[PATCH 0/8] i40iw: Fixes in i40iw for 4.15

[PATCH 0/8] i40iw: Fixes in i40iw for 4.15

[Henry Orosco]

The following patches are for the 4.15 kernel. Most are
fixes, one is for code cleanup (i40iw: Change accelerated
flag to bool).

Chien Tin Tung (2):
  i40iw: Use sqsize to initialize cqp_requests elements
  i40iw: Allocate a sdbuf per CQP WQE

Henry Orosco (2):
  i40iw: Notify user of established connection after QP in RTS
  i40iw: Change accelerated flag to bool

Mustafa Ismail (3):
  i40iw: Do not free sqbuf when event is I40IW_TIMER_TYPE_CLOSE
  i40iw: Correct ARP index mask
  i40iw: Reinitialize add_sd_cnt

Tatyana Nikolova (1):
  i40iw: Move MPA request event for loopback after connect

 drivers/infiniband/hw/i40iw/i40iw_cm.c   | 30 +++++++++++++----------
 drivers/infiniband/hw/i40iw/i40iw_cm.h   |  2 +-
 drivers/infiniband/hw/i40iw/i40iw_ctrl.c | 41 ++++++++++++++++++++++----------
 drivers/infiniband/hw/i40iw/i40iw_d.h    |  6 +++--
 drivers/infiniband/hw/i40iw/i40iw_main.c |  3 ++-
 5 files changed, 53 insertions(+), 29 deletions(-)

-- 
2.8.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 1/8] i40iw: Use sqsize to initialize cqp_requests elements

[Henry Orosco]

From: Chien Tin Tung <chien.tin.tung-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Use sqsize instead of I40IW_CQP_SW_SQSIZE_2048 to initialize
cqp_requests elements in the for-loop as sqsize is used
to allocate memory for cqp_requests.

Fixes: 8e06af711bf2 ("i40iw: add main, hdr, status")
Signed-off-by: Chien Tin Tung <chien.tin.tung-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/infiniband/hw/i40iw/i40iw_main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c
index e824296..1f9d9c7 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_main.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_main.c
@@ -607,7 +607,7 @@ static enum i40iw_status_code i40iw_create_cqp(struct i40iw_device *iwdev)
 	INIT_LIST_HEAD(&cqp->cqp_avail_reqs);
 	INIT_LIST_HEAD(&cqp->cqp_pending_reqs);
 	/* init the waitq of the cqp_requests and add them to the list */
-	for (i = 0; i < I40IW_CQP_SW_SQSIZE_2048; i++) {
+	for (i = 0; i < sqsize; i++) {
 		init_waitqueue_head(&cqp->cqp_requests[i].waitq);
 		list_add_tail(&cqp->cqp_requests[i].list, &cqp->cqp_avail_reqs);
 	}
-- 
2.8.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 2/8] i40iw: Allocate a sdbuf per CQP WQE

[Henry Orosco]

From: Chien Tin Tung <chien.tin.tung-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Currently there is only one sdbuf per Control QP(CQP) for programming
Segment Descriptor(SD) command thus limiting the number of SD work
requests that can be posted to one.  Allocate enough memory for one
sdbuf per CQP SQ WQE to allow more than one SD command at a time.
When a SD command is posted, it will use the corresponding sdbuf for
the WQE.

Fixes: 86dbcd0f12e9 ("i40iw: add file to handle cqp calls")
Signed-off-by: Chien Tin Tung <chien.tin.tung-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/infiniband/hw/i40iw/i40iw_ctrl.c | 41 ++++++++++++++++++++++----------
 drivers/infiniband/hw/i40iw/i40iw_d.h    |  4 +++-
 2 files changed, 31 insertions(+), 14 deletions(-)

diff --git a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c
index d88c6cf..db922d4 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c
@@ -513,7 +513,7 @@ static enum i40iw_status_code i40iw_sc_cqp_create(struct i40iw_sc_cqp *cqp,
 
 	ret_code = i40iw_allocate_dma_mem(cqp->dev->hw,
 					  &cqp->sdbuf,
-					  128,
+					  I40IW_UPDATE_SD_BUF_SIZE * cqp->sq_size,
 					  I40IW_SD_BUF_ALIGNMENT);
 
 	if (ret_code)
@@ -596,14 +596,14 @@ void i40iw_sc_cqp_post_sq(struct i40iw_sc_cqp *cqp)
 }
 
 /**
- * i40iw_sc_cqp_get_next_send_wqe - get next wqe on cqp sq
- * @cqp: struct for cqp hw
- * @wqe_idx: we index of cqp ring
+ * i40iw_sc_cqp_get_next_send_wqe_idx - get next WQE on CQP SQ and pass back the index
+ * @cqp: pointer to CQP structure
+ * @scratch: private data for CQP WQE
+ * @wqe_idx: WQE index for next WQE on CQP SQ
  */
-u64 *i40iw_sc_cqp_get_next_send_wqe(struct i40iw_sc_cqp *cqp, u64 scratch)
+static u64 *i40iw_sc_cqp_get_next_send_wqe_idx(struct i40iw_sc_cqp *cqp, u64 scratch, u32 *wqe_idx)
 {
 	u64 *wqe = NULL;
-	u32	wqe_idx;
 	enum i40iw_status_code ret_code;
 
 	if (I40IW_RING_FULL_ERR(cqp->sq_ring)) {
@@ -616,21 +616,33 @@ u64 *i40iw_sc_cqp_get_next_send_wqe(struct i40iw_sc_cqp *cqp, u64 scratch)
 			    cqp->sq_ring.size);
 		return NULL;
 	}
-	I40IW_ATOMIC_RING_MOVE_HEAD(cqp->sq_ring, wqe_idx, ret_code);
+	I40IW_ATOMIC_RING_MOVE_HEAD(cqp->sq_ring, *wqe_idx, ret_code);
 	cqp->dev->cqp_cmd_stats[OP_REQUESTED_COMMANDS]++;
 	if (ret_code)
 		return NULL;
-	if (!wqe_idx)
+	if (!*wqe_idx)
 		cqp->polarity = !cqp->polarity;
 
-	wqe = cqp->sq_base[wqe_idx].elem;
-	cqp->scratch_array[wqe_idx] = scratch;
+	wqe = cqp->sq_base[*wqe_idx].elem;
+	cqp->scratch_array[*wqe_idx] = scratch;
 	I40IW_CQP_INIT_WQE(wqe);
 
 	return wqe;
 }
 
 /**
+ * i40iw_sc_cqp_get_next_send_wqe - get next wqe on cqp sq
+ * @cqp: struct for cqp hw
+ * @scratch: private data for CQP WQE
+ */
+u64 *i40iw_sc_cqp_get_next_send_wqe(struct i40iw_sc_cqp *cqp, u64 scratch)
+{
+	u32 wqe_idx;
+
+	return i40iw_sc_cqp_get_next_send_wqe_idx(cqp, scratch, &wqe_idx);
+}
+
+/**
  * i40iw_sc_cqp_destroy - destroy cqp during close
  * @cqp: struct for cqp hw
  */
@@ -3587,8 +3599,10 @@ static enum i40iw_status_code cqp_sds_wqe_fill(struct i40iw_sc_cqp *cqp,
 	u64 *wqe;
 	int mem_entries, wqe_entries;
 	struct i40iw_dma_mem *sdbuf = &cqp->sdbuf;
+	u64 offset;
+	u32 wqe_idx;
 
-	wqe = i40iw_sc_cqp_get_next_send_wqe(cqp, scratch);
+	wqe = i40iw_sc_cqp_get_next_send_wqe_idx(cqp, scratch, &wqe_idx);
 	if (!wqe)
 		return I40IW_ERR_RING_FULL;
 
@@ -3601,8 +3615,9 @@ static enum i40iw_status_code cqp_sds_wqe_fill(struct i40iw_sc_cqp *cqp,
 		 LS_64(mem_entries, I40IW_CQPSQ_UPESD_ENTRY_COUNT);
 
 	if (mem_entries) {
-		memcpy(sdbuf->va, &info->entry[3], (mem_entries << 4));
-		data = sdbuf->pa;
+		offset = wqe_idx * I40IW_UPDATE_SD_BUF_SIZE;
+		memcpy((char *)sdbuf->va + offset, &info->entry[3], mem_entries << 4);
+		data = (u64)sdbuf->pa + offset;
 	} else {
 		data = 0;
 	}
diff --git a/drivers/infiniband/hw/i40iw/i40iw_d.h b/drivers/infiniband/hw/i40iw/i40iw_d.h
index 65ec39e..1077b78 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_d.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_d.h
@@ -1526,7 +1526,7 @@ enum i40iw_alignment {
 	I40IW_AEQ_ALIGNMENT =		0x100,
 	I40IW_CEQ_ALIGNMENT =		0x100,
 	I40IW_CQ0_ALIGNMENT =		0x100,
-	I40IW_SD_BUF_ALIGNMENT =	0x100
+	I40IW_SD_BUF_ALIGNMENT =	0x80
 };
 
 #define I40IW_WQE_SIZE_64	64
@@ -1534,6 +1534,8 @@ enum i40iw_alignment {
 #define I40IW_QP_WQE_MIN_SIZE	32
 #define I40IW_QP_WQE_MAX_SIZE	128
 
+#define I40IW_UPDATE_SD_BUF_SIZE 128
+
 #define I40IW_CQE_QTYPE_RQ 0
 #define I40IW_CQE_QTYPE_SQ 1
 
-- 
2.8.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/8] i40iw: Allocate a sdbuf per CQP WQE

[Jason Gunthorpe]

On Fri, Nov 17, 2017 at 10:46:51AM -0600, Henry Orosco wrote:
> From: Chien Tin Tung <chien.tin.tung-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> 
> Currently there is only one sdbuf per Control QP(CQP) for programming
> Segment Descriptor(SD) command thus limiting the number of SD work
> requests that can be posted to one.  Allocate enough memory for one
> sdbuf per CQP SQ WQE to allow more than one SD command at a time.
> When a SD command is posted, it will use the corresponding sdbuf for
> the WQE.

It is not clear to me from this description why this would be a rc
candidate?

Is this is a bug? or is it adding a feature of increasing the number
of 'SD work requests'?

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/8] i40iw: Allocate a sdbuf per CQP WQE

[Shiraz Saleem]

On Mon, Nov 20, 2017 at 11:38:00AM -0700, Jason Gunthorpe wrote:
> On Fri, Nov 17, 2017 at 10:46:51AM -0600, Henry Orosco wrote:
> > From: Chien Tin Tung <chien.tin.tung-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> > 
> > Currently there is only one sdbuf per Control QP(CQP) for programming
> > Segment Descriptor(SD) command thus limiting the number of SD work
> > requests that can be posted to one.  Allocate enough memory for one
> > sdbuf per CQP SQ WQE to allow more than one SD command at a time.
> > When a SD command is posted, it will use the corresponding sdbuf for
> > the WQE.
> 
> It is not clear to me from this description why this would be a rc
> candidate?
> 
> Is this is a bug? or is it adding a feature of increasing the number
> of 'SD work requests'?
>

Its a bug fix as we are reusing the same SD buffer
for all WQEs posted. This is incorrect and will become
a problem if we have multiple simultaneous SD WQEs 
being posted.

Shiraz

 
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/8] i40iw: Allocate a sdbuf per CQP WQE

[Jason Gunthorpe]

On Mon, Nov 20, 2017 at 03:29:59PM -0600, Shiraz Saleem wrote:
> On Mon, Nov 20, 2017 at 11:38:00AM -0700, Jason Gunthorpe wrote:
> > On Fri, Nov 17, 2017 at 10:46:51AM -0600, Henry Orosco wrote:
> > > From: Chien Tin Tung <chien.tin.tung-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> > > 
> > > Currently there is only one sdbuf per Control QP(CQP) for programming
> > > Segment Descriptor(SD) command thus limiting the number of SD work
> > > requests that can be posted to one.  Allocate enough memory for one
> > > sdbuf per CQP SQ WQE to allow more than one SD command at a time.
> > > When a SD command is posted, it will use the corresponding sdbuf for
> > > the WQE.
> > 
> > It is not clear to me from this description why this would be a rc
> > candidate?
> > 
> > Is this is a bug? or is it adding a feature of increasing the number
> > of 'SD work requests'?
> >
> 
> Its a bug fix as we are reusing the same SD buffer
> for all WQEs posted. This is incorrect and will become
> a problem if we have multiple simultaneous SD WQEs 
> being posted.

OK, can you send me a revised commit description making it clear what
the bug is and how it could be hit? Maybe something like:

 The driver only allocated one Segment Descriptor (SD) per ???, but there
 is nothing preventing multiple WQEs from trying to use the same
 SD. In this case new WQEs can corrupt past SD's resulting in
 mis-execution of the WQE.

 Fix this by allocating a SD for every possible WQE and blah blah

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/8] i40iw: Allocate a sdbuf per CQP WQE

[Shiraz Saleem]

On Mon, Nov 20, 2017 at 02:46:28PM -0700, Jason Gunthorpe wrote:
> On Mon, Nov 20, 2017 at 03:29:59PM -0600, Shiraz Saleem wrote:
> > On Mon, Nov 20, 2017 at 11:38:00AM -0700, Jason Gunthorpe wrote:
> > > On Fri, Nov 17, 2017 at 10:46:51AM -0600, Henry Orosco wrote:
> > > > From: Chien Tin Tung <chien.tin.tung-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> > > > 
> > > > Currently there is only one sdbuf per Control QP(CQP) for programming
> > > > Segment Descriptor(SD) command thus limiting the number of SD work
> > > > requests that can be posted to one.  Allocate enough memory for one
> > > > sdbuf per CQP SQ WQE to allow more than one SD command at a time.
> > > > When a SD command is posted, it will use the corresponding sdbuf for
> > > > the WQE.
> > > 
> > > It is not clear to me from this description why this would be a rc
> > > candidate?
> > > 
> > > Is this is a bug? or is it adding a feature of increasing the number
> > > of 'SD work requests'?
> > >
> > 
> > Its a bug fix as we are reusing the same SD buffer
> > for all WQEs posted. This is incorrect and will become
> > a problem if we have multiple simultaneous SD WQEs 
> > being posted.
> 
> OK, can you send me a revised commit description making it clear what
> the bug is and how it could be hit? Maybe something like:

Ok. Will resend with revised commit message.


--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 3/8] i40iw: Do not free sqbuf when event is I40IW_TIMER_TYPE_CLOSE

[Henry Orosco]

From: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

When the event type is I40IW_TIMER_TYPE_CLOSE, there is no sqbuf and
it should not be freed as one in i40iw_schedule_cm_timer().

Fixes: f27b4746f378 ("i40iw: add connection management code")
Signed-off-by: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/infiniband/hw/i40iw/i40iw_cm.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index 493d6ef..e145417 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -1043,7 +1043,7 @@ static int i40iw_parse_mpa(struct i40iw_cm_node *cm_node, u8 *buffer, u32 *type,
  * i40iw_schedule_cm_timer
  * @@cm_node: connection's node
  * @sqbuf: buffer to send
- * @type: if it es send ot close
+ * @type: if it is send or close
  * @send_retrans: if rexmits to be done
  * @close_when_complete: is cm_node to be removed
  *
@@ -1067,7 +1067,8 @@ int i40iw_schedule_cm_timer(struct i40iw_cm_node *cm_node,
 
 	new_send = kzalloc(sizeof(*new_send), GFP_ATOMIC);
 	if (!new_send) {
-		i40iw_free_sqbuf(vsi, (void *)sqbuf);
+		if (type != I40IW_TIMER_TYPE_CLOSE)
+			i40iw_free_sqbuf(vsi, (void *)sqbuf);
 		return -ENOMEM;
 	}
 	new_send->retrycount = I40IW_DEFAULT_RETRYS;
@@ -1082,7 +1083,6 @@ int i40iw_schedule_cm_timer(struct i40iw_cm_node *cm_node,
 		new_send->timetosend += (HZ / 10);
 		if (cm_node->close_entry) {
 			kfree(new_send);
-			i40iw_free_sqbuf(vsi, (void *)sqbuf);
 			i40iw_pr_err("already close entry\n");
 			return -EINVAL;
 		}
-- 
2.8.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 4/8] i40iw: Correct ARP index mask

[Henry Orosco]

From: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

ARP Index is 16 bits wide not 12 bits. Adjust
the mask for the full 16 bits.

Fixes: 4e9042e647ff ("i40iw: add hw and utils files")
Signed-off-by: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/infiniband/hw/i40iw/i40iw_d.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/i40iw/i40iw_d.h b/drivers/infiniband/hw/i40iw/i40iw_d.h
index 1077b78..029083c 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_d.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_d.h
@@ -1114,7 +1114,7 @@
 #define I40IWQPC_VLANTAG_MASK (0xffffULL << I40IWQPC_VLANTAG_SHIFT)
 
 #define I40IWQPC_ARPIDX_SHIFT 48
-#define I40IWQPC_ARPIDX_MASK (0xfffULL << I40IWQPC_ARPIDX_SHIFT)
+#define I40IWQPC_ARPIDX_MASK (0xffffULL << I40IWQPC_ARPIDX_SHIFT)
 
 #define I40IWQPC_FLOWLABEL_SHIFT 0
 #define I40IWQPC_FLOWLABEL_MASK (0xfffffUL << I40IWQPC_FLOWLABEL_SHIFT)
-- 
2.8.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 5/8] i40iw: Reinitialize add_sd_cnt

[Henry Orosco]

From: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

add_sd_cnt in info structure passed to i40iw_create_hmc_obj_type
must be 0 and since it is modified during the call, it must be
reset in the loop.

Fixes: 8e06af711bf2 ("i40iw: add main, hdr, status")
Signed-off-by: Mustafa Ismail <mustafa.ismail-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Shiraz Saleem <shiraz.saleem-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/infiniband/hw/i40iw/i40iw_main.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c
index 1f9d9c7..3aecf4c 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_main.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_main.c
@@ -483,6 +483,7 @@ static enum i40iw_status_code i40iw_create_hmc_objs(struct i40iw_device *iwdev,
 	for (i = 0; i < IW_HMC_OBJ_TYPE_NUM; i++) {
 		info.rsrc_type = iw_hmc_obj_types[i];
 		info.count = dev->hmc_info->hmc_obj[info.rsrc_type].cnt;
+		info.add_sd_cnt = 0;
 		status = i40iw_create_hmc_obj_type(dev, &info);
 		if (status) {
 			i40iw_pr_err("create obj type %d status = %d\n",
-- 
2.8.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 6/8] i40iw: Move MPA request event for loopback after connect

[Henry Orosco]

From: Tatyana Nikolova <tatyana.e.nikolova-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

For loopback, MPA request event is generated when cm_node
is initialized, which allows applications to act on the
connect request before i40iw_connect() has completed.
In some cases, the reject flow executes in parallel with
the connect flow and doesn't delete an APBVT entry,
because the apbvt_set variable is still not set by the
connect flow. Move the MPA request event to the end of
i40iw_connect() to notify application for a connect
request, after connect has completed.

Fixes: f27b4746f378 ("i40iw: add connection management code")
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Henry Orosco <henry.orosco-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/infiniband/hw/i40iw/i40iw_cm.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index e145417..3a35d56 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -2947,8 +2947,6 @@ static struct i40iw_cm_node *i40iw_create_cm_node(
 			loopback_remotenode->tcp_cntxt.snd_wnd = cm_node->tcp_cntxt.rcv_wnd;
 			cm_node->tcp_cntxt.snd_wscale = loopback_remotenode->tcp_cntxt.rcv_wscale;
 			loopback_remotenode->tcp_cntxt.snd_wscale = cm_node->tcp_cntxt.rcv_wscale;
-			loopback_remotenode->state = I40IW_CM_STATE_MPAREQ_RCVD;
-			i40iw_create_event(loopback_remotenode, I40IW_CM_EVENT_MPA_REQ);
 		}
 		return cm_node;
 	}
@@ -3864,6 +3862,11 @@ int i40iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
 			goto err;
 	}
 
+	if (cm_node->loopbackpartner) {
+		cm_node->loopbackpartner->state = I40IW_CM_STATE_MPAREQ_RCVD;
+		i40iw_create_event(cm_node->loopbackpartner, I40IW_CM_EVENT_MPA_REQ);
+	}
+
 	i40iw_debug(cm_node->dev,
 		    I40IW_DEBUG_CM,
 		    "Api - connect(): port=0x%04x, cm_node=%p, cm_id = %p.\n",
-- 
2.8.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 7/8] i40iw: Notify user of established connection after QP in RTS

[Henry Orosco]

Established CM event is sent prior to modifying QP to RTS state.
This can result in application closing the connection before the
QP is actually in RTS state. Move sending of established CM
event to after modify QP to RTS.

Fixes: f27b4746f378 ("i40iw: add connection management code")
Signed-off-by: Henry Orosco <henry.orosco-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/infiniband/hw/i40iw/i40iw_cm.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index 3a35d56..a2b242d 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -3687,11 +3687,15 @@ int i40iw_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
 	cm_id->add_ref(cm_id);
 	i40iw_add_ref(&iwqp->ibqp);
 
-	i40iw_send_cm_event(cm_node, cm_id, IW_CM_EVENT_ESTABLISHED, 0);
-
 	attr.qp_state = IB_QPS_RTS;
 	cm_node->qhash_set = false;
 	i40iw_modify_qp(&iwqp->ibqp, &attr, IB_QP_STATE, NULL);
+
+	cm_node->accelerated = 1;
+	status = i40iw_send_cm_event(cm_node, cm_id, IW_CM_EVENT_ESTABLISHED, 0);
+	if (status)
+		i40iw_debug(dev, I40IW_DEBUG_CM, "error sending cm event - ESTABLISHED\n");
+
 	if (cm_node->loopbackpartner) {
 		cm_node->loopbackpartner->pdata.size = conn_param->private_data_len;
 
@@ -3702,7 +3706,6 @@ int i40iw_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
 		i40iw_create_event(cm_node->loopbackpartner, I40IW_CM_EVENT_CONNECTED);
 	}
 
-	cm_node->accelerated = 1;
 	if (cm_node->accept_pend) {
 		atomic_dec(&cm_node->listener->pend_accepts_cnt);
 		cm_node->accept_pend = 0;
@@ -4047,9 +4050,6 @@ static void i40iw_cm_event_connected(struct i40iw_cm_event *event)
 	dev->iw_priv_qp_ops->qp_send_rtt(&iwqp->sc_qp, read0);
 	if (iwqp->page)
 		kunmap(iwqp->page);
-	status = i40iw_send_cm_event(cm_node, cm_id, IW_CM_EVENT_CONNECT_REPLY, 0);
-	if (status)
-		i40iw_pr_err("send cm event\n");
 
 	memset(&attr, 0, sizeof(attr));
 	attr.qp_state = IB_QPS_RTS;
@@ -4057,6 +4057,9 @@ static void i40iw_cm_event_connected(struct i40iw_cm_event *event)
 	i40iw_modify_qp(&iwqp->ibqp, &attr, IB_QP_STATE, NULL);
 
 	cm_node->accelerated = 1;
+	status = i40iw_send_cm_event(cm_node, cm_id, IW_CM_EVENT_CONNECT_REPLY, 0);
+	if (status)
+		i40iw_debug(dev, I40IW_DEBUG_CM, "error sending cm event - CONNECT_REPLY\n");
 
 	return;
 
-- 
2.8.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 8/8] i40iw: Change accelerated flag to bool

[Henry Orosco]

The accelerated flag only utilizes two values: 0 and 1.
Modify accelerated flag in struct i40iw_cm_node to bool.

Signed-off-by: Henry Orosco <henry.orosco-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/infiniband/hw/i40iw/i40iw_cm.c | 4 ++--
 drivers/infiniband/hw/i40iw/i40iw_cm.h | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index a2b242d..f41a302 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -3691,7 +3691,7 @@ int i40iw_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
 	cm_node->qhash_set = false;
 	i40iw_modify_qp(&iwqp->ibqp, &attr, IB_QP_STATE, NULL);
 
-	cm_node->accelerated = 1;
+	cm_node->accelerated = true;
 	status = i40iw_send_cm_event(cm_node, cm_id, IW_CM_EVENT_ESTABLISHED, 0);
 	if (status)
 		i40iw_debug(dev, I40IW_DEBUG_CM, "error sending cm event - ESTABLISHED\n");
@@ -4056,7 +4056,7 @@ static void i40iw_cm_event_connected(struct i40iw_cm_event *event)
 	cm_node->qhash_set = false;
 	i40iw_modify_qp(&iwqp->ibqp, &attr, IB_QP_STATE, NULL);
 
-	cm_node->accelerated = 1;
+	cm_node->accelerated = true;
 	status = i40iw_send_cm_event(cm_node, cm_id, IW_CM_EVENT_CONNECT_REPLY, 0);
 	if (status)
 		i40iw_debug(dev, I40IW_DEBUG_CM, "error sending cm event - CONNECT_REPLY\n");
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.h b/drivers/infiniband/hw/i40iw/i40iw_cm.h
index 0d5840d..e09b1f5 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.h
@@ -337,7 +337,7 @@ struct i40iw_cm_node {
 	u16     mpav2_ird_ord;
 	struct iw_cm_id *cm_id;
 	struct list_head list;
-	int accelerated;
+	bool accelerated;
 	struct i40iw_cm_listener *listener;
 	int apbvt_set;
 	int accept_pend;
-- 
2.8.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 8/8] i40iw: Change accelerated flag to bool

[Jason Gunthorpe]

On Fri, Nov 17, 2017 at 10:46:57AM -0600, Henry Orosco wrote:
> The accelerated flag only utilizes two values: 0 and 1.
> Modify accelerated flag in struct i40iw_cm_node to bool.

I think this is not rc material

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 8/8] i40iw: Change accelerated flag to bool

[Shiraz Saleem]

On Mon, Nov 20, 2017 at 11:39:54AM -0700, Jason Gunthorpe wrote:
> On Fri, Nov 17, 2017 at 10:46:57AM -0600, Henry Orosco wrote:
> > The accelerated flag only utilizes two values: 0 and 1.
> > Modify accelerated flag in struct i40iw_cm_node to bool.
> 
> I think this is not rc material
>

That's right. It can be deferred for 4.16

Shiraz
 
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html