Skip to contentSkip to conversation

staging/rdma/hfi1: Remove incorrect link credit check

PATCH New Jubin John 4 participants 7 messages Dec 18, 2015 – Dec 23, 2015 2 messages not in the archive

Subsystems (2)
lore
Share & tools
lore

Jubin John

author
#0 lore raw

PATCH 1/3 staging/rdma/hfi1: Remove incorrect link credit check

From: Dean Luick <[email protected]>

Remove an invalid sanity check that compares the local link credits with the peer link credits. The two have no dependency on each other.

---
 drivers/staging/rdma/hfi1/chip.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)
drivers/staging/rdma/hfi1/chip.c +1 -2
index dc69159..78a5f08 100644
@@ -7267,8 +7267,7 @@ static int set_buffer_control(struct hfi1_devdata *dd,
 		new_bc->vl[i].shared = 0;
 	}
 	new_total += be16_to_cpu(new_bc->overall_shared_limit);
-	if (new_total > (u32)dd->link_credits)
-		return -EINVAL;
+
 	/* fetch the current values */
 	get_buffer_control(dd, &cur_bc, &cur_total);
 
Signature · 1 line
1.7.1

A message that is not in the archive (sent off-list or not yet received): 1450398255-10346-1-git-send-email-jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org

#2 lore raw

PATCH 2/3 staging/rdma/hfi1: Fix module parameter spelling

Fix the spelling of user_credit_return_threshold, it was incorrectly spelled as user_credit_return_theshold causing two module parameters, one with typo, to be shown in modinfo

---
 drivers/staging/rdma/hfi1/init.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/staging/rdma/hfi1/init.c +1 -1
index 1c8286f..8f13d53 100644
@@ -113,7 +113,7 @@ MODULE_PARM_DESC(hdrq_entsize, "Size of header queue entries: 2 - 8B, 16 - 64B (
 
 unsigned int user_credit_return_threshold = 33;	/* default is 33% */
 module_param(user_credit_return_threshold, uint, S_IRUGO);
-MODULE_PARM_DESC(user_credit_return_theshold, "Credit return threshold for user send contexts, return when unreturned credits passes this many blocks (in percent of allocated blocks, 0 is off)");
+MODULE_PARM_DESC(user_credit_return_threshold, "Credit return threshold for user send contexts, return when unreturned credits passes this many blocks (in percent of allocated blocks, 0 is off)");
 
 static inline u64 encode_rcv_header_entry_size(u16);
 
Signature · 6 lines
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
#3 lore raw

Re: [PATCH 2/3] staging/rdma/hfi1: Fix module parameter spelling

Nice. Whenever you see a bug like this, you should report it to kernel-janitors because you know that 10 other people have made the same mistake.

I will take care of it.

regards,
dan carpenter

author
#4 lore raw

Re: [PATCH 2/3] staging/rdma/hfi1: Fix module parameter spelling

Quoted 3 lines · Dan Carpenter

Nice. Whenever you see a bug like this, you should report it to kernel-janitors because you know that 10 other people have made the same mistake.

Sure, I will do that for similar bugs.

Quoted 2 lines

I will take care of it.

Thank you.

Quoted 4 lines

regards,
dan carpenter

#5 lore raw

PATCH 3/3 staging/rdma/hfi1: check for ARMED->ACTIVE transition in receive interrupt

In reply to a message not in the archive

From: Jim Snow <[email protected]>

The link state will transition from ARMED to ACTIVE when a non-SC15 packet arrives, but the driver might not notice the change. With this fix, if the slowpath receive interrupt handler sees a non-SC15 packet while in the ARMED state, we queue work to call linkstate_active_work from process context to promote it to ACTIVE.

---
 drivers/staging/rdma/hfi1/chip.c   |    5 ++-
 drivers/staging/rdma/hfi1/chip.h   |    2 +
 drivers/staging/rdma/hfi1/driver.c |   66 ++++++++++++++++++++++++++++++++++++
 drivers/staging/rdma/hfi1/hfi.h    |   12 ++++++
 drivers/staging/rdma/hfi1/init.c   |    1 +
 5 files changed, 84 insertions(+), 2 deletions(-)
drivers/staging/rdma/hfi1/chip.c +3 -2
index 78a5f08..f82b848 100644
@@ -4606,7 +4606,7 @@ static inline void clear_recv_intr(struct hfi1_ctxtdata *rcd)
 }
 
 /* force the receive interrupt */
-static inline void force_recv_intr(struct hfi1_ctxtdata *rcd)
+void force_recv_intr(struct hfi1_ctxtdata *rcd)
 {
 	write_csr(rcd->dd, CCE_INT_FORCE + (8 * rcd->ireg), rcd->imask);
 }
@@ -4705,7 +4705,7 @@ u32 read_physical_state(struct hfi1_devdata *dd)
 				& DC_DC8051_STS_CUR_STATE_PORT_MASK;
 }
 
-static u32 read_logical_state(struct hfi1_devdata *dd)
+u32 read_logical_state(struct hfi1_devdata *dd)
 {
 	u64 reg;
 
@@ -6658,6 +6658,7 @@ int set_link_state(struct hfi1_pportdata *ppd, u32 state)
 				ppd->link_enabled = 1;
 		}
 
+		set_all_slowpath(ppd->dd);
 		ret = set_local_link_attributes(ppd);
 		if (ret)
 			break;
drivers/staging/rdma/hfi1/chip.h +2 -0
index d74aed8..620cf08 100644
@@ -691,6 +691,8 @@ u64 read_dev_cntr(struct hfi1_devdata *dd, int index, int vl);
 u64 write_dev_cntr(struct hfi1_devdata *dd, int index, int vl, u64 data);
 u64 read_port_cntr(struct hfi1_pportdata *ppd, int index, int vl);
 u64 write_port_cntr(struct hfi1_pportdata *ppd, int index, int vl, u64 data);
+u32 read_logical_state(struct hfi1_devdata *dd);
+void force_recv_intr(struct hfi1_ctxtdata *rcd);
 
 /* Per VL indexes */
 enum {
drivers/staging/rdma/hfi1/driver.c +66 -0

84 lines not on this page · raw message

drivers/staging/rdma/hfi1/hfi.h +12 -0
index 54ed6b3..bfd9723 100644
@@ -712,6 +712,7 @@ struct hfi1_pportdata {
 	u8 remote_link_down_reason;
 	/* Error events that will cause a port bounce. */
 	u32 port_error_action;
+	struct work_struct linkstate_active_work;
 };
 
 typedef int (*rhf_rcv_function_ptr)(struct hfi1_packet *packet);
@@ -1128,6 +1129,7 @@ void hfi1_free_ctxtdata(struct hfi1_devdata *, struct hfi1_ctxtdata *);
 int handle_receive_interrupt(struct hfi1_ctxtdata *, int);
 int handle_receive_interrupt_nodma_rtail(struct hfi1_ctxtdata *, int);
 int handle_receive_interrupt_dma_rtail(struct hfi1_ctxtdata *, int);
+void set_all_slowpath(struct hfi1_devdata *dd);
 
 /* receive packet handler dispositions */
 #define RCV_PKT_OK      0x0 /* keep going */
@@ -1148,6 +1150,16 @@ static inline u32 driver_lstate(struct hfi1_pportdata *ppd)
 	return ppd->lstate; /* use the cached value */
 }
 
+void receive_interrupt_work(struct work_struct *work);
+
+/* extract service channel from header and rhf */
+static inline int hdr2sc(struct hfi1_message_header *hdr, u64 rhf)
+{
+	return
+		((be16_to_cpu(hdr->lrh[0]) >> 12) & 0xf) |
+		((!!(rhf & RHF_DC_INFO_MASK)) << 4);
+}
+
 static inline u16 generate_jkey(kuid_t uid)
 {
 	return from_kuid(current_user_ns(), uid) & 0xffff;
drivers/staging/rdma/hfi1/init.c +1 -0
index 8f13d53..ee206ae 100644
@@ -498,6 +498,7 @@ void hfi1_init_pportdata(struct pci_dev *pdev, struct hfi1_pportdata *ppd,
 	INIT_WORK(&ppd->link_downgrade_work, handle_link_downgrade);
 	INIT_WORK(&ppd->sma_message_work, handle_sma_message);
 	INIT_WORK(&ppd->link_bounce_work, handle_link_bounce);
+	INIT_WORK(&ppd->linkstate_active_work, receive_interrupt_work);
 	mutex_init(&ppd->hls_lock);
 	spin_lock_init(&ppd->sdma_alllock);
 	spin_lock_init(&ppd->qsfp_info.qsfp_lock);
Signature · 6 lines
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

A message that is not in the archive (sent off-list or not yet received): 1450398255-10346-3-git-send-email-jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org

#7 lore raw

Re: [PATCH 3/3] staging/rdma/hfi1: check for ARMED->ACTIVE transition in receive interrupt

Possible off by one, but mostly the whitespace makes me itch.

Jim was not on the CC list.

Quoted 68 lines · Jubin John

From: Jim Snow <[email protected]>

The link state will transition from ARMED to ACTIVE when a non-SC15 packet arrives, but the driver might not notice the change. With this fix, if the slowpath receive interrupt handler sees a non-SC15 packet while in the ARMED state, we queue work to call linkstate_active_work from process context to promote it to ACTIVE.

---
 drivers/staging/rdma/hfi1/chip.c   |    5 ++-
 drivers/staging/rdma/hfi1/chip.h   |    2 +
 drivers/staging/rdma/hfi1/driver.c |   66 ++++++++++++++++++++++++++++++++++++
 drivers/staging/rdma/hfi1/hfi.h    |   12 ++++++
 drivers/staging/rdma/hfi1/init.c   |    1 +
 5 files changed, 84 insertions(+), 2 deletions(-)
drivers/staging/rdma/hfi1/chip.c +3 -2
index 78a5f08..f82b848 100644
@@ -4606,7 +4606,7 @@ static inline void clear_recv_intr(struct hfi1_ctxtdata *rcd)
 }
 
 /* force the receive interrupt */
-static inline void force_recv_intr(struct hfi1_ctxtdata *rcd)
+void force_recv_intr(struct hfi1_ctxtdata *rcd)
 {
 	write_csr(rcd->dd, CCE_INT_FORCE + (8 * rcd->ireg), rcd->imask);
 }
@@ -4705,7 +4705,7 @@ u32 read_physical_state(struct hfi1_devdata *dd)
 				& DC_DC8051_STS_CUR_STATE_PORT_MASK;
 }
 
-static u32 read_logical_state(struct hfi1_devdata *dd)
+u32 read_logical_state(struct hfi1_devdata *dd)
 {
 	u64 reg;
 
@@ -6658,6 +6658,7 @@ int set_link_state(struct hfi1_pportdata *ppd, u32 state)
 				ppd->link_enabled = 1;
 		}
 
+		set_all_slowpath(ppd->dd);
 		ret = set_local_link_attributes(ppd);
 		if (ret)
 			break;
drivers/staging/rdma/hfi1/chip.h +2 -0
index d74aed8..620cf08 100644
@@ -691,6 +691,8 @@ u64 read_dev_cntr(struct hfi1_devdata *dd, int index, int vl);
 u64 write_dev_cntr(struct hfi1_devdata *dd, int index, int vl, u64 data);
 u64 read_port_cntr(struct hfi1_pportdata *ppd, int index, int vl);
 u64 write_port_cntr(struct hfi1_pportdata *ppd, int index, int vl, u64 data);
+u32 read_logical_state(struct hfi1_devdata *dd);
+void force_recv_intr(struct hfi1_ctxtdata *rcd);
 
 /* Per VL indexes */
 enum {
drivers/staging/rdma/hfi1/driver.c +0 -0
index 4c52e78..16b1be1 100644

drivers/staging/rdma/hfi1/driver.c static inline void set_all_dma_rtail(struct hfi1_devdata *dd) · line 871

Quoted 10 lines
drivers/staging/rdma/hfi1/driver.c +7 -0
@@ -862,6 +862,15 @@ static inline void set_all_dma_rtail(struct hfi1_devdata *dd)
 			&handle_receive_interrupt_dma_rtail;
 }
 
+void set_all_slowpath(struct hfi1_devdata *dd)
+{
+	int i;
+
+	for (i = HFI1_CTRL_CTXT + 1; i < dd->first_user_ctxt; i++)
+		dd->rcd[i]->do_interrupt =
+		  &handle_receive_interrupt;

This fits within the 80 character limit.

We start counting from HFI1_CTRL_CTXT + 1 but in receive_interrupt_work() we start counting from HFI1_CTRL_CTXT. What's the story? It's either a bug or needs much better documentation.

Quoted 20 lines
drivers/staging/rdma/hfi1/driver.c +13 -0
+}
+
 /*
  * handle_receive_interrupt - receive a packet
  * @rcd: the context
@@ -929,6 +938,27 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 			last = skip_rcv_packet(&packet, thread);
 			skip_pkt = 0;
 		} else {
+			/* Auto activate link on non-SC15 packet receive */
+			if (unlikely(rcd->ppd->host_link_state ==
+				     HLS_UP_ARMED)) {
+				struct work_struct *lsaw =
+				  &rcd->ppd->linkstate_active_work;
+				struct hfi1_message_header *hdr =
+				  hfi1_get_msgheader(packet.rcd->dd,
+						     packet.rhf_addr);
+				if (hdr2sc(hdr, packet.rhf) != 0xf) {
+					int hwstate = read_logical_state(dd);
+

drivers/staging/rdma/hfi1/driver.c int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread) · line 961

Quoted 10 lines
drivers/staging/rdma/hfi1/driver.c +10 -0
+					if (hwstate != LSTATE_ACTIVE) {
+						dd_dev_info(dd, "Unexpected link state %d\n",
+							    hwstate);
+					} else {
+						queue_work(
+						  rcd->ppd->hfi1_wq, lsaw);
+						goto bail;
+					}
+				}
+			}

Can we make this an inline function?

Quoted 17 lines
drivers/staging/rdma/hfi1/driver.c +10 -0
 			last = process_rcv_packet(&packet, thread);
 		}
 
@@ -984,6 +1014,42 @@ bail:
 }
 
 /*
+ * We may discover in the interrupt that the hardware link state has
+ * changed from ARMED to ACTIVE (due to the arrival of a non-SC15 packet),
+ * and we need to update the driver's notion of the link state.  We cannot
+ * run set_link_state from interrupt context, so we queue this function on
+ * a workqueue.
+ *
+ * We delay the regular interrupt processing until after the state changes
+ * so that the link will be in the correct state by the time any application
+ * we wake up attempts to send a reply to any message it received.
+ * (Subsequent receive interrupts may possibly force the wakeup before we

drivers/staging/rdma/hfi1/driver.c bail: · line 1036

Quoted 10 lines
drivers/staging/rdma/hfi1/driver.c +10 -0
+ * update the link state.)
+ *
+ * The rcd is freed in hfi1_free_ctxtdata after hfi1_postinit_cleanup invokes
+ * dd->f_cleanup(dd) to disable the interrupt handler and flush workqueues,
+ * so we're safe from use-after-free of the rcd.
+ */
+void receive_interrupt_work(struct work_struct *work)
+{
+	struct hfi1_pportdata *ppd =
+	  container_of(work, struct hfi1_pportdata, linkstate_active_work);

Normally we would put mostly on the first line.

	struct hfi1_pportdata *ppd = container_of(work, struct hfi1_pportdata,
						  linkstate_active_work);

drivers/staging/rdma/hfi1/driver.c bail: · line 1046

Quoted 10 lines
drivers/staging/rdma/hfi1/driver.c +10 -0
+	struct hfi1_devdata *dd = ppd->dd;
+	int i;
+
+	/* Received non-SC15 packet implies neighbor_normal */
+	ppd->neighbor_normal = 1;
+	set_link_state(ppd, HLS_UP_ACTIVE);
+
+	/*
+	 * Interrupt all kernel contexts that could have had an
+	 *  interrupt during auto activation.

Remove the extra space before "interrupt".

Quoted 31 lines
drivers/staging/rdma/hfi1/driver.c +6 -0
+	 */
+	for (i = HFI1_CTRL_CTXT; i < dd->first_user_ctxt; i++)
+		force_recv_intr(dd->rcd[i]);
+}
+
+/*
  * Convert a given MTU size to the on-wire MAD packet enumeration.
  * Return -1 if the size is invalid.
  */
drivers/staging/rdma/hfi1/hfi.h +2 -0
index 54ed6b3..bfd9723 100644
@@ -712,6 +712,7 @@ struct hfi1_pportdata {
 	u8 remote_link_down_reason;
 	/* Error events that will cause a port bounce. */
 	u32 port_error_action;
+	struct work_struct linkstate_active_work;
 };
 
 typedef int (*rhf_rcv_function_ptr)(struct hfi1_packet *packet);
@@ -1128,6 +1129,7 @@ void hfi1_free_ctxtdata(struct hfi1_devdata *, struct hfi1_ctxtdata *);
 int handle_receive_interrupt(struct hfi1_ctxtdata *, int);
 int handle_receive_interrupt_nodma_rtail(struct hfi1_ctxtdata *, int);
 int handle_receive_interrupt_dma_rtail(struct hfi1_ctxtdata *, int);
+void set_all_slowpath(struct hfi1_devdata *dd);
 
 /* receive packet handler dispositions */
 #define RCV_PKT_OK      0x0 /* keep going */
@@ -1148,6 +1150,16 @@ static inline u32 driver_lstate(struct hfi1_pportdata *ppd)
 	return ppd->lstate; /* use the cached value */

drivers/staging/rdma/hfi1/hfi.h static inline u32 driver_lstate(struct hfi1_pportdata *ppd) · line 1160

Quoted 10 lines
drivers/staging/rdma/hfi1/hfi.h +8 -0
 }
 
+void receive_interrupt_work(struct work_struct *work);
+
+/* extract service channel from header and rhf */
+static inline int hdr2sc(struct hfi1_message_header *hdr, u64 rhf)
+{
+	return
+		((be16_to_cpu(hdr->lrh[0]) >> 12) & 0xf) |
+		((!!(rhf & RHF_DC_INFO_MASK)) << 4);

This should be:

	return ((be16_to_cpu(hdr->lrh[0]) >> 12) & 0xf) |
	       ((!!(rhf & RHF_DC_INFO_MASK)) << 4);
Quoted 24 lines
drivers/staging/rdma/hfi1/hfi.h +2 -0
+}
+
 static inline u16 generate_jkey(kuid_t uid)
 {
 	return from_kuid(current_user_ns(), uid) & 0xffff;
drivers/staging/rdma/hfi1/init.c +1 -0
index 8f13d53..ee206ae 100644
@@ -498,6 +498,7 @@ void hfi1_init_pportdata(struct pci_dev *pdev, struct hfi1_pportdata *ppd,
 	INIT_WORK(&ppd->link_downgrade_work, handle_link_downgrade);
 	INIT_WORK(&ppd->sma_message_work, handle_sma_message);
 	INIT_WORK(&ppd->link_bounce_work, handle_link_bounce);
+	INIT_WORK(&ppd->linkstate_active_work, receive_interrupt_work);
 	mutex_init(&ppd->hls_lock);
 	spin_lock_init(&ppd->sdma_alllock);
 	spin_lock_init(&ppd->qsfp_info.qsfp_lock);
Signature · 6 lines
1.7.1

_______________________________________________
devel mailing list
[email protected]
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

#8 lore raw

Re: [PATCH 3/3] staging/rdma/hfi1: check for ARMED->ACTIVE transition in receive interrupt

Quoted 76 lines · Dan Carpenter

Possible off by one, but mostly the whitespace makes me itch.

Jim was not on the CC list.

Quoted 71 lines · Jubin John

From: Jim Snow <[email protected]>

The link state will transition from ARMED to ACTIVE when a non-SC15 packet arrives, but the driver might not notice the change. With this fix, if the slowpath receive interrupt handler sees a non-SC15 packet while in the ARMED state, we queue work to call linkstate_active_work from process context to promote it to ACTIVE.

---
 drivers/staging/rdma/hfi1/chip.c   |    5 ++-
 drivers/staging/rdma/hfi1/chip.h   |    2 +
 drivers/staging/rdma/hfi1/driver.c |   66 ++++++++++++++++++++++++++++++++++++
 drivers/staging/rdma/hfi1/hfi.h    |   12 ++++++
 drivers/staging/rdma/hfi1/init.c   |    1 +
 5 files changed, 84 insertions(+), 2 deletions(-)
drivers/staging/rdma/hfi1/chip.c +3 -2
index 78a5f08..f82b848 100644
@@ -4606,7 +4606,7 @@ static inline void clear_recv_intr(struct hfi1_ctxtdata *rcd)
 }
 
 /* force the receive interrupt */
-static inline void force_recv_intr(struct hfi1_ctxtdata *rcd)
+void force_recv_intr(struct hfi1_ctxtdata *rcd)
 {
 	write_csr(rcd->dd, CCE_INT_FORCE + (8 * rcd->ireg), rcd->imask);
 }
@@ -4705,7 +4705,7 @@ u32 read_physical_state(struct hfi1_devdata *dd)
 				& DC_DC8051_STS_CUR_STATE_PORT_MASK;
 }
 
-static u32 read_logical_state(struct hfi1_devdata *dd)
+u32 read_logical_state(struct hfi1_devdata *dd)
 {
 	u64 reg;
 
@@ -6658,6 +6658,7 @@ int set_link_state(struct hfi1_pportdata *ppd, u32 state)
 				ppd->link_enabled = 1;
 		}
 
+		set_all_slowpath(ppd->dd);
 		ret = set_local_link_attributes(ppd);
 		if (ret)
 			break;
drivers/staging/rdma/hfi1/chip.h +2 -0
index d74aed8..620cf08 100644
@@ -691,6 +691,8 @@ u64 read_dev_cntr(struct hfi1_devdata *dd, int index, int vl);
 u64 write_dev_cntr(struct hfi1_devdata *dd, int index, int vl, u64 data);
 u64 read_port_cntr(struct hfi1_pportdata *ppd, int index, int vl);
 u64 write_port_cntr(struct hfi1_pportdata *ppd, int index, int vl, u64 data);
+u32 read_logical_state(struct hfi1_devdata *dd);
+void force_recv_intr(struct hfi1_ctxtdata *rcd);
 
 /* Per VL indexes */
 enum {
drivers/staging/rdma/hfi1/driver.c +0 -0
index 4c52e78..16b1be1 100644
@@ -862,6 +862,15 @@ static inline void set_all_dma_rtail(struct hfi1_devdata *dd)
 			&handle_receive_interrupt_dma_rtail;
 }
 
Quoted 9 lines

drivers/staging/rdma/hfi1/driver.c static inline void set_all_dma_rtail(struct hfi1_devdata *dd) · line 871

Quoted 7 lines
drivers/staging/rdma/hfi1/driver.c +7 -0
+void set_all_slowpath(struct hfi1_devdata *dd)
+{
+	int i;
+
+	for (i = HFI1_CTRL_CTXT + 1; i < dd->first_user_ctxt; i++)
+		dd->rcd[i]->do_interrupt =
+		  &handle_receive_interrupt;

This fits within the 80 character limit.

Will fix in v2.

Quoted 4 lines

We start counting from HFI1_CTRL_CTXT + 1 but in receive_interrupt_work() we start counting from HFI1_CTRL_CTXT. What's the story? It's either a bug or needs much better documentation.

In set_all_slowpath(), we exclude HFI1_CTRL_CTXT because HFI1_CTRL_CTXT (the SC15, error, and multicast context) must always use the slowpath handler. We skip HFI1_CTRL_CTXT in set_all_nodma_rtail() and set_all_dma_rtail() as well.

receive_interrupt_work() starts at HFI1_CTRL_CTXT because we want to trigger interrupts on all of the receive contexts, including HFI1_CTRL_CTXT in case any packets were received on any of the contexts between the interrupt and when the work function is called.

Quoted 25 lines
Quoted 25 lines
drivers/staging/rdma/hfi1/driver.c +17 -0
+}
+
 /*
  * handle_receive_interrupt - receive a packet
  * @rcd: the context
@@ -929,6 +938,27 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 			last = skip_rcv_packet(&packet, thread);
 			skip_pkt = 0;
 		} else {
+			/* Auto activate link on non-SC15 packet receive */
+			if (unlikely(rcd->ppd->host_link_state ==
+				     HLS_UP_ARMED)) {
+				struct work_struct *lsaw =
+				  &rcd->ppd->linkstate_active_work;
+				struct hfi1_message_header *hdr =
+				  hfi1_get_msgheader(packet.rcd->dd,
+						     packet.rhf_addr);
+				if (hdr2sc(hdr, packet.rhf) != 0xf) {
+					int hwstate = read_logical_state(dd);
+
+					if (hwstate != LSTATE_ACTIVE) {
+						dd_dev_info(dd, "Unexpected link state %d\n",
+							    hwstate);
+					} else {
Quoted 8 lines

drivers/staging/rdma/hfi1/driver.c int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread) · line 961

Quoted 6 lines
drivers/staging/rdma/hfi1/driver.c +6 -0
+						queue_work(
+						  rcd->ppd->hfi1_wq, lsaw);
+						goto bail;
+					}
+				}
+			}

Can we make this an inline function?

Yes, will make that change in v2.

Quoted 31 lines
Quoted 18 lines
drivers/staging/rdma/hfi1/driver.c +10 -0
 			last = process_rcv_packet(&packet, thread);
 		}
 
@@ -984,6 +1014,42 @@ bail:
 }
 
 /*
+ * We may discover in the interrupt that the hardware link state has
+ * changed from ARMED to ACTIVE (due to the arrival of a non-SC15 packet),
+ * and we need to update the driver's notion of the link state.  We cannot
+ * run set_link_state from interrupt context, so we queue this function on
+ * a workqueue.
+ *
+ * We delay the regular interrupt processing until after the state changes
+ * so that the link will be in the correct state by the time any application
+ * we wake up attempts to send a reply to any message it received.
+ * (Subsequent receive interrupts may possibly force the wakeup before we

drivers/staging/rdma/hfi1/driver.c bail: · line 1036

Quoted 10 lines
drivers/staging/rdma/hfi1/driver.c +10 -0
+ * update the link state.)
+ *
+ * The rcd is freed in hfi1_free_ctxtdata after hfi1_postinit_cleanup invokes
+ * dd->f_cleanup(dd) to disable the interrupt handler and flush workqueues,
+ * so we're safe from use-after-free of the rcd.
+ */
+void receive_interrupt_work(struct work_struct *work)
+{
+	struct hfi1_pportdata *ppd =
+	  container_of(work, struct hfi1_pportdata, linkstate_active_work);

Normally we would put mostly on the first line.

Quoted 2 lines
	struct hfi1_pportdata *ppd = container_of(work, struct hfi1_pportdata,
						  linkstate_active_work);

Fixed in v2.

Quoted 13 lines

drivers/staging/rdma/hfi1/driver.c bail: · line 1046

Quoted 11 lines
drivers/staging/rdma/hfi1/driver.c +10 -0
+	struct hfi1_devdata *dd = ppd->dd;
+	int i;
+
+	/* Received non-SC15 packet implies neighbor_normal */
+	ppd->neighbor_normal = 1;
+	set_link_state(ppd, HLS_UP_ACTIVE);
+
+	/*
+	 * Interrupt all kernel contexts that could have had an
+	 *  interrupt during auto activation.

Remove the extra space before "interrupt".

Fixed in v2.

Quoted 45 lines
Quoted 32 lines
drivers/staging/rdma/hfi1/driver.c +6 -0
+	 */
+	for (i = HFI1_CTRL_CTXT; i < dd->first_user_ctxt; i++)
+		force_recv_intr(dd->rcd[i]);
+}
+
+/*
  * Convert a given MTU size to the on-wire MAD packet enumeration.
  * Return -1 if the size is invalid.
  */
drivers/staging/rdma/hfi1/hfi.h +2 -0
index 54ed6b3..bfd9723 100644
@@ -712,6 +712,7 @@ struct hfi1_pportdata {
 	u8 remote_link_down_reason;
 	/* Error events that will cause a port bounce. */
 	u32 port_error_action;
+	struct work_struct linkstate_active_work;
 };
 
 typedef int (*rhf_rcv_function_ptr)(struct hfi1_packet *packet);
@@ -1128,6 +1129,7 @@ void hfi1_free_ctxtdata(struct hfi1_devdata *, struct hfi1_ctxtdata *);
 int handle_receive_interrupt(struct hfi1_ctxtdata *, int);
 int handle_receive_interrupt_nodma_rtail(struct hfi1_ctxtdata *, int);
 int handle_receive_interrupt_dma_rtail(struct hfi1_ctxtdata *, int);
+void set_all_slowpath(struct hfi1_devdata *dd);
 
 /* receive packet handler dispositions */
 #define RCV_PKT_OK      0x0 /* keep going */
@@ -1148,6 +1150,16 @@ static inline u32 driver_lstate(struct hfi1_pportdata *ppd)
 	return ppd->lstate; /* use the cached value */

drivers/staging/rdma/hfi1/hfi.h static inline u32 driver_lstate(struct hfi1_pportdata *ppd) · line 1160

Quoted 10 lines
drivers/staging/rdma/hfi1/hfi.h +8 -0
 }
 
+void receive_interrupt_work(struct work_struct *work);
+
+/* extract service channel from header and rhf */
+static inline int hdr2sc(struct hfi1_message_header *hdr, u64 rhf)
+{
+	return
+		((be16_to_cpu(hdr->lrh[0]) >> 12) & 0xf) |
+		((!!(rhf & RHF_DC_INFO_MASK)) << 4);

This should be:

Quoted 2 lines
	return ((be16_to_cpu(hdr->lrh[0]) >> 12) & 0xf) |
	       ((!!(rhf & RHF_DC_INFO_MASK)) << 4);

Fixed in v2.

Quoted 21 lines
Quoted 21 lines
drivers/staging/rdma/hfi1/hfi.h +2 -0
+}
+
 static inline u16 generate_jkey(kuid_t uid)
 {
 	return from_kuid(current_user_ns(), uid) & 0xffff;
drivers/staging/rdma/hfi1/init.c +1 -0
index 8f13d53..ee206ae 100644
@@ -498,6 +498,7 @@ void hfi1_init_pportdata(struct pci_dev *pdev, struct hfi1_pportdata *ppd,
 	INIT_WORK(&ppd->link_downgrade_work, handle_link_downgrade);
 	INIT_WORK(&ppd->sma_message_work, handle_sma_message);
 	INIT_WORK(&ppd->link_bounce_work, handle_link_bounce);
+	INIT_WORK(&ppd->linkstate_active_work, receive_interrupt_work);
 	mutex_init(&ppd->hls_lock);
 	spin_lock_init(&ppd->sdma_alllock);
 	spin_lock_init(&ppd->qsfp_info.qsfp_lock);
Signature · 1 line
1.7.1
Quoted 4 lines
Quoted 4 lines

_______________________________________________
devel mailing list [email protected]
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html