + return; + } + + spin_unlock_bh(&vscsi->intr_lock); + + switch (cmd->type) { + case SCSI_CDB: + ibmvscsis_parse_cmd(vscsi, cmd); + break; + case TASK_MANAGEMENT: + ibmvscsis_parse_task(vscsi, cmd); + break; + default: + dev_err(&vscsi->dev, "scheduler, invalid cmd type %d\n", + cmd->type); + spin_lock_bh(&vscsi->intr_lock); + ibmvscsis_free_cmd_resources(vscsi, cmd); + spin_unlock_bh(&vscsi->intr_lock); + break; + } +} + +static int ibmvscsis_alloc_cmds(struct scsi_info *vscsi, int num) +{ + struct ibmvscsis_cmd *cmd; + int i; + + INIT_LIST_HEAD(&vscsi->free_cmd); + vscsi->cmd_pool = kcalloc(num, sizeof(struct ibmvscsis_cmd), + GFP_KERNEL); + if (!vscsi->cmd_pool) + return -ENOMEM; + + for (i = 0, cmd = (struct ibmvscsis_cmd *)vscsi->cmd_pool; i < num; + i++, cmd++) { + cmd->adapter = vscsi; + INIT_WORK(&cmd->work, ibmvscsis_scheduler); + list_add_tail(&cmd->list, &vscsi->free_cmd); + } + + return 0; +} + +static void ibmvscsis_free_cmds(struct scsi_info *vscsi) +{ + kfree(vscsi->cmd_pool); + vscsi->cmd_pool = NULL; + INIT_LIST_HEAD(&vscsi->free_cmd); +} + +/** + * ibmvscsis_service_wait_q() - Service Waiting Queue + * @timer: Pointer to timer which has expired + * + * This routine is called when the timer pops to service the waiting + * queue. Elements on the queue have completed, their responses have been + * copied to the client, but the client's response queue was full so + * the queue message could not be sent. The routine grabs the proper locks + * and calls send messages. + * + * EXECUTION ENVIRONMENT: + * called at interrupt level + */ +static enum hrtimer_restart ibmvscsis_service_wait_q(struct hrtimer *timer) +{ + struct timer_cb *p_timer = container_of(timer, struct timer_cb, timer); + struct scsi_info *vscsi = container_of(p_timer, struct scsi_info, + rsp_q_timer); + + spin_lock_bh(&vscsi->intr_lock); + p_timer->timer_pops += 1; + p_timer->started = false; + ibmvscsis_send_messages(vscsi); + spin_unlock_bh(&vscsi->intr_lock); + + return HRTIMER_NORESTART; +} + +static long ibmvscsis_alloctimer(struct scsi_info *vscsi) +{ + struct timer_cb *p_timer; + + p_timer = &vscsi->rsp_q_timer; + hrtimer_init(&p_timer->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + + p_timer->timer.function = ibmvscsis_service_wait_q; + p_timer->started = false; + p_timer->timer_pops = 0; + + return ADAPT_SUCCESS; +} + +static void ibmvscsis_freetimer(struct scsi_info *vscsi) +{ + struct timer_cb *p_timer; + + p_timer = &vscsi->rsp_q_timer; + + (void)hrtimer_cancel(&p_timer->timer); + + p_timer->started = false; + p_timer->timer_pops = 0; +} + +static irqreturn_t ibmvscsis_interrupt(int dummy, void *data) +{ + struct scsi_info *vscsi = data; + + vio_disable_interrupts(vscsi->dma_dev); + tasklet_schedule(&vscsi->work_task); + + return IRQ_HANDLED; +} + +/** + * ibmvscsis_check_q() - Helper function to Check Init Message Valid + * @vscsi: Pointer to our adapter structure + * + * Checks if a initialize message was queued by the initiatior + * while the timing window was open. This function is called from + * probe after the CRQ is created and interrupts are enabled. + * It would only be used by adapters who wait for some event before + * completing the init handshake with the client. For ibmvscsi, this + * event is waiting for the port to be enabled. + * + * EXECUTION ENVIRONMENT: + * Process level only, interrupt lock held + */ +static long ibmvscsis_check_q(struct scsi_info *vscsi) +{ + uint format; + long rc; + + rc = ibmvscsis_check_init_msg(vscsi, &format); + if (rc) + ibmvscsis_post_disconnect(vscsi, ERR_DISCONNECT_RECONNECT, 0); + else if (format == UNUSED_FORMAT) + vscsi->state = WAIT_ENABLED; + else + vscsi->state = PART_UP_WAIT_ENAB; + + return rc; +} + +/** + * ibmvscsis_enable_change_state() - Set new state based on enabled status + * @vscsi: Pointer to our adapter structure + * + * This function determines our new state now that we are enabled. This + * may involve sending an Init Complete message to the client. + * + * Must be called with interrupt lock held. + */ +static long ibmvscsis_enable_change_state(struct scsi_info *vscsi) +{ + long rc = ADAPT_SUCCESS; + +handle_state_change: + switch (vscsi->state) { + case WAIT_ENABLED: + rc = ibmvscsis_send_init_message(vscsi, INIT_MSG); + switch (rc) { + case H_SUCCESS: + case H_DROPPED: + case H_CLOSED: + vscsi->state = WAIT_CONNECTION; + rc = ADAPT_SUCCESS; + break; + + case H_PARAMETER: + break; + + case H_HARDWARE: + break; + + default: + vscsi->state = UNDEFINED; + rc = H_HARDWARE; + break; + } + break; + case PART_UP_WAIT_ENAB: + rc = ibmvscsis_send_init_message(vscsi, INIT_COMPLETE_MSG); + switch (rc) { + case H_SUCCESS: + vscsi->state = CONNECTED; + rc = ADAPT_SUCCESS; + break; + + case H_DROPPED: + case H_CLOSED: + vscsi->state = WAIT_ENABLED; + goto handle_state_change; + + case H_PARAMETER: + break; + + case H_HARDWARE: + break; + + default: + rc = H_HARDWARE; + break; + } + break; + + case WAIT_CONNECTION: + case WAIT_IDLE: + case SRP_PROCESSING: + case CONNECTED: + rc = ADAPT_SUCCESS; + break; + /* should not be able to get here */ + case UNCONFIGURING: + rc = ERROR; + vscsi->state = UNDEFINED; + break; + + /* driver should never allow this to happen */ + case ERR_DISCONNECT: + case ERR_DISCONNECT_RECONNECT: + default: + dev_err(&vscsi->dev, "in invalid state %d during enable_change_state\n", + vscsi->state); + rc = ADAPT_SUCCESS; + break; + } + + return rc; +} + +/** + * ibmvscsis_create_command_q() - Create Command Queue + * @vscsi: Pointer to our adapter structure + * @num_cmds: Currently unused. In the future, may be used to determine + * the size of the CRQ. + * + * Allocates memory for command queue maps remote memory into an ioba + * initializes the command response queue + * + * EXECUTION ENVIRONMENT: + * Process level only + */ +static long ibmvscsis_create_command_q(struct scsi_info *vscsi, int num_cmds) +{ + long rc = 0; + int pages; + struct vio_dev *vdev = vscsi->dma_dev; + + /* We might support multiple pages in the future, but just 1 for now */ + pages = 1; + + vscsi->cmd_q.size = pages; + + vscsi->cmd_q.base_addr = + (struct viosrp_crq *)get_zeroed_page(GFP_KERNEL); + if (!vscsi->cmd_q.base_addr) + return -ENOMEM; + + vscsi->cmd_q.mask = ((uint)pages * CRQ_PER_PAGE) - 1; + + vscsi->cmd_q.crq_token = dma_map_single(&vdev->dev, + vscsi->cmd_q.base_addr, + PAGE_SIZE, DMA_BIDIRECTIONAL); + if (dma_mapping_error(&vdev->dev, vscsi->cmd_q.crq_token)) { + free_page((unsigned long)vscsi->cmd_q.base_addr); + return -ENOMEM; + } + + rc = h_reg_crq(vscsi->dds.unit_id, vscsi->cmd_q.crq_token, PAGE_SIZE); + if (rc) { + if (rc == H_CLOSED) { + vscsi->state = WAIT_ENABLED; + rc = 0; + } else { + dma_unmap_single(&vdev->dev, vscsi->cmd_q.crq_token, + PAGE_SIZE, DMA_BIDIRECTIONAL); + free_page((unsigned long)vscsi->cmd_q.base_addr); + rc = -ENODEV; + } + } else { + vscsi->state = WAIT_ENABLED; + } + + return rc; +} + +/** + * ibmvscsis_destroy_command_q - Destroy Command Queue + * @vscsi: Pointer to our adapter structure + * + * Releases memory for command queue and unmaps mapped remote memory. + * + * EXECUTION ENVIRONMENT: + * Process level only + */ +static void ibmvscsis_destroy_command_q(struct scsi_info *vscsi) +{ + dma_unmap_single(&vscsi->dma_dev->dev, vscsi->cmd_q.crq_token, + PAGE_SIZE, DMA_BIDIRECTIONAL); + free_page((unsigned long)vscsi->cmd_q.base_addr); + vscsi->cmd_q.base_addr = NULL; + vscsi->state = NO_QUEUE; +} + +static u8 ibmvscsis_fast_fail(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd) +{ + struct iu_entry *iue = cmd->iue; + struct se_cmd *se_cmd = &cmd->se_cmd; + struct srp_cmd *srp = (struct srp_cmd *)iue->sbuf->buf; + struct scsi_sense_hdr sshdr; + u8 rc = se_cmd->scsi_status; + + if (vscsi->fast_fail && (READ_CMD(srp->cdb) || WRITE_CMD(srp->cdb))) + if (scsi_normalize_sense(se_cmd->sense_buffer, + se_cmd->scsi_sense_length, &sshdr)) + if (sshdr.sense_key == HARDWARE_ERROR && + (se_cmd->residual_count == 0 || + se_cmd->residual_count == se_cmd->data_length)) { + rc = NO_SENSE; + cmd->flags |= CMD_FAST_FAIL; + } + + return rc; +} + +/** + * srp_build_response() - Build an SRP response buffer + * @vscsi: Pointer to our adapter structure + * @cmd: Pointer to command for which to send the response + * @len_p: Where to return the length of the IU response sent. This + * is needed to construct the CRQ response. + * + * Build the SRP response buffer and copy it to the client's memory space. + */ +static long srp_build_response(struct scsi_info *vscsi, + struct ibmvscsis_cmd *cmd, uint *len_p) +{ + struct iu_entry *iue = cmd->iue; + struct se_cmd *se_cmd = &cmd->se_cmd; + struct srp_rsp *rsp; + uint len; + u32 rsp_code; + char *data; + u32 *tsk_status; + long rc = ADAPT_SUCCESS; + + spin_lock_bh(&vscsi->intr_lock); + + rsp = &vio_iu(iue)->srp.rsp; + len = sizeof(*rsp); + memset(rsp, 0, len); + data = rsp->data; + + rsp->opcode = SRP_RSP; + + if (vscsi->credit > 0 && vscsi->state == SRP_PROCESSING) + rsp->req_lim_delta = cpu_to_be32(vscsi->credit); + else + rsp->req_lim_delta = cpu_to_be32(1 + vscsi->credit); + rsp->tag = cmd->rsp.tag; + rsp->flags = 0; + + if (cmd->type == SCSI_CDB) { + rsp->status = ibmvscsis_fast_fail(vscsi, cmd); + if (rsp->status) { + pr_debug("build_resp: cmd %p, scsi status %d\n", cmd, + (int)rsp->status); + ibmvscsis_determine_resid(se_cmd, rsp); + if (se_cmd->scsi_sense_length && se_cmd->sense_buffer) { + rsp->sense_data_len = + cpu_to_be32(se_cmd->scsi_sense_length); + rsp->flags |= SRP_RSP_FLAG_SNSVALID; + len += se_cmd->scsi_sense_length; + memcpy(data, se_cmd->sense_buffer, + se_cmd->scsi_sense_length); + } + rsp->sol_not = (cmd->rsp.sol_not & UCSOLNT) >> + UCSOLNT_RESP_SHIFT; + } else if (cmd->flags & CMD_FAST_FAIL) { + pr_debug("build_resp: cmd %p, fast fail\n", cmd); + rsp->sol_not = (cmd->rsp.sol_not & UCSOLNT) >> + UCSOLNT_RESP_SHIFT; + } else { + rsp->sol_not = (cmd->rsp.sol_not & SCSOLNT) >> + SCSOLNT_RESP_SHIFT; + } + } else { + /* this is task management */ + rsp->status = 0; + rsp->resp_data_len = cpu_to_be32(4); + rsp->flags |= SRP_RSP_FLAG_RSPVALID; + + switch (se_cmd->se_tmr_req->response) { + case TMR_FUNCTION_COMPLETE: + case TMR_TASK_DOES_NOT_EXIST: + rsp_code = SRP_TASK_MANAGEMENT_FUNCTION_COMPLETE; + rsp->sol_not = (cmd->rsp.sol_not & SCSOLNT) >> + SCSOLNT_RESP_SHIFT; + break; + case TMR_TASK_MGMT_FUNCTION_NOT_SUPPORTED: + case TMR_LUN_DOES_NOT_EXIST: + rsp_code = SRP_TASK_MANAGEMENT_FUNCTION_NOT_SUPPORTED; + rsp->sol_not = (cmd->rsp.sol_not & UCSOLNT) >> + UCSOLNT_RESP_SHIFT; + break; + case TMR_FUNCTION_FAILED: + case TMR_FUNCTION_REJECTED: + default: + rsp_code = SRP_TASK_MANAGEMENT_FUNCTION_FAILED; + rsp->sol_not = (cmd->rsp.sol_not & UCSOLNT) >> + UCSOLNT_RESP_SHIFT; + break; + } + + tsk_status = (u32 *)data; + *tsk_status = cpu_to_be32(rsp_code); + data = (char *)(tsk_status + 1); + len += 4; + } + + dma_wmb(); + rc = h_copy_rdma(len, vscsi->dds.window[LOCAL].liobn, iue->sbuf->dma, + vscsi->dds.window[REMOTE].liobn, + be64_to_cpu(iue->remote_token)); + + switch (rc) { + case H_SUCCESS: + vscsi->credit = 0; + *len_p = len; + break; + case H_PERMISSION: + if (connection_broken(vscsi)) + vscsi->flags |= RESPONSE_Q_DOWN | CLIENT_FAILED; + + dev_err(&vscsi->dev, "build_response: error copying to client, rc %ld, flags 0x%x, state 0x%hx\n", + rc, vscsi->flags, vscsi->state); + break; + case H_SOURCE_PARM: + case H_DEST_PARM: + default: + dev_err(&vscsi->dev, "build_response: error copying to client, rc %ld\n", + rc); + break; + } + + spin_unlock_bh(&vscsi->intr_lock); + + return rc; +} + +static int ibmvscsis_rdma(struct ibmvscsis_cmd *cmd, struct scatterlist *sg, + int nsg, struct srp_direct_buf *md, int nmd, + enum dma_data_direction dir, unsigned int bytes) +{ + struct iu_entry *iue = cmd->iue; + struct srp_target *target = iue->target; + struct scsi_info *vscsi = target->ldata; + struct scatterlist *sgp; + dma_addr_t client_ioba, server_ioba; + ulong buf_len; + ulong client_len, server_len; + int md_idx; + long tx_len; + long rc = 0; + + pr_debug("rdma: dir %d, bytes 0x%x\n", dir, bytes); + + if (bytes == 0) + return 0; + + sgp = sg; + client_len = 0; + server_len = 0; + md_idx = 0; + tx_len = bytes; + + do { + if (client_len == 0) { + if (md_idx >= nmd) { + dev_err(&vscsi->dev, "rdma: ran out of client memory descriptors\n"); + rc = -EIO; + break; + } + client_ioba = be64_to_cpu(md[md_idx].va); + client_len = be32_to_cpu(md[md_idx].len); + } + if (server_len == 0) { + if (!sgp) { + dev_err(&vscsi->dev, "rdma: ran out of scatter/gather list\n"); + rc = -EIO; + break; + } + server_ioba = sg_dma_address(sgp); + server_len = sg_dma_len(sgp); + } + + buf_len = tx_len; + + if (buf_len > client_len) + buf_len = client_len; + + if (buf_len > server_len) + buf_len = server_len; + + if (buf_len > max_vdma_size) + buf_len = max_vdma_size; + + if (dir == DMA_TO_DEVICE) { + /* read from client */ + rc = h_copy_rdma(buf_len, + vscsi->dds.window[REMOTE].liobn, + client_ioba, + vscsi->dds.window[LOCAL].liobn, + server_ioba); + } else { + /* write to client */ + struct srp_cmd *srp = (struct srp_cmd *)iue->sbuf->buf; + + if (!READ_CMD(srp->cdb)) + print_hex_dump_bytes(" data:", DUMP_PREFIX_NONE, + sg_virt(sgp), buf_len); + /* The h_copy_rdma will cause phyp, running in another + * partition, to read memory, so we need to make sure + * the data has been written out, hence these syncs. + */ + /* ensure that everything is in memory */ + isync(); + /* ensure that memory has been made visible */ + dma_wmb(); + rc = h_copy_rdma(buf_len, + vscsi->dds.window[LOCAL].liobn, + server_ioba, + vscsi->dds.window[REMOTE].liobn, + client_ioba); + } + switch (rc) { + case H_SUCCESS: + break; + case H_PERMISSION: + case H_SOURCE_PARM: + case H_DEST_PARM: + if (connection_broken(vscsi)) { + spin_lock_bh(&vscsi->intr_lock); + vscsi->flags |= + (RESPONSE_Q_DOWN | CLIENT_FAILED); + spin_unlock_bh(&vscsi->intr_lock); + } + dev_err(&vscsi->dev, "rdma: h_copy_rdma failed, rc %ld\n", + rc); + break; + + default: + dev_err(&vscsi->dev, "rdma: unknown error %ld from h_copy_rdma\n", + rc); + break; + } + + if (!rc) { + tx_len -= buf_len; + if (tx_len) { + client_len -= buf_len; + if (client_len == 0) + md_idx++; + else + client_ioba += buf_len; + + server_len -= buf_len; + if (server_len == 0) + sgp = sg_next(sgp); + else + server_ioba += buf_len; + } else { + break; + } + } + } while (!rc); + + return rc; +} + +/** + * ibmvscsis_handle_crq() - Handle CRQ + * @data: Pointer to our adapter structure + * + * Read the command elements from the command queue and copy the payloads + * associated with the command elements to local memory and execute the + * SRP requests. + * + * Note: this is an edge triggered interrupt. It can not be shared. + */ +static void ibmvscsis_handle_crq(unsigned long data) +{ + struct scsi_info *vscsi = (struct scsi_info *)data; + struct viosrp_crq *crq; + long rc; + bool ack = true; + volatile u8 valid; + + spin_lock_bh(&vscsi->intr_lock); + + pr_debug("got interrupt\n"); + + /* + * if we are in a path where we are waiting for all pending commands + * to complete because we received a transport event and anything in + * the command queue is for a new connection, do nothing + */ + if (TARGET_STOP(vscsi)) { + vio_enable_interrupts(vscsi->dma_dev); + + pr_debug("handle_crq, don't process: flags 0x%x, state 0x%hx\n", + vscsi->flags, vscsi->state); + spin_unlock_bh(&vscsi->intr_lock); + return; + } + + rc = vscsi->flags & SCHEDULE_DISCONNECT; + crq = vscsi->cmd_q.base_addr + vscsi->cmd_q.index; + valid = crq->valid; + dma_rmb(); + + while (valid) { + /* + * These are edege triggered interrupts. After dropping out of + * the while loop, the code must check for work since an + * interrupt could be lost, and an elment be left on the queue, + * hence the label. + */ +cmd_work: + vscsi->cmd_q.index = + (vscsi->cmd_q.index + 1) & vscsi->cmd_q.mask; + + if (!rc) { + rc = ibmvscsis_parse_command(vscsi, crq); + } else { + if ((uint)crq->valid == VALID_TRANS_EVENT) { + /* + * must service the transport layer events even + * in an error state, dont break out until all + * the consecutive transport events have been + * processed + */ + rc = ibmvscsis_trans_event(vscsi, crq); + } else if (vscsi->flags & TRANS_EVENT) { + /* + * if a tranport event has occurred leave + * everything but transport events on the queue + */ + pr_debug("handle_crq, ignoring\n"); + + /* + * need to decrement the queue index so we can + * look at the elment again + */ + if (vscsi->cmd_q.index) + vscsi->cmd_q.index -= 1; + else + /* + * index is at 0 it just wrapped. + * have it index last element in q + */ + vscsi->cmd_q.index = vscsi->cmd_q.mask; + break; + } + } + + crq->valid = INVALIDATE_CMD_RESP_EL; + + crq = vscsi->cmd_q.base_addr + vscsi->cmd_q.index; + valid = crq->valid; + dma_rmb(); + } + + if (!rc) { + if (ack) { + vio_enable_interrupts(vscsi->dma_dev); + ack = false; + pr_debug("handle_crq, reenabling interrupts\n"); + } + valid = crq->valid; + dma_rmb(); + if (valid) + goto cmd_work; + } else { + pr_debug("handle_crq, error: flags 0x%x, state 0x%hx, crq index 0x%x\n", + vscsi->flags, vscsi->state, vscsi->cmd_q.index); + } + + pr_debug("Leaving handle_crq: schedule_q empty %d, flags 0x%x, state 0x%hx\n", + (int)list_empty(&vscsi->schedule_q), vscsi->flags, + vscsi->state); + + spin_unlock_bh(&vscsi->intr_lock); +} + +static int ibmvscsis_probe(struct vio_dev *vdev, + const struct vio_device_id *id) +{ + struct scsi_info *vscsi; + int rc = 0; + long hrc = 0; + char wq_name[24]; + + vscsi = kzalloc(sizeof(*vscsi), GFP_KERNEL); + if (!vscsi) { + rc = -ENOMEM; + pr_err("probe: allocation of adapter failed\n"); + return rc; + } + + vscsi->dma_dev = vdev; + vscsi->dev = vdev->dev; + INIT_LIST_HEAD(&vscsi->schedule_q); + INIT_LIST_HEAD(&vscsi->waiting_rsp); + INIT_LIST_HEAD(&vscsi->active_q); + + snprintf(vscsi->tport.tport_name, 256, "%s", dev_name(&vdev->dev)); + + pr_debug("probe tport_name: %s\n", vscsi->tport.tport_name); + + rc = read_dma_window(vscsi); + if (rc) + goto free_adapter; + pr_debug("Probe: liobn 0x%x, riobn 0x%x\n", + vscsi->dds.window[LOCAL].liobn, + vscsi->dds.window[REMOTE].liobn); + + strcpy(vscsi->eye, "VSCSI "); + strncat(vscsi->eye, vdev->name, MAX_EYE); + + vscsi->dds.unit_id = vdev->unit_address; + + spin_lock_bh(&ibmvscsis_dev_lock); + list_add_tail(&vscsi->list, &ibmvscsis_dev_list); + spin_unlock_bh(&ibmvscsis_dev_lock); + + /* + * TBD: How do we determine # of cmds to request? Do we know how + * many "children" we have? + */ + vscsi->request_limit = INITIAL_SRP_LIMIT; + rc = srp_target_alloc(&vscsi->target, &vdev->dev, vscsi->request_limit, + SRP_MAX_IU_LEN); + if (rc) + goto rem_list; + + vscsi->target.ldata = vscsi; + + rc = ibmvscsis_alloc_cmds(vscsi, vscsi->request_limit); + if (rc) { + dev_err(&vscsi->dev, "alloc_cmds failed, rc %d, num %d\n", + rc, vscsi->request_limit); + goto free_target; + } + + /* + * Note: the lock is used in freeing timers, so must initialize + * first so that ordering in case of error is correct. + */ + spin_lock_init(&vscsi->intr_lock); + + rc = ibmvscsis_alloctimer(vscsi); + if (rc) { + dev_err(&vscsi->dev, "probe: alloctimer failed, rc %d\n", rc); + goto free_cmds; + } + + rc = ibmvscsis_create_command_q(vscsi, 256); + if (rc) { + dev_err(&vscsi->dev, "probe: create_command_q failed, rc %d\n", + rc); + goto free_timer; + } + + vscsi->map_buf = kzalloc(PAGE_SIZE, GFP_KERNEL); + if (!vscsi->map_buf) { + rc = -ENOMEM; + dev_err(&vscsi->dev, "probe: allocating cmd buffer failed\n"); + goto destroy_queue; + } + + vscsi->map_ioba = dma_map_single(&vdev->dev, vscsi->map_buf, PAGE_SIZE, + DMA_BIDIRECTIONAL); + if (dma_mapping_error(&vdev->dev, vscsi->map_ioba)) { + dev_err(&vscsi->dev, "probe: error mapping command buffer\n"); + goto free_buf; + } + + hrc = h_vioctl(vscsi->dds.unit_id, H_GET_PARTNER_INFO, + (u64)vscsi->map_ioba | ((u64)PAGE_SIZE << 32), 0, 0, 0, + 0); + if (hrc == H_SUCCESS) + vscsi->client_data.partition_number = + be64_to_cpu(*(u64 *)vscsi->map_buf); + /* + * We expect the VIOCTL to fail if we're configured as "any + * client can connect" and the client isn't activated yet. + * We'll make the call again when he sends an init msg. + */ + pr_debug("probe hrc %ld, client partition num %d\n", + hrc, vscsi->client_data.partition_number); + + tasklet_init(&vscsi->work_task, ibmvscsis_handle_crq, + (unsigned long)vscsi); + + init_completion(&vscsi->wait_idle); + + snprintf(wq_name, 24, "ibmvscsis%s", dev_name(&vdev->dev)); + vscsi->work_q = create_workqueue(wq_name); + if (!vscsi->work_q) { + rc = -ENOMEM; + dev_err(&vscsi->dev, "create_workqueue failed\n"); + goto unmap_buf; + } + + rc = request_irq(vdev->irq, ibmvscsis_interrupt, 0, "ibmvscsis", vscsi); + if (rc) { + rc = -EPERM; + dev_err(&vscsi->dev, "probe: request_irq failed, rc %d\n", rc); + goto destroy_WQ; + } + + spin_lock_bh(&vscsi->intr_lock); + vio_enable_interrupts(vdev); + if (rc) { + dev_err(&vscsi->dev, "enabling interrupts failed, rc %d\n", rc); + rc = -ENODEV; + spin_unlock_bh(&vscsi->intr_lock); + goto free_irq; + } + + if (ibmvscsis_check_q(vscsi)) { + rc = ERROR; + dev_err(&vscsi->dev, "probe: check_q failed, rc %d\n", rc); + spin_unlock_bh(&vscsi->intr_lock); + goto disable_interrupt; + } + spin_unlock_bh(&vscsi->intr_lock); + + dev_set_drvdata(&vdev->dev, vscsi); + + return 0; + +disable_interrupt: + vio_disable_interrupts(vdev); +free_irq: + free_irq(vdev->irq, vscsi); +destroy_WQ: + destroy_workqueue(vscsi->work_q); +unmap_buf: + dma_unmap_single(&vdev->dev, vscsi->map_ioba, PAGE_SIZE, + DMA_BIDIRECTIONAL); +free_buf: + kfree(vscsi->map_buf); +destroy_queue: + tasklet_kill(&vscsi->work_task); + ibmvscsis_unregister_command_q(vscsi); + ibmvscsis_destroy_command_q(vscsi); +free_timer: + ibmvscsis_freetimer(vscsi); +free_cmds: + ibmvscsis_free_cmds(vscsi); +free_target: + srp_target_free(&vscsi->target); +rem_list: + spin_lock_bh(&ibmvscsis_dev_lock); + list_del(&vscsi->list); + spin_unlock_bh(&ibmvscsis_dev_lock); +free_adapter: + kfree(vscsi); + + return rc; +} + +static int ibmvscsis_remove(struct vio_dev *vdev) +{ + struct scsi_info *vscsi = dev_get_drvdata(&vdev->dev); + + pr_debug("remove (%s)\n", dev_name(&vscsi->dma_dev->dev)); + + /* + * TBD: Need to handle if there are commands on the waiting_rsp q + * Actually, can there still be cmds outstanding to tcm? + */ + + vio_disable_interrupts(vdev); + free_irq(vdev->irq, vscsi); + destroy_workqueue(vscsi->work_q); + dma_unmap_single(&vdev->dev, vscsi->map_ioba, PAGE_SIZE, + DMA_BIDIRECTIONAL); + kfree(vscsi->map_buf); + tasklet_kill(&vscsi->work_task); + ibmvscsis_unregister_command_q(vscsi); + ibmvscsis_destroy_command_q(vscsi); + ibmvscsis_freetimer(vscsi); + ibmvscsis_free_cmds(vscsi); + srp_target_free(&vscsi->target); + spin_lock_bh(&ibmvscsis_dev_lock); + list_del(&vscsi->list); + spin_unlock_bh(&ibmvscsis_dev_lock); + kfree(vscsi); + + return 0; +} + +static ssize_t system_id_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + return snprintf(buf, PAGE_SIZE, "%s\n", system_id); +} + +static ssize_t partition_number_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + return snprintf(buf, PAGE_SIZE, "%x\n", partition_number); +} + +static ssize_t unit_address_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct scsi_info *vscsi = container_of(dev, struct scsi_info, dev); + + return snprintf(buf, PAGE_SIZE, "%x\n", vscsi->dma_dev->unit_address); +} + +static int ibmvscsis_get_system_info(void) +{ + struct device_node *rootdn, *vdevdn; + const char *id, *model, *name; + const uint *num; + + rootdn = of_find_node_by_path("/"); + if (!rootdn) + return -ENOENT; + + model = of_get_property(rootdn, "model", NULL); + id = of_get_property(rootdn, "system-id", NULL); + if (model && id) + snprintf(system_id, sizeof(system_id), "%s-%s", model, id); + + name = of_get_property(rootdn, "ibm,partition-name", NULL); + if (name) + strncpy(partition_name, name, sizeof(partition_name)); + + num = of_get_property(rootdn, "ibm,partition-no", NULL); + if (num) + partition_number = *num; + + of_node_put(rootdn); + + vdevdn = of_find_node_by_path("/vdevice"); + if (vdevdn) { + const uint *mvds; + + mvds = of_get_property(vdevdn, "ibm,max-virtual-dma-size", + NULL); + if (mvds) + max_vdma_size = *mvds; + of_node_put(vdevdn); + } + + return 0; +} + +static char *ibmvscsis_get_fabric_name(void) +{ + return "ibmvscsis"; +} + +static char *ibmvscsis_get_fabric_wwn(struct se_portal_group *se_tpg) +{ + struct ibmvscsis_tport *tport = + container_of(se_tpg, struct ibmvscsis_tport, se_tpg); + + return tport->tport_name; +} + +static u16 ibmvscsis_get_tag(struct se_portal_group *se_tpg) +{ + struct ibmvscsis_tport *tport = + container_of(se_tpg, struct ibmvscsis_tport, se_tpg); + + return tport->tport_tpgt; +} + +static u32 ibmvscsis_get_default_depth(struct se_portal_group *se_tpg) +{ + return 1; +} + +static int ibmvscsis_check_true(struct se_portal_group *se_tpg) +{ + return 1; +} + +static int ibmvscsis_check_false(struct se_portal_group *se_tpg) +{ + return 0; +} + +static u32 ibmvscsis_tpg_get_inst_index(struct se_portal_group *se_tpg) +{ + return 1; +} + +static int ibmvscsis_check_stop_free(struct se_cmd *se_cmd) +{ + return target_put_sess_cmd(se_cmd); +} + +static void ibmvscsis_release_cmd(struct se_cmd *se_cmd) +{ + struct ibmvscsis_cmd *cmd = container_of(se_cmd, struct ibmvscsis_cmd, + se_cmd); + struct scsi_info *vscsi = cmd->adapter; + + pr_debug("release_cmd %p, flags %d\n", se_cmd, cmd->flags); + + spin_lock_bh(&vscsi->intr_lock); + /* Remove from active_q */ + list_del(&cmd->list); + list_add_tail(&cmd->list, &vscsi->waiting_rsp); + ibmvscsis_send_messages(vscsi); + spin_unlock_bh(&vscsi->intr_lock); +} + +static u32 ibmvscsis_sess_get_index(struct se_session *se_sess) +{ + return 0; +} + +static int ibmvscsis_write_pending(struct se_cmd *se_cmd) +{ + struct ibmvscsis_cmd *cmd = container_of(se_cmd, struct ibmvscsis_cmd, + se_cmd); + struct iu_entry *iue = cmd->iue; + int rc; + + pr_debug("write_pending, se_cmd %p, length 0x%x\n", + se_cmd, se_cmd->data_length); + + rc = srp_transfer_data(cmd, &vio_iu(iue)->srp.cmd, ibmvscsis_rdma, + 1, 1); + if (rc) { + pr_err("srp_transfer_data() failed: %d\n", rc); + return -EAGAIN; + } + /* + * We now tell TCM to add this WRITE CDB directly into the TCM storage + * object execution queue. + */ + target_execute_cmd(se_cmd); + return 0; +} + +static int ibmvscsis_write_pending_status(struct se_cmd *se_cmd) +{ + return 0; +} + +static void ibmvscsis_set_default_node_attrs(struct se_node_acl *nacl) +{ +} + +static int ibmvscsis_get_cmd_state(struct se_cmd *se_cmd) +{ + return 0; +} + +static int ibmvscsis_queue_data_in(struct se_cmd *se_cmd) +{ + struct ibmvscsis_cmd *cmd = container_of(se_cmd, struct ibmvscsis_cmd, + se_cmd); + struct iu_entry *iue = cmd->iue; + struct scsi_info *vscsi = cmd->adapter; + char *sd; + uint len = 0; + int rc; + + pr_debug("queue_data_in, se_cmd %p, length 0x%x\n", + se_cmd, se_cmd->data_length); + + rc = srp_transfer_data(cmd, &vio_iu(iue)->srp.cmd, ibmvscsis_rdma, 1, + 1); + if (rc) { + pr_err("srp_transfer_data failed: %d\n", rc); + sd = se_cmd->sense_buffer; + se_cmd->scsi_sense_length = 18; + memset(se_cmd->sense_buffer, 0, se_cmd->scsi_sense_length); + /* Logical Unit Communication Time-out asc/ascq = 0x0801 */ + scsi_build_sense_buffer(0, se_cmd->sense_buffer, MEDIUM_ERROR, + 0x08, 0x01); + } + + srp_build_response(vscsi, cmd, &len); + cmd->rsp.format = SRP_FORMAT; + cmd->rsp.len = len; + + return 0; +} + +static int ibmvscsis_queue_status(struct se_cmd *se_cmd) +{ + struct ibmvscsis_cmd *cmd = container_of(se_cmd, struct ibmvscsis_cmd, + se_cmd); + struct scsi_info *vscsi = cmd->adapter; + uint len; + + pr_debug("queue_status %p\n", se_cmd); + + srp_build_response(vscsi, cmd, &len); + cmd->rsp.format = SRP_FORMAT; + cmd->rsp.len = len; + + return 0; +} + +static void ibmvscsis_queue_tm_rsp(struct se_cmd *se_cmd) +{ + struct ibmvscsis_cmd *cmd = container_of(se_cmd, struct ibmvscsis_cmd, + se_cmd); + struct scsi_info *vscsi = cmd->adapter; + uint len; + + pr_debug("queue_tm_rsp %p, status %d\n", + se_cmd, (int)se_cmd->se_tmr_req->response); + + srp_build_response(vscsi, cmd, &len); + cmd->rsp.format = SRP_FORMAT; + cmd->rsp.len = len; +} + +static void ibmvscsis_aborted_task(struct se_cmd *se_cmd) +{ + /* TBD: What (if anything) should we do here? */ + pr_debug("ibmvscsis_aborted_task %p\n", se_cmd); +} + +static struct se_wwn *ibmvscsis_make_tport(struct target_fabric_configfs *tf, + struct config_group *group, + const char *name) +{ + struct ibmvscsis_tport *tport; + + tport = ibmvscsis_lookup_port(name); + if (tport) { + tport->tport_proto_id = SCSI_PROTOCOL_SRP; + pr_debug("make_tport(%s), pointer:%p, tport_id:%x\n", + name, tport, tport->tport_proto_id); + return &tport->tport_wwn; + } + + return ERR_PTR(-EINVAL); +} + +static void ibmvscsis_drop_tport(struct se_wwn *wwn) +{ + struct ibmvscsis_tport *tport = container_of(wwn, + struct ibmvscsis_tport, + tport_wwn); + + pr_debug("drop_tport(%s)\n", + config_item_name(&tport->tport_wwn.wwn_group.cg_item)); +} + +static struct se_portal_group *ibmvscsis_make_tpg(struct se_wwn *wwn, + struct config_group *group, + const char *name) +{ + struct ibmvscsis_tport *tport = + container_of(wwn, struct ibmvscsis_tport, tport_wwn); + int rc; + + tport->releasing = false; + + rc = core_tpg_register(&tport->tport_wwn, &tport->se_tpg, + tport->tport_proto_id); + if (rc) + return ERR_PTR(rc); + + return &tport->se_tpg; +} + +static void ibmvscsis_drop_tpg(struct se_portal_group *se_tpg) +{ + struct ibmvscsis_tport *tport = container_of(se_tpg, + struct ibmvscsis_tport, + se_tpg); + + tport->releasing = true; + tport->enabled = false; + + /* + * Release the virtual I_T Nexus for this ibmvscsis TPG + */ + ibmvscsis_drop_nexus(tport); + /* + * Deregister the se_tpg from TCM.. + */ + core_tpg_deregister(se_tpg); +} + +static ssize_t ibmvscsis_wwn_version_show(struct config_item *item, + char *page) +{ + return scnprintf(page, PAGE_SIZE, "%s\n", IBMVSCSIS_VERSION); +} +CONFIGFS_ATTR_RO(ibmvscsis_wwn_, version); + +static struct configfs_attribute *ibmvscsis_wwn_attrs[] = { + &ibmvscsis_wwn_attr_version, + NULL, +}; + +static ssize_t ibmvscsis_tpg_enable_show(struct config_item *item, + char *page) +{ + struct se_portal_group *se_tpg = to_tpg(item); + struct ibmvscsis_tport *tport = container_of(se_tpg, + struct ibmvscsis_tport, + se_tpg); + + return snprintf(page, PAGE_SIZE, "%d\n", (tport->enabled) ? 1 : 0); +} + +static ssize_t ibmvscsis_tpg_enable_store(struct config_item *item, + const char *page, size_t count) +{ + struct se_portal_group *se_tpg = to_tpg(item); + struct ibmvscsis_tport *tport = container_of(se_tpg, + struct ibmvscsis_tport, + se_tpg); + struct scsi_info *vscsi = container_of(tport, struct scsi_info, tport); + unsigned long tmp; + int rc; + long lrc; + + rc = kstrtoul(page, 0, &tmp); + if (rc < 0) { + pr_err("Unable to extract srpt_tpg_store_enable\n"); + return -EINVAL; + } + + if ((tmp != 0) && (tmp != 1)) { + pr_err("Illegal value for srpt_tpg_store_enable\n"); + return -EINVAL; + } + + if (tmp) { + tport->enabled = true; + spin_lock_bh(&vscsi->intr_lock); + lrc = ibmvscsis_enable_change_state(vscsi); + if (lrc) + pr_err("enable_change_state failed, rc %ld state %d\n", + lrc, vscsi->state); + spin_unlock_bh(&vscsi->intr_lock); + } else { + tport->enabled = false; + } + + pr_debug("tpg_enable_store, state %d\n", vscsi->state); + + return count; +} +CONFIGFS_ATTR(ibmvscsis_tpg_, enable); + +static struct configfs_attribute *ibmvscsis_tpg_attrs[] = { + &ibmvscsis_tpg_attr_enable, + NULL, +}; + +static const struct target_core_fabric_ops ibmvscsis_ops = { + .module = THIS_MODULE, + .name = "ibmvscsis", + .get_fabric_name = ibmvscsis_get_fabric_name, + .tpg_get_wwn = ibmvscsis_get_fabric_wwn, + .tpg_get_tag = ibmvscsis_get_tag, + .tpg_get_default_depth = ibmvscsis_get_default_depth, + .tpg_check_demo_mode = ibmvscsis_check_true, + .tpg_check_demo_mode_cache = ibmvscsis_check_true, + .tpg_check_demo_mode_write_protect = ibmvscsis_check_false, + .tpg_check_prod_mode_write_protect = ibmvscsis_check_false, + .tpg_get_inst_index = ibmvscsis_tpg_get_inst_index, + .check_stop_free = ibmvscsis_check_stop_free, + .release_cmd = ibmvscsis_release_cmd, + .sess_get_index = ibmvscsis_sess_get_index, + .write_pending = ibmvscsis_write_pending, + .write_pending_status = ibmvscsis_write_pending_status, + .set_default_node_attributes = ibmvscsis_set_default_node_attrs, + .get_cmd_state = ibmvscsis_get_cmd_state, + .queue_data_in = ibmvscsis_queue_data_in, + .queue_status = ibmvscsis_queue_status, + .queue_tm_rsp = ibmvscsis_queue_tm_rsp, + .aborted_task = ibmvscsis_aborted_task, + /* + * Setup function pointers for logic in target_core_fabric_configfs.c + */ + .fabric_make_wwn = ibmvscsis_make_tport, + .fabric_drop_wwn = ibmvscsis_drop_tport, + .fabric_make_tpg = ibmvscsis_make_tpg, + .fabric_drop_tpg = ibmvscsis_drop_tpg, + + .tfc_wwn_attrs = ibmvscsis_wwn_attrs, + .tfc_tpg_base_attrs = ibmvscsis_tpg_attrs, +}; + +static void ibmvscsis_dev_release(struct device *dev) {}; + +static struct class_attribute ibmvscsis_class_attrs[] = { + __ATTR_NULL, +}; + +static struct device_attribute dev_attr_system_id = + __ATTR(system_id, S_IRUGO, system_id_show, NULL); + +static struct device_attribute dev_attr_partition_number = + __ATTR(partition_number, S_IRUGO, partition_number_show, NULL); + +static struct device_attribute dev_attr_unit_address = + __ATTR(unit_address, S_IRUGO, unit_address_show, NULL); + +static struct attribute *ibmvscsis_dev_attrs[] = { + &dev_attr_system_id.attr, + &dev_attr_partition_number.attr, + &dev_attr_unit_address.attr, +}; +ATTRIBUTE_GROUPS(ibmvscsis_dev); + +static struct class ibmvscsis_class = { + .name = "ibmvscsis", + .dev_release = ibmvscsis_dev_release, + .class_attrs = ibmvscsis_class_attrs, + .dev_groups = ibmvscsis_dev_groups, +}; + +static struct vio_device_id ibmvscsis_device_table[] = { + { "v-scsi-host", "IBM,v-scsi-host" }, + { "", "" } +}; +MODULE_DEVICE_TABLE(vio, ibmvscsis_device_table); + +static struct vio_driver ibmvscsis_driver = { + .name = "ibmvscsis", + .id_table = ibmvscsis_device_table, + .probe = ibmvscsis_probe, + .remove = ibmvscsis_remove, +}; + +/* + * ibmvscsis_init() - Kernel Module initialization + * + * Note: vio_register_driver() registers callback functions, and at least one + * of those callback functions calls TCM - Linux IO Target Subsystem, thus + * the SCSI Target template must be registered before vio_register_driver() + * is called. + */ +static int __init ibmvscsis_init(void) +{ + int rc = 0; + + rc = ibmvscsis_get_system_info(); + if (rc) { + pr_err("rc %d from get_system_info\n", rc); + goto out; + } + + rc = class_register(&ibmvscsis_class); + if (rc) { + pr_err("failed class register\n"); + goto out; + } + + rc = target_register_template(&ibmvscsis_ops); + if (rc) { + pr_err("rc %d from target_register_template\n", rc); + goto unregister_class; + } + + rc = vio_register_driver(&ibmvscsis_driver); + if (rc) { + pr_err("rc %d from vio_register_driver\n", rc); + goto unregister_target; + } + + return 0; + +unregister_target: + target_unregister_template(&ibmvscsis_ops); +unregister_class: + class_unregister(&ibmvscsis_class); +out: + return rc; +} + +static void __exit ibmvscsis_exit(void) +{ + pr_info("Unregister IBM virtual SCSI host driver\n"); + vio_unregister_driver(&ibmvscsis_driver); + target_unregister_template(&ibmvscsis_ops); + class_unregister(&ibmvscsis_class); +} + +MODULE_DESCRIPTION("IBMVSCSIS fabric driver"); +MODULE_AUTHOR("Bryant G. Ly and Michael Cyr"); +MODULE_LICENSE("GPL"); +MODULE_VERSION(IBMVSCSIS_VERSION); +module_init(ibmvscsis_init); +module_exit(ibmvscsis_exit); diff --git a/drivers/scsi/ibmvscsi_tgt/ibmvscsi_tgt.h b/drivers/scsi/ibmvscsi_tgt/ibmvscsi_tgt.h new file mode 100644 index 000000000000..981a0c992b6c --- /dev/null +++ b/drivers/scsi/ibmvscsi_tgt/ibmvscsi_tgt.h @@ -0,0 +1,346 @@ +/******************************************************************************* + * IBM Virtual SCSI Target Driver + * Copyright (C) 2003-2005 Dave Boutcher (boutcher@us.ibm.com) IBM Corp. + * Santiago Leon (santil@us.ibm.com) IBM Corp. + * Linda Xie (lxie@us.ibm.com) IBM Corp. + * + * Copyright (C) 2005-2011 FUJITA Tomonori <tomof@acm.org> + * Copyright (C) 2010 Nicholas A. Bellinger <nab@kernel.org> + * Copyright (C) 2016 Bryant G. Ly <bryantly@linux.vnet.ibm.com> IBM Corp. + * + * Authors: Bryant G. Ly <bryantly@linux.vnet.ibm.com> + * Authors: Michael Cyr <mikecyr@linux.vnet.ibm.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + ****************************************************************************/ + +#ifndef __H_IBMVSCSI_TGT +#define __H_IBMVSCSI_TGT + +#include "libsrp.h" + +#define SYS_ID_NAME_LEN 64 +#define PARTITION_NAMELEN 96 +#define IBMVSCSIS_NAMELEN 32 + +#define MSG_HI 0 +#define MSG_LOW 1 + +#define MAX_CMD_Q_PAGES 4 +#define CRQ_PER_PAGE (PAGE_SIZE / sizeof(struct viosrp_crq)) +/* in terms of number of elements */ +#define DEFAULT_CMD_Q_SIZE CRQ_PER_PAGE +#define MAX_CMD_Q_SIZE (DEFAULT_CMD_Q_SIZE * MAX_CMD_Q_PAGES) + +#define SRP_VIOLATION 0x102 /* general error code */ + +/* + * SRP buffer formats defined as of 16.a supported by this driver. + */ +#define SUPPORTED_FORMATS ((SRP_DATA_DESC_DIRECT << 1) | \ + (SRP_DATA_DESC_INDIRECT << 1)) + +#define SCSI_LUN_ADDR_METHOD_FLAT 1 + +struct dma_window { + u32 liobn; /* Unique per vdevice */ + u64 tce_base; /* Physical location of the TCE table */ + u64 tce_size; /* Size of the TCE table in bytes */ +}; + +struct target_dds { + u64 unit_id; /* 64 bit will force alignment */ +#define NUM_DMA_WINDOWS 2 +#define LOCAL 0 +#define REMOTE 1 + struct dma_window window[NUM_DMA_WINDOWS]; + + /* root node property "ibm,partition-no" */ + uint partition_num; + char partition_name[PARTITION_NAMELEN]; +}; + +#define MAX_NUM_PORTS 1 +#define MAX_H_COPY_RDMA (128 * 1024) + +#define MAX_EYE 64 + +/* Return codes */ +#define ADAPT_SUCCESS 0L +/* choose error codes that do not conflict with PHYP */ +#define ERROR -40L + +struct format_code { + u8 reserved; + u8 buffers; +}; + +struct client_info { +#define SRP_VERSION "16.a" + char srp_version[8]; + /* root node property ibm,partition-name */ + char partition_name[PARTITION_NAMELEN]; + /* root node property ibm,partition-no */ + u32 partition_number; + /* initially 1 */ + u32 mad_version; + u32 os_type; +}; + +/* + * Changing this constant changes the number of seconds to wait before + * considering the client will never service its queue again. + */ +#define SECONDS_TO_CONSIDER_FAILED 30 +/* + * These constants set the polling period used to determine if the client + * has freed at least one element in the response queue. + */ +#define WAIT_SECONDS 1 +#define WAIT_NANO_SECONDS 5000 +#define MAX_TIMER_POPS ((1000000 / WAIT_NANO_SECONDS) * \ + SECONDS_TO_CONSIDER_FAILED) +/* + * general purpose timer control block + * which can be used for multiple functions + */ +struct timer_cb { + struct hrtimer timer; + /* + * how long has it been since the client + * serviced the queue. The variable is incrmented + * in the service_wait_q routine and cleared + * in send messages + */ + int timer_pops; + /* the timer is started */ + bool started; +}; + +struct cmd_queue { + /* kva */ + struct viosrp_crq *base_addr; + dma_addr_t crq_token; + /* used to maintain index */ + uint mask; + /* current element */ + uint index; + int size; +}; + +#define SCSOLNT_RESP_SHIFT 1 +#define UCSOLNT_RESP_SHIFT 2 + +#define SCSOLNT BIT(SCSOLNT_RESP_SHIFT) +#define UCSOLNT BIT(UCSOLNT_RESP_SHIFT) + +enum cmd_type { + SCSI_CDB = 0x01, + TASK_MANAGEMENT = 0x02, + /* MAD or addressed to port 0 */ + ADAPTER_MAD = 0x04, + UNSET_TYPE = 0x08, +}; + +struct iu_rsp { + u8 format; + u8 sol_not; + u16 len; + /* tag is just to help client identify cmd, so don't translate be/le */ + u64 tag; +}; + +struct ibmvscsis_cmd { + struct list_head list; + /* Used for TCM Core operations */ + struct se_cmd se_cmd; + struct iu_entry *iue; + struct iu_rsp rsp; + struct work_struct work; + struct scsi_info *adapter; + /* Sense buffer that will be mapped into outgoing status */ + unsigned char sense_buf[TRANSPORT_SENSE_BUFFER]; + u64 init_time; +#define CMD_FAST_FAIL BIT(0) + u32 flags; + char type; +}; + +struct ibmvscsis_nexus { + struct se_session *se_sess; +}; + +struct ibmvscsis_tport { + /* SCSI protocol the tport is providing */ + u8 tport_proto_id; + /* ASCII formatted WWPN for SRP Target port */ + char tport_name[IBMVSCSIS_NAMELEN]; + /* Returned by ibmvscsis_make_tport() */ + struct se_wwn tport_wwn; + /* Returned by ibmvscsis_make_tpg() */ + struct se_portal_group se_tpg; + /* ibmvscsis port target portal group tag for TCM */ + u16 tport_tpgt; + /* Pointer to TCM session for I_T Nexus */ + struct ibmvscsis_nexus *ibmv_nexus; + bool enabled; + bool releasing; +}; + +struct scsi_info { + struct list_head list; + char eye[MAX_EYE]; + + /* commands waiting for space on repsonse queue */ + struct list_head waiting_rsp; +#define NO_QUEUE 0x00 +#define WAIT_ENABLED 0X01 + /* driver has received an initialize command */ +#define PART_UP_WAIT_ENAB 0x02 +#define WAIT_CONNECTION 0x04 + /* have established a connection */ +#define CONNECTED 0x08 + /* at least one port is processing SRP IU */ +#define SRP_PROCESSING 0x10 + /* remove request received */ +#define UNCONFIGURING 0x20 + /* disconnect by letting adapter go idle, no error */ +#define WAIT_IDLE 0x40 + /* disconnecting to clear an error */ +#define ERR_DISCONNECT 0x80 + /* disconnect to clear error state, then come back up */ +#define ERR_DISCONNECT_RECONNECT 0x100 + /* disconnected after clearing an error */ +#define ERR_DISCONNECTED 0x200 + /* A series of errors caused unexpected errors */ +#define UNDEFINED 0x400 + u16 state; + int fast_fail; + struct target_dds dds; + char *cmd_pool; + /* list of free commands */ + struct list_head free_cmd; + /* command elements ready for scheduler */ + struct list_head schedule_q; + /* commands sent to TCM */ + struct list_head active_q; + caddr_t *map_buf; + /* ioba of map buffer */ + dma_addr_t map_ioba; + /* allowable number of outstanding SRP requests */ + int request_limit; + /* extra credit */ + int credit; + /* outstanding transactions against credit limit */ + int debit; + + /* allow only one outstanding mad request */ +#define PROCESSING_MAD 0x00002 + /* Waiting to go idle */ +#define WAIT_FOR_IDLE 0x00004 + /* H_REG_CRQ called */ +#define CRQ_CLOSED 0x00010 + /* detected that client has failed */ +#define CLIENT_FAILED 0x00040 + /* detected that transport event occurred */ +#define TRANS_EVENT 0x00080 + /* don't attempt to send anything to the client */ +#define RESPONSE_Q_DOWN 0x00100 + /* request made to schedule disconnect handler */ +#define SCHEDULE_DISCONNECT 0x00400 + /* disconnect handler is scheduled */ +#define DISCONNECT_SCHEDULED 0x00800 + u32 flags; + /* adapter lock */ + spinlock_t intr_lock; + /* information needed to manage command queue */ + struct cmd_queue cmd_q; + /* used in hcall to copy response back into srp buffer */ + u64 empty_iu_id; + /* used in crq, to tag what iu the response is for */ + u64 empty_iu_tag; + uint new_state; + /* control block for the response queue timer */ + struct timer_cb rsp_q_timer; + /* keep last client to enable proper accounting */ + struct client_info client_data; + /* what can this client do */ + u32 client_cap; + /* + * The following two fields capture state and flag changes that + * can occur when the lock is given up. In the orginal design, + * the lock was held during calls into phyp; + * however, phyp did not meet PAPR architecture. This is + * a work around. + */ + u16 phyp_acr_state; + u32 phyp_acr_flags; + + struct workqueue_struct *work_q; + struct completion wait_idle; + struct device dev; + struct vio_dev *dma_dev; + struct srp_target target; + struct ibmvscsis_tport tport; + struct tasklet_struct work_task; + struct work_struct proc_work; +}; + +/* + * Provide a constant that allows software to detect the adapter is + * disconnecting from the client from one of several states. + */ +#define IS_DISCONNECTING (UNCONFIGURING | ERR_DISCONNECT_RECONNECT | \ + ERR_DISCONNECT) + +/* + * Provide a constant that can be used with interrupt handling that + * essentially lets the interrupt handler know that all requests should + * be thrown out, + */ +#define DONT_PROCESS_STATE (IS_DISCONNECTING | UNDEFINED | \ + ERR_DISCONNECTED | WAIT_IDLE) + +/* + * If any of these flag bits are set then do not allow the interrupt + * handler to schedule the off level handler. + */ +#define BLOCK (DISCONNECT_SCHEDULED) + +/* State and transition events that stop the interrupt handler */ +#define TARGET_STOP(VSCSI) (long)(((VSCSI)->state & DONT_PROCESS_STATE) | \ + ((VSCSI)->flags & BLOCK)) + +/* flag bit that are not reset during disconnect */ +#define PRESERVE_FLAG_FIELDS 0 + +#define vio_iu(IUE) ((union viosrp_iu *)((IUE)->sbuf->buf)) + +#define READ_CMD(cdb) (((cdb)[0] & 0x1F) == 8) +#define WRITE_CMD(cdb) (((cdb)[0] & 0x1F) == 0xA) + +#ifndef H_GET_PARTNER_INFO +#define H_GET_PARTNER_INFO 0x0000000000000008LL +#endif + +#define h_copy_rdma(l, sa, sb, da, db) \ + plpar_hcall_norets(H_COPY_RDMA, l, sa, sb, da, db) +#define h_vioctl(u, o, a, u1, u2, u3, u4) \ + plpar_hcall_norets(H_VIOCTL, u, o, a, u1, u2) +#define h_reg_crq(ua, tok, sz) \ + plpar_hcall_norets(H_REG_CRQ, ua, tok, sz) +#define h_free_crq(ua) \ + plpar_hcall_norets(H_FREE_CRQ, ua) +#define h_send_crq(ua, d1, d2) \ + plpar_hcall_norets(H_SEND_CRQ, ua, d1, d2) + +#endif diff --git a/drivers/scsi/ibmvscsi_tgt/libsrp.c b/drivers/scsi/ibmvscsi_tgt/libsrp.c new file mode 100644 index 000000000000..5a4cc28ca5ff --- /dev/null +++ b/drivers/scsi/ibmvscsi_tgt/libsrp.c @@ -0,0 +1,427 @@ +/******************************************************************************* + * SCSI RDMA Protocol lib functions + * + * Copyright (C) 2006 FUJITA Tomonori <tomof@acm.org> + * Copyright (C) 2016 Bryant G. Ly <bryantly@linux.vnet.ibm.com> IBM Corp. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + ***********************************************************************/ + +#define pr_fmt(fmt) "libsrp: " fmt + +#include <linux/printk.h> +#include <linux/err.h> +#include <linux/slab.h> +#include <linux/kfifo.h> +#include <linux/scatterlist.h> +#include <linux/dma-mapping.h> +#include <linux/module.h> +#include <scsi/srp.h> +#include <target/target_core_base.h> +#include "libsrp.h" +#include "ibmvscsi_tgt.h" + +static int srp_iu_pool_alloc(struct srp_queue *q, size_t max, + struct srp_buf **ring) +{ + struct iu_entry *iue; + int i; + + q->pool = kcalloc(max, sizeof(struct iu_entry *), GFP_KERNEL); + if (!q->pool) + return -ENOMEM; + q->items = kcalloc(max, sizeof(struct iu_entry), GFP_KERNEL); + if (!q->items) + goto free_pool; + + spin_lock_init(&q->lock); + kfifo_init(&q->queue, (void *)q->pool, max * sizeof(void *)); + + for (i = 0, iue = q->items; i < max; i++) { + kfifo_in(&q->queue, (void *)&iue, sizeof(void *)); + iue->sbuf = ring[i]; + iue++; + } + return 0; + +free_pool: + kfree(q->pool); + return -ENOMEM; +} + +static void srp_iu_pool_free(struct srp_queue *q) +{ + kfree(q->items); + kfree(q->pool); +} + +static struct srp_buf **srp_ring_alloc(struct device *dev, + size_t max, size_t size) +{ + struct srp_buf **ring; + int i; + + ring = kcalloc(max, sizeof(struct srp_buf *), GFP_KERNEL); + if (!ring) + return NULL; + + for (i = 0; i < max; i++) { + ring[i] = kzalloc(sizeof(*ring[i]), GFP_KERNEL); + if (!ring[i]) + goto out; + ring[i]->buf = dma_alloc_coherent(dev, size, &ring[i]->dma, + GFP_KERNEL); + if (!ring[i]->buf) + goto out; + } + return ring; + +out: + for (i = 0; i < max && ring[i]; i++) { + if (ring[i]->buf) { + dma_free_coherent(dev, size, ring[i]->buf, + ring[i]->dma); + } + kfree(ring[i]); + } + kfree(ring); + + return NULL; +} + +static void srp_ring_free(struct device *dev, struct srp_buf **ring, + size_t max, size_t size) +{ + int i; + + for (i = 0; i < max; i++) { + dma_free_coherent(dev, size, ring[i]->buf, ring[i]->dma); + kfree(ring[i]); + } + kfree(ring); +} + +int srp_target_alloc(struct srp_target *target, struct device *dev, + size_t nr, size_t iu_size) +{ + int err; + + spin_lock_init(&target->lock); + + target->dev = dev; + + target->srp_iu_size = iu_size; + target->rx_ring_size = nr; + target->rx_ring = srp_ring_alloc(target->dev, nr, iu_size); + if (!target->rx_ring) + return -ENOMEM; + err = srp_iu_pool_alloc(&target->iu_queue, nr, target->rx_ring); + if (err) + goto free_ring; + + dev_set_drvdata(target->dev, target); + return 0; + +free_ring: + srp_ring_free(target->dev, target->rx_ring, nr, iu_size); + return -ENOMEM; +} + +void srp_target_free(struct srp_target *target) +{ + dev_set_drvdata(target->dev, NULL); + srp_ring_free(target->dev, target->rx_ring, target->rx_ring_size, + target->srp_iu_size); + srp_iu_pool_free(&target->iu_queue); +} + +struct iu_entry *srp_iu_get(struct srp_target *target) +{ + struct iu_entry *iue = NULL; + + if (kfifo_out_locked(&target->iu_queue.queue, (void *)&iue, + sizeof(void *), + &target->iu_queue.lock) != sizeof(void *)) { + WARN_ONCE(1, "unexpected fifo state"); + return NULL; + } + if (!iue) + return iue; + iue->target = target; + iue->flags = 0; + return iue; +} + +void srp_iu_put(struct iu_entry *iue) +{ + kfifo_in_locked(&iue->target->iu_queue.queue, (void *)&iue, + sizeof(void *), &iue->target->iu_queue.lock); +} + +static int srp_direct_data(struct ibmvscsis_cmd *cmd, struct srp_direct_buf *md, + enum dma_data_direction dir, srp_rdma_t rdma_io, + int dma_map, int ext_desc) +{ + struct iu_entry *iue = NULL; + struct scatterlist *sg = NULL; + int err, nsg = 0, len; + + if (dma_map) { + iue = cmd->iue; + sg = cmd->se_cmd.t_data_sg; + nsg = dma_map_sg(iue->target->dev, sg, cmd->se_cmd.t_data_nents, + DMA_BIDIRECTIONAL); + if (!nsg) { + pr_err("fail to map %p %d\n", iue, + cmd->se_cmd.t_data_nents); + return 0; + } + len = min(cmd->se_cmd.data_length, be32_to_cpu(md->len)); + } else { + len = be32_to_cpu(md->len); + } + + err = rdma_io(cmd, sg, nsg, md, 1, dir, len); + + if (dma_map) + dma_unmap_sg(iue->target->dev, sg, nsg, DMA_BIDIRECTIONAL); + + return err; +} + +static int srp_indirect_data(struct ibmvscsis_cmd *cmd, struct srp_cmd *srp_cmd, + struct srp_indirect_buf *id, + enum dma_data_direction dir, srp_rdma_t rdma_io, + int dma_map, int ext_desc) +{ + struct iu_entry *iue = NULL; + struct srp_direct_buf *md = NULL; + struct scatterlist dummy, *sg = NULL; + dma_addr_t token = 0; + int err = 0; + int nmd, nsg = 0, len; + + if (dma_map || ext_desc) { + iue = cmd->iue; + sg = cmd->se_cmd.t_data_sg; + } + + nmd = be32_to_cpu(id->table_desc.len) / sizeof(struct srp_direct_buf); + + if ((dir == DMA_FROM_DEVICE && nmd == srp_cmd->data_in_desc_cnt) || + (dir == DMA_TO_DEVICE && nmd == srp_cmd->data_out_desc_cnt)) { + md = &id->desc_list[0]; + goto rdma; + } + + if (ext_desc && dma_map) { + md = dma_alloc_coherent(iue->target->dev, + be32_to_cpu(id->table_desc.len), + &token, GFP_KERNEL); + if (!md) { + pr_err("Can't get dma memory %u\n", + be32_to_cpu(id->table_desc.len)); + return -ENOMEM; + } + + sg_init_one(&dummy, md, be32_to_cpu(id->table_desc.len)); + sg_dma_address(&dummy) = token; + sg_dma_len(&dummy) = be32_to_cpu(id->table_desc.len); + err = rdma_io(cmd, &dummy, 1, &id->table_desc, 1, DMA_TO_DEVICE, + be32_to_cpu(id->table_desc.len)); + if (err) { + pr_err("Error copying indirect table %d\n", err); + goto free_mem; + } + } else { + pr_err("This command uses external indirect buffer\n"); + return -EINVAL; + } + +rdma: + if (dma_map) { + nsg = dma_map_sg(iue->target->dev, sg, cmd->se_cmd.t_data_nents, + DMA_BIDIRECTIONAL); + if (!nsg) { + pr_err("fail to map %p %d\n", iue, + cmd->se_cmd.t_data_nents); + err = -EIO; + goto free_mem; + } + len = min(cmd->se_cmd.data_length, be32_to_cpu(id->len)); + } else { + len = be32_to_cpu(id->len); + } + + err = rdma_io(cmd, sg, nsg, md, nmd, dir, len); + + if (dma_map) + dma_unmap_sg(iue->target->dev, sg, nsg, DMA_BIDIRECTIONAL); + +free_mem: + if (token && dma_map) { + dma_free_coherent(iue->target->dev, + be32_to_cpu(id->table_desc.len), md, token); + } + return err; +} + +static int data_out_desc_size(struct srp_cmd *cmd) +{ + int size = 0; + u8 fmt = cmd->buf_fmt >> 4; + + switch (fmt) { + case SRP_NO_DATA_DESC: + break; + case SRP_DATA_DESC_DIRECT: + size = sizeof(struct srp_direct_buf); + break; + case SRP_DATA_DESC_INDIRECT: + size = sizeof(struct srp_indirect_buf) + + sizeof(struct srp_direct_buf) * cmd->data_out_desc_cnt; + break; + default: + pr_err("client error. Invalid data_out_format %x\n", fmt); + break; + } + return size; +} + +/* + * TODO: this can be called multiple times for a single command if it + * has very long data. + */ +int srp_transfer_data(struct ibmvscsis_cmd *cmd, struct srp_cmd *srp_cmd, + srp_rdma_t rdma_io, int dma_map, int ext_desc) +{ + struct srp_direct_buf *md; + struct srp_indirect_buf *id; + enum dma_data_direction dir; + int offset, err = 0; + u8 format; + + if (!cmd->se_cmd.t_data_nents) + return 0; + + offset = srp_cmd->add_cdb_len & ~3; + + dir = srp_cmd_direction(srp_cmd); + if (dir == DMA_FROM_DEVICE) + offset += data_out_desc_size(srp_cmd); + + if (dir == DMA_TO_DEVICE) + format = srp_cmd->buf_fmt >> 4; + else + format = srp_cmd->buf_fmt & ((1U << 4) - 1); + + switch (format) { + case SRP_NO_DATA_DESC: + break; + case SRP_DATA_DESC_DIRECT: + md = (struct srp_direct_buf *)(srp_cmd->add_data + offset); + err = srp_direct_data(cmd, md, dir, rdma_io, dma_map, ext_desc); + break; + case SRP_DATA_DESC_INDIRECT: + id = (struct srp_indirect_buf *)(srp_cmd->add_data + offset); + err = srp_indirect_data(cmd, srp_cmd, id, dir, rdma_io, dma_map, + ext_desc); + break; + default: + pr_err("Unknown format %d %x\n", dir, format); + err = -EINVAL; + } + + return err; +} + +u64 srp_data_length(struct srp_cmd *cmd, enum dma_data_direction dir) +{ + struct srp_direct_buf *md; + struct srp_indirect_buf *id; + u64 len = 0; + uint offset = cmd->add_cdb_len & ~3; + u8 fmt; + + if (dir == DMA_TO_DEVICE) { + fmt = cmd->buf_fmt >> 4; + } else { + fmt = cmd->buf_fmt & ((1U << 4) - 1); + offset += data_out_desc_size(cmd); + } + + switch (fmt) { + case SRP_NO_DATA_DESC: + break; + case SRP_DATA_DESC_DIRECT: + md = (struct srp_direct_buf *)(cmd->add_data + offset); + len = be32_to_cpu(md->len); + break; + case SRP_DATA_DESC_INDIRECT: + id = (struct srp_indirect_buf *)(cmd->add_data + offset); + len = be32_to_cpu(id->len); + break; + default: + pr_err("invalid data format %x\n", fmt); + break; + } + return len; +} + +int srp_get_desc_table(struct srp_cmd *srp_cmd, enum dma_data_direction *dir, + u64 *data_len) +{ + struct srp_indirect_buf *idb; + struct srp_direct_buf *db; + uint add_cdb_offset; + int rc; + + /* + * The pointer computations below will only be compiled correctly + * if srp_cmd::add_data is declared as s8*, u8*, s8[] or u8[], so check + * whether srp_cmd::add_data has been declared as a byte pointer. + */ + BUILD_BUG_ON(!__same_type(srp_cmd->add_data[0], (s8)0) + && !__same_type(srp_cmd->add_data[0], (u8)0)); + + BUG_ON(!dir); + BUG_ON(!data_len); + + rc = 0; + *data_len = 0; + + *dir = DMA_NONE; + + if (srp_cmd->buf_fmt & 0xf) + *dir = DMA_FROM_DEVICE; + else if (srp_cmd->buf_fmt >> 4) + *dir = DMA_TO_DEVICE; + + add_cdb_offset = srp_cmd->add_cdb_len & ~3; + if (((srp_cmd->buf_fmt & 0xf) == SRP_DATA_DESC_DIRECT) || + ((srp_cmd->buf_fmt >> 4) == SRP_DATA_DESC_DIRECT)) { + db = (struct srp_direct_buf *)(srp_cmd->add_data + + add_cdb_offset); + *data_len = be32_to_cpu(db->len); + } else if (((srp_cmd->buf_fmt & 0xf) == SRP_DATA_DESC_INDIRECT) || + ((srp_cmd->buf_fmt >> 4) == SRP_DATA_DESC_INDIRECT)) { + idb = (struct srp_indirect_buf *)(srp_cmd->add_data + + add_cdb_offset); + + *data_len = be32_to_cpu(idb->len); + } + return rc; +} + +MODULE_DESCRIPTION("SCSI RDMA Protocol lib functions"); +MODULE_AUTHOR("FUJITA Tomonori"); +MODULE_LICENSE("GPL"); diff --git a/drivers/scsi/ibmvscsi_tgt/libsrp.h b/drivers/scsi/ibmvscsi_tgt/libsrp.h new file mode 100644 index 000000000000..4696f331453e --- /dev/null +++ b/drivers/scsi/ibmvscsi_tgt/libsrp.h @@ -0,0 +1,123 @@ +#ifndef __LIBSRP_H__ +#define __LIBSRP_H__ + +#include <linux/list.h> +#include <linux/kfifo.h> +#include <scsi/srp.h> + +enum srp_valid { + INVALIDATE_CMD_RESP_EL = 0, + VALID_CMD_RESP_EL = 0x80, + VALID_INIT_MSG = 0xC0, + VALID_TRANS_EVENT = 0xFF +}; + +enum srp_format { + SRP_FORMAT = 1, + MAD_FORMAT = 2, + OS400_FORMAT = 3, + AIX_FORMAT = 4, + LINUX_FORMAT = 5, + MESSAGE_IN_CRQ = 6 +}; + +enum srp_init_msg { + INIT_MSG = 1, + INIT_COMPLETE_MSG = 2 +}; + +enum srp_trans_event { + UNUSED_FORMAT = 0, + PARTNER_FAILED = 1, + PARTNER_DEREGISTER = 2, + MIGRATED = 6 +}; + +enum srp_status { + HEADER_DESCRIPTOR = 0xF1, + PING = 0xF5, + PING_RESPONSE = 0xF6 +}; + +enum srp_mad_version { + MAD_VERSION_1 = 1 +}; + +enum srp_os_type { + OS400 = 1, + LINUX = 2, + AIX = 3, + OFW = 4 +}; + +enum srp_task_attributes { + SRP_SIMPLE_TASK = 0, + SRP_HEAD_TASK = 1, + SRP_ORDERED_TASK = 2, + SRP_ACA_TASK = 4 +}; + +enum { + SRP_TASK_MANAGEMENT_FUNCTION_COMPLETE = 0, + SRP_REQUEST_FIELDS_INVALID = 2, + SRP_TASK_MANAGEMENT_FUNCTION_NOT_SUPPORTED = 4, + SRP_TASK_MANAGEMENT_FUNCTION_FAILED = 5 +}; + +struct srp_buf { + dma_addr_t dma; + void *buf; +}; + +struct srp_queue { + void *pool; + void *items; + struct kfifo queue; + spinlock_t lock; +}; + +struct srp_target { + struct device *dev; + + spinlock_t lock; + struct list_head cmd_queue; + + size_t srp_iu_size; + struct srp_queue iu_queue; + size_t rx_ring_size; + struct srp_buf **rx_ring; + + void *ldata; +}; + +struct iu_entry { + struct srp_target *target; + + struct list_head ilist; + dma_addr_t remote_token; + unsigned long flags; + + struct srp_buf *sbuf; + u16 iu_len; +}; + +struct ibmvscsis_cmd; + +typedef int (srp_rdma_t)(struct ibmvscsis_cmd *, struct scatterlist *, int, + struct srp_direct_buf *, int, + enum dma_data_direction, unsigned int); +int srp_target_alloc(struct srp_target *, struct device *, size_t, size_t); +void srp_target_free(struct srp_target *); +struct iu_entry *srp_iu_get(struct srp_target *); +void srp_iu_put(struct iu_entry *); +int srp_transfer_data(struct ibmvscsis_cmd *, struct srp_cmd *, + srp_rdma_t, int, int); +u64 srp_data_length(struct srp_cmd *cmd, enum dma_data_direction dir); +int srp_get_desc_table(struct srp_cmd *srp_cmd, enum dma_data_direction *dir, + u64 *data_len); +static inline int srp_cmd_direction(struct srp_cmd *cmd) +{ + return (cmd->buf_fmt >> 4) ? DMA_TO_DEVICE : DMA_FROM_DEVICE; +} + +#endif diff --git a/drivers/scsi/libsas/sas_ata.c b/drivers/scsi/libsas/sas_ata.c index 596a5450f0eb..763f012fdeca 100644 --- a/drivers/scsi/libsas/sas_ata.c +++ b/drivers/scsi/libsas/sas_ata.c @@ -233,15 +233,8 @@ static unsigned int sas_ata_qc_issue(struct ata_queued_cmd *qc) task->task_state_flags = SAS_TASK_STATE_PENDING; qc->lldd_task = task; - switch (qc->tf.protocol) { - case ATA_PROT_NCQ: - task->ata_task.use_ncq = 1; - /* fall through */ - case ATAPI_PROT_DMA: - case ATA_PROT_DMA: - task->ata_task.dma_xfer = 1; - break; - } + task->ata_task.use_ncq = ata_is_ncq(qc->tf.protocol); + task->ata_task.dma_xfer = ata_is_dma(qc->tf.protocol); if (qc->scsicmd) ASSIGN_SAS_TASK(qc->scsicmd, task); diff --git a/drivers/scsi/lpfc/lpfc_init.c b/drivers/scsi/lpfc/lpfc_init.c index adf61b43eb70..734a0428ef0e 100644 --- a/drivers/scsi/lpfc/lpfc_init.c +++ b/drivers/scsi/lpfc/lpfc_init.c @@ -4854,20 +4854,17 @@ static int lpfc_enable_pci_dev(struct lpfc_hba *phba) { struct pci_dev *pdev; - int bars = 0; /* Obtain PCI device reference */ if (!phba->pcidev) goto out_error; else pdev = phba->pcidev; - /* Select PCI BARs */ - bars = pci_select_bars(pdev, IORESOURCE_MEM); /* Enable PCI device */ if (pci_enable_device_mem(pdev)) goto out_error; /* Request PCI resource for the device */ - if (pci_request_selected_regions(pdev, bars, LPFC_DRIVER_NAME)) + if (pci_request_mem_regions(pdev, LPFC_DRIVER_NAME)) goto out_disable_device; /* Set up device as PCI master and save state for EEH */ pci_set_master(pdev); @@ -4884,7 +4881,7 @@ out_disable_device: pci_disable_device(pdev); out_error: lpfc_printf_log(phba, KERN_ERR, LOG_INIT, - "1401 Failed to enable pci device, bars:x%x\n", bars); + "1401 Failed to enable pci device\n"); return -ENODEV; } @@ -4899,17 +4896,14 @@ static void lpfc_disable_pci_dev(struct lpfc_hba *phba) { struct pci_dev *pdev; - int bars; /* Obtain PCI device reference */ if (!phba->pcidev) return; else pdev = phba->pcidev; - /* Select PCI BARs */ - bars = pci_select_bars(pdev, IORESOURCE_MEM); /* Release PCI resource and disable PCI device */ - pci_release_selected_regions(pdev, bars); + pci_release_mem_regions(pdev); pci_disable_device(pdev); return; @@ -9811,7 +9805,6 @@ lpfc_pci_remove_one_s3(struct pci_dev *pdev) struct lpfc_vport **vports; struct lpfc_hba *phba = vport->phba; int i; - int bars = pci_select_bars(pdev, IORESOURCE_MEM); spin_lock_irq(&phba->hbalock); vport->load_flag |= FC_UNLOADING; @@ -9886,7 +9879,7 @@ lpfc_pci_remove_one_s3(struct pci_dev *pdev) lpfc_hba_free(phba); - pci_release_selected_regions(pdev, bars); + pci_release_mem_regions(pdev); pci_disable_device(pdev); } diff --git a/drivers/scsi/osd/osd_initiator.c b/drivers/scsi/osd/osd_initiator.c index 3b11aad03752..2f2a9910e30e 100644 --- a/drivers/scsi/osd/osd_initiator.c +++ b/drivers/scsi/osd/osd_initiator.c @@ -726,7 +726,7 @@ static int _osd_req_list_objects(struct osd_request *or, return PTR_ERR(bio); } - bio->bi_rw &= ~REQ_WRITE; + bio_set_op_attrs(bio, REQ_OP_READ, 0); or->in.bio = bio; or->in.total_bytes = bio->bi_iter.bi_size; return 0; @@ -824,7 +824,7 @@ void osd_req_write(struct osd_request *or, { _osd_req_encode_common(or, OSD_ACT_WRITE, obj, offset, len); WARN_ON(or->out.bio || or->out.total_bytes); - WARN_ON(0 == (bio->bi_rw & REQ_WRITE)); + WARN_ON(!op_is_write(bio_op(bio))); or->out.bio = bio; or->out.total_bytes = len; } @@ -839,7 +839,7 @@ int osd_req_write_kern(struct osd_request *or, if (IS_ERR(bio)) return PTR_ERR(bio); - bio->bi_rw |= REQ_WRITE; /* FIXME: bio_set_dir() */ + bio_set_op_attrs(bio, REQ_OP_WRITE, 0); osd_req_write(or, obj, offset, bio, len); return 0; } @@ -875,7 +875,7 @@ void osd_req_read(struct osd_request *or, { _osd_req_encode_common(or, OSD_ACT_READ, obj, offset, len); WARN_ON(or->in.bio || or->in.total_bytes); - WARN_ON(bio->bi_rw & REQ_WRITE); + WARN_ON(op_is_write(bio_op(bio))); or->in.bio = bio; or->in.total_bytes = len; } @@ -956,7 +956,7 @@ static int _osd_req_finalize_cdb_cont(struct osd_request *or, const u8 *cap_key) if (IS_ERR(bio)) return PTR_ERR(bio); - bio->bi_rw |= REQ_WRITE; + bio_set_op_attrs(bio, REQ_OP_WRITE, 0); /* integrity check the continuation before the bio is linked * with the other data segments since the continuation @@ -1077,7 +1077,7 @@ int osd_req_write_sg_kern(struct osd_request *or, if (IS_ERR(bio)) return PTR_ERR(bio); - bio->bi_rw |= REQ_WRITE; + bio_set_op_attrs(bio, REQ_OP_WRITE, 0); osd_req_write_sg(or, obj, bio, sglist, numentries); return 0; @@ -1558,18 +1558,25 @@ static int _osd_req_finalize_data_integrity(struct osd_request *or, static struct request *_make_request(struct request_queue *q, bool has_write, struct _osd_io_info *oii, gfp_t flags) { - if (oii->bio) - return blk_make_request(q, oii->bio, flags); - else { - struct request *req; - - req = blk_get_request(q, has_write ? WRITE : READ, flags); - if (IS_ERR(req)) - return req; + struct request *req; + struct bio *bio = oii->bio; + int ret; - blk_rq_set_block_pc(req); + req = blk_get_request(q, has_write ? WRITE : READ, flags); + if (IS_ERR(req)) return req; + blk_rq_set_block_pc(req); + + for_each_bio(bio) { + struct bio *bounce_bio = bio; + + blk_queue_bounce(req->q, &bounce_bio); + ret = blk_rq_append_bio(req, bounce_bio); + if (ret) + return ERR_PTR(ret); } + + return req; } static int _init_blk_request(struct osd_request *or, diff --git a/drivers/scsi/qla2xxx/qla_nx.h b/drivers/scsi/qla2xxx/qla_nx.h index 59c477883a73..6201dce3553b 100644 --- a/drivers/scsi/qla2xxx/qla_nx.h +++ b/drivers/scsi/qla2xxx/qla_nx.h @@ -1183,7 +1183,6 @@ static const int MD_MIU_TEST_AGT_RDDATA[] = { 0x410000A8, 0x410000AC, #define CRB_NIU_XG_PAUSE_CTL_P1 0x8 #define qla82xx_get_temp_val(x) ((x) >> 16) -#define qla82xx_get_temp_val1(x) ((x) && 0x0000FFFF) #define qla82xx_get_temp_state(x) ((x) & 0xffff) #define qla82xx_encode_temp(val, state) (((val) << 16) | (state)) diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c index 60bff78e9ead..d3e852ad5aa3 100644 --- a/drivers/scsi/sd.c +++ b/drivers/scsi/sd.c @@ -1012,7 +1012,8 @@ static int sd_setup_read_write_cmnd(struct scsi_cmnd *SCpnt) } else if (rq_data_dir(rq) == READ) { SCpnt->cmnd[0] = READ_6; } else { - scmd_printk(KERN_ERR, SCpnt, "Unknown command %llx\n", (unsigned long long) rq->cmd_flags); + scmd_printk(KERN_ERR, SCpnt, "Unknown command %llu,%llx\n", + req_op(rq), (unsigned long long) rq->cmd_flags); goto out; } @@ -1137,21 +1138,26 @@ static int sd_init_command(struct scsi_cmnd *cmd) { struct request *rq = cmd->request; - if (rq->cmd_flags & REQ_DISCARD) + switch (req_op(rq)) { + case REQ_OP_DISCARD: return sd_setup_discard_cmnd(cmd); - else if (rq->cmd_flags & REQ_WRITE_SAME) + case REQ_OP_WRITE_SAME: return sd_setup_write_same_cmnd(cmd); - else if (rq->cmd_flags & REQ_FLUSH) + case REQ_OP_FLUSH: return sd_setup_flush_cmnd(cmd); - else + case REQ_OP_READ: + case REQ_OP_WRITE: return sd_setup_read_write_cmnd(cmd); + default: + BUG(); + } } static void sd_uninit_command(struct scsi_cmnd *SCpnt) { struct request *rq = SCpnt->request; - if (rq->cmd_flags & REQ_DISCARD) + if (req_op(rq) == REQ_OP_DISCARD) __free_page(rq->completion_data); if (SCpnt->cmnd != rq->cmd) { @@ -1613,8 +1619,7 @@ static int sd_pr_register(struct block_device *bdev, u64 old_key, u64 new_key, return -EOPNOTSUPP; return sd_pr_command(bdev, (flags & PR_FL_IGNORE_KEY) ? 0x06 : 0x00, old_key, new_key, 0, - (1 << 0) /* APTPL */ | - (1 << 2) /* ALL_TG_PT */); + (1 << 0) /* APTPL */); } static int sd_pr_reserve(struct block_device *bdev, u64 key, enum pr_type type, @@ -1774,7 +1779,7 @@ static int sd_done(struct scsi_cmnd *SCpnt) unsigned char op = SCpnt->cmnd[0]; unsigned char unmap = SCpnt->cmnd[1] & 8; - if (req->cmd_flags & REQ_DISCARD || req->cmd_flags & REQ_WRITE_SAME) { + if (req_op(req) == REQ_OP_DISCARD || req_op(req) == REQ_OP_WRITE_SAME) { if (!result) { good_bytes = blk_rq_bytes(req); scsi_set_resid(SCpnt, 0); @@ -2988,7 +2993,6 @@ static void sd_probe_async(void *data, async_cookie_t cookie) sd_revalidate_disk(gd); - gd->driverfs_dev = &sdp->sdev_gendev; gd->flags = GENHD_FL_EXT_DEVT; if (sdp->removable) { gd->flags |= GENHD_FL_REMOVABLE; @@ -2996,7 +3000,7 @@ static void sd_probe_async(void *data, async_cookie_t cookie) } blk_pm_runtime_init(sdp->request_queue, dev); - add_disk(gd); + device_add_disk(dev, gd); if (sdkp->capacity) sd_dif_config_host(sdkp); diff --git a/drivers/scsi/sr.c b/drivers/scsi/sr.c index 64c867405ad4..ed179348de80 100644 --- a/drivers/scsi/sr.c +++ b/drivers/scsi/sr.c @@ -713,7 +713,6 @@ static int sr_probe(struct device *dev) get_capabilities(cd); sr_vendor_init(cd); - disk->driverfs_dev = &sdev->sdev_gendev; set_capacity(disk, cd->capacity); disk->private_data = &cd->driver; disk->queue = sdev->request_queue; @@ -730,7 +729,7 @@ static int sr_probe(struct device *dev) dev_set_drvdata(dev, cd); disk->flags |= GENHD_FL_REMOVABLE; - add_disk(disk); + device_add_disk(&sdev->sdev_gendev, disk); sdev_printk(KERN_DEBUG, sdev, "Attached scsi CD-ROM %s\n", cd->cdi.name); |