xiaomi-sm8450/android_kernel_xiaomi_sm8450
1
0
Fork 0
Koodi Ongelmat Pull-pyynnöt Actions Paketit Projektit Julkaisut Wiki Toiminta

Merge tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband

Selaa lähdekoodia 
Pull infiniband updates from Roland Dreier:
 "Main batch of InfiniBand/RDMA changes for 3.15:

   - The biggest change is core API extensions and mlx5 low-level driver
     support for handling DIF/DIX-style protection information, and the
     addition of PI support to the iSER initiator.  Target support will
     be arriving shortly through the SCSI target tree.

   - A nice simplification to the "umem" memory pinning library now that
     we have chained sg lists.  Kudos to Yishai Hadas for realizing our
     code didn't have to be so crazy.

   - Another nice simplification to the sg wrappers used by qib, ipath
     and ehca to handle their mapping of memory to adapter.

   - The usual batch of fixes to bugs found by static checkers etc.
     from intrepid people like Dan Carpenter and Yann Droneaud.

   - A large batch of cxgb4, ocrdma, qib driver updates"

* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (102 commits)
  RDMA/ocrdma: Unregister inet notifier when unloading ocrdma
  RDMA/ocrdma: Fix warnings about pointer <-> integer casts
  RDMA/ocrdma: Code clean-up
  RDMA/ocrdma: Display FW version
  RDMA/ocrdma: Query controller information
  RDMA/ocrdma: Support non-embedded mailbox commands
  RDMA/ocrdma: Handle CQ overrun error
  RDMA/ocrdma: Display proper value for max_mw
  RDMA/ocrdma: Use non-zero tag in SRQ posting
  RDMA/ocrdma: Memory leak fix in ocrdma_dereg_mr()
  RDMA/ocrdma: Increment abi version count
  RDMA/ocrdma: Update version string
  be2net: Add abi version between be2net and ocrdma
  RDMA/ocrdma: ABI versioning between ocrdma and be2net
  RDMA/ocrdma: Allow DPP QP creation
  RDMA/ocrdma: Read ASIC_ID register to select asic_gen
  RDMA/ocrdma: SQ and RQ doorbell offset clean up
  RDMA/ocrdma: EQ full catastrophe avoidance
  RDMA/cxgb4: Disable DSGL use by default
  RDMA/cxgb4: rx_data() needs to hold the ep mutex
  ...
This commit is contained in:
Linus Torvalds
2014-04-03 16:57:19 -07:00
vanhempi 3cf59142bc f7eaa7ed8f
commit 877f075aac
88 muutettua tiedostoa jossa 4635 lisäystä ja 1476 poistoa
Download Patch File Download Diff File Expand all files Collapse all files

93
drivers/infiniband/ulp/iser/iscsi_iser.c
Näytä tiedosto

@@ -5,7 +5,7 @@
* Copyright (C) 2004 Alex Aizman
* Copyright (C) 2005 Mike Christie
* Copyright (c) 2005, 2006 Voltaire, Inc. All rights reserved.
* Copyright (c) 2013 Mellanox Technologies. All rights reserved.
* Copyright (c) 2013-2014 Mellanox Technologies. All rights reserved.
* maintained by openib-general@openib.org
*
* This software is available to you under a choice of one of two
@@ -82,6 +82,8 @@ static unsigned int iscsi_max_lun = 512;
module_param_named(max_lun, iscsi_max_lun, uint, S_IRUGO);
int iser_debug_level = 0;
bool iser_pi_enable = false;
int iser_pi_guard = 0;
MODULE_DESCRIPTION("iSER (iSCSI Extensions for RDMA) Datamover");
MODULE_LICENSE("Dual BSD/GPL");
@@ -91,6 +93,12 @@ MODULE_VERSION(DRV_VER);
module_param_named(debug_level, iser_debug_level, int, 0644);
MODULE_PARM_DESC(debug_level, "Enable debug tracing if > 0 (default:disabled)");
module_param_named(pi_enable, iser_pi_enable, bool, 0644);
MODULE_PARM_DESC(pi_enable, "Enable T10-PI offload support (default:disabled)");
module_param_named(pi_guard, iser_pi_guard, int, 0644);
MODULE_PARM_DESC(pi_guard, "T10-PI guard_type, 0:CRC|1:IP_CSUM (default:CRC)");
struct iser_global ig;
void
@@ -138,8 +146,8 @@ static int iscsi_iser_pdu_alloc(struct iscsi_task *task, uint8_t opcode)
int iser_initialize_task_headers(struct iscsi_task *task,
struct iser_tx_desc *tx_desc)
{
struct iscsi_iser_conn *iser_conn = task->conn->dd_data;
struct iser_device *device = iser_conn->ib_conn->device;
struct iser_conn *ib_conn = task->conn->dd_data;
struct iser_device *device = ib_conn->device;
struct iscsi_iser_task *iser_task = task->dd_data;
u64 dma_addr;
@@ -153,7 +161,7 @@ int iser_initialize_task_headers(struct iscsi_task *task,
tx_desc->tx_sg[0].length = ISER_HEADERS_LEN;
tx_desc->tx_sg[0].lkey = device->mr->lkey;
iser_task->iser_conn = iser_conn;
iser_task->ib_conn = ib_conn;
return 0;
}
/**
@@ -176,6 +184,8 @@ iscsi_iser_task_init(struct iscsi_task *task)
iser_task->command_sent = 0;
iser_task_rdma_init(iser_task);
iser_task->sc = task->sc;
return 0;
}
@@ -278,10 +288,9 @@ iscsi_iser_task_xmit(struct iscsi_task *task)
static void iscsi_iser_cleanup_task(struct iscsi_task *task)
{
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_tx_desc *tx_desc = &iser_task->desc;
struct iscsi_iser_conn *iser_conn = task->conn->dd_data;
struct iser_device *device = iser_conn->ib_conn->device;
struct iser_tx_desc *tx_desc = &iser_task->desc;
struct iser_conn *ib_conn = task->conn->dd_data;
struct iser_device *device = ib_conn->device;
ib_dma_unmap_single(device->ib_device,
tx_desc->dma_addr, ISER_HEADERS_LEN, DMA_TO_DEVICE);
@@ -296,14 +305,25 @@ static void iscsi_iser_cleanup_task(struct iscsi_task *task)
}
}
static u8 iscsi_iser_check_protection(struct iscsi_task *task, sector_t *sector)
{
struct iscsi_iser_task *iser_task = task->dd_data;
if (iser_task->dir[ISER_DIR_IN])
return iser_check_task_pi_status(iser_task, ISER_DIR_IN,
sector);
else
return iser_check_task_pi_status(iser_task, ISER_DIR_OUT,
sector);
}
static struct iscsi_cls_conn *
iscsi_iser_conn_create(struct iscsi_cls_session *cls_session, uint32_t conn_idx)
{
struct iscsi_conn *conn;
struct iscsi_cls_conn *cls_conn;
struct iscsi_iser_conn *iser_conn;
cls_conn = iscsi_conn_setup(cls_session, sizeof(*iser_conn), conn_idx);
cls_conn = iscsi_conn_setup(cls_session, 0, conn_idx);
if (!cls_conn)
return NULL;
conn = cls_conn->dd_data;
@@ -314,10 +334,6 @@ iscsi_iser_conn_create(struct iscsi_cls_session *cls_session, uint32_t conn_idx)
*/
conn->max_recv_dlength = ISER_RECV_DATA_SEG_LEN;
iser_conn = conn->dd_data;
conn->dd_data = iser_conn;
iser_conn->iscsi_conn = conn;
return cls_conn;
}
@@ -325,8 +341,7 @@ static void
iscsi_iser_conn_destroy(struct iscsi_cls_conn *cls_conn)
{
struct iscsi_conn *conn = cls_conn->dd_data;
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iser_conn *ib_conn = iser_conn->ib_conn;
struct iser_conn *ib_conn = conn->dd_data;
iscsi_conn_teardown(cls_conn);
/*
@@ -335,7 +350,7 @@ iscsi_iser_conn_destroy(struct iscsi_cls_conn *cls_conn)
* we free it here.
*/
if (ib_conn) {
ib_conn->iser_conn = NULL;
ib_conn->iscsi_conn = NULL;
iser_conn_put(ib_conn, 1); /* deref iscsi/ib conn unbinding */
}
}
@@ -346,7 +361,6 @@ iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session,
int is_leading)
{
struct iscsi_conn *conn = cls_conn->dd_data;
struct iscsi_iser_conn *iser_conn;
struct iscsi_session *session;
struct iser_conn *ib_conn;
struct iscsi_endpoint *ep;
@@ -373,11 +387,11 @@ iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session,
/* binds the iSER connection retrieved from the previously
* connected ep_handle to the iSCSI layer connection. exchanges
* connection pointers */
iser_info("binding iscsi/iser conn %p %p to ib_conn %p\n",
conn, conn->dd_data, ib_conn);
iser_conn = conn->dd_data;
ib_conn->iser_conn = iser_conn;
iser_conn->ib_conn = ib_conn;
iser_info("binding iscsi conn %p to ib_conn %p\n", conn, ib_conn);
conn->dd_data = ib_conn;
ib_conn->iscsi_conn = conn;
iser_conn_get(ib_conn); /* ref iscsi/ib conn binding */
return 0;
}
@@ -386,8 +400,7 @@ static void
iscsi_iser_conn_stop(struct iscsi_cls_conn *cls_conn, int flag)
{
struct iscsi_conn *conn = cls_conn->dd_data;
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iser_conn *ib_conn = iser_conn->ib_conn;
struct iser_conn *ib_conn = conn->dd_data;
/*
* Userspace may have goofed up and not bound the connection or
@@ -401,7 +414,7 @@ iscsi_iser_conn_stop(struct iscsi_cls_conn *cls_conn, int flag)
*/
iser_conn_put(ib_conn, 1); /* deref iscsi/ib conn unbinding */
}
iser_conn->ib_conn = NULL;
conn->dd_data = NULL;
}
static void iscsi_iser_session_destroy(struct iscsi_cls_session *cls_session)
@@ -413,6 +426,17 @@ static void iscsi_iser_session_destroy(struct iscsi_cls_session *cls_session)
iscsi_host_free(shost);
}
static inline unsigned int
iser_dif_prot_caps(int prot_caps)
{
return ((prot_caps & IB_PROT_T10DIF_TYPE_1) ? SHOST_DIF_TYPE1_PROTECTION |
SHOST_DIX_TYPE1_PROTECTION : 0) |
((prot_caps & IB_PROT_T10DIF_TYPE_2) ? SHOST_DIF_TYPE2_PROTECTION |
SHOST_DIX_TYPE2_PROTECTION : 0) |
((prot_caps & IB_PROT_T10DIF_TYPE_3) ? SHOST_DIF_TYPE3_PROTECTION |
SHOST_DIX_TYPE3_PROTECTION : 0);
}
static struct iscsi_cls_session *
iscsi_iser_session_create(struct iscsi_endpoint *ep,
uint16_t cmds_max, uint16_t qdepth,
@@ -437,8 +461,18 @@ iscsi_iser_session_create(struct iscsi_endpoint *ep,
* older userspace tools (before 2.0-870) did not pass us
* the leading conn's ep so this will be NULL;
*/
if (ep)
if (ep) {
ib_conn = ep->dd_data;
if (ib_conn->pi_support) {
u32 sig_caps = ib_conn->device->dev_attr.sig_prot_cap;
scsi_host_set_prot(shost, iser_dif_prot_caps(sig_caps));
if (iser_pi_guard)
scsi_host_set_guard(shost, SHOST_DIX_GUARD_IP);
else
scsi_host_set_guard(shost, SHOST_DIX_GUARD_CRC);
}
}
if (iscsi_host_add(shost,
ep ? ib_conn->device->ib_device->dma_device : NULL))
@@ -618,7 +652,7 @@ iscsi_iser_ep_disconnect(struct iscsi_endpoint *ep)
struct iser_conn *ib_conn;
ib_conn = ep->dd_data;
if (ib_conn->iser_conn)
if (ib_conn->iscsi_conn)
/*
* Must suspend xmit path if the ep is bound to the
* iscsi_conn, so we know we are not accessing the ib_conn
@@ -626,7 +660,7 @@ iscsi_iser_ep_disconnect(struct iscsi_endpoint *ep)
*
* This may not be bound if the ep poll failed.
*/
iscsi_suspend_tx(ib_conn->iser_conn->iscsi_conn);
iscsi_suspend_tx(ib_conn->iscsi_conn);
iser_info("ib conn %p state %d\n", ib_conn, ib_conn->state);
@@ -732,6 +766,7 @@ static struct iscsi_transport iscsi_iser_transport = {
.xmit_task = iscsi_iser_task_xmit,
.cleanup_task = iscsi_iser_cleanup_task,
.alloc_pdu = iscsi_iser_pdu_alloc,
.check_protection = iscsi_iser_check_protection,
/* recovery */
.session_recovery_timedout = iscsi_session_recovery_timedout,

85
drivers/infiniband/ulp/iser/iscsi_iser.h
Näytä tiedosto

@@ -8,7 +8,7 @@
*
* Copyright (c) 2004, 2005, 2006 Voltaire, Inc. All rights reserved.
* Copyright (c) 2005, 2006 Cisco Systems. All rights reserved.
* Copyright (c) 2013 Mellanox Technologies. All rights reserved.
* Copyright (c) 2013-2014 Mellanox Technologies. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -46,6 +46,8 @@
#include <linux/printk.h>
#include <scsi/libiscsi.h>
#include <scsi/scsi_transport_iscsi.h>
#include <scsi/scsi_cmnd.h>
#include <scsi/scsi_device.h>
#include <linux/interrupt.h>
#include <linux/wait.h>
@@ -67,7 +69,7 @@
#define DRV_NAME "iser"
#define PFX DRV_NAME ": "
#define DRV_VER "1.1"
#define DRV_VER "1.3"
#define iser_dbg(fmt, arg...) \
do { \
@@ -134,10 +136,21 @@
ISER_MAX_TX_MISC_PDUS + \
ISER_MAX_RX_MISC_PDUS)
/* Max registration work requests per command */
#define ISER_MAX_REG_WR_PER_CMD 5
/* For Signature we don't support DATAOUTs so no need to make room for them */
#define ISER_QP_SIG_MAX_REQ_DTOS (ISER_DEF_XMIT_CMDS_MAX * \
(1 + ISER_MAX_REG_WR_PER_CMD) + \
ISER_MAX_TX_MISC_PDUS + \
ISER_MAX_RX_MISC_PDUS)
#define ISER_VER 0x10
#define ISER_WSV 0x08
#define ISER_RSV 0x04
#define ISER_FASTREG_LI_WRID 0xffffffffffffffffULL
struct iser_hdr {
u8 flags;
u8 rsvd[3];
@@ -201,7 +214,6 @@ struct iser_data_buf {
/* fwd declarations */
struct iser_device;
struct iser_cq_desc;
struct iscsi_iser_conn;
struct iscsi_iser_task;
struct iscsi_endpoint;
@@ -258,6 +270,7 @@ struct iscsi_iser_task;
struct iser_device {
struct ib_device *ib_device;
struct ib_pd *pd;
struct ib_device_attr dev_attr;
struct ib_cq *rx_cq[ISER_MAX_CQ];
struct ib_cq *tx_cq[ISER_MAX_CQ];
struct ib_mr *mr;
@@ -277,17 +290,35 @@ struct iser_device {
enum iser_data_dir cmd_dir);
};
#define ISER_CHECK_GUARD 0xc0
#define ISER_CHECK_REFTAG 0x0f
#define ISER_CHECK_APPTAG 0x30
enum iser_reg_indicator {
ISER_DATA_KEY_VALID = 1 << 0,
ISER_PROT_KEY_VALID = 1 << 1,
ISER_SIG_KEY_VALID = 1 << 2,
ISER_FASTREG_PROTECTED = 1 << 3,
};
struct iser_pi_context {
struct ib_mr *prot_mr;
struct ib_fast_reg_page_list *prot_frpl;
struct ib_mr *sig_mr;
};
struct fast_reg_descriptor {
struct list_head list;
/* For fast registration - FRWR */
struct ib_mr *data_mr;
struct ib_fast_reg_page_list *data_frpl;
/* Valid for fast registration flag */
bool valid;
struct iser_pi_context *pi_ctx;
/* registration indicators container */
u8 reg_indicators;
};
struct iser_conn {
struct iscsi_iser_conn *iser_conn; /* iser conn for upcalls */
struct iscsi_conn *iscsi_conn;
struct iscsi_endpoint *ep;
enum iser_ib_conn_state state; /* rdma connection state */
atomic_t refcount;
@@ -310,6 +341,9 @@ struct iser_conn {
unsigned int rx_desc_head;
struct iser_rx_desc *rx_descs;
struct ib_recv_wr rx_wr[ISER_MIN_POSTED_RX];
bool pi_support;
/* Connection memory registration pool */
union {
struct {
struct ib_fmr_pool *pool; /* pool of IB FMRs */
@@ -319,24 +353,22 @@ struct iser_conn {
struct {
struct list_head pool;
int pool_size;
} frwr;
} fastreg;
};
struct iscsi_iser_conn {
struct iscsi_conn *iscsi_conn;/* ptr to iscsi conn */
struct iser_conn *ib_conn; /* iSER IB conn */
} fastreg;
};
};
struct iscsi_iser_task {
struct iser_tx_desc desc;
struct iscsi_iser_conn *iser_conn;
struct iser_conn *ib_conn;
enum iser_task_status status;
struct scsi_cmnd *sc;
int command_sent; /* set if command sent */
int dir[ISER_DIRS_NUM]; /* set if dir use*/
struct iser_regd_buf rdma_regd[ISER_DIRS_NUM];/* regd rdma buf */
struct iser_data_buf data[ISER_DIRS_NUM]; /* orig. data des*/
struct iser_data_buf data_copy[ISER_DIRS_NUM];/* contig. copy */
struct iser_data_buf prot[ISER_DIRS_NUM]; /* prot desc */
struct iser_data_buf prot_copy[ISER_DIRS_NUM];/* prot copy */
};
struct iser_page_vec {
@@ -362,6 +394,8 @@ struct iser_global {
extern struct iser_global ig;
extern int iser_debug_level;
extern bool iser_pi_enable;
extern int iser_pi_guard;
/* allocate connection resources needed for rdma functionality */
int iser_conn_set_full_featured_mode(struct iscsi_conn *conn);
@@ -401,13 +435,15 @@ void iser_task_rdma_finalize(struct iscsi_iser_task *task);
void iser_free_rx_descriptors(struct iser_conn *ib_conn);
void iser_finalize_rdma_unaligned_sg(struct iscsi_iser_task *task,
enum iser_data_dir cmd_dir);
void iser_finalize_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
struct iser_data_buf *mem,
struct iser_data_buf *mem_copy,
enum iser_data_dir cmd_dir);
int iser_reg_rdma_mem_fmr(struct iscsi_iser_task *task,
enum iser_data_dir cmd_dir);
int iser_reg_rdma_mem_frwr(struct iscsi_iser_task *task,
enum iser_data_dir cmd_dir);
int iser_reg_rdma_mem_fastreg(struct iscsi_iser_task *task,
enum iser_data_dir cmd_dir);
int iser_connect(struct iser_conn *ib_conn,
struct sockaddr_in *src_addr,
@@ -420,8 +456,8 @@ int iser_reg_page_vec(struct iser_conn *ib_conn,
void iser_unreg_mem_fmr(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir);
void iser_unreg_mem_frwr(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir);
void iser_unreg_mem_fastreg(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir);
int iser_post_recvl(struct iser_conn *ib_conn);
int iser_post_recvm(struct iser_conn *ib_conn, int count);
@@ -432,12 +468,15 @@ int iser_dma_map_task_data(struct iscsi_iser_task *iser_task,
enum iser_data_dir iser_dir,
enum dma_data_direction dma_dir);
void iser_dma_unmap_task_data(struct iscsi_iser_task *iser_task);
void iser_dma_unmap_task_data(struct iscsi_iser_task *iser_task,
struct iser_data_buf *data);
int iser_initialize_task_headers(struct iscsi_task *task,
struct iser_tx_desc *tx_desc);
int iser_alloc_rx_descriptors(struct iser_conn *ib_conn, struct iscsi_session *session);
int iser_create_fmr_pool(struct iser_conn *ib_conn, unsigned cmds_max);
void iser_free_fmr_pool(struct iser_conn *ib_conn);
int iser_create_frwr_pool(struct iser_conn *ib_conn, unsigned cmds_max);
void iser_free_frwr_pool(struct iser_conn *ib_conn);
int iser_create_fastreg_pool(struct iser_conn *ib_conn, unsigned cmds_max);
void iser_free_fastreg_pool(struct iser_conn *ib_conn);
u8 iser_check_task_pi_status(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir, sector_t *sector);
#endif

154
drivers/infiniband/ulp/iser/iser_initiator.c
Näytä tiedosto

@@ -1,6 +1,6 @@
/*
* Copyright (c) 2004, 2005, 2006 Voltaire, Inc. All rights reserved.
* Copyright (c) 2013 Mellanox Technologies. All rights reserved.
* Copyright (c) 2013-2014 Mellanox Technologies. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -49,7 +49,7 @@ static int iser_prepare_read_cmd(struct iscsi_task *task,
{
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_device *device = iser_task->iser_conn->ib_conn->device;
struct iser_device *device = iser_task->ib_conn->device;
struct iser_regd_buf *regd_buf;
int err;
struct iser_hdr *hdr = &iser_task->desc.iser_header;
@@ -62,11 +62,22 @@ static int iser_prepare_read_cmd(struct iscsi_task *task,
if (err)
return err;
if (scsi_prot_sg_count(iser_task->sc)) {
struct iser_data_buf *pbuf_in = &iser_task->prot[ISER_DIR_IN];
err = iser_dma_map_task_data(iser_task,
pbuf_in,
ISER_DIR_IN,
DMA_FROM_DEVICE);
if (err)
return err;
}
if (edtl > iser_task->data[ISER_DIR_IN].data_len) {
iser_err("Total data length: %ld, less than EDTL: "
"%d, in READ cmd BHS itt: %d, conn: 0x%p\n",
iser_task->data[ISER_DIR_IN].data_len, edtl,
task->itt, iser_task->iser_conn);
task->itt, iser_task->ib_conn);
return -EINVAL;
}
@@ -99,7 +110,7 @@ iser_prepare_write_cmd(struct iscsi_task *task,
unsigned int edtl)
{
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_device *device = iser_task->iser_conn->ib_conn->device;
struct iser_device *device = iser_task->ib_conn->device;
struct iser_regd_buf *regd_buf;
int err;
struct iser_hdr *hdr = &iser_task->desc.iser_header;
@@ -113,6 +124,17 @@ iser_prepare_write_cmd(struct iscsi_task *task,
if (err)
return err;
if (scsi_prot_sg_count(iser_task->sc)) {
struct iser_data_buf *pbuf_out = &iser_task->prot[ISER_DIR_OUT];
err = iser_dma_map_task_data(iser_task,
pbuf_out,
ISER_DIR_OUT,
DMA_TO_DEVICE);
if (err)
return err;
}
if (edtl > iser_task->data[ISER_DIR_OUT].data_len) {
iser_err("Total data length: %ld, less than EDTL: %d, "
"in WRITE cmd BHS itt: %d, conn: 0x%p\n",
@@ -327,7 +349,7 @@ free_login_buf:
static int iser_post_rx_bufs(struct iscsi_conn *conn, struct iscsi_hdr *req)
{
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iser_conn *ib_conn = conn->dd_data;
struct iscsi_session *session = conn->session;
iser_dbg("req op %x flags %x\n", req->opcode, req->flags);
@@ -340,19 +362,18 @@ static int iser_post_rx_bufs(struct iscsi_conn *conn, struct iscsi_hdr *req)
* response) and no posted send buffers left - they must have been
* consumed during previous login phases.
*/
WARN_ON(iser_conn->ib_conn->post_recv_buf_count != 1);
WARN_ON(atomic_read(&iser_conn->ib_conn->post_send_buf_count) != 0);
WARN_ON(ib_conn->post_recv_buf_count != 1);
WARN_ON(atomic_read(&ib_conn->post_send_buf_count) != 0);
if (session->discovery_sess) {
iser_info("Discovery session, re-using login RX buffer\n");
return 0;
} else
iser_info("Normal session, posting batch of RX %d buffers\n",
iser_conn->ib_conn->min_posted_rx);
ib_conn->min_posted_rx);
/* Initial post receive buffers */
if (iser_post_recvm(iser_conn->ib_conn,
iser_conn->ib_conn->min_posted_rx))
if (iser_post_recvm(ib_conn, ib_conn->min_posted_rx))
return -ENOMEM;
return 0;
@@ -364,11 +385,11 @@ static int iser_post_rx_bufs(struct iscsi_conn *conn, struct iscsi_hdr *req)
int iser_send_command(struct iscsi_conn *conn,
struct iscsi_task *task)
{
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iser_conn *ib_conn = conn->dd_data;
struct iscsi_iser_task *iser_task = task->dd_data;
unsigned long edtl;
int err;
struct iser_data_buf *data_buf;
struct iser_data_buf *data_buf, *prot_buf;
struct iscsi_scsi_req *hdr = (struct iscsi_scsi_req *)task->hdr;
struct scsi_cmnd *sc = task->sc;
struct iser_tx_desc *tx_desc = &iser_task->desc;
@@ -377,20 +398,28 @@ 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;
iser_create_send_desc(iser_conn->ib_conn, tx_desc);
iser_create_send_desc(ib_conn, tx_desc);
if (hdr->flags & ISCSI_FLAG_CMD_READ)
if (hdr->flags & ISCSI_FLAG_CMD_READ) {
data_buf = &iser_task->data[ISER_DIR_IN];
else
prot_buf = &iser_task->prot[ISER_DIR_IN];
} else {
data_buf = &iser_task->data[ISER_DIR_OUT];
prot_buf = &iser_task->prot[ISER_DIR_OUT];
}
if (scsi_sg_count(sc)) { /* using a scatter list */
data_buf->buf = scsi_sglist(sc);
data_buf->size = scsi_sg_count(sc);
}
data_buf->data_len = scsi_bufflen(sc);
if (scsi_prot_sg_count(sc)) {
prot_buf->buf = scsi_prot_sglist(sc);
prot_buf->size = scsi_prot_sg_count(sc);
prot_buf->data_len = sc->prot_sdb->length;
}
if (hdr->flags & ISCSI_FLAG_CMD_READ) {
err = iser_prepare_read_cmd(task, edtl);
if (err)
@@ -408,7 +437,7 @@ int iser_send_command(struct iscsi_conn *conn,
iser_task->status = ISER_TASK_STATUS_STARTED;
err = iser_post_send(iser_conn->ib_conn, tx_desc);
err = iser_post_send(ib_conn, tx_desc);
if (!err)
return 0;
@@ -424,7 +453,7 @@ int iser_send_data_out(struct iscsi_conn *conn,
struct iscsi_task *task,
struct iscsi_data *hdr)
{
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iser_conn *ib_conn = conn->dd_data;
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_tx_desc *tx_desc = NULL;
struct iser_regd_buf *regd_buf;
@@ -473,7 +502,7 @@ int iser_send_data_out(struct iscsi_conn *conn,
itt, buf_offset, data_seg_len);
err = iser_post_send(iser_conn->ib_conn, tx_desc);
err = iser_post_send(ib_conn, tx_desc);
if (!err)
return 0;
@@ -486,19 +515,18 @@ send_data_out_error:
int iser_send_control(struct iscsi_conn *conn,
struct iscsi_task *task)
{
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iser_conn *ib_conn = conn->dd_data;
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_tx_desc *mdesc = &iser_task->desc;
unsigned long data_seg_len;
int err = 0;
struct iser_device *device;
struct iser_conn *ib_conn = iser_conn->ib_conn;
/* build the tx desc regd header and add it to the tx desc dto */
mdesc->type = ISCSI_TX_CONTROL;
iser_create_send_desc(iser_conn->ib_conn, mdesc);
iser_create_send_desc(ib_conn, mdesc);
device = iser_conn->ib_conn->device;
device = ib_conn->device;
data_seg_len = ntoh24(task->hdr->dlength);
@@ -513,14 +541,13 @@ int iser_send_control(struct iscsi_conn *conn,
ib_conn->login_req_dma, task->data_count,
DMA_TO_DEVICE);
memcpy(iser_conn->ib_conn->login_req_buf, task->data,
task->data_count);
memcpy(ib_conn->login_req_buf, task->data, task->data_count);
ib_dma_sync_single_for_device(device->ib_device,
ib_conn->login_req_dma, task->data_count,
DMA_TO_DEVICE);
tx_dsg->addr = iser_conn->ib_conn->login_req_dma;
tx_dsg->addr = ib_conn->login_req_dma;
tx_dsg->length = task->data_count;
tx_dsg->lkey = device->mr->lkey;
mdesc->num_sge = 2;
@@ -529,7 +556,7 @@ int iser_send_control(struct iscsi_conn *conn,
if (task == conn->login_task) {
iser_dbg("op %x dsl %lx, posting login rx buffer\n",
task->hdr->opcode, data_seg_len);
err = iser_post_recvl(iser_conn->ib_conn);
err = iser_post_recvl(ib_conn);
if (err)
goto send_control_error;
err = iser_post_rx_bufs(conn, task->hdr);
@@ -537,7 +564,7 @@ int iser_send_control(struct iscsi_conn *conn,
goto send_control_error;
}
err = iser_post_send(iser_conn->ib_conn, mdesc);
err = iser_post_send(ib_conn, mdesc);
if (!err)
return 0;
@@ -553,7 +580,6 @@ void iser_rcv_completion(struct iser_rx_desc *rx_desc,
unsigned long rx_xfer_len,
struct iser_conn *ib_conn)
{
struct iscsi_iser_conn *conn = ib_conn->iser_conn;
struct iscsi_hdr *hdr;
u64 rx_dma;
int rx_buflen, outstanding, count, err;
@@ -575,17 +601,17 @@ void iser_rcv_completion(struct iser_rx_desc *rx_desc,
iser_dbg("op 0x%x itt 0x%x dlen %d\n", hdr->opcode,
hdr->itt, (int)(rx_xfer_len - ISER_HEADERS_LEN));
iscsi_iser_recv(conn->iscsi_conn, hdr,
rx_desc->data, rx_xfer_len - ISER_HEADERS_LEN);
iscsi_iser_recv(ib_conn->iscsi_conn, hdr, rx_desc->data,
rx_xfer_len - ISER_HEADERS_LEN);
ib_dma_sync_single_for_device(ib_conn->device->ib_device, rx_dma,
rx_buflen, DMA_FROM_DEVICE);
rx_buflen, 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 *
* parallel to the execution of iser_conn_term. So the code that waits *
* for the posted rx bufs refcount to become zero handles everything */
conn->ib_conn->post_recv_buf_count--;
ib_conn->post_recv_buf_count--;
if (rx_dma == ib_conn->login_resp_dma)
return;
@@ -635,6 +661,9 @@ void iser_task_rdma_init(struct iscsi_iser_task *iser_task)
iser_task->data[ISER_DIR_IN].data_len = 0;
iser_task->data[ISER_DIR_OUT].data_len = 0;
iser_task->prot[ISER_DIR_IN].data_len = 0;
iser_task->prot[ISER_DIR_OUT].data_len = 0;
memset(&iser_task->rdma_regd[ISER_DIR_IN], 0,
sizeof(struct iser_regd_buf));
memset(&iser_task->rdma_regd[ISER_DIR_OUT], 0,
@@ -643,28 +672,63 @@ void iser_task_rdma_init(struct iscsi_iser_task *iser_task)
void iser_task_rdma_finalize(struct iscsi_iser_task *iser_task)
{
struct iser_device *device = iser_task->iser_conn->ib_conn->device;
int is_rdma_aligned = 1;
struct iser_device *device = iser_task->ib_conn->device;
int is_rdma_data_aligned = 1;
int is_rdma_prot_aligned = 1;
int prot_count = scsi_prot_sg_count(iser_task->sc);
/* if we were reading, copy back to unaligned sglist,
* anyway dma_unmap and free the copy
*/
if (iser_task->data_copy[ISER_DIR_IN].copy_buf != NULL) {
is_rdma_aligned = 0;
iser_finalize_rdma_unaligned_sg(iser_task, ISER_DIR_IN);
is_rdma_data_aligned = 0;
iser_finalize_rdma_unaligned_sg(iser_task,
&iser_task->data[ISER_DIR_IN],
&iser_task->data_copy[ISER_DIR_IN],
ISER_DIR_IN);
}
if (iser_task->data_copy[ISER_DIR_OUT].copy_buf != NULL) {
is_rdma_aligned = 0;
iser_finalize_rdma_unaligned_sg(iser_task, ISER_DIR_OUT);
is_rdma_data_aligned = 0;
iser_finalize_rdma_unaligned_sg(iser_task,
&iser_task->data[ISER_DIR_OUT],
&iser_task->data_copy[ISER_DIR_OUT],
ISER_DIR_OUT);
}
if (iser_task->dir[ISER_DIR_IN])
if (iser_task->prot_copy[ISER_DIR_IN].copy_buf != NULL) {
is_rdma_prot_aligned = 0;
iser_finalize_rdma_unaligned_sg(iser_task,
&iser_task->prot[ISER_DIR_IN],
&iser_task->prot_copy[ISER_DIR_IN],
ISER_DIR_IN);
}
if (iser_task->prot_copy[ISER_DIR_OUT].copy_buf != NULL) {
is_rdma_prot_aligned = 0;
iser_finalize_rdma_unaligned_sg(iser_task,
&iser_task->prot[ISER_DIR_OUT],
&iser_task->prot_copy[ISER_DIR_OUT],
ISER_DIR_OUT);
}
if (iser_task->dir[ISER_DIR_IN]) {
device->iser_unreg_rdma_mem(iser_task, ISER_DIR_IN);
if (is_rdma_data_aligned)
iser_dma_unmap_task_data(iser_task,
&iser_task->data[ISER_DIR_IN]);
if (prot_count && is_rdma_prot_aligned)
iser_dma_unmap_task_data(iser_task,
&iser_task->prot[ISER_DIR_IN]);
}
if (iser_task->dir[ISER_DIR_OUT])
if (iser_task->dir[ISER_DIR_OUT]) {
device->iser_unreg_rdma_mem(iser_task, ISER_DIR_OUT);
/* if the data was unaligned, it was already unmapped and then copied */
if (is_rdma_aligned)
iser_dma_unmap_task_data(iser_task);
if (is_rdma_data_aligned)
iser_dma_unmap_task_data(iser_task,
&iser_task->data[ISER_DIR_OUT]);
if (prot_count && is_rdma_prot_aligned)
iser_dma_unmap_task_data(iser_task,
&iser_task->prot[ISER_DIR_OUT]);
}
}

470
drivers/infiniband/ulp/iser/iser_memory.c
Näytä tiedosto

@@ -1,6 +1,6 @@
/*
* Copyright (c) 2004, 2005, 2006 Voltaire, Inc. All rights reserved.
* Copyright (c) 2013 Mellanox Technologies. All rights reserved.
* Copyright (c) 2013-2014 Mellanox Technologies. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -45,13 +45,19 @@
* iser_start_rdma_unaligned_sg
*/
static int iser_start_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
struct iser_data_buf *data,
struct iser_data_buf *data_copy,
enum iser_data_dir cmd_dir)
{
int dma_nents;
struct ib_device *dev;
struct ib_device *dev = iser_task->ib_conn->device->ib_device;
struct scatterlist *sgl = (struct scatterlist *)data->buf;
struct scatterlist *sg;
char *mem = NULL;
struct iser_data_buf *data = &iser_task->data[cmd_dir];
unsigned long cmd_data_len = data->data_len;
unsigned long cmd_data_len = 0;
int dma_nents, i;
for_each_sg(sgl, sg, data->size, i)
cmd_data_len += ib_sg_dma_len(dev, sg);
if (cmd_data_len > ISER_KMALLOC_THRESHOLD)
mem = (void *)__get_free_pages(GFP_ATOMIC,
@@ -61,17 +67,16 @@ static int iser_start_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
if (mem == NULL) {
iser_err("Failed to allocate mem size %d %d for copying sglist\n",
data->size,(int)cmd_data_len);
data->size, (int)cmd_data_len);
return -ENOMEM;
}
if (cmd_dir == ISER_DIR_OUT) {
/* copy the unaligned sg the buffer which is used for RDMA */
struct scatterlist *sgl = (struct scatterlist *)data->buf;
struct scatterlist *sg;
int i;
char *p, *from;
sgl = (struct scatterlist *)data->buf;
p = mem;
for_each_sg(sgl, sg, data->size, i) {
from = kmap_atomic(sg_page(sg));
@@ -83,39 +88,37 @@ static int iser_start_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
}
}
sg_init_one(&iser_task->data_copy[cmd_dir].sg_single, mem, cmd_data_len);
iser_task->data_copy[cmd_dir].buf =
&iser_task->data_copy[cmd_dir].sg_single;
iser_task->data_copy[cmd_dir].size = 1;
sg_init_one(&data_copy->sg_single, mem, cmd_data_len);
data_copy->buf = &data_copy->sg_single;
data_copy->size = 1;
data_copy->copy_buf = mem;
iser_task->data_copy[cmd_dir].copy_buf = mem;
dev = iser_task->iser_conn->ib_conn->device->ib_device;
dma_nents = ib_dma_map_sg(dev,
&iser_task->data_copy[cmd_dir].sg_single,
1,
dma_nents = ib_dma_map_sg(dev, &data_copy->sg_single, 1,
(cmd_dir == ISER_DIR_OUT) ?
DMA_TO_DEVICE : DMA_FROM_DEVICE);
BUG_ON(dma_nents == 0);
iser_task->data_copy[cmd_dir].dma_nents = dma_nents;
data_copy->dma_nents = dma_nents;
data_copy->data_len = cmd_data_len;
return 0;
}
/**
* iser_finalize_rdma_unaligned_sg
*/
void iser_finalize_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir)
struct iser_data_buf *data,
struct iser_data_buf *data_copy,
enum iser_data_dir cmd_dir)
{
struct ib_device *dev;
struct iser_data_buf *mem_copy;
unsigned long cmd_data_len;
dev = iser_task->iser_conn->ib_conn->device->ib_device;
mem_copy = &iser_task->data_copy[cmd_dir];
dev = iser_task->ib_conn->device->ib_device;
ib_dma_unmap_sg(dev, &mem_copy->sg_single, 1,
ib_dma_unmap_sg(dev, &data_copy->sg_single, 1,
(cmd_dir == ISER_DIR_OUT) ?
DMA_TO_DEVICE : DMA_FROM_DEVICE);
@@ -127,10 +130,10 @@ void iser_finalize_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
int i;
/* copy back read RDMA to unaligned sg */
mem = mem_copy->copy_buf;
mem = data_copy->copy_buf;
sgl = (struct scatterlist *)iser_task->data[ISER_DIR_IN].buf;
sg_size = iser_task->data[ISER_DIR_IN].size;
sgl = (struct scatterlist *)data->buf;
sg_size = data->size;
p = mem;
for_each_sg(sgl, sg, sg_size, i) {
@@ -143,15 +146,15 @@ void iser_finalize_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
}
}
cmd_data_len = iser_task->data[cmd_dir].data_len;
cmd_data_len = data->data_len;
if (cmd_data_len > ISER_KMALLOC_THRESHOLD)
free_pages((unsigned long)mem_copy->copy_buf,
free_pages((unsigned long)data_copy->copy_buf,
ilog2(roundup_pow_of_two(cmd_data_len)) - PAGE_SHIFT);
else
kfree(mem_copy->copy_buf);
kfree(data_copy->copy_buf);
mem_copy->copy_buf = NULL;
data_copy->copy_buf = NULL;
}
#define IS_4K_ALIGNED(addr) ((((unsigned long)addr) & ~MASK_4K) == 0)
@@ -319,7 +322,7 @@ int iser_dma_map_task_data(struct iscsi_iser_task *iser_task,
struct ib_device *dev;
iser_task->dir[iser_dir] = 1;
dev = iser_task->iser_conn->ib_conn->device->ib_device;
dev = iser_task->ib_conn->device->ib_device;
data->dma_nents = ib_dma_map_sg(dev, data->buf, data->size, dma_dir);
if (data->dma_nents == 0) {
@@ -329,31 +332,23 @@ int iser_dma_map_task_data(struct iscsi_iser_task *iser_task,
return 0;
}
void iser_dma_unmap_task_data(struct iscsi_iser_task *iser_task)
void iser_dma_unmap_task_data(struct iscsi_iser_task *iser_task,
struct iser_data_buf *data)
{
struct ib_device *dev;
struct iser_data_buf *data;
dev = iser_task->iser_conn->ib_conn->device->ib_device;
if (iser_task->dir[ISER_DIR_IN]) {
data = &iser_task->data[ISER_DIR_IN];
ib_dma_unmap_sg(dev, data->buf, data->size, DMA_FROM_DEVICE);
}
if (iser_task->dir[ISER_DIR_OUT]) {
data = &iser_task->data[ISER_DIR_OUT];
ib_dma_unmap_sg(dev, data->buf, data->size, DMA_TO_DEVICE);
}
dev = iser_task->ib_conn->device->ib_device;
ib_dma_unmap_sg(dev, data->buf, data->size, DMA_FROM_DEVICE);
}
static int fall_to_bounce_buf(struct iscsi_iser_task *iser_task,
struct ib_device *ibdev,
struct iser_data_buf *mem,
struct iser_data_buf *mem_copy,
enum iser_data_dir cmd_dir,
int aligned_len)
{
struct iscsi_conn *iscsi_conn = iser_task->iser_conn->iscsi_conn;
struct iser_data_buf *mem = &iser_task->data[cmd_dir];
struct iscsi_conn *iscsi_conn = iser_task->ib_conn->iscsi_conn;
iscsi_conn->fmr_unalign_cnt++;
iser_warn("rdma alignment violation (%d/%d aligned) or FMR not supported\n",
@@ -363,12 +358,12 @@ static int fall_to_bounce_buf(struct iscsi_iser_task *iser_task,
iser_data_buf_dump(mem, ibdev);
/* unmap the command data before accessing it */
iser_dma_unmap_task_data(iser_task);
iser_dma_unmap_task_data(iser_task, mem);
/* allocate copy buf, if we are writing, copy the */
/* unaligned scatterlist, dma map the copy */
if (iser_start_rdma_unaligned_sg(iser_task, cmd_dir) != 0)
return -ENOMEM;
if (iser_start_rdma_unaligned_sg(iser_task, mem, mem_copy, cmd_dir) != 0)
return -ENOMEM;
return 0;
}
@@ -382,7 +377,7 @@ static int fall_to_bounce_buf(struct iscsi_iser_task *iser_task,
int iser_reg_rdma_mem_fmr(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir)
{
struct iser_conn *ib_conn = iser_task->iser_conn->ib_conn;
struct iser_conn *ib_conn = iser_task->ib_conn;
struct iser_device *device = ib_conn->device;
struct ib_device *ibdev = device->ib_device;
struct iser_data_buf *mem = &iser_task->data[cmd_dir];
@@ -396,7 +391,8 @@ int iser_reg_rdma_mem_fmr(struct iscsi_iser_task *iser_task,
aligned_len = iser_data_buf_aligned_len(mem, ibdev);
if (aligned_len != mem->dma_nents) {
err = fall_to_bounce_buf(iser_task, ibdev,
err = fall_to_bounce_buf(iser_task, ibdev, mem,
&iser_task->data_copy[cmd_dir],
cmd_dir, aligned_len);
if (err) {
iser_err("failed to allocate bounce buffer\n");
@@ -422,8 +418,8 @@ int iser_reg_rdma_mem_fmr(struct iscsi_iser_task *iser_task,
(unsigned long)regd_buf->reg.va,
(unsigned long)regd_buf->reg.len);
} else { /* use FMR for multiple dma entries */
iser_page_vec_build(mem, ib_conn->fastreg.fmr.page_vec, ibdev);
err = iser_reg_page_vec(ib_conn, ib_conn->fastreg.fmr.page_vec,
iser_page_vec_build(mem, ib_conn->fmr.page_vec, ibdev);
err = iser_reg_page_vec(ib_conn, ib_conn->fmr.page_vec,
&regd_buf->reg);
if (err && err != -EAGAIN) {
iser_data_buf_dump(mem, ibdev);
@@ -431,12 +427,12 @@ int iser_reg_rdma_mem_fmr(struct iscsi_iser_task *iser_task,
mem->dma_nents,
ntoh24(iser_task->desc.iscsi_header.dlength));
iser_err("page_vec: data_size = 0x%x, length = %d, offset = 0x%x\n",
ib_conn->fastreg.fmr.page_vec->data_size,
ib_conn->fastreg.fmr.page_vec->length,
ib_conn->fastreg.fmr.page_vec->offset);
for (i = 0; i < ib_conn->fastreg.fmr.page_vec->length; i++)
ib_conn->fmr.page_vec->data_size,
ib_conn->fmr.page_vec->length,
ib_conn->fmr.page_vec->offset);
for (i = 0; i < ib_conn->fmr.page_vec->length; i++)
iser_err("page_vec[%d] = 0x%llx\n", i,
(unsigned long long) ib_conn->fastreg.fmr.page_vec->pages[i]);
(unsigned long long) ib_conn->fmr.page_vec->pages[i]);
}
if (err)
return err;
@@ -444,94 +440,280 @@ int iser_reg_rdma_mem_fmr(struct iscsi_iser_task *iser_task,
return 0;
}
static int iser_fast_reg_mr(struct fast_reg_descriptor *desc,
struct iser_conn *ib_conn,
struct iser_regd_buf *regd_buf,
u32 offset, unsigned int data_size,
unsigned int page_list_len)
static inline enum ib_t10_dif_type
scsi2ib_prot_type(unsigned char prot_type)
{
switch (prot_type) {
case SCSI_PROT_DIF_TYPE0:
return IB_T10DIF_NONE;
case SCSI_PROT_DIF_TYPE1:
return IB_T10DIF_TYPE1;
case SCSI_PROT_DIF_TYPE2:
return IB_T10DIF_TYPE2;
case SCSI_PROT_DIF_TYPE3:
return IB_T10DIF_TYPE3;
default:
return IB_T10DIF_NONE;
}
}
static int
iser_set_sig_attrs(struct scsi_cmnd *sc, struct ib_sig_attrs *sig_attrs)
{
unsigned char scsi_ptype = scsi_get_prot_type(sc);
sig_attrs->mem.sig_type = IB_SIG_TYPE_T10_DIF;
sig_attrs->wire.sig_type = IB_SIG_TYPE_T10_DIF;
sig_attrs->mem.sig.dif.pi_interval = sc->device->sector_size;
sig_attrs->wire.sig.dif.pi_interval = sc->device->sector_size;
switch (scsi_get_prot_op(sc)) {
case SCSI_PROT_WRITE_INSERT:
case SCSI_PROT_READ_STRIP:
sig_attrs->mem.sig.dif.type = IB_T10DIF_NONE;
sig_attrs->wire.sig.dif.type = scsi2ib_prot_type(scsi_ptype);
sig_attrs->wire.sig.dif.bg_type = IB_T10DIF_CRC;
sig_attrs->wire.sig.dif.ref_tag = scsi_get_lba(sc) &
0xffffffff;
break;
case SCSI_PROT_READ_INSERT:
case SCSI_PROT_WRITE_STRIP:
sig_attrs->mem.sig.dif.type = scsi2ib_prot_type(scsi_ptype);
sig_attrs->mem.sig.dif.bg_type = IB_T10DIF_CRC;
sig_attrs->mem.sig.dif.ref_tag = scsi_get_lba(sc) &
0xffffffff;
sig_attrs->wire.sig.dif.type = IB_T10DIF_NONE;
break;
case SCSI_PROT_READ_PASS:
case SCSI_PROT_WRITE_PASS:
sig_attrs->mem.sig.dif.type = scsi2ib_prot_type(scsi_ptype);
sig_attrs->mem.sig.dif.bg_type = IB_T10DIF_CRC;
sig_attrs->mem.sig.dif.ref_tag = scsi_get_lba(sc) &
0xffffffff;
sig_attrs->wire.sig.dif.type = scsi2ib_prot_type(scsi_ptype);
sig_attrs->wire.sig.dif.bg_type = IB_T10DIF_CRC;
sig_attrs->wire.sig.dif.ref_tag = scsi_get_lba(sc) &
0xffffffff;
break;
default:
iser_err("Unsupported PI operation %d\n",
scsi_get_prot_op(sc));
return -EINVAL;
}
return 0;
}
static int
iser_set_prot_checks(struct scsi_cmnd *sc, u8 *mask)
{
switch (scsi_get_prot_type(sc)) {
case SCSI_PROT_DIF_TYPE0:
*mask = 0x0;
break;
case SCSI_PROT_DIF_TYPE1:
case SCSI_PROT_DIF_TYPE2:
*mask = ISER_CHECK_GUARD | ISER_CHECK_REFTAG;
break;
case SCSI_PROT_DIF_TYPE3:
*mask = ISER_CHECK_GUARD;
break;
default:
iser_err("Unsupported protection type %d\n",
scsi_get_prot_type(sc));
return -EINVAL;
}
return 0;
}
static int
iser_reg_sig_mr(struct iscsi_iser_task *iser_task,
struct fast_reg_descriptor *desc, struct ib_sge *data_sge,
struct ib_sge *prot_sge, struct ib_sge *sig_sge)
{
struct iser_conn *ib_conn = iser_task->ib_conn;
struct iser_pi_context *pi_ctx = desc->pi_ctx;
struct ib_send_wr sig_wr, inv_wr;
struct ib_send_wr *bad_wr, *wr = NULL;
struct ib_sig_attrs sig_attrs;
int ret;
u32 key;
memset(&sig_attrs, 0, sizeof(sig_attrs));
ret = iser_set_sig_attrs(iser_task->sc, &sig_attrs);
if (ret)
goto err;
ret = iser_set_prot_checks(iser_task->sc, &sig_attrs.check_mask);
if (ret)
goto err;
if (!(desc->reg_indicators & ISER_SIG_KEY_VALID)) {
memset(&inv_wr, 0, sizeof(inv_wr));
inv_wr.opcode = IB_WR_LOCAL_INV;
inv_wr.wr_id = ISER_FASTREG_LI_WRID;
inv_wr.ex.invalidate_rkey = pi_ctx->sig_mr->rkey;
wr = &inv_wr;
/* Bump the key */
key = (u8)(pi_ctx->sig_mr->rkey & 0x000000FF);
ib_update_fast_reg_key(pi_ctx->sig_mr, ++key);
}
memset(&sig_wr, 0, sizeof(sig_wr));
sig_wr.opcode = IB_WR_REG_SIG_MR;
sig_wr.wr_id = ISER_FASTREG_LI_WRID;
sig_wr.sg_list = data_sge;
sig_wr.num_sge = 1;
sig_wr.wr.sig_handover.sig_attrs = &sig_attrs;
sig_wr.wr.sig_handover.sig_mr = pi_ctx->sig_mr;
if (scsi_prot_sg_count(iser_task->sc))
sig_wr.wr.sig_handover.prot = prot_sge;
sig_wr.wr.sig_handover.access_flags = IB_ACCESS_LOCAL_WRITE |
IB_ACCESS_REMOTE_READ |
IB_ACCESS_REMOTE_WRITE;
if (!wr)
wr = &sig_wr;
else
wr->next = &sig_wr;
ret = ib_post_send(ib_conn->qp, wr, &bad_wr);
if (ret) {
iser_err("reg_sig_mr failed, ret:%d\n", ret);
goto err;
}
desc->reg_indicators &= ~ISER_SIG_KEY_VALID;
sig_sge->lkey = pi_ctx->sig_mr->lkey;
sig_sge->addr = 0;
sig_sge->length = data_sge->length + prot_sge->length;
if (scsi_get_prot_op(iser_task->sc) == SCSI_PROT_WRITE_INSERT ||
scsi_get_prot_op(iser_task->sc) == SCSI_PROT_READ_STRIP) {
sig_sge->length += (data_sge->length /
iser_task->sc->device->sector_size) * 8;
}
iser_dbg("sig_sge: addr: 0x%llx length: %u lkey: 0x%x\n",
sig_sge->addr, sig_sge->length,
sig_sge->lkey);
err:
return ret;
}
static int iser_fast_reg_mr(struct iscsi_iser_task *iser_task,
struct iser_regd_buf *regd_buf,
struct iser_data_buf *mem,
enum iser_reg_indicator ind,
struct ib_sge *sge)
{
struct fast_reg_descriptor *desc = regd_buf->reg.mem_h;
struct iser_conn *ib_conn = iser_task->ib_conn;
struct iser_device *device = ib_conn->device;
struct ib_device *ibdev = device->ib_device;
struct ib_mr *mr;
struct ib_fast_reg_page_list *frpl;
struct ib_send_wr fastreg_wr, inv_wr;
struct ib_send_wr *bad_wr, *wr = NULL;
u8 key;
int ret;
int ret, offset, size, plen;
if (!desc->valid) {
/* if there a single dma entry, dma mr suffices */
if (mem->dma_nents == 1) {
struct scatterlist *sg = (struct scatterlist *)mem->buf;
sge->lkey = device->mr->lkey;
sge->addr = ib_sg_dma_address(ibdev, &sg[0]);
sge->length = ib_sg_dma_len(ibdev, &sg[0]);
iser_dbg("Single DMA entry: lkey=0x%x, addr=0x%llx, length=0x%x\n",
sge->lkey, sge->addr, sge->length);
return 0;
}
if (ind == ISER_DATA_KEY_VALID) {
mr = desc->data_mr;
frpl = desc->data_frpl;
} else {
mr = desc->pi_ctx->prot_mr;
frpl = desc->pi_ctx->prot_frpl;
}
plen = iser_sg_to_page_vec(mem, device->ib_device, frpl->page_list,
&offset, &size);
if (plen * SIZE_4K < size) {
iser_err("fast reg page_list too short to hold this SG\n");
return -EINVAL;
}
if (!(desc->reg_indicators & ind)) {
memset(&inv_wr, 0, sizeof(inv_wr));
inv_wr.wr_id = ISER_FASTREG_LI_WRID;
inv_wr.opcode = IB_WR_LOCAL_INV;
inv_wr.send_flags = IB_SEND_SIGNALED;
inv_wr.ex.invalidate_rkey = desc->data_mr->rkey;
inv_wr.ex.invalidate_rkey = mr->rkey;
wr = &inv_wr;
/* Bump the key */
key = (u8)(desc->data_mr->rkey & 0x000000FF);
ib_update_fast_reg_key(desc->data_mr, ++key);
key = (u8)(mr->rkey & 0x000000FF);
ib_update_fast_reg_key(mr, ++key);
}
/* Prepare FASTREG WR */
memset(&fastreg_wr, 0, sizeof(fastreg_wr));
fastreg_wr.wr_id = ISER_FASTREG_LI_WRID;
fastreg_wr.opcode = IB_WR_FAST_REG_MR;
fastreg_wr.send_flags = IB_SEND_SIGNALED;
fastreg_wr.wr.fast_reg.iova_start = desc->data_frpl->page_list[0] + offset;
fastreg_wr.wr.fast_reg.page_list = desc->data_frpl;
fastreg_wr.wr.fast_reg.page_list_len = page_list_len;
fastreg_wr.wr.fast_reg.iova_start = frpl->page_list[0] + offset;
fastreg_wr.wr.fast_reg.page_list = frpl;
fastreg_wr.wr.fast_reg.page_list_len = plen;
fastreg_wr.wr.fast_reg.page_shift = SHIFT_4K;
fastreg_wr.wr.fast_reg.length = data_size;
fastreg_wr.wr.fast_reg.rkey = desc->data_mr->rkey;
fastreg_wr.wr.fast_reg.length = size;
fastreg_wr.wr.fast_reg.rkey = mr->rkey;
fastreg_wr.wr.fast_reg.access_flags = (IB_ACCESS_LOCAL_WRITE |
IB_ACCESS_REMOTE_WRITE |
IB_ACCESS_REMOTE_READ);
if (!wr) {
if (!wr)
wr = &fastreg_wr;
atomic_inc(&ib_conn->post_send_buf_count);
} else {
else
wr->next = &fastreg_wr;
atomic_add(2, &ib_conn->post_send_buf_count);
}
ret = ib_post_send(ib_conn->qp, wr, &bad_wr);
if (ret) {
if (bad_wr->next)
atomic_sub(2, &ib_conn->post_send_buf_count);
else
atomic_dec(&ib_conn->post_send_buf_count);
iser_err("fast registration failed, ret:%d\n", ret);
return ret;
}
desc->valid = false;
desc->reg_indicators &= ~ind;
regd_buf->reg.mem_h = desc;
regd_buf->reg.lkey = desc->data_mr->lkey;
regd_buf->reg.rkey = desc->data_mr->rkey;
regd_buf->reg.va = desc->data_frpl->page_list[0] + offset;
regd_buf->reg.len = data_size;
regd_buf->reg.is_mr = 1;
sge->lkey = mr->lkey;
sge->addr = frpl->page_list[0] + offset;
sge->length = size;
return ret;
}
/**
* iser_reg_rdma_mem_frwr - Registers memory intended for RDMA,
* iser_reg_rdma_mem_fastreg - Registers memory intended for RDMA,
* using Fast Registration WR (if possible) obtaining rkey and va
*
* returns 0 on success, errno code on failure
*/
int iser_reg_rdma_mem_frwr(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir)
int iser_reg_rdma_mem_fastreg(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir)
{
struct iser_conn *ib_conn = iser_task->iser_conn->ib_conn;
struct iser_conn *ib_conn = iser_task->ib_conn;
struct iser_device *device = ib_conn->device;
struct ib_device *ibdev = device->ib_device;
struct iser_data_buf *mem = &iser_task->data[cmd_dir];
struct iser_regd_buf *regd_buf = &iser_task->rdma_regd[cmd_dir];
struct fast_reg_descriptor *desc;
unsigned int data_size, page_list_len;
struct fast_reg_descriptor *desc = NULL;
struct ib_sge data_sge;
int err, aligned_len;
unsigned long flags;
u32 offset;
aligned_len = iser_data_buf_aligned_len(mem, ibdev);
if (aligned_len != mem->dma_nents) {
err = fall_to_bounce_buf(iser_task, ibdev,
err = fall_to_bounce_buf(iser_task, ibdev, mem,
&iser_task->data_copy[cmd_dir],
cmd_dir, aligned_len);
if (err) {
iser_err("failed to allocate bounce buffer\n");
@@ -540,41 +722,79 @@ int iser_reg_rdma_mem_frwr(struct iscsi_iser_task *iser_task,
mem = &iser_task->data_copy[cmd_dir];
}
/* if there a single dma entry, dma mr suffices */
if (mem->dma_nents == 1) {
struct scatterlist *sg = (struct scatterlist *)mem->buf;
regd_buf->reg.lkey = device->mr->lkey;
regd_buf->reg.rkey = device->mr->rkey;
regd_buf->reg.len = ib_sg_dma_len(ibdev, &sg[0]);
regd_buf->reg.va = ib_sg_dma_address(ibdev, &sg[0]);
regd_buf->reg.is_mr = 0;
} else {
if (mem->dma_nents != 1 ||
scsi_get_prot_op(iser_task->sc) != SCSI_PROT_NORMAL) {
spin_lock_irqsave(&ib_conn->lock, flags);
desc = list_first_entry(&ib_conn->fastreg.frwr.pool,
desc = list_first_entry(&ib_conn->fastreg.pool,
struct fast_reg_descriptor, list);
list_del(&desc->list);
spin_unlock_irqrestore(&ib_conn->lock, flags);
page_list_len = iser_sg_to_page_vec(mem, device->ib_device,
desc->data_frpl->page_list,
&offset, &data_size);
regd_buf->reg.mem_h = desc;
}
if (page_list_len * SIZE_4K < data_size) {
iser_err("fast reg page_list too short to hold this SG\n");
err = -EINVAL;
goto err_reg;
err = iser_fast_reg_mr(iser_task, regd_buf, mem,
ISER_DATA_KEY_VALID, &data_sge);
if (err)
goto err_reg;
if (scsi_get_prot_op(iser_task->sc) != SCSI_PROT_NORMAL) {
struct ib_sge prot_sge, sig_sge;
memset(&prot_sge, 0, sizeof(prot_sge));
if (scsi_prot_sg_count(iser_task->sc)) {
mem = &iser_task->prot[cmd_dir];
aligned_len = iser_data_buf_aligned_len(mem, ibdev);
if (aligned_len != mem->dma_nents) {
err = fall_to_bounce_buf(iser_task, ibdev, mem,
&iser_task->prot_copy[cmd_dir],
cmd_dir, aligned_len);
if (err) {
iser_err("failed to allocate bounce buffer\n");
return err;
}
mem = &iser_task->prot_copy[cmd_dir];
}
err = iser_fast_reg_mr(iser_task, regd_buf, mem,
ISER_PROT_KEY_VALID, &prot_sge);
if (err)
goto err_reg;
}
err = iser_fast_reg_mr(desc, ib_conn, regd_buf,
offset, data_size, page_list_len);
if (err)
goto err_reg;
err = iser_reg_sig_mr(iser_task, desc, &data_sge,
&prot_sge, &sig_sge);
if (err) {
iser_err("Failed to register signature mr\n");
return err;
}
desc->reg_indicators |= ISER_FASTREG_PROTECTED;
regd_buf->reg.lkey = sig_sge.lkey;
regd_buf->reg.rkey = desc->pi_ctx->sig_mr->rkey;
regd_buf->reg.va = sig_sge.addr;
regd_buf->reg.len = sig_sge.length;
regd_buf->reg.is_mr = 1;
} else {
if (desc) {
regd_buf->reg.rkey = desc->data_mr->rkey;
regd_buf->reg.is_mr = 1;
} else {
regd_buf->reg.rkey = device->mr->rkey;
regd_buf->reg.is_mr = 0;
}
regd_buf->reg.lkey = data_sge.lkey;
regd_buf->reg.va = data_sge.addr;
regd_buf->reg.len = data_sge.length;
}
return 0;
err_reg:
spin_lock_irqsave(&ib_conn->lock, flags);
list_add_tail(&desc->list, &ib_conn->fastreg.frwr.pool);
spin_unlock_irqrestore(&ib_conn->lock, flags);
if (desc) {
spin_lock_irqsave(&ib_conn->lock, flags);
list_add_tail(&desc->list, &ib_conn->fastreg.pool);
spin_unlock_irqrestore(&ib_conn->lock, flags);
}
return err;
}

325
drivers/infiniband/ulp/iser/iser_verbs.c
Näytä tiedosto

@@ -1,7 +1,7 @@
/*
* Copyright (c) 2004, 2005, 2006 Voltaire, Inc. All rights reserved.
* Copyright (c) 2005, 2006 Cisco Systems. All rights reserved.
* Copyright (c) 2013 Mellanox Technologies. All rights reserved.
* Copyright (c) 2013-2014 Mellanox Technologies. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -71,17 +71,14 @@ static void iser_event_handler(struct ib_event_handler *handler,
*/
static int iser_create_device_ib_res(struct iser_device *device)
{
int i, j;
struct iser_cq_desc *cq_desc;
struct ib_device_attr *dev_attr;
struct ib_device_attr *dev_attr = &device->dev_attr;
int ret, i, j;
dev_attr = kmalloc(sizeof(*dev_attr), GFP_KERNEL);
if (!dev_attr)
return -ENOMEM;
if (ib_query_device(device->ib_device, dev_attr)) {
ret = ib_query_device(device->ib_device, dev_attr);
if (ret) {
pr_warn("Query device failed for %s\n", device->ib_device->name);
goto dev_attr_err;
return ret;
}
/* Assign function handles - based on FMR support */
@@ -94,14 +91,14 @@ static int iser_create_device_ib_res(struct iser_device *device)
device->iser_unreg_rdma_mem = iser_unreg_mem_fmr;
} else
if (dev_attr->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) {
iser_info("FRWR supported, using FRWR for registration\n");
device->iser_alloc_rdma_reg_res = iser_create_frwr_pool;
device->iser_free_rdma_reg_res = iser_free_frwr_pool;
device->iser_reg_rdma_mem = iser_reg_rdma_mem_frwr;
device->iser_unreg_rdma_mem = iser_unreg_mem_frwr;
iser_info("FastReg supported, using FastReg for registration\n");
device->iser_alloc_rdma_reg_res = iser_create_fastreg_pool;
device->iser_free_rdma_reg_res = iser_free_fastreg_pool;
device->iser_reg_rdma_mem = iser_reg_rdma_mem_fastreg;
device->iser_unreg_rdma_mem = iser_unreg_mem_fastreg;
} else {
iser_err("IB device does not support FMRs nor FRWRs, can't register memory\n");
goto dev_attr_err;
iser_err("IB device does not support FMRs nor FastRegs, can't register memory\n");
return -1;
}
device->cqs_used = min(ISER_MAX_CQ, device->ib_device->num_comp_vectors);
@@ -158,7 +155,6 @@ static int iser_create_device_ib_res(struct iser_device *device)
if (ib_register_event_handler(&device->event_handler))
goto handler_err;
kfree(dev_attr);
return 0;
handler_err:
@@ -178,8 +174,6 @@ pd_err:
kfree(device->cq_desc);
cq_desc_err:
iser_err("failed to allocate an IB resource\n");
dev_attr_err:
kfree(dev_attr);
return -1;
}
@@ -221,13 +215,13 @@ int iser_create_fmr_pool(struct iser_conn *ib_conn, unsigned cmds_max)
struct ib_fmr_pool_param params;
int ret = -ENOMEM;
ib_conn->fastreg.fmr.page_vec = kmalloc(sizeof(struct iser_page_vec) +
(sizeof(u64)*(ISCSI_ISER_SG_TABLESIZE + 1)),
GFP_KERNEL);
if (!ib_conn->fastreg.fmr.page_vec)
ib_conn->fmr.page_vec = kmalloc(sizeof(*ib_conn->fmr.page_vec) +
(sizeof(u64)*(ISCSI_ISER_SG_TABLESIZE + 1)),
GFP_KERNEL);
if (!ib_conn->fmr.page_vec)
return ret;
ib_conn->fastreg.fmr.page_vec->pages = (u64 *)(ib_conn->fastreg.fmr.page_vec + 1);
ib_conn->fmr.page_vec->pages = (u64 *)(ib_conn->fmr.page_vec + 1);
params.page_shift = SHIFT_4K;
/* when the first/last SG element are not start/end *
@@ -243,16 +237,16 @@ int iser_create_fmr_pool(struct iser_conn *ib_conn, unsigned cmds_max)
IB_ACCESS_REMOTE_WRITE |
IB_ACCESS_REMOTE_READ);
ib_conn->fastreg.fmr.pool = ib_create_fmr_pool(device->pd, &params);
if (!IS_ERR(ib_conn->fastreg.fmr.pool))
ib_conn->fmr.pool = ib_create_fmr_pool(device->pd, &params);
if (!IS_ERR(ib_conn->fmr.pool))
return 0;
/* no FMR => no need for page_vec */
kfree(ib_conn->fastreg.fmr.page_vec);
ib_conn->fastreg.fmr.page_vec = NULL;
kfree(ib_conn->fmr.page_vec);
ib_conn->fmr.page_vec = NULL;
ret = PTR_ERR(ib_conn->fastreg.fmr.pool);
ib_conn->fastreg.fmr.pool = NULL;
ret = PTR_ERR(ib_conn->fmr.pool);
ib_conn->fmr.pool = NULL;
if (ret != -ENOSYS) {
iser_err("FMR allocation failed, err %d\n", ret);
return ret;
@@ -268,93 +262,173 @@ int iser_create_fmr_pool(struct iser_conn *ib_conn, unsigned cmds_max)
void iser_free_fmr_pool(struct iser_conn *ib_conn)
{
iser_info("freeing conn %p fmr pool %p\n",
ib_conn, ib_conn->fastreg.fmr.pool);
ib_conn, ib_conn->fmr.pool);
if (ib_conn->fastreg.fmr.pool != NULL)
ib_destroy_fmr_pool(ib_conn->fastreg.fmr.pool);
if (ib_conn->fmr.pool != NULL)
ib_destroy_fmr_pool(ib_conn->fmr.pool);
ib_conn->fastreg.fmr.pool = NULL;
ib_conn->fmr.pool = NULL;
kfree(ib_conn->fastreg.fmr.page_vec);
ib_conn->fastreg.fmr.page_vec = NULL;
kfree(ib_conn->fmr.page_vec);
ib_conn->fmr.page_vec = NULL;
}
static int
iser_create_fastreg_desc(struct ib_device *ib_device, struct ib_pd *pd,
bool pi_enable, struct fast_reg_descriptor *desc)
{
int ret;
desc->data_frpl = ib_alloc_fast_reg_page_list(ib_device,
ISCSI_ISER_SG_TABLESIZE + 1);
if (IS_ERR(desc->data_frpl)) {
ret = PTR_ERR(desc->data_frpl);
iser_err("Failed to allocate ib_fast_reg_page_list err=%d\n",
ret);
return PTR_ERR(desc->data_frpl);
}
desc->data_mr = ib_alloc_fast_reg_mr(pd, ISCSI_ISER_SG_TABLESIZE + 1);
if (IS_ERR(desc->data_mr)) {
ret = PTR_ERR(desc->data_mr);
iser_err("Failed to allocate ib_fast_reg_mr err=%d\n", ret);
goto fast_reg_mr_failure;
}
desc->reg_indicators |= ISER_DATA_KEY_VALID;
if (pi_enable) {
struct ib_mr_init_attr mr_init_attr = {0};
struct iser_pi_context *pi_ctx = NULL;
desc->pi_ctx = kzalloc(sizeof(*desc->pi_ctx), GFP_KERNEL);
if (!desc->pi_ctx) {
iser_err("Failed to allocate pi context\n");
ret = -ENOMEM;
goto pi_ctx_alloc_failure;
}
pi_ctx = desc->pi_ctx;
pi_ctx->prot_frpl = ib_alloc_fast_reg_page_list(ib_device,
ISCSI_ISER_SG_TABLESIZE);
if (IS_ERR(pi_ctx->prot_frpl)) {
ret = PTR_ERR(pi_ctx->prot_frpl);
iser_err("Failed to allocate prot frpl ret=%d\n",
ret);
goto prot_frpl_failure;
}
pi_ctx->prot_mr = ib_alloc_fast_reg_mr(pd,
ISCSI_ISER_SG_TABLESIZE + 1);
if (IS_ERR(pi_ctx->prot_mr)) {
ret = PTR_ERR(pi_ctx->prot_mr);
iser_err("Failed to allocate prot frmr ret=%d\n",
ret);
goto prot_mr_failure;
}
desc->reg_indicators |= ISER_PROT_KEY_VALID;
mr_init_attr.max_reg_descriptors = 2;
mr_init_attr.flags |= IB_MR_SIGNATURE_EN;
pi_ctx->sig_mr = ib_create_mr(pd, &mr_init_attr);
if (IS_ERR(pi_ctx->sig_mr)) {
ret = PTR_ERR(pi_ctx->sig_mr);
iser_err("Failed to allocate signature enabled mr err=%d\n",
ret);
goto sig_mr_failure;
}
desc->reg_indicators |= ISER_SIG_KEY_VALID;
}
desc->reg_indicators &= ~ISER_FASTREG_PROTECTED;
iser_dbg("Create fr_desc %p page_list %p\n",
desc, desc->data_frpl->page_list);
return 0;
sig_mr_failure:
ib_dereg_mr(desc->pi_ctx->prot_mr);
prot_mr_failure:
ib_free_fast_reg_page_list(desc->pi_ctx->prot_frpl);
prot_frpl_failure:
kfree(desc->pi_ctx);
pi_ctx_alloc_failure:
ib_dereg_mr(desc->data_mr);
fast_reg_mr_failure:
ib_free_fast_reg_page_list(desc->data_frpl);
return ret;
}
/**
* iser_create_frwr_pool - Creates pool of fast_reg descriptors
* iser_create_fastreg_pool - Creates pool of fast_reg descriptors
* for fast registration work requests.
* returns 0 on success, or errno code on failure
*/
int iser_create_frwr_pool(struct iser_conn *ib_conn, unsigned cmds_max)
int iser_create_fastreg_pool(struct iser_conn *ib_conn, unsigned cmds_max)
{
struct iser_device *device = ib_conn->device;
struct fast_reg_descriptor *desc;
int i, ret;
INIT_LIST_HEAD(&ib_conn->fastreg.frwr.pool);
ib_conn->fastreg.frwr.pool_size = 0;
INIT_LIST_HEAD(&ib_conn->fastreg.pool);
ib_conn->fastreg.pool_size = 0;
for (i = 0; i < cmds_max; i++) {
desc = kmalloc(sizeof(*desc), GFP_KERNEL);
desc = kzalloc(sizeof(*desc), GFP_KERNEL);
if (!desc) {
iser_err("Failed to allocate a new fast_reg descriptor\n");
ret = -ENOMEM;
goto err;
}
desc->data_frpl = ib_alloc_fast_reg_page_list(device->ib_device,
ISCSI_ISER_SG_TABLESIZE + 1);
if (IS_ERR(desc->data_frpl)) {
ret = PTR_ERR(desc->data_frpl);
iser_err("Failed to allocate ib_fast_reg_page_list err=%d\n", ret);
goto fast_reg_page_failure;
ret = iser_create_fastreg_desc(device->ib_device, device->pd,
ib_conn->pi_support, desc);
if (ret) {
iser_err("Failed to create fastreg descriptor err=%d\n",
ret);
kfree(desc);
goto err;
}
desc->data_mr = ib_alloc_fast_reg_mr(device->pd,
ISCSI_ISER_SG_TABLESIZE + 1);
if (IS_ERR(desc->data_mr)) {
ret = PTR_ERR(desc->data_mr);
iser_err("Failed to allocate ib_fast_reg_mr err=%d\n", ret);
goto fast_reg_mr_failure;
}
desc->valid = true;
list_add_tail(&desc->list, &ib_conn->fastreg.frwr.pool);
ib_conn->fastreg.frwr.pool_size++;
list_add_tail(&desc->list, &ib_conn->fastreg.pool);
ib_conn->fastreg.pool_size++;
}
return 0;
fast_reg_mr_failure:
ib_free_fast_reg_page_list(desc->data_frpl);
fast_reg_page_failure:
kfree(desc);
err:
iser_free_frwr_pool(ib_conn);
iser_free_fastreg_pool(ib_conn);
return ret;
}
/**
* iser_free_frwr_pool - releases the pool of fast_reg descriptors
* iser_free_fastreg_pool - releases the pool of fast_reg descriptors
*/
void iser_free_frwr_pool(struct iser_conn *ib_conn)
void iser_free_fastreg_pool(struct iser_conn *ib_conn)
{
struct fast_reg_descriptor *desc, *tmp;
int i = 0;
if (list_empty(&ib_conn->fastreg.frwr.pool))
if (list_empty(&ib_conn->fastreg.pool))
return;
iser_info("freeing conn %p frwr pool\n", ib_conn);
iser_info("freeing conn %p fr pool\n", ib_conn);
list_for_each_entry_safe(desc, tmp, &ib_conn->fastreg.frwr.pool, list) {
list_for_each_entry_safe(desc, tmp, &ib_conn->fastreg.pool, list) {
list_del(&desc->list);
ib_free_fast_reg_page_list(desc->data_frpl);
ib_dereg_mr(desc->data_mr);
if (desc->pi_ctx) {
ib_free_fast_reg_page_list(desc->pi_ctx->prot_frpl);
ib_dereg_mr(desc->pi_ctx->prot_mr);
ib_destroy_mr(desc->pi_ctx->sig_mr);
kfree(desc->pi_ctx);
}
kfree(desc);
++i;
}
if (i < ib_conn->fastreg.frwr.pool_size)
if (i < ib_conn->fastreg.pool_size)
iser_warn("pool still has %d regions registered\n",
ib_conn->fastreg.frwr.pool_size - i);
ib_conn->fastreg.pool_size - i);
}
/**
@@ -389,12 +463,17 @@ static int iser_create_ib_conn_res(struct iser_conn *ib_conn)
init_attr.qp_context = (void *)ib_conn;
init_attr.send_cq = device->tx_cq[min_index];
init_attr.recv_cq = device->rx_cq[min_index];
init_attr.cap.max_send_wr = ISER_QP_MAX_REQ_DTOS;
init_attr.cap.max_recv_wr = ISER_QP_MAX_RECV_DTOS;
init_attr.cap.max_send_sge = 2;
init_attr.cap.max_recv_sge = 1;
init_attr.sq_sig_type = IB_SIGNAL_REQ_WR;
init_attr.qp_type = IB_QPT_RC;
if (ib_conn->pi_support) {
init_attr.cap.max_send_wr = ISER_QP_SIG_MAX_REQ_DTOS;
init_attr.create_flags |= IB_QP_CREATE_SIGNATURE_EN;
} else {
init_attr.cap.max_send_wr = ISER_QP_MAX_REQ_DTOS;
}
ret = rdma_create_qp(ib_conn->cma_id, device->pd, &init_attr);
if (ret)
@@ -591,6 +670,19 @@ static int iser_addr_handler(struct rdma_cm_id *cma_id)
ib_conn = (struct iser_conn *)cma_id->context;
ib_conn->device = device;
/* connection T10-PI support */
if (iser_pi_enable) {
if (!(device->dev_attr.device_cap_flags &
IB_DEVICE_SIGNATURE_HANDOVER)) {
iser_warn("T10-PI requested but not supported on %s, "
"continue without T10-PI\n",
ib_conn->device->ib_device->name);
ib_conn->pi_support = false;
} else {
ib_conn->pi_support = true;
}
}
ret = rdma_resolve_route(cma_id, 1000);
if (ret) {
iser_err("resolve route failed: %d\n", ret);
@@ -636,6 +728,11 @@ failure:
static void iser_connected_handler(struct rdma_cm_id *cma_id)
{
struct iser_conn *ib_conn;
struct ib_qp_attr attr;
struct ib_qp_init_attr init_attr;
(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);
ib_conn = (struct iser_conn *)cma_id->context;
ib_conn->state = ISER_CONN_UP;
@@ -653,9 +750,8 @@ static int iser_disconnected_handler(struct rdma_cm_id *cma_id)
* terminated asynchronously from the iSCSI layer's perspective. */
if (iser_conn_state_comp_exch(ib_conn, ISER_CONN_UP,
ISER_CONN_TERMINATING)){
if (ib_conn->iser_conn)
iscsi_conn_failure(ib_conn->iser_conn->iscsi_conn,
ISCSI_ERR_CONN_FAILED);
if (ib_conn->iscsi_conn)
iscsi_conn_failure(ib_conn->iscsi_conn, ISCSI_ERR_CONN_FAILED);
else
iser_err("iscsi_iser connection isn't bound\n");
}
@@ -801,7 +897,7 @@ int iser_reg_page_vec(struct iser_conn *ib_conn,
page_list = page_vec->pages;
io_addr = page_list[0];
mem = ib_fmr_pool_map_phys(ib_conn->fastreg.fmr.pool,
mem = ib_fmr_pool_map_phys(ib_conn->fmr.pool,
page_list,
page_vec->length,
io_addr);
@@ -855,11 +951,11 @@ void iser_unreg_mem_fmr(struct iscsi_iser_task *iser_task,
reg->mem_h = NULL;
}
void iser_unreg_mem_frwr(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir)
void iser_unreg_mem_fastreg(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir)
{
struct iser_mem_reg *reg = &iser_task->rdma_regd[cmd_dir].reg;
struct iser_conn *ib_conn = iser_task->iser_conn->ib_conn;
struct iser_conn *ib_conn = iser_task->ib_conn;
struct fast_reg_descriptor *desc = reg->mem_h;
if (!reg->is_mr)
@@ -868,7 +964,7 @@ void iser_unreg_mem_frwr(struct iscsi_iser_task *iser_task,
reg->mem_h = NULL;
reg->is_mr = 0;
spin_lock_bh(&ib_conn->lock);
list_add_tail(&desc->list, &ib_conn->fastreg.frwr.pool);
list_add_tail(&desc->list, &ib_conn->fastreg.pool);
spin_unlock_bh(&ib_conn->lock);
}
@@ -969,7 +1065,7 @@ static void iser_handle_comp_error(struct iser_tx_desc *desc,
* perspective. */
if (iser_conn_state_comp_exch(ib_conn, ISER_CONN_UP,
ISER_CONN_TERMINATING))
iscsi_conn_failure(ib_conn->iser_conn->iscsi_conn,
iscsi_conn_failure(ib_conn->iscsi_conn,
ISCSI_ERR_CONN_FAILED);
/* no more non completed posts to the QP, complete the
@@ -993,18 +1089,16 @@ static int iser_drain_tx_cq(struct iser_device *device, int cq_index)
if (wc.status == IB_WC_SUCCESS) {
if (wc.opcode == IB_WC_SEND)
iser_snd_completion(tx_desc, ib_conn);
else if (wc.opcode == IB_WC_LOCAL_INV ||
wc.opcode == IB_WC_FAST_REG_MR) {
atomic_dec(&ib_conn->post_send_buf_count);
continue;
} else
else
iser_err("expected opcode %d got %d\n",
IB_WC_SEND, wc.opcode);
} else {
iser_err("tx id %llx status %d vend_err %x\n",
wc.wr_id, wc.status, wc.vendor_err);
atomic_dec(&ib_conn->post_send_buf_count);
iser_handle_comp_error(tx_desc, ib_conn);
wc.wr_id, wc.status, wc.vendor_err);
if (wc.wr_id != ISER_FASTREG_LI_WRID) {
atomic_dec(&ib_conn->post_send_buf_count);
iser_handle_comp_error(tx_desc, ib_conn);
}
}
completed_tx++;
}
@@ -1022,8 +1116,12 @@ static void iser_cq_tasklet_fn(unsigned long data)
struct iser_rx_desc *desc;
unsigned long xfer_len;
struct iser_conn *ib_conn;
int completed_tx, completed_rx;
completed_tx = completed_rx = 0;
int completed_tx, completed_rx = 0;
/* First do tx drain, so in a case where we have rx flushes and a successful
* tx completion we will still go through completion error handling.
*/
completed_tx = iser_drain_tx_cq(device, cq_index);
while (ib_poll_cq(cq, 1, &wc) == 1) {
desc = (struct iser_rx_desc *) (unsigned long) wc.wr_id;
@@ -1051,7 +1149,6 @@ static void iser_cq_tasklet_fn(unsigned long data)
* " would not cause interrupts to be missed" */
ib_req_notify_cq(cq, IB_CQ_NEXT_COMP);
completed_tx += iser_drain_tx_cq(device, cq_index);
iser_dbg("got %d rx %d tx completions\n", completed_rx, completed_tx);
}
@@ -1063,3 +1160,51 @@ static void iser_cq_callback(struct ib_cq *cq, void *cq_context)
tasklet_schedule(&device->cq_tasklet[cq_index]);
}
u8 iser_check_task_pi_status(struct iscsi_iser_task *iser_task,
enum iser_data_dir cmd_dir, sector_t *sector)
{
struct iser_mem_reg *reg = &iser_task->rdma_regd[cmd_dir].reg;
struct fast_reg_descriptor *desc = reg->mem_h;
unsigned long sector_size = iser_task->sc->device->sector_size;
struct ib_mr_status mr_status;
int ret;
if (desc && desc->reg_indicators & ISER_FASTREG_PROTECTED) {
desc->reg_indicators &= ~ISER_FASTREG_PROTECTED;
ret = ib_check_mr_status(desc->pi_ctx->sig_mr,
IB_MR_CHECK_SIG_STATUS, &mr_status);
if (ret) {
pr_err("ib_check_mr_status failed, ret %d\n", ret);
goto err;
}
if (mr_status.fail_status & IB_MR_CHECK_SIG_STATUS) {
sector_t sector_off = mr_status.sig_err.sig_err_offset;
do_div(sector_off, sector_size + 8);
*sector = scsi_get_lba(iser_task->sc) + sector_off;
pr_err("PI error found type %d at sector %llx "
"expected %x vs actual %x\n",
mr_status.sig_err.err_type,
(unsigned long long)*sector,
mr_status.sig_err.expected,
mr_status.sig_err.actual);
switch (mr_status.sig_err.err_type) {
case IB_SIG_BAD_GUARD:
return 0x1;
case IB_SIG_BAD_REFTAG:
return 0x3;
case IB_SIG_BAD_APPTAG:
return 0x2;
}
}
}
return 0;
err:
/* Not alot we can do here, return ambiguous guard error */
return 0x1;
}

83
drivers/infiniband/ulp/srp/ib_srp.c
Näytä tiedosto

@@ -411,6 +411,8 @@ static void srp_path_rec_completion(int status,
static int srp_lookup_path(struct srp_target_port *target)
{
int ret;
target->path.numb_path = 1;
init_completion(&target->done);
@@ -431,7 +433,9 @@ static int srp_lookup_path(struct srp_target_port *target)
if (target->path_query_id < 0)
return target->path_query_id;
wait_for_completion(&target->done);
ret = wait_for_completion_interruptible(&target->done);
if (ret < 0)
return ret;
if (target->status < 0)
shost_printk(KERN_WARNING, target->scsi_host,
@@ -710,7 +714,9 @@ static int srp_connect_target(struct srp_target_port *target)
ret = srp_send_req(target);
if (ret)
return ret;
wait_for_completion(&target->done);
ret = wait_for_completion_interruptible(&target->done);
if (ret < 0)
return ret;
/*
* The CM event handling code will set status to
@@ -777,6 +783,7 @@ static void srp_unmap_data(struct scsi_cmnd *scmnd,
* srp_claim_req - Take ownership of the scmnd associated with a request.
* @target: SRP target port.
* @req: SRP request.
* @sdev: If not NULL, only take ownership for this SCSI device.
* @scmnd: If NULL, take ownership of @req->scmnd. If not NULL, only take
* ownership of @req->scmnd if it equals @scmnd.
*
@@ -785,16 +792,17 @@ static void srp_unmap_data(struct scsi_cmnd *scmnd,
*/
static struct scsi_cmnd *srp_claim_req(struct srp_target_port *target,
struct srp_request *req,
struct scsi_device *sdev,
struct scsi_cmnd *scmnd)
{
unsigned long flags;
spin_lock_irqsave(&target->lock, flags);
if (!scmnd) {
if (req->scmnd &&
(!sdev || req->scmnd->device == sdev) &&
(!scmnd || req->scmnd == scmnd)) {
scmnd = req->scmnd;
req->scmnd = NULL;
} else if (req->scmnd == scmnd) {
req->scmnd = NULL;
} else {
scmnd = NULL;
}
@@ -821,9 +829,10 @@ static void srp_free_req(struct srp_target_port *target,
}
static void srp_finish_req(struct srp_target_port *target,
struct srp_request *req, int result)
struct srp_request *req, struct scsi_device *sdev,
int result)
{
struct scsi_cmnd *scmnd = srp_claim_req(target, req, NULL);
struct scsi_cmnd *scmnd = srp_claim_req(target, req, sdev, NULL);
if (scmnd) {
srp_free_req(target, req, scmnd, 0);
@@ -835,11 +844,20 @@ static void srp_finish_req(struct srp_target_port *target,
static void srp_terminate_io(struct srp_rport *rport)
{
struct srp_target_port *target = rport->lld_data;
struct Scsi_Host *shost = target->scsi_host;
struct scsi_device *sdev;
int i;
/*
* Invoking srp_terminate_io() while srp_queuecommand() is running
* is not safe. Hence the warning statement below.
*/
shost_for_each_device(sdev, shost)
WARN_ON_ONCE(sdev->request_queue->request_fn_active);
for (i = 0; i < target->req_ring_size; ++i) {
struct srp_request *req = &target->req_ring[i];
srp_finish_req(target, req, DID_TRANSPORT_FAILFAST << 16);
srp_finish_req(target, req, NULL, DID_TRANSPORT_FAILFAST << 16);
}
}
@@ -876,7 +894,7 @@ static int srp_rport_reconnect(struct srp_rport *rport)
for (i = 0; i < target->req_ring_size; ++i) {
struct srp_request *req = &target->req_ring[i];
srp_finish_req(target, req, DID_RESET << 16);
srp_finish_req(target, req, NULL, DID_RESET << 16);
}
INIT_LIST_HEAD(&target->free_tx);
@@ -1284,7 +1302,7 @@ static void srp_process_rsp(struct srp_target_port *target, struct srp_rsp *rsp)
complete(&target->tsk_mgmt_done);
} else {
req = &target->req_ring[rsp->tag];
scmnd = srp_claim_req(target, req, NULL);
scmnd = srp_claim_req(target, req, NULL, NULL);
if (!scmnd) {
shost_printk(KERN_ERR, target->scsi_host,
"Null scmnd for RSP w/tag %016llx\n",
@@ -1804,8 +1822,10 @@ static void srp_cm_rej_handler(struct ib_cm_id *cm_id,
shost_printk(KERN_WARNING, shost,
PFX "SRP_LOGIN_REJ: requested max_it_iu_len too large\n");
else
shost_printk(KERN_WARNING, shost,
PFX "SRP LOGIN REJECTED, reason 0x%08x\n", reason);
shost_printk(KERN_WARNING, shost, PFX
"SRP LOGIN from %pI6 to %pI6 REJECTED, reason 0x%08x\n",
target->path.sgid.raw,
target->orig_dgid, reason);
} else
shost_printk(KERN_WARNING, shost,
" REJ reason: IB_CM_REJ_CONSUMER_DEFINED,"
@@ -1863,6 +1883,7 @@ static int srp_cm_handler(struct ib_cm_id *cm_id, struct ib_cm_event *event)
case IB_CM_TIMEWAIT_EXIT:
shost_printk(KERN_ERR, target->scsi_host,
PFX "connection closed\n");
comp = 1;
target->status = 0;
break;
@@ -1999,7 +2020,7 @@ static int srp_abort(struct scsi_cmnd *scmnd)
shost_printk(KERN_ERR, target->scsi_host, "SRP abort called\n");
if (!req || !srp_claim_req(target, req, scmnd))
if (!req || !srp_claim_req(target, req, NULL, scmnd))
return SUCCESS;
if (srp_send_tsk_mgmt(target, req->index, scmnd->device->lun,
SRP_TSK_ABORT_TASK) == 0)
@@ -2030,8 +2051,7 @@ static int srp_reset_device(struct scsi_cmnd *scmnd)
for (i = 0; i < target->req_ring_size; ++i) {
struct srp_request *req = &target->req_ring[i];
if (req->scmnd && req->scmnd->device == scmnd->device)
srp_finish_req(target, req, DID_RESET << 16);
srp_finish_req(target, req, scmnd->device, DID_RESET << 16);
}
return SUCCESS;
@@ -2612,6 +2632,8 @@ static ssize_t srp_create_target(struct device *dev,
target->tl_retry_count = 7;
target->queue_size = SRP_DEFAULT_QUEUE_SIZE;
mutex_lock(&host->add_target_mutex);
ret = srp_parse_options(buf, target);
if (ret)
goto err;
@@ -2649,16 +2671,9 @@ static ssize_t srp_create_target(struct device *dev,
if (ret)
goto err_free_mem;
ib_query_gid(ibdev, host->port, 0, &target->path.sgid);
shost_printk(KERN_DEBUG, target->scsi_host, PFX
"new target: id_ext %016llx ioc_guid %016llx pkey %04x "
"service_id %016llx dgid %pI6\n",
(unsigned long long) be64_to_cpu(target->id_ext),
(unsigned long long) be64_to_cpu(target->ioc_guid),
be16_to_cpu(target->path.pkey),
(unsigned long long) be64_to_cpu(target->service_id),
target->path.dgid.raw);
ret = ib_query_gid(ibdev, host->port, 0, &target->path.sgid);
if (ret)
goto err_free_mem;
ret = srp_create_target_ib(target);
if (ret)
@@ -2679,7 +2694,19 @@ static ssize_t srp_create_target(struct device *dev,
if (ret)
goto err_disconnect;
return count;
shost_printk(KERN_DEBUG, target->scsi_host, PFX
"new target: id_ext %016llx ioc_guid %016llx pkey %04x service_id %016llx sgid %pI6 dgid %pI6\n",
be64_to_cpu(target->id_ext),
be64_to_cpu(target->ioc_guid),
be16_to_cpu(target->path.pkey),
be64_to_cpu(target->service_id),
target->path.sgid.raw, target->path.dgid.raw);
ret = count;
out:
mutex_unlock(&host->add_target_mutex);
return ret;
err_disconnect:
srp_disconnect_target(target);
@@ -2695,8 +2722,7 @@ err_free_mem:
err:
scsi_host_put(target_host);
return ret;
goto out;
}
static DEVICE_ATTR(add_target, S_IWUSR, NULL, srp_create_target);
@@ -2732,6 +2758,7 @@ static struct srp_host *srp_add_port(struct srp_device *device, u8 port)
INIT_LIST_HEAD(&host->target_list);
spin_lock_init(&host->target_lock);
init_completion(&host->released);
mutex_init(&host->add_target_mutex);
host->srp_dev = device;
host->port = port;

1
drivers/infiniband/ulp/srp/ib_srp.h
Näytä tiedosto

@@ -105,6 +105,7 @@ struct srp_host {
spinlock_t target_lock;
struct completion released;
struct list_head list;
struct mutex add_target_mutex;
};
struct srp_request {