2006-09-22 22:22:48 +00:00
|
|
|
/*
|
|
|
|
* Copyright (c) 2004, 2005 Topspin Communications. All rights reserved.
|
|
|
|
* Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
|
|
|
|
* Copyright (c) 2005 Cisco Systems, Inc. All rights reserved.
|
|
|
|
* Copyright (c) 2005 Mellanox Technologies. All rights reserved.
|
|
|
|
* Copyright (c) 2004 Voltaire, Inc. All rights reserved.
|
|
|
|
* Copyright (c) 2005 Open Grid Computing, Inc. 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
|
|
|
|
* General Public License (GPL) Version 2, available from the file
|
|
|
|
* COPYING in the main directory of this source tree, or the
|
|
|
|
* OpenIB.org BSD license below:
|
|
|
|
*
|
|
|
|
* Redistribution and use in source and binary forms, with or
|
|
|
|
* without modification, are permitted provided that the following
|
|
|
|
* conditions are met:
|
|
|
|
*
|
|
|
|
* - Redistributions of source code must retain the above
|
|
|
|
* copyright notice, this list of conditions and the following
|
|
|
|
* disclaimer.
|
|
|
|
*
|
|
|
|
* - Redistributions in binary form must reproduce the above
|
|
|
|
* copyright notice, this list of conditions and the following
|
|
|
|
* disclaimer in the documentation and/or other materials
|
|
|
|
* provided with the distribution.
|
|
|
|
*
|
|
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
|
|
|
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
|
|
|
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
|
|
|
|
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
|
|
|
|
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
|
|
|
|
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
|
|
|
|
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
|
|
* SOFTWARE.
|
|
|
|
*
|
|
|
|
*/
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
|
|
|
#include <linux/gfp.h>
|
|
|
|
|
2006-09-22 22:22:48 +00:00
|
|
|
#include "c2.h"
|
|
|
|
#include "c2_vq.h"
|
|
|
|
#include "c2_status.h"
|
|
|
|
|
|
|
|
#define C2_CQ_MSG_SIZE ((sizeof(struct c2wr_ce) + 32-1) & ~(32-1))
|
|
|
|
|
|
|
|
static struct c2_cq *c2_cq_get(struct c2_dev *c2dev, int cqn)
|
|
|
|
{
|
|
|
|
struct c2_cq *cq;
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&c2dev->lock, flags);
|
|
|
|
cq = c2dev->qptr_array[cqn];
|
|
|
|
if (!cq) {
|
|
|
|
spin_unlock_irqrestore(&c2dev->lock, flags);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
atomic_inc(&cq->refcount);
|
|
|
|
spin_unlock_irqrestore(&c2dev->lock, flags);
|
|
|
|
return cq;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void c2_cq_put(struct c2_cq *cq)
|
|
|
|
{
|
|
|
|
if (atomic_dec_and_test(&cq->refcount))
|
|
|
|
wake_up(&cq->wait);
|
|
|
|
}
|
|
|
|
|
|
|
|
void c2_cq_event(struct c2_dev *c2dev, u32 mq_index)
|
|
|
|
{
|
|
|
|
struct c2_cq *cq;
|
|
|
|
|
|
|
|
cq = c2_cq_get(c2dev, mq_index);
|
|
|
|
if (!cq) {
|
|
|
|
printk("discarding events on destroyed CQN=%d\n", mq_index);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
(*cq->ibcq.comp_handler) (&cq->ibcq, cq->ibcq.cq_context);
|
|
|
|
c2_cq_put(cq);
|
|
|
|
}
|
|
|
|
|
|
|
|
void c2_cq_clean(struct c2_dev *c2dev, struct c2_qp *qp, u32 mq_index)
|
|
|
|
{
|
|
|
|
struct c2_cq *cq;
|
|
|
|
struct c2_mq *q;
|
|
|
|
|
|
|
|
cq = c2_cq_get(c2dev, mq_index);
|
|
|
|
if (!cq)
|
|
|
|
return;
|
|
|
|
|
|
|
|
spin_lock_irq(&cq->lock);
|
|
|
|
q = &cq->mq;
|
|
|
|
if (q && !c2_mq_empty(q)) {
|
|
|
|
u16 priv = q->priv;
|
|
|
|
struct c2wr_ce *msg;
|
|
|
|
|
|
|
|
while (priv != be16_to_cpu(*q->shared)) {
|
|
|
|
msg = (struct c2wr_ce *)
|
|
|
|
(q->msg_pool.host + priv * q->msg_size);
|
|
|
|
if (msg->qp_user_context == (u64) (unsigned long) qp) {
|
|
|
|
msg->qp_user_context = (u64) 0;
|
|
|
|
}
|
|
|
|
priv = (priv + 1) % q->q_size;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
spin_unlock_irq(&cq->lock);
|
|
|
|
c2_cq_put(cq);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline enum ib_wc_status c2_cqe_status_to_openib(u8 status)
|
|
|
|
{
|
|
|
|
switch (status) {
|
|
|
|
case C2_OK:
|
|
|
|
return IB_WC_SUCCESS;
|
|
|
|
case CCERR_FLUSHED:
|
|
|
|
return IB_WC_WR_FLUSH_ERR;
|
|
|
|
case CCERR_BASE_AND_BOUNDS_VIOLATION:
|
|
|
|
return IB_WC_LOC_PROT_ERR;
|
|
|
|
case CCERR_ACCESS_VIOLATION:
|
|
|
|
return IB_WC_LOC_ACCESS_ERR;
|
|
|
|
case CCERR_TOTAL_LENGTH_TOO_BIG:
|
|
|
|
return IB_WC_LOC_LEN_ERR;
|
|
|
|
case CCERR_INVALID_WINDOW:
|
|
|
|
return IB_WC_MW_BIND_ERR;
|
|
|
|
default:
|
|
|
|
return IB_WC_GENERAL_ERR;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static inline int c2_poll_one(struct c2_dev *c2dev,
|
|
|
|
struct c2_cq *cq, struct ib_wc *entry)
|
|
|
|
{
|
|
|
|
struct c2wr_ce *ce;
|
|
|
|
struct c2_qp *qp;
|
|
|
|
int is_recv = 0;
|
|
|
|
|
2009-05-13 23:53:39 +00:00
|
|
|
ce = c2_mq_consume(&cq->mq);
|
2006-09-22 22:22:48 +00:00
|
|
|
if (!ce) {
|
|
|
|
return -EAGAIN;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if the qp returned is null then this qp has already
|
|
|
|
* been freed and we are unable process the completion.
|
|
|
|
* try pulling the next message
|
|
|
|
*/
|
|
|
|
while ((qp =
|
|
|
|
(struct c2_qp *) (unsigned long) ce->qp_user_context) == NULL) {
|
|
|
|
c2_mq_free(&cq->mq);
|
2009-05-13 23:53:39 +00:00
|
|
|
ce = c2_mq_consume(&cq->mq);
|
2006-09-22 22:22:48 +00:00
|
|
|
if (!ce)
|
|
|
|
return -EAGAIN;
|
|
|
|
}
|
|
|
|
|
|
|
|
entry->status = c2_cqe_status_to_openib(c2_wr_get_result(ce));
|
|
|
|
entry->wr_id = ce->hdr.context;
|
2006-12-31 19:09:42 +00:00
|
|
|
entry->qp = &qp->ibqp;
|
2006-09-22 22:22:48 +00:00
|
|
|
entry->wc_flags = 0;
|
|
|
|
entry->slid = 0;
|
|
|
|
entry->sl = 0;
|
|
|
|
entry->src_qp = 0;
|
|
|
|
entry->dlid_path_bits = 0;
|
|
|
|
entry->pkey_index = 0;
|
|
|
|
|
|
|
|
switch (c2_wr_get_id(ce)) {
|
|
|
|
case C2_WR_TYPE_SEND:
|
|
|
|
entry->opcode = IB_WC_SEND;
|
|
|
|
break;
|
|
|
|
case C2_WR_TYPE_RDMA_WRITE:
|
|
|
|
entry->opcode = IB_WC_RDMA_WRITE;
|
|
|
|
break;
|
|
|
|
case C2_WR_TYPE_RDMA_READ:
|
|
|
|
entry->opcode = IB_WC_RDMA_READ;
|
|
|
|
break;
|
|
|
|
case C2_WR_TYPE_BIND_MW:
|
|
|
|
entry->opcode = IB_WC_BIND_MW;
|
|
|
|
break;
|
|
|
|
case C2_WR_TYPE_RECV:
|
|
|
|
entry->byte_len = be32_to_cpu(ce->bytes_rcvd);
|
|
|
|
entry->opcode = IB_WC_RECV;
|
|
|
|
is_recv = 1;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* consume the WQEs */
|
|
|
|
if (is_recv)
|
|
|
|
c2_mq_lconsume(&qp->rq_mq, 1);
|
|
|
|
else
|
|
|
|
c2_mq_lconsume(&qp->sq_mq,
|
|
|
|
be32_to_cpu(c2_wr_get_wqe_count(ce)) + 1);
|
|
|
|
|
|
|
|
/* free the message */
|
|
|
|
c2_mq_free(&cq->mq);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
int c2_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry)
|
|
|
|
{
|
|
|
|
struct c2_dev *c2dev = to_c2dev(ibcq->device);
|
|
|
|
struct c2_cq *cq = to_c2cq(ibcq);
|
|
|
|
unsigned long flags;
|
|
|
|
int npolled, err;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&cq->lock, flags);
|
|
|
|
|
|
|
|
for (npolled = 0; npolled < num_entries; ++npolled) {
|
|
|
|
|
|
|
|
err = c2_poll_one(c2dev, cq, entry + npolled);
|
|
|
|
if (err)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
spin_unlock_irqrestore(&cq->lock, flags);
|
|
|
|
|
|
|
|
return npolled;
|
|
|
|
}
|
|
|
|
|
IB: Return "maybe missed event" hint from ib_req_notify_cq()
The semantics defined by the InfiniBand specification say that
completion events are only generated when a completions is added to a
completion queue (CQ) after completion notification is requested. In
other words, this means that the following race is possible:
while (CQ is not empty)
ib_poll_cq(CQ);
// new completion is added after while loop is exited
ib_req_notify_cq(CQ);
// no event is generated for the existing completion
To close this race, the IB spec recommends doing another poll of the
CQ after requesting notification.
However, it is not always possible to arrange code this way (for
example, we have found that NAPI for IPoIB cannot poll after
requesting notification). Also, some hardware (eg Mellanox HCAs)
actually will generate an event for completions added before the call
to ib_req_notify_cq() -- which is allowed by the spec, since there's
no way for any upper-layer consumer to know exactly when a completion
was really added -- so the extra poll of the CQ is just a waste.
Motivated by this, we add a new flag "IB_CQ_REPORT_MISSED_EVENTS" for
ib_req_notify_cq() so that it can return a hint about whether the a
completion may have been added before the request for notification.
The return value of ib_req_notify_cq() is extended so:
< 0 means an error occurred while requesting notification
== 0 means notification was requested successfully, and if
IB_CQ_REPORT_MISSED_EVENTS was passed in, then no
events were missed and it is safe to wait for another
event.
> 0 is only returned if IB_CQ_REPORT_MISSED_EVENTS was
passed in. It means that the consumer must poll the
CQ again to make sure it is empty to avoid the race
described above.
We add a flag to enable this behavior rather than turning it on
unconditionally, because checking for missed events may incur
significant overhead for some low-level drivers, and consumers that
don't care about the results of this test shouldn't be forced to pay
for the test.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-07 04:02:48 +00:00
|
|
|
int c2_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags notify_flags)
|
2006-09-22 22:22:48 +00:00
|
|
|
{
|
|
|
|
struct c2_mq_shared __iomem *shared;
|
|
|
|
struct c2_cq *cq;
|
IB: Return "maybe missed event" hint from ib_req_notify_cq()
The semantics defined by the InfiniBand specification say that
completion events are only generated when a completions is added to a
completion queue (CQ) after completion notification is requested. In
other words, this means that the following race is possible:
while (CQ is not empty)
ib_poll_cq(CQ);
// new completion is added after while loop is exited
ib_req_notify_cq(CQ);
// no event is generated for the existing completion
To close this race, the IB spec recommends doing another poll of the
CQ after requesting notification.
However, it is not always possible to arrange code this way (for
example, we have found that NAPI for IPoIB cannot poll after
requesting notification). Also, some hardware (eg Mellanox HCAs)
actually will generate an event for completions added before the call
to ib_req_notify_cq() -- which is allowed by the spec, since there's
no way for any upper-layer consumer to know exactly when a completion
was really added -- so the extra poll of the CQ is just a waste.
Motivated by this, we add a new flag "IB_CQ_REPORT_MISSED_EVENTS" for
ib_req_notify_cq() so that it can return a hint about whether the a
completion may have been added before the request for notification.
The return value of ib_req_notify_cq() is extended so:
< 0 means an error occurred while requesting notification
== 0 means notification was requested successfully, and if
IB_CQ_REPORT_MISSED_EVENTS was passed in, then no
events were missed and it is safe to wait for another
event.
> 0 is only returned if IB_CQ_REPORT_MISSED_EVENTS was
passed in. It means that the consumer must poll the
CQ again to make sure it is empty to avoid the race
described above.
We add a flag to enable this behavior rather than turning it on
unconditionally, because checking for missed events may incur
significant overhead for some low-level drivers, and consumers that
don't care about the results of this test shouldn't be forced to pay
for the test.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-07 04:02:48 +00:00
|
|
|
unsigned long flags;
|
|
|
|
int ret = 0;
|
2006-09-22 22:22:48 +00:00
|
|
|
|
|
|
|
cq = to_c2cq(ibcq);
|
|
|
|
shared = cq->mq.peer;
|
|
|
|
|
IB: Return "maybe missed event" hint from ib_req_notify_cq()
The semantics defined by the InfiniBand specification say that
completion events are only generated when a completions is added to a
completion queue (CQ) after completion notification is requested. In
other words, this means that the following race is possible:
while (CQ is not empty)
ib_poll_cq(CQ);
// new completion is added after while loop is exited
ib_req_notify_cq(CQ);
// no event is generated for the existing completion
To close this race, the IB spec recommends doing another poll of the
CQ after requesting notification.
However, it is not always possible to arrange code this way (for
example, we have found that NAPI for IPoIB cannot poll after
requesting notification). Also, some hardware (eg Mellanox HCAs)
actually will generate an event for completions added before the call
to ib_req_notify_cq() -- which is allowed by the spec, since there's
no way for any upper-layer consumer to know exactly when a completion
was really added -- so the extra poll of the CQ is just a waste.
Motivated by this, we add a new flag "IB_CQ_REPORT_MISSED_EVENTS" for
ib_req_notify_cq() so that it can return a hint about whether the a
completion may have been added before the request for notification.
The return value of ib_req_notify_cq() is extended so:
< 0 means an error occurred while requesting notification
== 0 means notification was requested successfully, and if
IB_CQ_REPORT_MISSED_EVENTS was passed in, then no
events were missed and it is safe to wait for another
event.
> 0 is only returned if IB_CQ_REPORT_MISSED_EVENTS was
passed in. It means that the consumer must poll the
CQ again to make sure it is empty to avoid the race
described above.
We add a flag to enable this behavior rather than turning it on
unconditionally, because checking for missed events may incur
significant overhead for some low-level drivers, and consumers that
don't care about the results of this test shouldn't be forced to pay
for the test.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-07 04:02:48 +00:00
|
|
|
if ((notify_flags & IB_CQ_SOLICITED_MASK) == IB_CQ_NEXT_COMP)
|
2006-09-22 22:22:48 +00:00
|
|
|
writeb(C2_CQ_NOTIFICATION_TYPE_NEXT, &shared->notification_type);
|
IB: Return "maybe missed event" hint from ib_req_notify_cq()
The semantics defined by the InfiniBand specification say that
completion events are only generated when a completions is added to a
completion queue (CQ) after completion notification is requested. In
other words, this means that the following race is possible:
while (CQ is not empty)
ib_poll_cq(CQ);
// new completion is added after while loop is exited
ib_req_notify_cq(CQ);
// no event is generated for the existing completion
To close this race, the IB spec recommends doing another poll of the
CQ after requesting notification.
However, it is not always possible to arrange code this way (for
example, we have found that NAPI for IPoIB cannot poll after
requesting notification). Also, some hardware (eg Mellanox HCAs)
actually will generate an event for completions added before the call
to ib_req_notify_cq() -- which is allowed by the spec, since there's
no way for any upper-layer consumer to know exactly when a completion
was really added -- so the extra poll of the CQ is just a waste.
Motivated by this, we add a new flag "IB_CQ_REPORT_MISSED_EVENTS" for
ib_req_notify_cq() so that it can return a hint about whether the a
completion may have been added before the request for notification.
The return value of ib_req_notify_cq() is extended so:
< 0 means an error occurred while requesting notification
== 0 means notification was requested successfully, and if
IB_CQ_REPORT_MISSED_EVENTS was passed in, then no
events were missed and it is safe to wait for another
event.
> 0 is only returned if IB_CQ_REPORT_MISSED_EVENTS was
passed in. It means that the consumer must poll the
CQ again to make sure it is empty to avoid the race
described above.
We add a flag to enable this behavior rather than turning it on
unconditionally, because checking for missed events may incur
significant overhead for some low-level drivers, and consumers that
don't care about the results of this test shouldn't be forced to pay
for the test.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-07 04:02:48 +00:00
|
|
|
else if ((notify_flags & IB_CQ_SOLICITED_MASK) == IB_CQ_SOLICITED)
|
2006-09-22 22:22:48 +00:00
|
|
|
writeb(C2_CQ_NOTIFICATION_TYPE_NEXT_SE, &shared->notification_type);
|
|
|
|
else
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
writeb(CQ_WAIT_FOR_DMA | CQ_ARMED, &shared->armed);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Now read back shared->armed to make the PCI
|
|
|
|
* write synchronous. This is necessary for
|
|
|
|
* correct cq notification semantics.
|
|
|
|
*/
|
|
|
|
readb(&shared->armed);
|
|
|
|
|
IB: Return "maybe missed event" hint from ib_req_notify_cq()
The semantics defined by the InfiniBand specification say that
completion events are only generated when a completions is added to a
completion queue (CQ) after completion notification is requested. In
other words, this means that the following race is possible:
while (CQ is not empty)
ib_poll_cq(CQ);
// new completion is added after while loop is exited
ib_req_notify_cq(CQ);
// no event is generated for the existing completion
To close this race, the IB spec recommends doing another poll of the
CQ after requesting notification.
However, it is not always possible to arrange code this way (for
example, we have found that NAPI for IPoIB cannot poll after
requesting notification). Also, some hardware (eg Mellanox HCAs)
actually will generate an event for completions added before the call
to ib_req_notify_cq() -- which is allowed by the spec, since there's
no way for any upper-layer consumer to know exactly when a completion
was really added -- so the extra poll of the CQ is just a waste.
Motivated by this, we add a new flag "IB_CQ_REPORT_MISSED_EVENTS" for
ib_req_notify_cq() so that it can return a hint about whether the a
completion may have been added before the request for notification.
The return value of ib_req_notify_cq() is extended so:
< 0 means an error occurred while requesting notification
== 0 means notification was requested successfully, and if
IB_CQ_REPORT_MISSED_EVENTS was passed in, then no
events were missed and it is safe to wait for another
event.
> 0 is only returned if IB_CQ_REPORT_MISSED_EVENTS was
passed in. It means that the consumer must poll the
CQ again to make sure it is empty to avoid the race
described above.
We add a flag to enable this behavior rather than turning it on
unconditionally, because checking for missed events may incur
significant overhead for some low-level drivers, and consumers that
don't care about the results of this test shouldn't be forced to pay
for the test.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-05-07 04:02:48 +00:00
|
|
|
if (notify_flags & IB_CQ_REPORT_MISSED_EVENTS) {
|
|
|
|
spin_lock_irqsave(&cq->lock, flags);
|
|
|
|
ret = !c2_mq_empty(&cq->mq);
|
|
|
|
spin_unlock_irqrestore(&cq->lock, flags);
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
2006-09-22 22:22:48 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void c2_free_cq_buf(struct c2_dev *c2dev, struct c2_mq *mq)
|
|
|
|
{
|
2006-10-27 22:28:35 +00:00
|
|
|
dma_free_coherent(&c2dev->pcidev->dev, mq->q_size * mq->msg_size,
|
2010-04-21 22:23:10 +00:00
|
|
|
mq->msg_pool.host, dma_unmap_addr(mq, mapping));
|
2006-09-22 22:22:48 +00:00
|
|
|
}
|
|
|
|
|
2014-08-12 23:20:04 +00:00
|
|
|
static int c2_alloc_cq_buf(struct c2_dev *c2dev, struct c2_mq *mq,
|
|
|
|
size_t q_size, size_t msg_size)
|
2006-09-22 22:22:48 +00:00
|
|
|
{
|
2006-10-27 22:28:35 +00:00
|
|
|
u8 *pool_start;
|
2006-09-22 22:22:48 +00:00
|
|
|
|
2014-08-12 23:20:04 +00:00
|
|
|
if (q_size > SIZE_MAX / msg_size)
|
|
|
|
return -EINVAL;
|
|
|
|
|
2006-10-27 22:28:35 +00:00
|
|
|
pool_start = dma_alloc_coherent(&c2dev->pcidev->dev, q_size * msg_size,
|
|
|
|
&mq->host_dma, GFP_KERNEL);
|
2006-09-22 22:22:48 +00:00
|
|
|
if (!pool_start)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
c2_mq_rep_init(mq,
|
|
|
|
0, /* index (currently unknown) */
|
|
|
|
q_size,
|
|
|
|
msg_size,
|
2006-10-27 22:28:35 +00:00
|
|
|
pool_start,
|
2006-09-22 22:22:48 +00:00
|
|
|
NULL, /* peer (currently unknown) */
|
|
|
|
C2_MQ_HOST_TARGET);
|
|
|
|
|
2010-04-21 22:23:10 +00:00
|
|
|
dma_unmap_addr_set(mq, mapping, mq->host_dma);
|
2006-09-22 22:22:48 +00:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
int c2_init_cq(struct c2_dev *c2dev, int entries,
|
|
|
|
struct c2_ucontext *ctx, struct c2_cq *cq)
|
|
|
|
{
|
|
|
|
struct c2wr_cq_create_req wr;
|
|
|
|
struct c2wr_cq_create_rep *reply;
|
|
|
|
unsigned long peer_pa;
|
|
|
|
struct c2_vq_req *vq_req;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
might_sleep();
|
|
|
|
|
|
|
|
cq->ibcq.cqe = entries - 1;
|
|
|
|
cq->is_kernel = !ctx;
|
|
|
|
|
|
|
|
/* Allocate a shared pointer */
|
|
|
|
cq->mq.shared = c2_alloc_mqsp(c2dev, c2dev->kern_mqsp_pool,
|
|
|
|
&cq->mq.shared_dma, GFP_KERNEL);
|
|
|
|
if (!cq->mq.shared)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
/* Allocate pages for the message pool */
|
|
|
|
err = c2_alloc_cq_buf(c2dev, &cq->mq, entries + 1, C2_CQ_MSG_SIZE);
|
|
|
|
if (err)
|
|
|
|
goto bail0;
|
|
|
|
|
|
|
|
vq_req = vq_req_alloc(c2dev);
|
|
|
|
if (!vq_req) {
|
|
|
|
err = -ENOMEM;
|
|
|
|
goto bail1;
|
|
|
|
}
|
|
|
|
|
|
|
|
memset(&wr, 0, sizeof(wr));
|
|
|
|
c2_wr_set_id(&wr, CCWR_CQ_CREATE);
|
|
|
|
wr.hdr.context = (unsigned long) vq_req;
|
|
|
|
wr.rnic_handle = c2dev->adapter_handle;
|
|
|
|
wr.msg_size = cpu_to_be32(cq->mq.msg_size);
|
|
|
|
wr.depth = cpu_to_be32(cq->mq.q_size);
|
|
|
|
wr.shared_ht = cpu_to_be64(cq->mq.shared_dma);
|
|
|
|
wr.msg_pool = cpu_to_be64(cq->mq.host_dma);
|
|
|
|
wr.user_context = (u64) (unsigned long) (cq);
|
|
|
|
|
|
|
|
vq_req_get(c2dev, vq_req);
|
|
|
|
|
|
|
|
err = vq_send_wr(c2dev, (union c2wr *) & wr);
|
|
|
|
if (err) {
|
|
|
|
vq_req_put(c2dev, vq_req);
|
|
|
|
goto bail2;
|
|
|
|
}
|
|
|
|
|
|
|
|
err = vq_wait_for_reply(c2dev, vq_req);
|
|
|
|
if (err)
|
|
|
|
goto bail2;
|
|
|
|
|
|
|
|
reply = (struct c2wr_cq_create_rep *) (unsigned long) (vq_req->reply_msg);
|
|
|
|
if (!reply) {
|
|
|
|
err = -ENOMEM;
|
|
|
|
goto bail2;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((err = c2_errno(reply)) != 0)
|
|
|
|
goto bail3;
|
|
|
|
|
|
|
|
cq->adapter_handle = reply->cq_handle;
|
|
|
|
cq->mq.index = be32_to_cpu(reply->mq_index);
|
|
|
|
|
|
|
|
peer_pa = c2dev->pa + be32_to_cpu(reply->adapter_shared);
|
|
|
|
cq->mq.peer = ioremap_nocache(peer_pa, PAGE_SIZE);
|
|
|
|
if (!cq->mq.peer) {
|
|
|
|
err = -ENOMEM;
|
|
|
|
goto bail3;
|
|
|
|
}
|
|
|
|
|
|
|
|
vq_repbuf_free(c2dev, reply);
|
|
|
|
vq_req_free(c2dev, vq_req);
|
|
|
|
|
|
|
|
spin_lock_init(&cq->lock);
|
|
|
|
atomic_set(&cq->refcount, 1);
|
|
|
|
init_waitqueue_head(&cq->wait);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Use the MQ index allocated by the adapter to
|
|
|
|
* store the CQ in the qptr_array
|
|
|
|
*/
|
|
|
|
cq->cqn = cq->mq.index;
|
|
|
|
c2dev->qptr_array[cq->cqn] = cq;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
bail3:
|
|
|
|
vq_repbuf_free(c2dev, reply);
|
|
|
|
bail2:
|
|
|
|
vq_req_free(c2dev, vq_req);
|
|
|
|
bail1:
|
|
|
|
c2_free_cq_buf(c2dev, &cq->mq);
|
|
|
|
bail0:
|
|
|
|
c2_free_mqsp(cq->mq.shared);
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
void c2_free_cq(struct c2_dev *c2dev, struct c2_cq *cq)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
struct c2_vq_req *vq_req;
|
|
|
|
struct c2wr_cq_destroy_req wr;
|
|
|
|
struct c2wr_cq_destroy_rep *reply;
|
|
|
|
|
|
|
|
might_sleep();
|
|
|
|
|
|
|
|
/* Clear CQ from the qptr array */
|
|
|
|
spin_lock_irq(&c2dev->lock);
|
|
|
|
c2dev->qptr_array[cq->mq.index] = NULL;
|
|
|
|
atomic_dec(&cq->refcount);
|
|
|
|
spin_unlock_irq(&c2dev->lock);
|
|
|
|
|
|
|
|
wait_event(cq->wait, !atomic_read(&cq->refcount));
|
|
|
|
|
|
|
|
vq_req = vq_req_alloc(c2dev);
|
|
|
|
if (!vq_req) {
|
|
|
|
goto bail0;
|
|
|
|
}
|
|
|
|
|
|
|
|
memset(&wr, 0, sizeof(wr));
|
|
|
|
c2_wr_set_id(&wr, CCWR_CQ_DESTROY);
|
|
|
|
wr.hdr.context = (unsigned long) vq_req;
|
|
|
|
wr.rnic_handle = c2dev->adapter_handle;
|
|
|
|
wr.cq_handle = cq->adapter_handle;
|
|
|
|
|
|
|
|
vq_req_get(c2dev, vq_req);
|
|
|
|
|
|
|
|
err = vq_send_wr(c2dev, (union c2wr *) & wr);
|
|
|
|
if (err) {
|
|
|
|
vq_req_put(c2dev, vq_req);
|
|
|
|
goto bail1;
|
|
|
|
}
|
|
|
|
|
|
|
|
err = vq_wait_for_reply(c2dev, vq_req);
|
|
|
|
if (err)
|
|
|
|
goto bail1;
|
|
|
|
|
|
|
|
reply = (struct c2wr_cq_destroy_rep *) (unsigned long) (vq_req->reply_msg);
|
2008-04-17 04:09:34 +00:00
|
|
|
if (reply)
|
|
|
|
vq_repbuf_free(c2dev, reply);
|
2006-09-22 22:22:48 +00:00
|
|
|
bail1:
|
|
|
|
vq_req_free(c2dev, vq_req);
|
|
|
|
bail0:
|
|
|
|
if (cq->is_kernel) {
|
|
|
|
c2_free_cq_buf(c2dev, &cq->mq);
|
|
|
|
}
|
|
|
|
|
|
|
|
return;
|
|
|
|
}
|