int phys_mem_open(struct inode *inode, struct file *filp)
{
struct phys_mem_dev *dev; /* device information */
struct phys_mem_session *session; /* the to-be created session */
/* Find the device */
dev = container_of(inode->i_cdev, struct phys_mem_dev, cdev);
session = kmem_cache_alloc(session_mem_cache, GFP_KERNEL);
if (!session)
return -ENOMEM;
session->session_id = atomic64_add_return(1, &session_counter);
session->device = dev;
sema_init(&session->sem, 1);
session->vmas = 0;
session->num_frame_stati = 0;
session->frame_stati = NULL;
session->status.state = SESSION_STATE_INVALID;
SET_STATE(session, SESSION_STATE_OPEN);
/* and use filp->private_data to point to the device data */
filp->private_data = session;
return 0; /* success */
}
static inline bool nft_overquota(struct nft_quota *priv,
const struct sk_buff *skb)
{
return atomic64_add_return(skb->len, &priv->consumed) >= priv->quota;
}
u64 notrace trace_clock_counter(void)
{
return atomic64_add_return(1, &trace_counter);
}
static __init int test_atomic64(void)
{
long long v0 = 0xaaa31337c001d00dLL;
long long v1 = 0xdeadbeefdeafcafeLL;
long long v2 = 0xfaceabadf00df001LL;
long long onestwos = 0x1111111122222222LL;
long long one = 1LL;
atomic64_t v = ATOMIC64_INIT(v0);
long long r = v0;
BUG_ON(v.counter != r);
atomic64_set(&v, v1);
r = v1;
BUG_ON(v.counter != r);
BUG_ON(atomic64_read(&v) != r);
INIT(v0);
atomic64_add(onestwos, &v);
r += onestwos;
BUG_ON(v.counter != r);
INIT(v0);
atomic64_add(-one, &v);
r += -one;
BUG_ON(v.counter != r);
INIT(v0);
r += onestwos;
BUG_ON(atomic64_add_return(onestwos, &v) != r);
BUG_ON(v.counter != r);
INIT(v0);
r += -one;
BUG_ON(atomic64_add_return(-one, &v) != r);
BUG_ON(v.counter != r);
INIT(v0);
atomic64_sub(onestwos, &v);
r -= onestwos;
BUG_ON(v.counter != r);
INIT(v0);
atomic64_sub(-one, &v);
r -= -one;
BUG_ON(v.counter != r);
INIT(v0);
r -= onestwos;
BUG_ON(atomic64_sub_return(onestwos, &v) != r);
BUG_ON(v.counter != r);
INIT(v0);
r -= -one;
BUG_ON(atomic64_sub_return(-one, &v) != r);
BUG_ON(v.counter != r);
INIT(v0);
atomic64_inc(&v);
r += one;
BUG_ON(v.counter != r);
INIT(v0);
r += one;
BUG_ON(atomic64_inc_return(&v) != r);
BUG_ON(v.counter != r);
INIT(v0);
atomic64_dec(&v);
r -= one;
BUG_ON(v.counter != r);
INIT(v0);
r -= one;
BUG_ON(atomic64_dec_return(&v) != r);
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(atomic64_xchg(&v, v1) != v0);
r = v1;
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(atomic64_cmpxchg(&v, v0, v1) != v0);
r = v1;
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(atomic64_cmpxchg(&v, v2, v1) != v0);
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(atomic64_add_unless(&v, one, v0));
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(!atomic64_add_unless(&v, one, v1));
r += one;
BUG_ON(v.counter != r);
#ifdef CONFIG_ARCH_HAS_ATOMIC64_DEC_IF_POSITIVE
INIT(onestwos);
BUG_ON(atomic64_dec_if_positive(&v) != (onestwos - 1));
r -= one;
BUG_ON(v.counter != r);
INIT(0);
BUG_ON(atomic64_dec_if_positive(&v) != -one);
BUG_ON(v.counter != r);
INIT(-one);
BUG_ON(atomic64_dec_if_positive(&v) != (-one - one));
BUG_ON(v.counter != r);
#else
#warning Please implement atomic64_dec_if_positive for your architecture and select the above Kconfig symbol
#endif
INIT(onestwos);
BUG_ON(!atomic64_inc_not_zero(&v));
r += one;
BUG_ON(v.counter != r);
INIT(0);
BUG_ON(atomic64_inc_not_zero(&v));
BUG_ON(v.counter != r);
INIT(-one);
BUG_ON(!atomic64_inc_not_zero(&v));
r += one;
BUG_ON(v.counter != r);
#ifdef CONFIG_X86
pr_info("passed for %s platform %s CX8 and %s SSE\n",
#ifdef CONFIG_X86_64
"x86-64",
#elif defined(CONFIG_X86_CMPXCHG64)
"i586+",
#else
"i386+",
#endif
boot_cpu_has(X86_FEATURE_CX8) ? "with" : "without",
boot_cpu_has(X86_FEATURE_XMM) ? "with" : "without");
#else
pr_info("passed\n");
#endif
return 0;
}
/**
* ipath_ruc_loopback - handle UC and RC lookback requests
* @sqp: the sending QP
*
* This is called from ipath_do_send() to
* forward a WQE addressed to the same HCA.
* Note that although we are single threaded due to the tasklet, we still
* have to protect against post_send(). We don't have to worry about
* receive interrupts since this is a connected protocol and all packets
* will pass through here.
*/
static void ipath_ruc_loopback(struct ipath_qp *sqp)
{
struct ipath_ibdev *dev = to_idev(sqp->ibqp.device);
struct ipath_qp *qp;
struct ipath_swqe *wqe;
struct ipath_sge *sge;
unsigned long flags;
struct ib_wc wc;
u64 sdata;
atomic64_t *maddr;
enum ib_wc_status send_status;
/*
* Note that we check the responder QP state after
* checking the requester's state.
*/
qp = ipath_lookup_qpn(&dev->qp_table, sqp->remote_qpn);
spin_lock_irqsave(&sqp->s_lock, flags);
/* Return if we are already busy processing a work request. */
if ((sqp->s_flags & (IPATH_S_BUSY | IPATH_S_ANY_WAIT)) ||
!(ib_ipath_state_ops[sqp->state] & IPATH_PROCESS_OR_FLUSH_SEND))
goto unlock;
sqp->s_flags |= IPATH_S_BUSY;
again:
if (sqp->s_last == sqp->s_head)
goto clr_busy;
wqe = get_swqe_ptr(sqp, sqp->s_last);
/* Return if it is not OK to start a new work reqeust. */
if (!(ib_ipath_state_ops[sqp->state] & IPATH_PROCESS_NEXT_SEND_OK)) {
if (!(ib_ipath_state_ops[sqp->state] & IPATH_FLUSH_SEND))
goto clr_busy;
/* We are in the error state, flush the work request. */
send_status = IB_WC_WR_FLUSH_ERR;
goto flush_send;
}
/*
* We can rely on the entry not changing without the s_lock
* being held until we update s_last.
* We increment s_cur to indicate s_last is in progress.
*/
if (sqp->s_last == sqp->s_cur) {
if (++sqp->s_cur >= sqp->s_size)
sqp->s_cur = 0;
}
spin_unlock_irqrestore(&sqp->s_lock, flags);
if (!qp || !(ib_ipath_state_ops[qp->state] & IPATH_PROCESS_RECV_OK)) {
dev->n_pkt_drops++;
/*
* For RC, the requester would timeout and retry so
* shortcut the timeouts and just signal too many retries.
*/
if (sqp->ibqp.qp_type == IB_QPT_RC)
send_status = IB_WC_RETRY_EXC_ERR;
else
send_status = IB_WC_SUCCESS;
goto serr;
}
memset(&wc, 0, sizeof wc);
send_status = IB_WC_SUCCESS;
sqp->s_sge.sge = wqe->sg_list[0];
sqp->s_sge.sg_list = wqe->sg_list + 1;
sqp->s_sge.num_sge = wqe->wr.num_sge;
sqp->s_len = wqe->length;
switch (wqe->wr.opcode) {
case IB_WR_SEND_WITH_IMM:
wc.wc_flags = IB_WC_WITH_IMM;
wc.imm_data = wqe->wr.ex.imm_data;
/* FALLTHROUGH */
case IB_WR_SEND:
if (!ipath_get_rwqe(qp, 0))
goto rnr_nak;
break;
case IB_WR_RDMA_WRITE_WITH_IMM:
if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_WRITE)))
goto inv_err;
wc.wc_flags = IB_WC_WITH_IMM;
wc.imm_data = wqe->wr.ex.imm_data;
if (!ipath_get_rwqe(qp, 1))
goto rnr_nak;
/* FALLTHROUGH */
case IB_WR_RDMA_WRITE:
if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_WRITE)))
goto inv_err;
if (wqe->length == 0)
break;
if (unlikely(!ipath_rkey_ok(qp, &qp->r_sge, wqe->length,
wqe->wr.wr.rdma.remote_addr,
wqe->wr.wr.rdma.rkey,
IB_ACCESS_REMOTE_WRITE)))
goto acc_err;
break;
case IB_WR_RDMA_READ:
if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_READ)))
goto inv_err;
if (unlikely(!ipath_rkey_ok(qp, &sqp->s_sge, wqe->length,
wqe->wr.wr.rdma.remote_addr,
wqe->wr.wr.rdma.rkey,
IB_ACCESS_REMOTE_READ)))
goto acc_err;
qp->r_sge.sge = wqe->sg_list[0];
qp->r_sge.sg_list = wqe->sg_list + 1;
qp->r_sge.num_sge = wqe->wr.num_sge;
break;
case IB_WR_ATOMIC_CMP_AND_SWP:
case IB_WR_ATOMIC_FETCH_AND_ADD:
if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_ATOMIC)))
goto inv_err;
if (unlikely(!ipath_rkey_ok(qp, &qp->r_sge, sizeof(u64),
wqe->wr.wr.atomic.remote_addr,
wqe->wr.wr.atomic.rkey,
IB_ACCESS_REMOTE_ATOMIC)))
goto acc_err;
/* Perform atomic OP and save result. */
maddr = (atomic64_t *) qp->r_sge.sge.vaddr;
sdata = wqe->wr.wr.atomic.compare_add;
*(u64 *) sqp->s_sge.sge.vaddr =
(wqe->wr.opcode == IB_WR_ATOMIC_FETCH_AND_ADD) ?
(u64) atomic64_add_return(sdata, maddr) - sdata :
(u64) cmpxchg((u64 *) qp->r_sge.sge.vaddr,
sdata, wqe->wr.wr.atomic.swap);
goto send_comp;
default:
send_status = IB_WC_LOC_QP_OP_ERR;
goto serr;
}
sge = &sqp->s_sge.sge;
while (sqp->s_len) {
u32 len = sqp->s_len;
if (len > sge->length)
len = sge->length;
if (len > sge->sge_length)
len = sge->sge_length;
BUG_ON(len == 0);
ipath_copy_sge(&qp->r_sge, sge->vaddr, len);
sge->vaddr += len;
sge->length -= len;
sge->sge_length -= len;
if (sge->sge_length == 0) {
if (--sqp->s_sge.num_sge)
*sge = *sqp->s_sge.sg_list++;
} else if (sge->length == 0 && sge->mr != NULL) {
if (++sge->n >= IPATH_SEGSZ) {
if (++sge->m >= sge->mr->mapsz)
break;
sge->n = 0;
}
sge->vaddr =
sge->mr->map[sge->m]->segs[sge->n].vaddr;
sge->length =
sge->mr->map[sge->m]->segs[sge->n].length;
}
sqp->s_len -= len;
}
if (!test_and_clear_bit(IPATH_R_WRID_VALID, &qp->r_aflags))
goto send_comp;
if (wqe->wr.opcode == IB_WR_RDMA_WRITE_WITH_IMM)
wc.opcode = IB_WC_RECV_RDMA_WITH_IMM;
else
wc.opcode = IB_WC_RECV;
wc.wr_id = qp->r_wr_id;
wc.status = IB_WC_SUCCESS;
wc.byte_len = wqe->length;
wc.qp = &qp->ibqp;
wc.src_qp = qp->remote_qpn;
wc.slid = qp->remote_ah_attr.dlid;
wc.sl = qp->remote_ah_attr.sl;
wc.port_num = 1;
/* Signal completion event if the solicited bit is set. */
ipath_cq_enter(to_icq(qp->ibqp.recv_cq), &wc,
wqe->wr.send_flags & IB_SEND_SOLICITED);
send_comp:
spin_lock_irqsave(&sqp->s_lock, flags);
flush_send:
sqp->s_rnr_retry = sqp->s_rnr_retry_cnt;
ipath_send_complete(sqp, wqe, send_status);
goto again;
rnr_nak:
/* Handle RNR NAK */
if (qp->ibqp.qp_type == IB_QPT_UC)
goto send_comp;
/*
* Note: we don't need the s_lock held since the BUSY flag
* makes this single threaded.
*/
if (sqp->s_rnr_retry == 0) {
send_status = IB_WC_RNR_RETRY_EXC_ERR;
goto serr;
}
if (sqp->s_rnr_retry_cnt < 7)
sqp->s_rnr_retry--;
spin_lock_irqsave(&sqp->s_lock, flags);
if (!(ib_ipath_state_ops[sqp->state] & IPATH_PROCESS_RECV_OK))
goto clr_busy;
sqp->s_flags |= IPATH_S_WAITING;
dev->n_rnr_naks++;
sqp->s_rnr_timeout = ib_ipath_rnr_table[qp->r_min_rnr_timer];
ipath_insert_rnr_queue(sqp);
goto clr_busy;
inv_err:
send_status = IB_WC_REM_INV_REQ_ERR;
wc.status = IB_WC_LOC_QP_OP_ERR;
goto err;
acc_err:
send_status = IB_WC_REM_ACCESS_ERR;
wc.status = IB_WC_LOC_PROT_ERR;
err:
/* responder goes to error state */
ipath_rc_error(qp, wc.status);
serr:
spin_lock_irqsave(&sqp->s_lock, flags);
ipath_send_complete(sqp, wqe, send_status);
if (sqp->ibqp.qp_type == IB_QPT_RC) {
int lastwqe = ipath_error_qp(sqp, IB_WC_WR_FLUSH_ERR);
sqp->s_flags &= ~IPATH_S_BUSY;
spin_unlock_irqrestore(&sqp->s_lock, flags);
if (lastwqe) {
struct ib_event ev;
ev.device = sqp->ibqp.device;
ev.element.qp = &sqp->ibqp;
ev.event = IB_EVENT_QP_LAST_WQE_REACHED;
sqp->ibqp.event_handler(&ev, sqp->ibqp.qp_context);
}
goto done;
}
clr_busy:
sqp->s_flags &= ~IPATH_S_BUSY;
unlock:
spin_unlock_irqrestore(&sqp->s_lock, flags);
done:
if (qp && atomic_dec_and_test(&qp->refcount))
wake_up(&qp->wait);
}
static __init int test_atomic64(void)
{
long long v0 = 0xaaa31337c001d00dLL;
long long v1 = 0xdeadbeefdeafcafeLL;
long long v2 = 0xfaceabadf00df001LL;
long long onestwos = 0x1111111122222222LL;
long long one = 1LL;
atomic64_t v = ATOMIC64_INIT(v0);
long long r = v0;
BUG_ON(v.counter != r);
atomic64_set(&v, v1);
r = v1;
BUG_ON(v.counter != r);
BUG_ON(atomic64_read(&v) != r);
INIT(v0);
atomic64_add(onestwos, &v);
r += onestwos;
BUG_ON(v.counter != r);
INIT(v0);
atomic64_add(-one, &v);
r += -one;
BUG_ON(v.counter != r);
INIT(v0);
r += onestwos;
BUG_ON(atomic64_add_return(onestwos, &v) != r);
BUG_ON(v.counter != r);
INIT(v0);
r += -one;
BUG_ON(atomic64_add_return(-one, &v) != r);
BUG_ON(v.counter != r);
INIT(v0);
atomic64_sub(onestwos, &v);
r -= onestwos;
BUG_ON(v.counter != r);
INIT(v0);
atomic64_sub(-one, &v);
r -= -one;
BUG_ON(v.counter != r);
INIT(v0);
r -= onestwos;
BUG_ON(atomic64_sub_return(onestwos, &v) != r);
BUG_ON(v.counter != r);
INIT(v0);
r -= -one;
BUG_ON(atomic64_sub_return(-one, &v) != r);
BUG_ON(v.counter != r);
INIT(v0);
atomic64_inc(&v);
r += one;
BUG_ON(v.counter != r);
INIT(v0);
r += one;
BUG_ON(atomic64_inc_return(&v) != r);
BUG_ON(v.counter != r);
INIT(v0);
atomic64_dec(&v);
r -= one;
BUG_ON(v.counter != r);
INIT(v0);
r -= one;
BUG_ON(atomic64_dec_return(&v) != r);
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(atomic64_xchg(&v, v1) != v0);
r = v1;
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(atomic64_cmpxchg(&v, v0, v1) != v0);
r = v1;
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(atomic64_cmpxchg(&v, v2, v1) != v0);
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(atomic64_add_unless(&v, one, v0));
BUG_ON(v.counter != r);
INIT(v0);
BUG_ON(!atomic64_add_unless(&v, one, v1));
r += one;
BUG_ON(v.counter != r);
#if defined(CONFIG_X86) || defined(CONFIG_MIPS) || defined(CONFIG_PPC) || \
defined(CONFIG_S390) || defined(_ASM_GENERIC_ATOMIC64_H) || defined(CONFIG_ARM)
INIT(onestwos);
BUG_ON(atomic64_dec_if_positive(&v) != (onestwos - 1));
r -= one;
BUG_ON(v.counter != r);
INIT(0);
BUG_ON(atomic64_dec_if_positive(&v) != -one);
BUG_ON(v.counter != r);
INIT(-one);
BUG_ON(atomic64_dec_if_positive(&v) != (-one - one));
BUG_ON(v.counter != r);
#else
#warning Please implement atomic64_dec_if_positive for your architecture, and add it to the IF above
#endif
INIT(onestwos);
BUG_ON(!atomic64_inc_not_zero(&v));
r += one;
BUG_ON(v.counter != r);
INIT(0);
BUG_ON(atomic64_inc_not_zero(&v));
BUG_ON(v.counter != r);
INIT(-one);
BUG_ON(!atomic64_inc_not_zero(&v));
r += one;
BUG_ON(v.counter != r);
#ifdef CONFIG_X86
#ifdef CONFIG_DEBUG_PRINTK
printk(KERN_INFO "atomic64 test passed for %s platform %s CX8 and %s SSE\n",
#ifdef CONFIG_X86_64
"x86-64",
#elif defined(CONFIG_X86_CMPXCHG64)
"i586+",
#else
"i386+",
#endif
boot_cpu_has(X86_FEATURE_CX8) ? "with" : "without",
boot_cpu_has(X86_FEATURE_XMM) ? "with" : "without");
#else
;
#endif
#else
#ifdef CONFIG_DEBUG_PRINTK
printk(KERN_INFO "atomic64 test passed\n");
#else
;
#endif
#endif
return 0;
}
/**
* ipath_ruc_loopback - handle UC and RC lookback requests
* @sqp: the loopback QP
*
* This is called from ipath_do_uc_send() or ipath_do_rc_send() to
* forward a WQE addressed to the same HCA.
* Note that although we are single threaded due to the tasklet, we still
* have to protect against post_send(). We don't have to worry about
* receive interrupts since this is a connected protocol and all packets
* will pass through here.
*/
static void ipath_ruc_loopback(struct ipath_qp *sqp)
{
struct ipath_ibdev *dev = to_idev(sqp->ibqp.device);
struct ipath_qp *qp;
struct ipath_swqe *wqe;
struct ipath_sge *sge;
unsigned long flags;
struct ib_wc wc;
u64 sdata;
atomic64_t *maddr;
qp = ipath_lookup_qpn(&dev->qp_table, sqp->remote_qpn);
if (!qp) {
dev->n_pkt_drops++;
return;
}
again:
spin_lock_irqsave(&sqp->s_lock, flags);
if (!(ib_ipath_state_ops[sqp->state] & IPATH_PROCESS_SEND_OK) ||
qp->s_rnr_timeout) {
spin_unlock_irqrestore(&sqp->s_lock, flags);
goto done;
}
/* Get the next send request. */
if (sqp->s_last == sqp->s_head) {
/* Send work queue is empty. */
spin_unlock_irqrestore(&sqp->s_lock, flags);
goto done;
}
/*
* We can rely on the entry not changing without the s_lock
* being held until we update s_last.
*/
wqe = get_swqe_ptr(sqp, sqp->s_last);
spin_unlock_irqrestore(&sqp->s_lock, flags);
wc.wc_flags = 0;
wc.imm_data = 0;
sqp->s_sge.sge = wqe->sg_list[0];
sqp->s_sge.sg_list = wqe->sg_list + 1;
sqp->s_sge.num_sge = wqe->wr.num_sge;
sqp->s_len = wqe->length;
switch (wqe->wr.opcode) {
case IB_WR_SEND_WITH_IMM:
wc.wc_flags = IB_WC_WITH_IMM;
wc.imm_data = wqe->wr.imm_data;
/* FALLTHROUGH */
case IB_WR_SEND:
if (!ipath_get_rwqe(qp, 0)) {
rnr_nak:
/* Handle RNR NAK */
if (qp->ibqp.qp_type == IB_QPT_UC)
goto send_comp;
if (sqp->s_rnr_retry == 0) {
wc.status = IB_WC_RNR_RETRY_EXC_ERR;
goto err;
}
if (sqp->s_rnr_retry_cnt < 7)
sqp->s_rnr_retry--;
dev->n_rnr_naks++;
sqp->s_rnr_timeout =
ib_ipath_rnr_table[qp->r_min_rnr_timer];
ipath_insert_rnr_queue(sqp);
goto done;
}
break;
case IB_WR_RDMA_WRITE_WITH_IMM:
wc.wc_flags = IB_WC_WITH_IMM;
wc.imm_data = wqe->wr.imm_data;
if (!ipath_get_rwqe(qp, 1))
goto rnr_nak;
/* FALLTHROUGH */
case IB_WR_RDMA_WRITE:
if (wqe->length == 0)
break;
if (unlikely(!ipath_rkey_ok(qp, &qp->r_sge, wqe->length,
wqe->wr.wr.rdma.remote_addr,
wqe->wr.wr.rdma.rkey,
IB_ACCESS_REMOTE_WRITE))) {
acc_err:
wc.status = IB_WC_REM_ACCESS_ERR;
err:
wc.wr_id = wqe->wr.wr_id;
wc.opcode = ib_ipath_wc_opcode[wqe->wr.opcode];
wc.vendor_err = 0;
wc.byte_len = 0;
wc.qp = &sqp->ibqp;
wc.src_qp = sqp->remote_qpn;
wc.pkey_index = 0;
wc.slid = sqp->remote_ah_attr.dlid;
wc.sl = sqp->remote_ah_attr.sl;
wc.dlid_path_bits = 0;
wc.port_num = 0;
spin_lock_irqsave(&sqp->s_lock, flags);
ipath_sqerror_qp(sqp, &wc);
spin_unlock_irqrestore(&sqp->s_lock, flags);
goto done;
}
break;
case IB_WR_RDMA_READ:
if (unlikely(!(qp->qp_access_flags &
IB_ACCESS_REMOTE_READ)))
goto acc_err;
if (unlikely(!ipath_rkey_ok(qp, &sqp->s_sge, wqe->length,
wqe->wr.wr.rdma.remote_addr,
wqe->wr.wr.rdma.rkey,
IB_ACCESS_REMOTE_READ)))
goto acc_err;
qp->r_sge.sge = wqe->sg_list[0];
qp->r_sge.sg_list = wqe->sg_list + 1;
qp->r_sge.num_sge = wqe->wr.num_sge;
break;
case IB_WR_ATOMIC_CMP_AND_SWP:
case IB_WR_ATOMIC_FETCH_AND_ADD:
if (unlikely(!(qp->qp_access_flags &
IB_ACCESS_REMOTE_ATOMIC)))
goto acc_err;
if (unlikely(!ipath_rkey_ok(qp, &qp->r_sge, sizeof(u64),
wqe->wr.wr.atomic.remote_addr,
wqe->wr.wr.atomic.rkey,
IB_ACCESS_REMOTE_ATOMIC)))
goto acc_err;
/* Perform atomic OP and save result. */
maddr = (atomic64_t *) qp->r_sge.sge.vaddr;
sdata = wqe->wr.wr.atomic.compare_add;
*(u64 *) sqp->s_sge.sge.vaddr =
(wqe->wr.opcode == IB_WR_ATOMIC_FETCH_AND_ADD) ?
(u64) atomic64_add_return(sdata, maddr) - sdata :
(u64) cmpxchg((u64 *) qp->r_sge.sge.vaddr,
sdata, wqe->wr.wr.atomic.swap);
goto send_comp;
default:
goto done;
}
sge = &sqp->s_sge.sge;
while (sqp->s_len) {
u32 len = sqp->s_len;
if (len > sge->length)
len = sge->length;
BUG_ON(len == 0);
ipath_copy_sge(&qp->r_sge, sge->vaddr, len);
sge->vaddr += len;
sge->length -= len;
sge->sge_length -= len;
if (sge->sge_length == 0) {
if (--sqp->s_sge.num_sge)
*sge = *sqp->s_sge.sg_list++;
} else if (sge->length == 0 && sge->mr != NULL) {
if (++sge->n >= IPATH_SEGSZ) {
if (++sge->m >= sge->mr->mapsz)
break;
sge->n = 0;
}
sge->vaddr =
sge->mr->map[sge->m]->segs[sge->n].vaddr;
sge->length =
sge->mr->map[sge->m]->segs[sge->n].length;
}
sqp->s_len -= len;
}
if (wqe->wr.opcode == IB_WR_RDMA_WRITE ||
wqe->wr.opcode == IB_WR_RDMA_READ)
goto send_comp;
if (wqe->wr.opcode == IB_WR_RDMA_WRITE_WITH_IMM)
wc.opcode = IB_WC_RECV_RDMA_WITH_IMM;
else
wc.opcode = IB_WC_RECV;
wc.wr_id = qp->r_wr_id;
wc.status = IB_WC_SUCCESS;
wc.vendor_err = 0;
wc.byte_len = wqe->length;
wc.qp = &qp->ibqp;
wc.src_qp = qp->remote_qpn;
/* XXX do we know which pkey matched? Only needed for GSI. */
wc.pkey_index = 0;
wc.slid = qp->remote_ah_attr.dlid;
wc.sl = qp->remote_ah_attr.sl;
wc.dlid_path_bits = 0;
/* Signal completion event if the solicited bit is set. */
ipath_cq_enter(to_icq(qp->ibqp.recv_cq), &wc,
wqe->wr.send_flags & IB_SEND_SOLICITED);
send_comp:
sqp->s_rnr_retry = sqp->s_rnr_retry_cnt;
if (!(sqp->s_flags & IPATH_S_SIGNAL_REQ_WR) ||
(wqe->wr.send_flags & IB_SEND_SIGNALED)) {
wc.wr_id = wqe->wr.wr_id;
wc.status = IB_WC_SUCCESS;
wc.opcode = ib_ipath_wc_opcode[wqe->wr.opcode];
wc.vendor_err = 0;
wc.byte_len = wqe->length;
wc.qp = &sqp->ibqp;
wc.src_qp = 0;
wc.pkey_index = 0;
wc.slid = 0;
wc.sl = 0;
wc.dlid_path_bits = 0;
wc.port_num = 0;
ipath_cq_enter(to_icq(sqp->ibqp.send_cq), &wc, 0);
}
/* Update s_last now that we are finished with the SWQE */
spin_lock_irqsave(&sqp->s_lock, flags);
if (++sqp->s_last >= sqp->s_size)
sqp->s_last = 0;
spin_unlock_irqrestore(&sqp->s_lock, flags);
goto again;
done:
if (atomic_dec_and_test(&qp->refcount))
wake_up(&qp->wait);
}
static void ipath_ruc_loopback(struct ipath_qp *sqp)
{
struct ipath_ibdev *dev = to_idev(sqp->ibqp.device);
struct ipath_qp *qp;
struct ipath_swqe *wqe;
struct ipath_sge *sge;
unsigned long flags;
struct ib_wc wc;
u64 sdata;
atomic64_t *maddr;
enum ib_wc_status send_status;
qp = ipath_lookup_qpn(&dev->qp_table, sqp->remote_qpn);
spin_lock_irqsave(&sqp->s_lock, flags);
if ((sqp->s_flags & (IPATH_S_BUSY | IPATH_S_ANY_WAIT)) ||
!(ib_ipath_state_ops[sqp->state] & IPATH_PROCESS_OR_FLUSH_SEND))
goto unlock;
sqp->s_flags |= IPATH_S_BUSY;
again:
if (sqp->s_last == sqp->s_head)
goto clr_busy;
wqe = get_swqe_ptr(sqp, sqp->s_last);
if (!(ib_ipath_state_ops[sqp->state] & IPATH_PROCESS_NEXT_SEND_OK)) {
if (!(ib_ipath_state_ops[sqp->state] & IPATH_FLUSH_SEND))
goto clr_busy;
send_status = IB_WC_WR_FLUSH_ERR;
goto flush_send;
}
if (sqp->s_last == sqp->s_cur) {
if (++sqp->s_cur >= sqp->s_size)
sqp->s_cur = 0;
}
spin_unlock_irqrestore(&sqp->s_lock, flags);
if (!qp || !(ib_ipath_state_ops[qp->state] & IPATH_PROCESS_RECV_OK)) {
dev->n_pkt_drops++;
if (sqp->ibqp.qp_type == IB_QPT_RC)
send_status = IB_WC_RETRY_EXC_ERR;
else
send_status = IB_WC_SUCCESS;
goto serr;
}
memset(&wc, 0, sizeof wc);
send_status = IB_WC_SUCCESS;
sqp->s_sge.sge = wqe->sg_list[0];
sqp->s_sge.sg_list = wqe->sg_list + 1;
sqp->s_sge.num_sge = wqe->wr.num_sge;
sqp->s_len = wqe->length;
switch (wqe->wr.opcode) {
case IB_WR_SEND_WITH_IMM:
wc.wc_flags = IB_WC_WITH_IMM;
wc.ex.imm_data = wqe->wr.ex.imm_data;
case IB_WR_SEND:
if (!ipath_get_rwqe(qp, 0))
goto rnr_nak;
break;
case IB_WR_RDMA_WRITE_WITH_IMM:
if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_WRITE)))
goto inv_err;
wc.wc_flags = IB_WC_WITH_IMM;
wc.ex.imm_data = wqe->wr.ex.imm_data;
if (!ipath_get_rwqe(qp, 1))
goto rnr_nak;
case IB_WR_RDMA_WRITE:
if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_WRITE)))
goto inv_err;
if (wqe->length == 0)
break;
if (unlikely(!ipath_rkey_ok(qp, &qp->r_sge, wqe->length,
wqe->wr.wr.rdma.remote_addr,
wqe->wr.wr.rdma.rkey,
IB_ACCESS_REMOTE_WRITE)))
goto acc_err;
break;
case IB_WR_RDMA_READ:
if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_READ)))
goto inv_err;
if (unlikely(!ipath_rkey_ok(qp, &sqp->s_sge, wqe->length,
wqe->wr.wr.rdma.remote_addr,
wqe->wr.wr.rdma.rkey,
IB_ACCESS_REMOTE_READ)))
goto acc_err;
qp->r_sge.sge = wqe->sg_list[0];
qp->r_sge.sg_list = wqe->sg_list + 1;
qp->r_sge.num_sge = wqe->wr.num_sge;
break;
case IB_WR_ATOMIC_CMP_AND_SWP:
case IB_WR_ATOMIC_FETCH_AND_ADD:
if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_ATOMIC)))
goto inv_err;
if (unlikely(!ipath_rkey_ok(qp, &qp->r_sge, sizeof(u64),
wqe->wr.wr.atomic.remote_addr,
wqe->wr.wr.atomic.rkey,
IB_ACCESS_REMOTE_ATOMIC)))
goto acc_err;
maddr = (atomic64_t *) qp->r_sge.sge.vaddr;
sdata = wqe->wr.wr.atomic.compare_add;
*(u64 *) sqp->s_sge.sge.vaddr =
(wqe->wr.opcode == IB_WR_ATOMIC_FETCH_AND_ADD) ?
(u64) atomic64_add_return(sdata, maddr) - sdata :
(u64) cmpxchg((u64 *) qp->r_sge.sge.vaddr,
sdata, wqe->wr.wr.atomic.swap);
goto send_comp;
default:
send_status = IB_WC_LOC_QP_OP_ERR;
goto serr;
}
sge = &sqp->s_sge.sge;
while (sqp->s_len) {
u32 len = sqp->s_len;
if (len > sge->length)
len = sge->length;
if (len > sge->sge_length)
len = sge->sge_length;
BUG_ON(len == 0);
ipath_copy_sge(&qp->r_sge, sge->vaddr, len);
sge->vaddr += len;
sge->length -= len;
sge->sge_length -= len;
if (sge->sge_length == 0) {
if (--sqp->s_sge.num_sge)
*sge = *sqp->s_sge.sg_list++;
} else if (sge->length == 0 && sge->mr != NULL) {
if (++sge->n >= IPATH_SEGSZ) {
if (++sge->m >= sge->mr->mapsz)
break;
sge->n = 0;
}
sge->vaddr =
sge->mr->map[sge->m]->segs[sge->n].vaddr;
sge->length =
sge->mr->map[sge->m]->segs[sge->n].length;
}
sqp->s_len -= len;
}
if (!test_and_clear_bit(IPATH_R_WRID_VALID, &qp->r_aflags))
goto send_comp;
if (wqe->wr.opcode == IB_WR_RDMA_WRITE_WITH_IMM)
wc.opcode = IB_WC_RECV_RDMA_WITH_IMM;
else
wc.opcode = IB_WC_RECV;
wc.wr_id = qp->r_wr_id;
wc.status = IB_WC_SUCCESS;
wc.byte_len = wqe->length;
wc.qp = &qp->ibqp;
wc.src_qp = qp->remote_qpn;
wc.slid = qp->remote_ah_attr.dlid;
wc.sl = qp->remote_ah_attr.sl;
wc.port_num = 1;
ipath_cq_enter(to_icq(qp->ibqp.recv_cq), &wc,
wqe->wr.send_flags & IB_SEND_SOLICITED);
send_comp:
spin_lock_irqsave(&sqp->s_lock, flags);
flush_send:
sqp->s_rnr_retry = sqp->s_rnr_retry_cnt;
ipath_send_complete(sqp, wqe, send_status);
goto again;
rnr_nak:
if (qp->ibqp.qp_type == IB_QPT_UC)
goto send_comp;
if (sqp->s_rnr_retry == 0) {
send_status = IB_WC_RNR_RETRY_EXC_ERR;
goto serr;
}
if (sqp->s_rnr_retry_cnt < 7)
sqp->s_rnr_retry--;
spin_lock_irqsave(&sqp->s_lock, flags);
if (!(ib_ipath_state_ops[sqp->state] & IPATH_PROCESS_RECV_OK))
goto clr_busy;
sqp->s_flags |= IPATH_S_WAITING;
dev->n_rnr_naks++;
sqp->s_rnr_timeout = ib_ipath_rnr_table[qp->r_min_rnr_timer];
ipath_insert_rnr_queue(sqp);
goto clr_busy;
inv_err:
send_status = IB_WC_REM_INV_REQ_ERR;
wc.status = IB_WC_LOC_QP_OP_ERR;
goto err;
acc_err:
send_status = IB_WC_REM_ACCESS_ERR;
wc.status = IB_WC_LOC_PROT_ERR;
err:
ipath_rc_error(qp, wc.status);
serr:
spin_lock_irqsave(&sqp->s_lock, flags);
ipath_send_complete(sqp, wqe, send_status);
if (sqp->ibqp.qp_type == IB_QPT_RC) {
int lastwqe = ipath_error_qp(sqp, IB_WC_WR_FLUSH_ERR);
sqp->s_flags &= ~IPATH_S_BUSY;
spin_unlock_irqrestore(&sqp->s_lock, flags);
if (lastwqe) {
struct ib_event ev;
ev.device = sqp->ibqp.device;
ev.element.qp = &sqp->ibqp;
ev.event = IB_EVENT_QP_LAST_WQE_REACHED;
sqp->ibqp.event_handler(&ev, sqp->ibqp.qp_context);
}
goto done;
}
clr_busy:
sqp->s_flags &= ~IPATH_S_BUSY;
unlock:
spin_unlock_irqrestore(&sqp->s_lock, flags);
done:
if (qp && atomic_dec_and_test(&qp->refcount))
wake_up(&qp->wait);
}
.unlocked_ioctl = dispatch_ioctl,
.open = phys_mem_open,
.release = phys_mem_release,
.mmap = dispatch_mmap,
};
首先来看open函数，open函数为phys_mem_open，其主要作用是创建一个
session，并将其初始化，将该session与filp相关联，核心代码如下：
dev = container_of(inode->i_cdev, struct phys_mem_dev, cdev);
/* 从slab分配器中分配一个session */
session = kmem_cache_alloc(session_mem_cache, GFP_KERNEL);
...
/* 初始化session */
session->session_id = atomic64_add_return(1, &session_counter);
session->device = dev;
sema_init(&session->sem, 1);
session->vmas = 0;
session->num_frame_stati = 0;
session->frame_stati = NULL;
session->status.state = SESSION_STATE_INVALID;
SET_STATE(session, SESSION_STATE_OPEN);
/* 关联filp与session */
filp->private_data = session;
release函数为phys_mem_release，它的主要作用是关闭设备文件，并
释放相关资源。release函数的实现使用了一个循环，每次循环都让状
态机（前面所示）回退一个状态，直到状态机的状态变为SESSION_STATE_CLOSED
