summaryrefslogtreecommitdiff
path: root/drivers/infiniband/ulp/iser
diff options
context:
space:
mode:
authorAndré Fabian Silva Delgado <emulatorman@parabola.nu>2016-03-25 03:53:42 -0300
committerAndré Fabian Silva Delgado <emulatorman@parabola.nu>2016-03-25 03:53:42 -0300
commit03dd4cb26d967f9588437b0fc9cc0e8353322bb7 (patch)
treefa581f6dc1c0596391690d1f67eceef3af8246dc /drivers/infiniband/ulp/iser
parentd4e493caf788ef44982e131ff9c786546904d934 (diff)
Linux-libre 4.5-gnu
Diffstat (limited to 'drivers/infiniband/ulp/iser')
-rw-r--r--drivers/infiniband/ulp/iser/iscsi_iser.c13
-rw-r--r--drivers/infiniband/ulp/iser/iscsi_iser.h156
-rw-r--r--drivers/infiniband/ulp/iser/iser_initiator.c323
-rw-r--r--drivers/infiniband/ulp/iser/iser_memory.c179
-rw-r--r--drivers/infiniband/ulp/iser/iser_verbs.c337
5 files changed, 463 insertions, 545 deletions
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c b/drivers/infiniband/ulp/iser/iscsi_iser.c
index 9080161e0..c827c93f4 100644
--- a/drivers/infiniband/ulp/iser/iscsi_iser.c
+++ b/drivers/infiniband/ulp/iser/iscsi_iser.c
@@ -644,7 +644,7 @@ iscsi_iser_session_create(struct iscsi_endpoint *ep,
ib_conn = &iser_conn->ib_conn;
if (ib_conn->pi_support) {
- u32 sig_caps = ib_conn->device->dev_attr.sig_prot_cap;
+ u32 sig_caps = ib_conn->device->ib_device->attrs.sig_prot_cap;
scsi_host_set_prot(shost, iser_dif_prot_caps(sig_caps));
scsi_host_set_guard(shost, SHOST_DIX_GUARD_IP |
@@ -656,7 +656,7 @@ iscsi_iser_session_create(struct iscsi_endpoint *ep,
* max fastreg page list length.
*/
shost->sg_tablesize = min_t(unsigned short, shost->sg_tablesize,
- ib_conn->device->dev_attr.max_fast_reg_page_list_len);
+ ib_conn->device->ib_device->attrs.max_fast_reg_page_list_len);
shost->max_sectors = min_t(unsigned int,
1024, (shost->sg_tablesize * PAGE_SIZE) >> 9);
@@ -1059,7 +1059,8 @@ static int __init iser_init(void)
release_wq = alloc_workqueue("release workqueue", 0, 0);
if (!release_wq) {
iser_err("failed to allocate release workqueue\n");
- return -ENOMEM;
+ err = -ENOMEM;
+ goto err_alloc_wq;
}
iscsi_iser_scsi_transport = iscsi_register_transport(
@@ -1067,12 +1068,14 @@ static int __init iser_init(void)
if (!iscsi_iser_scsi_transport) {
iser_err("iscsi_register_transport failed\n");
err = -EINVAL;
- goto register_transport_failure;
+ goto err_reg;
}
return 0;
-register_transport_failure:
+err_reg:
+ destroy_workqueue(release_wq);
+err_alloc_wq:
kmem_cache_destroy(ig.desc_cache);
return err;
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h b/drivers/infiniband/ulp/iser/iscsi_iser.h
index 8a5998e6a..95f0a64e0 100644
--- a/drivers/infiniband/ulp/iser/iscsi_iser.h
+++ b/drivers/infiniband/ulp/iser/iscsi_iser.h
@@ -48,6 +48,7 @@
#include <scsi/scsi_transport_iscsi.h>
#include <scsi/scsi_cmnd.h>
#include <scsi/scsi_device.h>
+#include <scsi/iser.h>
#include <linux/interrupt.h>
#include <linux/wait.h>
@@ -151,46 +152,10 @@
- ISER_MAX_RX_MISC_PDUS) / \
(1 + ISER_INFLIGHT_DATAOUTS))
-#define ISER_WC_BATCH_COUNT 16
#define ISER_SIGNAL_CMD_COUNT 32
-#define ISER_VER 0x10
-#define ISER_WSV 0x08
-#define ISER_RSV 0x04
-
-#define ISER_FASTREG_LI_WRID 0xffffffffffffffffULL
-#define ISER_BEACON_WRID 0xfffffffffffffffeULL
-
-/**
- * struct iser_hdr - iSER header
- *
- * @flags: flags support (zbva, remote_inv)
- * @rsvd: reserved
- * @write_stag: write rkey
- * @write_va: write virtual address
- * @reaf_stag: read rkey
- * @read_va: read virtual address
- */
-struct iser_hdr {
- u8 flags;
- u8 rsvd[3];
- __be32 write_stag;
- __be64 write_va;
- __be32 read_stag;
- __be64 read_va;
-} __attribute__((packed));
-
-
-#define ISER_ZBVA_NOT_SUPPORTED 0x80
-#define ISER_SEND_W_INV_NOT_SUPPORTED 0x40
-
-struct iser_cm_hdr {
- u8 flags;
- u8 rsvd[3];
-} __packed;
-
/* Constant PDU lengths calculations */
-#define ISER_HEADERS_LEN (sizeof(struct iser_hdr) + sizeof(struct iscsi_hdr))
+#define ISER_HEADERS_LEN (sizeof(struct iser_ctrl) + sizeof(struct iscsi_hdr))
#define ISER_RECV_DATA_SEG_LEN 128
#define ISER_RX_PAYLOAD_SIZE (ISER_HEADERS_LEN + ISER_RECV_DATA_SEG_LEN)
@@ -269,7 +234,7 @@ enum iser_desc_type {
#define ISER_MAX_WRS 7
/**
- * struct iser_tx_desc - iSER TX descriptor (for send wr_id)
+ * struct iser_tx_desc - iSER TX descriptor
*
* @iser_header: iser header
* @iscsi_header: iscsi header
@@ -287,12 +252,13 @@ enum iser_desc_type {
* @sig_attrs: Signature attributes
*/
struct iser_tx_desc {
- struct iser_hdr iser_header;
+ struct iser_ctrl iser_header;
struct iscsi_hdr iscsi_header;
enum iser_desc_type type;
u64 dma_addr;
struct ib_sge tx_sg[2];
int num_sge;
+ struct ib_cqe cqe;
bool mapped;
u8 wr_idx;
union iser_wr {
@@ -306,9 +272,10 @@ struct iser_tx_desc {
};
#define ISER_RX_PAD_SIZE (256 - (ISER_RX_PAYLOAD_SIZE + \
- sizeof(u64) + sizeof(struct ib_sge)))
+ sizeof(u64) + sizeof(struct ib_sge) + \
+ sizeof(struct ib_cqe)))
/**
- * struct iser_rx_desc - iSER RX descriptor (for recv wr_id)
+ * struct iser_rx_desc - iSER RX descriptor
*
* @iser_header: iser header
* @iscsi_header: iscsi header
@@ -318,12 +285,32 @@ struct iser_tx_desc {
* @pad: for sense data TODO: Modify to maximum sense length supported
*/
struct iser_rx_desc {
- struct iser_hdr iser_header;
+ struct iser_ctrl iser_header;
struct iscsi_hdr iscsi_header;
char data[ISER_RECV_DATA_SEG_LEN];
u64 dma_addr;
struct ib_sge rx_sg;
+ struct ib_cqe cqe;
char pad[ISER_RX_PAD_SIZE];
+} __packed;
+
+/**
+ * struct iser_login_desc - iSER login descriptor
+ *
+ * @req: pointer to login request buffer
+ * @resp: pointer to login response buffer
+ * @req_dma: DMA address of login request buffer
+ * @rsp_dma: DMA address of login response buffer
+ * @sge: IB sge for login post recv
+ * @cqe: completion handler
+ */
+struct iser_login_desc {
+ void *req;
+ void *rsp;
+ u64 req_dma;
+ u64 rsp_dma;
+ struct ib_sge sge;
+ struct ib_cqe cqe;
} __attribute__((packed));
struct iser_conn;
@@ -333,18 +320,12 @@ struct iscsi_iser_task;
/**
* struct iser_comp - iSER completion context
*
- * @device: pointer to device handle
* @cq: completion queue
- * @wcs: work completion array
- * @tasklet: Tasklet handle
* @active_qps: Number of active QPs attached
* to completion context
*/
struct iser_comp {
- struct iser_device *device;
struct ib_cq *cq;
- struct ib_wc wcs[ISER_WC_BATCH_COUNT];
- struct tasklet_struct tasklet;
int active_qps;
};
@@ -380,7 +361,6 @@ struct iser_reg_ops {
*
* @ib_device: RDMA device
* @pd: Protection Domain for this device
- * @dev_attr: Device attributes container
* @mr: Global DMA memory region
* @event_handler: IB events handle routine
* @ig_list: entry in devices list
@@ -389,18 +369,19 @@ struct iser_reg_ops {
* cpus and device max completion vectors
* @comps: Dinamically allocated array of completion handlers
* @reg_ops: Registration ops
+ * @remote_inv_sup: Remote invalidate is supported on this device
*/
struct iser_device {
struct ib_device *ib_device;
struct ib_pd *pd;
- struct ib_device_attr dev_attr;
struct ib_mr *mr;
struct ib_event_handler event_handler;
struct list_head ig_list;
int refcount;
int comps_used;
struct iser_comp *comps;
- struct iser_reg_ops *reg_ops;
+ const struct iser_reg_ops *reg_ops;
+ bool remote_inv_sup;
};
#define ISER_CHECK_GUARD 0xc0
@@ -475,10 +456,11 @@ struct iser_fr_pool {
* @rx_wr: receive work request for batch posts
* @device: reference to iser device
* @comp: iser completion context
- * @pi_support: Indicate device T10-PI support
- * @beacon: beacon send wr to signal all flush errors were drained
- * @flush_comp: completes when all connection completions consumed
* @fr_pool: connection fast registration poool
+ * @pi_support: Indicate device T10-PI support
+ * @last: last send wr to signal all flush errors were drained
+ * @last_cqe: cqe handler for last wr
+ * @last_comp: completes when all connection completions consumed
*/
struct ib_conn {
struct rdma_cm_id *cma_id;
@@ -488,10 +470,12 @@ struct ib_conn {
struct ib_recv_wr rx_wr[ISER_MIN_POSTED_RX];
struct iser_device *device;
struct iser_comp *comp;
- bool pi_support;
- struct ib_send_wr beacon;
- struct completion flush_comp;
struct iser_fr_pool fr_pool;
+ bool pi_support;
+ struct ib_send_wr last;
+ struct ib_cqe last_cqe;
+ struct ib_cqe reg_cqe;
+ struct completion last_comp;
};
/**
@@ -514,11 +498,7 @@ struct ib_conn {
* @up_completion: connection establishment completed
* (state is ISER_CONN_UP)
* @conn_list: entry in ig conn list
- * @login_buf: login data buffer (stores login parameters)
- * @login_req_buf: login request buffer
- * @login_req_dma: login request buffer dma address
- * @login_resp_buf: login response buffer
- * @login_resp_dma: login response buffer dma address
+ * @login_desc: login descriptor
* @rx_desc_head: head of rx_descs cyclic buffer
* @rx_descs: rx buffers array (cyclic buffer)
* @num_rx_descs: number of rx descriptors
@@ -541,15 +521,13 @@ struct iser_conn {
struct completion ib_completion;
struct completion up_completion;
struct list_head conn_list;
-
- char *login_buf;
- char *login_req_buf, *login_resp_buf;
- u64 login_req_dma, login_resp_dma;
+ struct iser_login_desc login_desc;
unsigned int rx_desc_head;
struct iser_rx_desc *rx_descs;
u32 num_rx_descs;
unsigned short scsi_sg_tablesize;
unsigned int scsi_max_sectors;
+ bool snd_w_inv;
};
/**
@@ -579,9 +557,8 @@ struct iscsi_iser_task {
struct iser_page_vec {
u64 *pages;
- int length;
- int offset;
- int data_size;
+ int npages;
+ struct ib_mr fake_mr;
};
/**
@@ -633,12 +610,14 @@ int iser_conn_terminate(struct iser_conn *iser_conn);
void iser_release_work(struct work_struct *work);
-void iser_rcv_completion(struct iser_rx_desc *desc,
- unsigned long dto_xfer_len,
- struct ib_conn *ib_conn);
-
-void iser_snd_completion(struct iser_tx_desc *desc,
- struct ib_conn *ib_conn);
+void iser_err_comp(struct ib_wc *wc, const char *type);
+void iser_login_rsp(struct ib_cq *cq, struct ib_wc *wc);
+void iser_task_rsp(struct ib_cq *cq, struct ib_wc *wc);
+void iser_cmd_comp(struct ib_cq *cq, struct ib_wc *wc);
+void iser_ctrl_comp(struct ib_cq *cq, struct ib_wc *wc);
+void iser_dataout_comp(struct ib_cq *cq, struct ib_wc *wc);
+void iser_reg_comp(struct ib_cq *cq, struct ib_wc *wc);
+void iser_last_comp(struct ib_cq *cq, struct ib_wc *wc);
void iser_task_rdma_init(struct iscsi_iser_task *task);
@@ -651,7 +630,8 @@ void iser_finalize_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir);
int iser_reg_rdma_mem(struct iscsi_iser_task *task,
- enum iser_data_dir dir);
+ enum iser_data_dir dir,
+ bool all_imm);
void iser_unreg_rdma_mem(struct iscsi_iser_task *task,
enum iser_data_dir dir);
@@ -719,4 +699,28 @@ iser_tx_next_wr(struct iser_tx_desc *tx_desc)
return cur_wr;
}
+static inline struct iser_conn *
+to_iser_conn(struct ib_conn *ib_conn)
+{
+ return container_of(ib_conn, struct iser_conn, ib_conn);
+}
+
+static inline struct iser_rx_desc *
+iser_rx(struct ib_cqe *cqe)
+{
+ return container_of(cqe, struct iser_rx_desc, cqe);
+}
+
+static inline struct iser_tx_desc *
+iser_tx(struct ib_cqe *cqe)
+{
+ return container_of(cqe, struct iser_tx_desc, cqe);
+}
+
+static inline struct iser_login_desc *
+iser_login(struct ib_cqe *cqe)
+{
+ return container_of(cqe, struct iser_login_desc, cqe);
+}
+
#endif
diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c
index ffd00c420..ed54b388e 100644
--- a/drivers/infiniband/ulp/iser/iser_initiator.c
+++ b/drivers/infiniband/ulp/iser/iser_initiator.c
@@ -51,7 +51,7 @@ static int iser_prepare_read_cmd(struct iscsi_task *task)
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_mem_reg *mem_reg;
int err;
- struct iser_hdr *hdr = &iser_task->desc.iser_header;
+ struct iser_ctrl *hdr = &iser_task->desc.iser_header;
struct iser_data_buf *buf_in = &iser_task->data[ISER_DIR_IN];
err = iser_dma_map_task_data(iser_task,
@@ -72,7 +72,7 @@ static int iser_prepare_read_cmd(struct iscsi_task *task)
return err;
}
- err = iser_reg_rdma_mem(iser_task, ISER_DIR_IN);
+ err = iser_reg_rdma_mem(iser_task, ISER_DIR_IN, false);
if (err) {
iser_err("Failed to set up Data-IN RDMA\n");
return err;
@@ -104,7 +104,7 @@ iser_prepare_write_cmd(struct iscsi_task *task,
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_mem_reg *mem_reg;
int err;
- struct iser_hdr *hdr = &iser_task->desc.iser_header;
+ struct iser_ctrl *hdr = &iser_task->desc.iser_header;
struct iser_data_buf *buf_out = &iser_task->data[ISER_DIR_OUT];
struct ib_sge *tx_dsg = &iser_task->desc.tx_sg[1];
@@ -126,7 +126,8 @@ iser_prepare_write_cmd(struct iscsi_task *task,
return err;
}
- err = iser_reg_rdma_mem(iser_task, ISER_DIR_OUT);
+ err = iser_reg_rdma_mem(iser_task, ISER_DIR_OUT,
+ buf_out->data_len == imm_sz);
if (err != 0) {
iser_err("Failed to register write cmd RDMA mem\n");
return err;
@@ -166,7 +167,7 @@ static void iser_create_send_desc(struct iser_conn *iser_conn,
ib_dma_sync_single_for_cpu(device->ib_device,
tx_desc->dma_addr, ISER_HEADERS_LEN, DMA_TO_DEVICE);
- memset(&tx_desc->iser_header, 0, sizeof(struct iser_hdr));
+ memset(&tx_desc->iser_header, 0, sizeof(struct iser_ctrl));
tx_desc->iser_header.flags = ISER_VER;
tx_desc->num_sge = 1;
}
@@ -174,73 +175,63 @@ static void iser_create_send_desc(struct iser_conn *iser_conn,
static void iser_free_login_buf(struct iser_conn *iser_conn)
{
struct iser_device *device = iser_conn->ib_conn.device;
+ struct iser_login_desc *desc = &iser_conn->login_desc;
- if (!iser_conn->login_buf)
+ if (!desc->req)
return;
- if (iser_conn->login_req_dma)
- ib_dma_unmap_single(device->ib_device,
- iser_conn->login_req_dma,
- ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_TO_DEVICE);
+ ib_dma_unmap_single(device->ib_device, desc->req_dma,
+ ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_TO_DEVICE);
- if (iser_conn->login_resp_dma)
- ib_dma_unmap_single(device->ib_device,
- iser_conn->login_resp_dma,
- ISER_RX_LOGIN_SIZE, DMA_FROM_DEVICE);
+ ib_dma_unmap_single(device->ib_device, desc->rsp_dma,
+ ISER_RX_LOGIN_SIZE, DMA_FROM_DEVICE);
- kfree(iser_conn->login_buf);
+ kfree(desc->req);
+ kfree(desc->rsp);
/* make sure we never redo any unmapping */
- iser_conn->login_req_dma = 0;
- iser_conn->login_resp_dma = 0;
- iser_conn->login_buf = NULL;
+ desc->req = NULL;
+ desc->rsp = NULL;
}
static int iser_alloc_login_buf(struct iser_conn *iser_conn)
{
struct iser_device *device = iser_conn->ib_conn.device;
- int req_err, resp_err;
-
- BUG_ON(device == NULL);
-
- iser_conn->login_buf = kmalloc(ISCSI_DEF_MAX_RECV_SEG_LEN +
- ISER_RX_LOGIN_SIZE, GFP_KERNEL);
- if (!iser_conn->login_buf)
- goto out_err;
-
- iser_conn->login_req_buf = iser_conn->login_buf;
- iser_conn->login_resp_buf = iser_conn->login_buf +
- ISCSI_DEF_MAX_RECV_SEG_LEN;
-
- iser_conn->login_req_dma = ib_dma_map_single(device->ib_device,
- iser_conn->login_req_buf,
- ISCSI_DEF_MAX_RECV_SEG_LEN,
- DMA_TO_DEVICE);
-
- iser_conn->login_resp_dma = ib_dma_map_single(device->ib_device,
- iser_conn->login_resp_buf,
- ISER_RX_LOGIN_SIZE,
- DMA_FROM_DEVICE);
-
- req_err = ib_dma_mapping_error(device->ib_device,
- iser_conn->login_req_dma);
- resp_err = ib_dma_mapping_error(device->ib_device,
- iser_conn->login_resp_dma);
-
- if (req_err || resp_err) {
- if (req_err)
- iser_conn->login_req_dma = 0;
- if (resp_err)
- iser_conn->login_resp_dma = 0;
- goto free_login_buf;
- }
+ struct iser_login_desc *desc = &iser_conn->login_desc;
+
+ desc->req = kmalloc(ISCSI_DEF_MAX_RECV_SEG_LEN, GFP_KERNEL);
+ if (!desc->req)
+ return -ENOMEM;
+
+ desc->req_dma = ib_dma_map_single(device->ib_device, desc->req,
+ ISCSI_DEF_MAX_RECV_SEG_LEN,
+ DMA_TO_DEVICE);
+ if (ib_dma_mapping_error(device->ib_device,
+ desc->req_dma))
+ goto free_req;
+
+ desc->rsp = kmalloc(ISER_RX_LOGIN_SIZE, GFP_KERNEL);
+ if (!desc->rsp)
+ goto unmap_req;
+
+ desc->rsp_dma = ib_dma_map_single(device->ib_device, desc->rsp,
+ ISER_RX_LOGIN_SIZE,
+ DMA_FROM_DEVICE);
+ if (ib_dma_mapping_error(device->ib_device,
+ desc->rsp_dma))
+ goto free_rsp;
+
return 0;
-free_login_buf:
- iser_free_login_buf(iser_conn);
+free_rsp:
+ kfree(desc->rsp);
+unmap_req:
+ ib_dma_unmap_single(device->ib_device, desc->req_dma,
+ ISCSI_DEF_MAX_RECV_SEG_LEN,
+ DMA_TO_DEVICE);
+free_req:
+ kfree(desc->req);
-out_err:
- iser_err("unable to alloc or map login buf\n");
return -ENOMEM;
}
@@ -280,11 +271,11 @@ int iser_alloc_rx_descriptors(struct iser_conn *iser_conn,
goto rx_desc_dma_map_failed;
rx_desc->dma_addr = dma_addr;
-
+ rx_desc->cqe.done = iser_task_rsp;
rx_sg = &rx_desc->rx_sg;
- rx_sg->addr = rx_desc->dma_addr;
+ rx_sg->addr = rx_desc->dma_addr;
rx_sg->length = ISER_RX_PAYLOAD_SIZE;
- rx_sg->lkey = device->pd->local_dma_lkey;
+ rx_sg->lkey = device->pd->local_dma_lkey;
}
iser_conn->rx_desc_head = 0;
@@ -383,6 +374,7 @@ int iser_send_command(struct iscsi_conn *conn,
/* build the tx desc regd header and add it to the tx desc dto */
tx_desc->type = ISCSI_TX_SCSI_COMMAND;
+ tx_desc->cqe.done = iser_cmd_comp;
iser_create_send_desc(iser_conn, tx_desc);
if (hdr->flags & ISCSI_FLAG_CMD_READ) {
@@ -464,6 +456,7 @@ int iser_send_data_out(struct iscsi_conn *conn,
}
tx_desc->type = ISCSI_TX_DATAOUT;
+ tx_desc->cqe.done = iser_dataout_comp;
tx_desc->iser_header.flags = ISER_VER;
memcpy(&tx_desc->iscsi_header, hdr, sizeof(struct iscsi_hdr));
@@ -513,6 +506,7 @@ int iser_send_control(struct iscsi_conn *conn,
/* build the tx desc regd header and add it to the tx desc dto */
mdesc->type = ISCSI_TX_CONTROL;
+ mdesc->cqe.done = iser_ctrl_comp;
iser_create_send_desc(iser_conn, mdesc);
device = iser_conn->ib_conn.device;
@@ -520,25 +514,25 @@ int iser_send_control(struct iscsi_conn *conn,
data_seg_len = ntoh24(task->hdr->dlength);
if (data_seg_len > 0) {
+ struct iser_login_desc *desc = &iser_conn->login_desc;
struct ib_sge *tx_dsg = &mdesc->tx_sg[1];
+
if (task != conn->login_task) {
iser_err("data present on non login task!!!\n");
goto send_control_error;
}
- ib_dma_sync_single_for_cpu(device->ib_device,
- iser_conn->login_req_dma, task->data_count,
- DMA_TO_DEVICE);
+ ib_dma_sync_single_for_cpu(device->ib_device, desc->req_dma,
+ task->data_count, DMA_TO_DEVICE);
- memcpy(iser_conn->login_req_buf, task->data, task->data_count);
+ memcpy(desc->req, task->data, task->data_count);
- ib_dma_sync_single_for_device(device->ib_device,
- iser_conn->login_req_dma, task->data_count,
- DMA_TO_DEVICE);
+ ib_dma_sync_single_for_device(device->ib_device, desc->req_dma,
+ task->data_count, DMA_TO_DEVICE);
- tx_dsg->addr = iser_conn->login_req_dma;
- tx_dsg->length = task->data_count;
- tx_dsg->lkey = device->pd->local_dma_lkey;
+ tx_dsg->addr = desc->req_dma;
+ tx_dsg->length = task->data_count;
+ tx_dsg->lkey = device->pd->local_dma_lkey;
mdesc->num_sge = 2;
}
@@ -562,41 +556,126 @@ send_control_error:
return err;
}
-/**
- * iser_rcv_dto_completion - recv DTO completion
- */
-void iser_rcv_completion(struct iser_rx_desc *rx_desc,
- unsigned long rx_xfer_len,
- struct ib_conn *ib_conn)
+void iser_login_rsp(struct ib_cq *cq, struct ib_wc *wc)
{
- struct iser_conn *iser_conn = container_of(ib_conn, struct iser_conn,
- ib_conn);
+ struct ib_conn *ib_conn = wc->qp->qp_context;
+ struct iser_conn *iser_conn = to_iser_conn(ib_conn);
+ struct iser_login_desc *desc = iser_login(wc->wr_cqe);
struct iscsi_hdr *hdr;
- u64 rx_dma;
- int rx_buflen, outstanding, count, err;
+ char *data;
+ int length;
- /* differentiate between login to all other PDUs */
- if ((char *)rx_desc == iser_conn->login_resp_buf) {
- rx_dma = iser_conn->login_resp_dma;
- rx_buflen = ISER_RX_LOGIN_SIZE;
- } else {
- rx_dma = rx_desc->dma_addr;
- rx_buflen = ISER_RX_PAYLOAD_SIZE;
+ if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ iser_err_comp(wc, "login_rsp");
+ return;
+ }
+
+ ib_dma_sync_single_for_cpu(ib_conn->device->ib_device,
+ desc->rsp_dma, ISER_RX_LOGIN_SIZE,
+ DMA_FROM_DEVICE);
+
+ hdr = desc->rsp + sizeof(struct iser_ctrl);
+ data = desc->rsp + ISER_HEADERS_LEN;
+ length = wc->byte_len - ISER_HEADERS_LEN;
+
+ iser_dbg("op 0x%x itt 0x%x dlen %d\n", hdr->opcode,
+ hdr->itt, length);
+
+ iscsi_iser_recv(iser_conn->iscsi_conn, hdr, data, length);
+
+ ib_dma_sync_single_for_device(ib_conn->device->ib_device,
+ desc->rsp_dma, ISER_RX_LOGIN_SIZE,
+ DMA_FROM_DEVICE);
+
+ ib_conn->post_recv_buf_count--;
+}
+
+static inline void
+iser_inv_desc(struct iser_fr_desc *desc, u32 rkey)
+{
+ if (likely(rkey == desc->rsc.mr->rkey))
+ desc->rsc.mr_valid = 0;
+ else if (likely(rkey == desc->pi_ctx->sig_mr->rkey))
+ desc->pi_ctx->sig_mr_valid = 0;
+}
+
+static int
+iser_check_remote_inv(struct iser_conn *iser_conn,
+ struct ib_wc *wc,
+ struct iscsi_hdr *hdr)
+{
+ if (wc->wc_flags & IB_WC_WITH_INVALIDATE) {
+ struct iscsi_task *task;
+ u32 rkey = wc->ex.invalidate_rkey;
+
+ iser_dbg("conn %p: remote invalidation for rkey %#x\n",
+ iser_conn, rkey);
+
+ if (unlikely(!iser_conn->snd_w_inv)) {
+ iser_err("conn %p: unexepected remote invalidation, "
+ "terminating connection\n", iser_conn);
+ return -EPROTO;
+ }
+
+ task = iscsi_itt_to_ctask(iser_conn->iscsi_conn, hdr->itt);
+ if (likely(task)) {
+ struct iscsi_iser_task *iser_task = task->dd_data;
+ struct iser_fr_desc *desc;
+
+ if (iser_task->dir[ISER_DIR_IN]) {
+ desc = iser_task->rdma_reg[ISER_DIR_IN].mem_h;
+ iser_inv_desc(desc, rkey);
+ }
+
+ if (iser_task->dir[ISER_DIR_OUT]) {
+ desc = iser_task->rdma_reg[ISER_DIR_OUT].mem_h;
+ iser_inv_desc(desc, rkey);
+ }
+ } else {
+ iser_err("failed to get task for itt=%d\n", hdr->itt);
+ return -EINVAL;
+ }
}
- ib_dma_sync_single_for_cpu(ib_conn->device->ib_device, rx_dma,
- rx_buflen, DMA_FROM_DEVICE);
+ return 0;
+}
- hdr = &rx_desc->iscsi_header;
+
+void iser_task_rsp(struct ib_cq *cq, struct ib_wc *wc)
+{
+ struct ib_conn *ib_conn = wc->qp->qp_context;
+ struct iser_conn *iser_conn = to_iser_conn(ib_conn);
+ struct iser_rx_desc *desc = iser_rx(wc->wr_cqe);
+ struct iscsi_hdr *hdr;
+ int length;
+ int outstanding, count, err;
+
+ if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ iser_err_comp(wc, "task_rsp");
+ return;
+ }
+
+ ib_dma_sync_single_for_cpu(ib_conn->device->ib_device,
+ desc->dma_addr, ISER_RX_PAYLOAD_SIZE,
+ DMA_FROM_DEVICE);
+
+ hdr = &desc->iscsi_header;
+ length = wc->byte_len - ISER_HEADERS_LEN;
iser_dbg("op 0x%x itt 0x%x dlen %d\n", hdr->opcode,
- hdr->itt, (int)(rx_xfer_len - ISER_HEADERS_LEN));
+ hdr->itt, length);
+
+ if (iser_check_remote_inv(iser_conn, wc, hdr)) {
+ iscsi_conn_failure(iser_conn->iscsi_conn,
+ ISCSI_ERR_CONN_FAILED);
+ return;
+ }
- iscsi_iser_recv(iser_conn->iscsi_conn, hdr, rx_desc->data,
- rx_xfer_len - ISER_HEADERS_LEN);
+ iscsi_iser_recv(iser_conn->iscsi_conn, hdr, desc->data, length);
- ib_dma_sync_single_for_device(ib_conn->device->ib_device, rx_dma,
- rx_buflen, DMA_FROM_DEVICE);
+ ib_dma_sync_single_for_device(ib_conn->device->ib_device,
+ desc->dma_addr, ISER_RX_PAYLOAD_SIZE,
+ DMA_FROM_DEVICE);
/* decrementing conn->post_recv_buf_count only --after-- freeing the *
* task eliminates the need to worry on tasks which are completed in *
@@ -604,9 +683,6 @@ void iser_rcv_completion(struct iser_rx_desc *rx_desc,
* for the posted rx bufs refcount to become zero handles everything */
ib_conn->post_recv_buf_count--;
- if (rx_dma == iser_conn->login_resp_dma)
- return;
-
outstanding = ib_conn->post_recv_buf_count;
if (outstanding + iser_conn->min_posted_rx <= iser_conn->qp_max_recv_dtos) {
count = min(iser_conn->qp_max_recv_dtos - outstanding,
@@ -617,26 +693,47 @@ void iser_rcv_completion(struct iser_rx_desc *rx_desc,
}
}
-void iser_snd_completion(struct iser_tx_desc *tx_desc,
- struct ib_conn *ib_conn)
+void iser_cmd_comp(struct ib_cq *cq, struct ib_wc *wc)
{
+ if (unlikely(wc->status != IB_WC_SUCCESS))
+ iser_err_comp(wc, "command");
+}
+
+void iser_ctrl_comp(struct ib_cq *cq, struct ib_wc *wc)
+{
+ struct iser_tx_desc *desc = iser_tx(wc->wr_cqe);
struct iscsi_task *task;
- struct iser_device *device = ib_conn->device;
- if (tx_desc->type == ISCSI_TX_DATAOUT) {
- ib_dma_unmap_single(device->ib_device, tx_desc->dma_addr,
- ISER_HEADERS_LEN, DMA_TO_DEVICE);
- kmem_cache_free(ig.desc_cache, tx_desc);
- tx_desc = NULL;
+ if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ iser_err_comp(wc, "control");
+ return;
}
- if (tx_desc && tx_desc->type == ISCSI_TX_CONTROL) {
- /* this arithmetic is legal by libiscsi dd_data allocation */
- task = (void *) ((long)(void *)tx_desc -
- sizeof(struct iscsi_task));
- if (task->hdr->itt == RESERVED_ITT)
- iscsi_put_task(task);
- }
+ /* this arithmetic is legal by libiscsi dd_data allocation */
+ task = (void *)desc - sizeof(struct iscsi_task);
+ if (task->hdr->itt == RESERVED_ITT)
+ iscsi_put_task(task);
+}
+
+void iser_dataout_comp(struct ib_cq *cq, struct ib_wc *wc)
+{
+ struct iser_tx_desc *desc = iser_tx(wc->wr_cqe);
+ struct ib_conn *ib_conn = wc->qp->qp_context;
+ struct iser_device *device = ib_conn->device;
+
+ if (unlikely(wc->status != IB_WC_SUCCESS))
+ iser_err_comp(wc, "dataout");
+
+ ib_dma_unmap_single(device->ib_device, desc->dma_addr,
+ ISER_HEADERS_LEN, DMA_TO_DEVICE);
+ kmem_cache_free(ig.desc_cache, desc);
+}
+
+void iser_last_comp(struct ib_cq *cq, struct ib_wc *wc)
+{
+ struct ib_conn *ib_conn = wc->qp->qp_context;
+
+ complete(&ib_conn->last_comp);
}
void iser_task_rdma_init(struct iscsi_iser_task *iser_task)
diff --git a/drivers/infiniband/ulp/iser/iser_memory.c b/drivers/infiniband/ulp/iser/iser_memory.c
index ea765fb96..9a391cc5b 100644
--- a/drivers/infiniband/ulp/iser/iser_memory.c
+++ b/drivers/infiniband/ulp/iser/iser_memory.c
@@ -49,7 +49,7 @@ int iser_fast_reg_mr(struct iscsi_iser_task *iser_task,
struct iser_reg_resources *rsc,
struct iser_mem_reg *mem_reg);
-static struct iser_reg_ops fastreg_ops = {
+static const struct iser_reg_ops fastreg_ops = {
.alloc_reg_res = iser_alloc_fastreg_pool,
.free_reg_res = iser_free_fastreg_pool,
.reg_mem = iser_fast_reg_mr,
@@ -58,7 +58,7 @@ static struct iser_reg_ops fastreg_ops = {
.reg_desc_put = iser_reg_desc_put_fr,
};
-static struct iser_reg_ops fmr_ops = {
+static const struct iser_reg_ops fmr_ops = {
.alloc_reg_res = iser_alloc_fmr_pool,
.free_reg_res = iser_free_fmr_pool,
.reg_mem = iser_fast_reg_fmr,
@@ -67,19 +67,24 @@ static struct iser_reg_ops fmr_ops = {
.reg_desc_put = iser_reg_desc_put_fmr,
};
+void iser_reg_comp(struct ib_cq *cq, struct ib_wc *wc)
+{
+ iser_err_comp(wc, "memreg");
+}
+
int iser_assign_reg_ops(struct iser_device *device)
{
- struct ib_device_attr *dev_attr = &device->dev_attr;
+ struct ib_device *ib_dev = device->ib_device;
/* Assign function handles - based on FMR support */
- if (device->ib_device->alloc_fmr && device->ib_device->dealloc_fmr &&
- device->ib_device->map_phys_fmr && device->ib_device->unmap_fmr) {
+ if (ib_dev->alloc_fmr && ib_dev->dealloc_fmr &&
+ ib_dev->map_phys_fmr && ib_dev->unmap_fmr) {
iser_info("FMR supported, using FMR for registration\n");
device->reg_ops = &fmr_ops;
- } else
- if (dev_attr->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) {
+ } else if (ib_dev->attrs.device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) {
iser_info("FastReg supported, using FastReg for registration\n");
device->reg_ops = &fastreg_ops;
+ device->remote_inv_sup = iser_always_reg;
} else {
iser_err("IB device does not support FMRs nor FastRegs, can't register memory\n");
return -1;
@@ -131,67 +136,6 @@ iser_reg_desc_put_fmr(struct ib_conn *ib_conn,
{
}
-#define IS_4K_ALIGNED(addr) ((((unsigned long)addr) & ~MASK_4K) == 0)
-
-/**
- * iser_sg_to_page_vec - Translates scatterlist entries to physical addresses
- * and returns the length of resulting physical address array (may be less than
- * the original due to possible compaction).
- *
- * we build a "page vec" under the assumption that the SG meets the RDMA
- * alignment requirements. Other then the first and last SG elements, all
- * the "internal" elements can be compacted into a list whose elements are
- * dma addresses of physical pages. The code supports also the weird case
- * where --few fragments of the same page-- are present in the SG as
- * consecutive elements. Also, it handles one entry SG.
- */
-
-static int iser_sg_to_page_vec(struct iser_data_buf *data,
- struct ib_device *ibdev, u64 *pages,
- int *offset, int *data_size)
-{
- struct scatterlist *sg, *sgl = data->sg;
- u64 start_addr, end_addr, page, chunk_start = 0;
- unsigned long total_sz = 0;
- unsigned int dma_len;
- int i, new_chunk, cur_page, last_ent = data->dma_nents - 1;
-
- /* compute the offset of first element */
- *offset = (u64) sgl[0].offset & ~MASK_4K;
-
- new_chunk = 1;
- cur_page = 0;
- for_each_sg(sgl, sg, data->dma_nents, i) {
- start_addr = ib_sg_dma_address(ibdev, sg);
- if (new_chunk)
- chunk_start = start_addr;
- dma_len = ib_sg_dma_len(ibdev, sg);
- end_addr = start_addr + dma_len;
- total_sz += dma_len;
-
- /* collect page fragments until aligned or end of SG list */
- if (!IS_4K_ALIGNED(end_addr) && i < last_ent) {
- new_chunk = 0;
- continue;
- }
- new_chunk = 1;
-
- /* address of the first page in the contiguous chunk;
- masking relevant for the very first SG entry,
- which might be unaligned */
- page = chunk_start & MASK_4K;
- do {
- pages[cur_page++] = page;
- page += SIZE_4K;
- } while (page < end_addr);
- }
-
- *data_size = total_sz;
- iser_dbg("page_vec->data_size:%d cur_page %d\n",
- *data_size, cur_page);
- return cur_page;
-}
-
static void iser_data_buf_dump(struct iser_data_buf *data,
struct ib_device *ibdev)
{
@@ -210,10 +154,10 @@ static void iser_dump_page_vec(struct iser_page_vec *page_vec)
{
int i;
- iser_err("page vec length %d data size %d\n",
- page_vec->length, page_vec->data_size);
- for (i = 0; i < page_vec->length; i++)
- iser_err("%d %lx\n",i,(unsigned long)page_vec->pages[i]);
+ iser_err("page vec npages %d data length %d\n",
+ page_vec->npages, page_vec->fake_mr.length);
+ for (i = 0; i < page_vec->npages; i++)
+ iser_err("vec[%d]: %llx\n", i, page_vec->pages[i]);
}
int iser_dma_map_task_data(struct iscsi_iser_task *iser_task,
@@ -251,7 +195,11 @@ iser_reg_dma(struct iser_device *device, struct iser_data_buf *mem,
struct scatterlist *sg = mem->sg;
reg->sge.lkey = device->pd->local_dma_lkey;
- reg->rkey = device->mr->rkey;
+ /*
+ * FIXME: rework the registration code path to differentiate
+ * rkey/lkey use cases
+ */
+ reg->rkey = device->mr ? device->mr->rkey : 0;
reg->sge.addr = ib_sg_dma_address(device->ib_device, &sg[0]);
reg->sge.length = ib_sg_dma_len(device->ib_device, &sg[0]);
@@ -262,11 +210,16 @@ iser_reg_dma(struct iser_device *device, struct iser_data_buf *mem,
return 0;
}
-/**
- * iser_reg_page_vec - Register physical memory
- *
- * returns: 0 on success, errno code on failure
- */
+static int iser_set_page(struct ib_mr *mr, u64 addr)
+{
+ struct iser_page_vec *page_vec =
+ container_of(mr, struct iser_page_vec, fake_mr);
+
+ page_vec->pages[page_vec->npages++] = addr;
+
+ return 0;
+}
+
static
int iser_fast_reg_fmr(struct iscsi_iser_task *iser_task,
struct iser_data_buf *mem,
@@ -280,22 +233,19 @@ int iser_fast_reg_fmr(struct iscsi_iser_task *iser_task,
struct ib_pool_fmr *fmr;
int ret, plen;
- plen = iser_sg_to_page_vec(mem, device->ib_device,
- page_vec->pages,
- &page_vec->offset,
- &page_vec->data_size);
- page_vec->length = plen;
- if (plen * SIZE_4K < page_vec->data_size) {
+ page_vec->npages = 0;
+ page_vec->fake_mr.page_size = SIZE_4K;
+ plen = ib_sg_to_pages(&page_vec->fake_mr, mem->sg,
+ mem->size, iser_set_page);
+ if (unlikely(plen < mem->size)) {
iser_err("page vec too short to hold this SG\n");
iser_data_buf_dump(mem, device->ib_device);
iser_dump_page_vec(page_vec);
return -EINVAL;
}
- fmr = ib_fmr_pool_map_phys(fmr_pool,
- page_vec->pages,
- page_vec->length,
- page_vec->pages[0]);
+ fmr = ib_fmr_pool_map_phys(fmr_pool, page_vec->pages,
+ page_vec->npages, page_vec->pages[0]);
if (IS_ERR(fmr)) {
ret = PTR_ERR(fmr);
iser_err("ib_fmr_pool_map_phys failed: %d\n", ret);
@@ -304,8 +254,8 @@ int iser_fast_reg_fmr(struct iscsi_iser_task *iser_task,
reg->sge.lkey = fmr->fmr->lkey;
reg->rkey = fmr->fmr->rkey;
- reg->sge.addr = page_vec->pages[0] + page_vec->offset;
- reg->sge.length = page_vec->data_size;
+ reg->sge.addr = page_vec->fake_mr.iova;
+ reg->sge.length = page_vec->fake_mr.length;
reg->mem_h = fmr;
iser_dbg("fmr reg: lkey=0x%x, rkey=0x%x, addr=0x%llx,"
@@ -413,19 +363,16 @@ iser_set_prot_checks(struct scsi_cmnd *sc, u8 *mask)
*mask |= ISER_CHECK_GUARD;
}
-static void
-iser_inv_rkey(struct ib_send_wr *inv_wr, struct ib_mr *mr)
+static inline void
+iser_inv_rkey(struct ib_send_wr *inv_wr,
+ struct ib_mr *mr,
+ struct ib_cqe *cqe)
{
- u32 rkey;
-
inv_wr->opcode = IB_WR_LOCAL_INV;
- inv_wr->wr_id = ISER_FASTREG_LI_WRID;
+ inv_wr->wr_cqe = cqe;
inv_wr->ex.invalidate_rkey = mr->rkey;
inv_wr->send_flags = 0;
inv_wr->num_sge = 0;
-
- rkey = ib_inc_rkey(mr->rkey);
- ib_update_fast_reg_key(mr, rkey);
}
static int
@@ -437,7 +384,9 @@ iser_reg_sig_mr(struct iscsi_iser_task *iser_task,
{
struct iser_tx_desc *tx_desc = &iser_task->desc;
struct ib_sig_attrs *sig_attrs = &tx_desc->sig_attrs;
+ struct ib_cqe *cqe = &iser_task->iser_conn->ib_conn.reg_cqe;
struct ib_sig_handover_wr *wr;
+ struct ib_mr *mr = pi_ctx->sig_mr;
int ret;
memset(sig_attrs, 0, sizeof(*sig_attrs));
@@ -447,17 +396,19 @@ iser_reg_sig_mr(struct iscsi_iser_task *iser_task,
iser_set_prot_checks(iser_task->sc, &sig_attrs->check_mask);
- if (!pi_ctx->sig_mr_valid)
- iser_inv_rkey(iser_tx_next_wr(tx_desc), pi_ctx->sig_mr);
+ if (pi_ctx->sig_mr_valid)
+ iser_inv_rkey(iser_tx_next_wr(tx_desc), mr, cqe);
+
+ ib_update_fast_reg_key(mr, ib_inc_rkey(mr->rkey));
wr = sig_handover_wr(iser_tx_next_wr(tx_desc));
wr->wr.opcode = IB_WR_REG_SIG_MR;
- wr->wr.wr_id = ISER_FASTREG_LI_WRID;
+ wr->wr.wr_cqe = cqe;
wr->wr.sg_list = &data_reg->sge;
wr->wr.num_sge = 1;
wr->wr.send_flags = 0;
wr->sig_attrs = sig_attrs;
- wr->sig_mr = pi_ctx->sig_mr;
+ wr->sig_mr = mr;
if (scsi_prot_sg_count(iser_task->sc))
wr->prot = &prot_reg->sge;
else
@@ -465,10 +416,10 @@ iser_reg_sig_mr(struct iscsi_iser_task *iser_task,
wr->access_flags = IB_ACCESS_LOCAL_WRITE |
IB_ACCESS_REMOTE_READ |
IB_ACCESS_REMOTE_WRITE;
- pi_ctx->sig_mr_valid = 0;
+ pi_ctx->sig_mr_valid = 1;
- sig_reg->sge.lkey = pi_ctx->sig_mr->lkey;
- sig_reg->rkey = pi_ctx->sig_mr->rkey;
+ sig_reg->sge.lkey = mr->lkey;
+ sig_reg->rkey = mr->rkey;
sig_reg->sge.addr = 0;
sig_reg->sge.length = scsi_transfer_length(iser_task->sc);
@@ -485,12 +436,15 @@ static int iser_fast_reg_mr(struct iscsi_iser_task *iser_task,
struct iser_mem_reg *reg)
{
struct iser_tx_desc *tx_desc = &iser_task->desc;
+ struct ib_cqe *cqe = &iser_task->iser_conn->ib_conn.reg_cqe;
struct ib_mr *mr = rsc->mr;
struct ib_reg_wr *wr;
int n;
- if (!rsc->mr_valid)
- iser_inv_rkey(iser_tx_next_wr(tx_desc), mr);
+ if (rsc->mr_valid)
+ iser_inv_rkey(iser_tx_next_wr(tx_desc), mr, cqe);
+
+ ib_update_fast_reg_key(mr, ib_inc_rkey(mr->rkey));
n = ib_map_mr_sg(mr, mem->sg, mem->size, SIZE_4K);
if (unlikely(n != mem->size)) {
@@ -501,7 +455,7 @@ static int iser_fast_reg_mr(struct iscsi_iser_task *iser_task,
wr = reg_wr(iser_tx_next_wr(tx_desc));
wr->wr.opcode = IB_WR_REG_MR;
- wr->wr.wr_id = ISER_FASTREG_LI_WRID;
+ wr->wr.wr_cqe = cqe;
wr->wr.send_flags = 0;
wr->wr.num_sge = 0;
wr->mr = mr;
@@ -510,7 +464,7 @@ static int iser_fast_reg_mr(struct iscsi_iser_task *iser_task,
IB_ACCESS_REMOTE_WRITE |
IB_ACCESS_REMOTE_READ;
- rsc->mr_valid = 0;
+ rsc->mr_valid = 1;
reg->sge.lkey = mr->lkey;
reg->rkey = mr->rkey;
@@ -554,7 +508,8 @@ iser_reg_data_sg(struct iscsi_iser_task *task,
}
int iser_reg_rdma_mem(struct iscsi_iser_task *task,
- enum iser_data_dir dir)
+ enum iser_data_dir dir,
+ bool all_imm)
{
struct ib_conn *ib_conn = &task->iser_conn->ib_conn;
struct iser_device *device = ib_conn->device;
@@ -565,8 +520,8 @@ int iser_reg_rdma_mem(struct iscsi_iser_task *task,
bool use_dma_key;
int err;
- use_dma_key = (mem->dma_nents == 1 && !iser_always_reg &&
- scsi_get_prot_op(task->sc) == SCSI_PROT_NORMAL);
+ use_dma_key = mem->dma_nents == 1 && (all_imm || !iser_always_reg) &&
+ scsi_get_prot_op(task->sc) == SCSI_PROT_NORMAL;
if (!use_dma_key) {
desc = device->reg_ops->reg_desc_get(ib_conn);
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c
index 42f4da620..40c0f4978 100644
--- a/drivers/infiniband/ulp/iser/iser_verbs.c
+++ b/drivers/infiniband/ulp/iser/iser_verbs.c
@@ -44,17 +44,6 @@
#define ISER_MAX_CQ_LEN (ISER_MAX_RX_LEN + ISER_MAX_TX_LEN + \
ISCSI_ISER_MAX_CONN)
-static int iser_cq_poll_limit = 512;
-
-static void iser_cq_tasklet_fn(unsigned long data);
-static void iser_cq_callback(struct ib_cq *cq, void *cq_context);
-
-static void iser_cq_event_callback(struct ib_event *cause, void *context)
-{
- iser_err("cq event %s (%d)\n",
- ib_event_msg(cause->event), cause->event);
-}
-
static void iser_qp_event_callback(struct ib_event *cause, void *context)
{
iser_err("qp event %s (%d)\n",
@@ -78,59 +67,40 @@ static void iser_event_handler(struct ib_event_handler *handler,
*/
static int iser_create_device_ib_res(struct iser_device *device)
{
- struct ib_device_attr *dev_attr = &device->dev_attr;
+ struct ib_device *ib_dev = device->ib_device;
int ret, i, max_cqe;
- ret = ib_query_device(device->ib_device, dev_attr);
- if (ret) {
- pr_warn("Query device failed for %s\n", device->ib_device->name);
- return ret;
- }
-
ret = iser_assign_reg_ops(device);
if (ret)
return ret;
device->comps_used = min_t(int, num_online_cpus(),
- device->ib_device->num_comp_vectors);
+ ib_dev->num_comp_vectors);
device->comps = kcalloc(device->comps_used, sizeof(*device->comps),
GFP_KERNEL);
if (!device->comps)
goto comps_err;
- max_cqe = min(ISER_MAX_CQ_LEN, dev_attr->max_cqe);
+ max_cqe = min(ISER_MAX_CQ_LEN, ib_dev->attrs.max_cqe);
iser_info("using %d CQs, device %s supports %d vectors max_cqe %d\n",
- device->comps_used, device->ib_device->name,
- device->ib_device->num_comp_vectors, max_cqe);
+ device->comps_used, ib_dev->name,
+ ib_dev->num_comp_vectors, max_cqe);
- device->pd = ib_alloc_pd(device->ib_device);
+ device->pd = ib_alloc_pd(ib_dev);
if (IS_ERR(device->pd))
goto pd_err;
for (i = 0; i < device->comps_used; i++) {
- struct ib_cq_init_attr cq_attr = {};
struct iser_comp *comp = &device->comps[i];
- comp->device = device;
- cq_attr.cqe = max_cqe;
- cq_attr.comp_vector = i;
- comp->cq = ib_create_cq(device->ib_device,
- iser_cq_callback,
- iser_cq_event_callback,
- (void *)comp,
- &cq_attr);
+ comp->cq = ib_alloc_cq(ib_dev, comp, max_cqe, i,
+ IB_POLL_SOFTIRQ);
if (IS_ERR(comp->cq)) {
comp->cq = NULL;
goto cq_err;
}
-
- if (ib_req_notify_cq(comp->cq, IB_CQ_NEXT_COMP))
- goto cq_err;
-
- tasklet_init(&comp->tasklet, iser_cq_tasklet_fn,
- (unsigned long)comp);
}
if (!iser_always_reg) {
@@ -140,11 +110,11 @@ static int iser_create_device_ib_res(struct iser_device *device)
device->mr = ib_get_dma_mr(device->pd, access);
if (IS_ERR(device->mr))
- goto dma_mr_err;
+ goto cq_err;
}
- INIT_IB_EVENT_HANDLER(&device->event_handler, device->ib_device,
- iser_event_handler);
+ INIT_IB_EVENT_HANDLER(&device->event_handler, ib_dev,
+ iser_event_handler);
if (ib_register_event_handler(&device->event_handler))
goto handler_err;
@@ -153,15 +123,12 @@ static int iser_create_device_ib_res(struct iser_device *device)
handler_err:
if (device->mr)
ib_dereg_mr(device->mr);
-dma_mr_err:
- for (i = 0; i < device->comps_used; i++)
- tasklet_kill(&device->comps[i].tasklet);
cq_err:
for (i = 0; i < device->comps_used; i++) {
struct iser_comp *comp = &device->comps[i];
if (comp->cq)
- ib_destroy_cq(comp->cq);
+ ib_free_cq(comp->cq);
}
ib_dealloc_pd(device->pd);
pd_err:
@@ -182,8 +149,7 @@ static void iser_free_device_ib_res(struct iser_device *device)
for (i = 0; i < device->comps_used; i++) {
struct iser_comp *comp = &device->comps[i];
- tasklet_kill(&comp->tasklet);
- ib_destroy_cq(comp->cq);
+ ib_free_cq(comp->cq);
comp->cq = NULL;
}
@@ -299,7 +265,7 @@ iser_alloc_reg_res(struct ib_device *ib_device,
iser_err("Failed to allocate ib_fast_reg_mr err=%d\n", ret);
return ret;
}
- res->mr_valid = 1;
+ res->mr_valid = 0;
return 0;
}
@@ -336,7 +302,7 @@ iser_alloc_pi_ctx(struct ib_device *ib_device,
ret = PTR_ERR(pi_ctx->sig_mr);
goto sig_mr_failure;
}
- pi_ctx->sig_mr_valid = 1;
+ pi_ctx->sig_mr_valid = 0;
desc->pi_ctx->sig_protected = 0;
return 0;
@@ -461,10 +427,9 @@ void iser_free_fastreg_pool(struct ib_conn *ib_conn)
*/
static int iser_create_ib_conn_res(struct ib_conn *ib_conn)
{
- struct iser_conn *iser_conn = container_of(ib_conn, struct iser_conn,
- ib_conn);
+ struct iser_conn *iser_conn = to_iser_conn(ib_conn);
struct iser_device *device;
- struct ib_device_attr *dev_attr;
+ struct ib_device *ib_dev;
struct ib_qp_init_attr init_attr;
int ret = -ENOMEM;
int index, min_index = 0;
@@ -472,7 +437,7 @@ static int iser_create_ib_conn_res(struct ib_conn *ib_conn)
BUG_ON(ib_conn->device == NULL);
device = ib_conn->device;
- dev_attr = &device->dev_attr;
+ ib_dev = device->ib_device;
memset(&init_attr, 0, sizeof init_attr);
@@ -503,16 +468,16 @@ static int iser_create_ib_conn_res(struct ib_conn *ib_conn)
iser_conn->max_cmds =
ISER_GET_MAX_XMIT_CMDS(ISER_QP_SIG_MAX_REQ_DTOS);
} else {
- if (dev_attr->max_qp_wr > ISER_QP_MAX_REQ_DTOS) {
+ if (ib_dev->attrs.max_qp_wr > ISER_QP_MAX_REQ_DTOS) {
init_attr.cap.max_send_wr = ISER_QP_MAX_REQ_DTOS + 1;
iser_conn->max_cmds =
ISER_GET_MAX_XMIT_CMDS(ISER_QP_MAX_REQ_DTOS);
} else {
- init_attr.cap.max_send_wr = dev_attr->max_qp_wr;
+ init_attr.cap.max_send_wr = ib_dev->attrs.max_qp_wr;
iser_conn->max_cmds =
- ISER_GET_MAX_XMIT_CMDS(dev_attr->max_qp_wr);
+ ISER_GET_MAX_XMIT_CMDS(ib_dev->attrs.max_qp_wr);
iser_dbg("device %s supports max_send_wr %d\n",
- device->ib_device->name, dev_attr->max_qp_wr);
+ device->ib_device->name, ib_dev->attrs.max_qp_wr);
}
}
@@ -724,13 +689,13 @@ int iser_conn_terminate(struct iser_conn *iser_conn)
iser_conn, err);
/* post an indication that all flush errors were consumed */
- err = ib_post_send(ib_conn->qp, &ib_conn->beacon, &bad_wr);
+ err = ib_post_send(ib_conn->qp, &ib_conn->last, &bad_wr);
if (err) {
- iser_err("conn %p failed to post beacon", ib_conn);
+ iser_err("conn %p failed to post last wr", ib_conn);
return 1;
}
- wait_for_completion(&ib_conn->flush_comp);
+ wait_for_completion(&ib_conn->last_comp);
}
return 1;
@@ -756,7 +721,7 @@ iser_calc_scsi_params(struct iser_conn *iser_conn,
sg_tablesize = DIV_ROUND_UP(max_sectors * 512, SIZE_4K);
sup_sg_tablesize = min_t(unsigned, ISCSI_ISER_MAX_SG_TABLESIZE,
- device->dev_attr.max_fast_reg_page_list_len);
+ device->ib_device->attrs.max_fast_reg_page_list_len);
if (sg_tablesize > sup_sg_tablesize) {
sg_tablesize = sup_sg_tablesize;
@@ -799,7 +764,7 @@ static void iser_addr_handler(struct rdma_cm_id *cma_id)
/* connection T10-PI support */
if (iser_pi_enable) {
- if (!(device->dev_attr.device_cap_flags &
+ if (!(device->ib_device->attrs.device_cap_flags &
IB_DEVICE_SIGNATURE_HANDOVER)) {
iser_warn("T10-PI requested but not supported on %s, "
"continue without T10-PI\n",
@@ -841,16 +806,17 @@ static void iser_route_handler(struct rdma_cm_id *cma_id)
goto failure;
memset(&conn_param, 0, sizeof conn_param);
- conn_param.responder_resources = device->dev_attr.max_qp_rd_atom;
+ conn_param.responder_resources = device->ib_device->attrs.max_qp_rd_atom;
conn_param.initiator_depth = 1;
conn_param.retry_count = 7;
conn_param.rnr_retry_count = 6;
memset(&req_hdr, 0, sizeof(req_hdr));
- req_hdr.flags = (ISER_ZBVA_NOT_SUPPORTED |
- ISER_SEND_W_INV_NOT_SUPPORTED);
- conn_param.private_data = (void *)&req_hdr;
- conn_param.private_data_len = sizeof(struct iser_cm_hdr);
+ req_hdr.flags = ISER_ZBVA_NOT_SUP;
+ if (!device->remote_inv_sup)
+ req_hdr.flags |= ISER_SEND_W_INV_NOT_SUP;
+ conn_param.private_data = (void *)&req_hdr;
+ conn_param.private_data_len = sizeof(struct iser_cm_hdr);
ret = rdma_connect(cma_id, &conn_param);
if (ret) {
@@ -863,7 +829,8 @@ failure:
iser_connect_error(cma_id);
}
-static void iser_connected_handler(struct rdma_cm_id *cma_id)
+static void iser_connected_handler(struct rdma_cm_id *cma_id,
+ const void *private_data)
{
struct iser_conn *iser_conn;
struct ib_qp_attr attr;
@@ -877,6 +844,15 @@ static void iser_connected_handler(struct rdma_cm_id *cma_id)
(void)ib_query_qp(cma_id->qp, &attr, ~0, &init_attr);
iser_info("remote qpn:%x my qpn:%x\n", attr.dest_qp_num, cma_id->qp->qp_num);
+ if (private_data) {
+ u8 flags = *(u8 *)private_data;
+
+ iser_conn->snd_w_inv = !(flags & ISER_SEND_W_INV_NOT_SUP);
+ }
+
+ iser_info("conn %p: negotiated %s invalidation\n",
+ iser_conn, iser_conn->snd_w_inv ? "remote" : "local");
+
iser_conn->state = ISER_CONN_UP;
complete(&iser_conn->up_completion);
}
@@ -928,7 +904,7 @@ static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *eve
iser_route_handler(cma_id);
break;
case RDMA_CM_EVENT_ESTABLISHED:
- iser_connected_handler(cma_id);
+ iser_connected_handler(cma_id, event->param.conn.private_data);
break;
case RDMA_CM_EVENT_ADDR_ERROR:
case RDMA_CM_EVENT_ROUTE_ERROR:
@@ -967,14 +943,21 @@ static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *eve
void iser_conn_init(struct iser_conn *iser_conn)
{
+ struct ib_conn *ib_conn = &iser_conn->ib_conn;
+
iser_conn->state = ISER_CONN_INIT;
- iser_conn->ib_conn.post_recv_buf_count = 0;
- init_completion(&iser_conn->ib_conn.flush_comp);
init_completion(&iser_conn->stop_completion);
init_completion(&iser_conn->ib_completion);
init_completion(&iser_conn->up_completion);
INIT_LIST_HEAD(&iser_conn->conn_list);
mutex_init(&iser_conn->state_mutex);
+
+ ib_conn->post_recv_buf_count = 0;
+ ib_conn->reg_cqe.done = iser_reg_comp;
+ ib_conn->last_cqe.done = iser_last_comp;
+ ib_conn->last.wr_cqe = &ib_conn->last_cqe;
+ ib_conn->last.opcode = IB_WR_SEND;
+ init_completion(&ib_conn->last_comp);
}
/**
@@ -1000,9 +983,6 @@ int iser_connect(struct iser_conn *iser_conn,
iser_conn->state = ISER_CONN_PENDING;
- ib_conn->beacon.wr_id = ISER_BEACON_WRID;
- ib_conn->beacon.opcode = IB_WR_SEND;
-
ib_conn->cma_id = rdma_create_id(&init_net, iser_cma_handler,
(void *)iser_conn,
RDMA_PS_TCP, IB_QPT_RC);
@@ -1045,56 +1025,60 @@ connect_failure:
int iser_post_recvl(struct iser_conn *iser_conn)
{
- struct ib_recv_wr rx_wr, *rx_wr_failed;
struct ib_conn *ib_conn = &iser_conn->ib_conn;
- struct ib_sge sge;
+ struct iser_login_desc *desc = &iser_conn->login_desc;
+ struct ib_recv_wr wr, *wr_failed;
int ib_ret;
- sge.addr = iser_conn->login_resp_dma;
- sge.length = ISER_RX_LOGIN_SIZE;
- sge.lkey = ib_conn->device->pd->local_dma_lkey;
+ desc->sge.addr = desc->rsp_dma;
+ desc->sge.length = ISER_RX_LOGIN_SIZE;
+ desc->sge.lkey = ib_conn->device->pd->local_dma_lkey;
- rx_wr.wr_id = (uintptr_t)iser_conn->login_resp_buf;
- rx_wr.sg_list = &sge;
- rx_wr.num_sge = 1;
- rx_wr.next = NULL;
+ desc->cqe.done = iser_login_rsp;
+ wr.wr_cqe = &desc->cqe;
+ wr.sg_list = &desc->sge;
+ wr.num_sge = 1;
+ wr.next = NULL;
ib_conn->post_recv_buf_count++;
- ib_ret = ib_post_recv(ib_conn->qp, &rx_wr, &rx_wr_failed);
+ ib_ret = ib_post_recv(ib_conn->qp, &wr, &wr_failed);
if (ib_ret) {
iser_err("ib_post_recv failed ret=%d\n", ib_ret);
ib_conn->post_recv_buf_count--;
}
+
return ib_ret;
}
int iser_post_recvm(struct iser_conn *iser_conn, int count)
{
- struct ib_recv_wr *rx_wr, *rx_wr_failed;
- int i, ib_ret;
struct ib_conn *ib_conn = &iser_conn->ib_conn;
unsigned int my_rx_head = iser_conn->rx_desc_head;
struct iser_rx_desc *rx_desc;
+ struct ib_recv_wr *wr, *wr_failed;
+ int i, ib_ret;
- for (rx_wr = ib_conn->rx_wr, i = 0; i < count; i++, rx_wr++) {
- rx_desc = &iser_conn->rx_descs[my_rx_head];
- rx_wr->wr_id = (uintptr_t)rx_desc;
- rx_wr->sg_list = &rx_desc->rx_sg;
- rx_wr->num_sge = 1;
- rx_wr->next = rx_wr + 1;
+ for (wr = ib_conn->rx_wr, i = 0; i < count; i++, wr++) {
+ rx_desc = &iser_conn->rx_descs[my_rx_head];
+ rx_desc->cqe.done = iser_task_rsp;
+ wr->wr_cqe = &rx_desc->cqe;
+ wr->sg_list = &rx_desc->rx_sg;
+ wr->num_sge = 1;
+ wr->next = wr + 1;
my_rx_head = (my_rx_head + 1) & iser_conn->qp_max_recv_dtos_mask;
}
- rx_wr--;
- rx_wr->next = NULL; /* mark end of work requests list */
+ wr--;
+ wr->next = NULL; /* mark end of work requests list */
ib_conn->post_recv_buf_count += count;
- ib_ret = ib_post_recv(ib_conn->qp, ib_conn->rx_wr, &rx_wr_failed);
+ ib_ret = ib_post_recv(ib_conn->qp, ib_conn->rx_wr, &wr_failed);
if (ib_ret) {
iser_err("ib_post_recv failed ret=%d\n", ib_ret);
ib_conn->post_recv_buf_count -= count;
} else
iser_conn->rx_desc_head = my_rx_head;
+
return ib_ret;
}
@@ -1115,7 +1099,7 @@ int iser_post_send(struct ib_conn *ib_conn, struct iser_tx_desc *tx_desc,
DMA_TO_DEVICE);
wr->next = NULL;
- wr->wr_id = (uintptr_t)tx_desc;
+ wr->wr_cqe = &tx_desc->cqe;
wr->sg_list = tx_desc->tx_sg;
wr->num_sge = tx_desc->num_sge;
wr->opcode = IB_WR_SEND;
@@ -1129,149 +1113,6 @@ int iser_post_send(struct ib_conn *ib_conn, struct iser_tx_desc *tx_desc,
return ib_ret;
}
-/**
- * is_iser_tx_desc - Indicate if the completion wr_id
- * is a TX descriptor or not.
- * @iser_conn: iser connection
- * @wr_id: completion WR identifier
- *
- * Since we cannot rely on wc opcode in FLUSH errors
- * we must work around it by checking if the wr_id address
- * falls in the iser connection rx_descs buffer. If so
- * it is an RX descriptor, otherwize it is a TX.
- */
-static inline bool
-is_iser_tx_desc(struct iser_conn *iser_conn, void *wr_id)
-{
- void *start = iser_conn->rx_descs;
- int len = iser_conn->num_rx_descs * sizeof(*iser_conn->rx_descs);
-
- if (wr_id >= start && wr_id < start + len)
- return false;
-
- return true;
-}
-
-/**
- * iser_handle_comp_error() - Handle error completion
- * @ib_conn: connection RDMA resources
- * @wc: work completion
- *
- * Notes: We may handle a FLUSH error completion and in this case
- * we only cleanup in case TX type was DATAOUT. For non-FLUSH
- * error completion we should also notify iscsi layer that
- * connection is failed (in case we passed bind stage).
- */
-static void
-iser_handle_comp_error(struct ib_conn *ib_conn,
- struct ib_wc *wc)
-{
- void *wr_id = (void *)(uintptr_t)wc->wr_id;
- struct iser_conn *iser_conn = container_of(ib_conn, struct iser_conn,
- ib_conn);
-
- if (wc->status != IB_WC_WR_FLUSH_ERR)
- if (iser_conn->iscsi_conn)
- iscsi_conn_failure(iser_conn->iscsi_conn,
- ISCSI_ERR_CONN_FAILED);
-
- if (wc->wr_id == ISER_FASTREG_LI_WRID)
- return;
-
- if (is_iser_tx_desc(iser_conn, wr_id)) {
- struct iser_tx_desc *desc = wr_id;
-
- if (desc->type == ISCSI_TX_DATAOUT)
- kmem_cache_free(ig.desc_cache, desc);
- } else {
- ib_conn->post_recv_buf_count--;
- }
-}
-
-/**
- * iser_handle_wc - handle a single work completion
- * @wc: work completion
- *
- * Soft-IRQ context, work completion can be either
- * SEND or RECV, and can turn out successful or
- * with error (or flush error).
- */
-static void iser_handle_wc(struct ib_wc *wc)
-{
- struct ib_conn *ib_conn;
- struct iser_tx_desc *tx_desc;
- struct iser_rx_desc *rx_desc;
-
- ib_conn = wc->qp->qp_context;
- if (likely(wc->status == IB_WC_SUCCESS)) {
- if (wc->opcode == IB_WC_RECV) {
- rx_desc = (struct iser_rx_desc *)(uintptr_t)wc->wr_id;
- iser_rcv_completion(rx_desc, wc->byte_len,
- ib_conn);
- } else
- if (wc->opcode == IB_WC_SEND) {
- tx_desc = (struct iser_tx_desc *)(uintptr_t)wc->wr_id;
- iser_snd_completion(tx_desc, ib_conn);
- } else {
- iser_err("Unknown wc opcode %d\n", wc->opcode);
- }
- } else {
- if (wc->status != IB_WC_WR_FLUSH_ERR)
- iser_err("%s (%d): wr id %llx vend_err %x\n",
- ib_wc_status_msg(wc->status), wc->status,
- wc->wr_id, wc->vendor_err);
- else
- iser_dbg("%s (%d): wr id %llx\n",
- ib_wc_status_msg(wc->status), wc->status,
- wc->wr_id);
-
- if (wc->wr_id == ISER_BEACON_WRID)
- /* all flush errors were consumed */
- complete(&ib_conn->flush_comp);
- else
- iser_handle_comp_error(ib_conn, wc);
- }
-}
-
-/**
- * iser_cq_tasklet_fn - iSER completion polling loop
- * @data: iSER completion context
- *
- * Soft-IRQ context, polling connection CQ until
- * either CQ was empty or we exausted polling budget
- */
-static void iser_cq_tasklet_fn(unsigned long data)
-{
- struct iser_comp *comp = (struct iser_comp *)data;
- struct ib_cq *cq = comp->cq;
- struct ib_wc *const wcs = comp->wcs;
- int i, n, completed = 0;
-
- while ((n = ib_poll_cq(cq, ARRAY_SIZE(comp->wcs), wcs)) > 0) {
- for (i = 0; i < n; i++)
- iser_handle_wc(&wcs[i]);
-
- completed += n;
- if (completed >= iser_cq_poll_limit)
- break;
- }
-
- /*
- * It is assumed here that arming CQ only once its empty
- * would not cause interrupts to be missed.
- */
- ib_req_notify_cq(cq, IB_CQ_NEXT_COMP);
-
- iser_dbg("got %d completions\n", completed);
-}
-
-static void iser_cq_callback(struct ib_cq *cq, void *cq_context)
-{
- struct iser_comp *comp = cq_context;
-
- tasklet_schedule(&comp->tasklet);
-}
-
u8 iser_check_task_pi_status(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir, sector_t *sector)
{
@@ -1319,3 +1160,21 @@ err:
/* Not alot we can do here, return ambiguous guard error */
return 0x1;
}
+
+void iser_err_comp(struct ib_wc *wc, const char *type)
+{
+ if (wc->status != IB_WC_WR_FLUSH_ERR) {
+ struct iser_conn *iser_conn = to_iser_conn(wc->qp->qp_context);
+
+ iser_err("%s failure: %s (%d) vend_err %x\n", type,
+ ib_wc_status_msg(wc->status), wc->status,
+ wc->vendor_err);
+
+ if (iser_conn->iscsi_conn)
+ iscsi_conn_failure(iser_conn->iscsi_conn,
+ ISCSI_ERR_CONN_FAILED);
+ } else {
+ iser_dbg("%s failure: %s (%d)\n", type,
+ ib_wc_status_msg(wc->status), wc->status);
+ }
+}