Skip to contentSkip to conversation

i40e/i40evf: refactor tx timeout logic

PATCH New Catherine Sullivan 2 participants 3 messages Sep 24, 2015 – Oct 22, 2015

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v3 2/2 i40e/i40evf: refactor tx timeout logic

From: Kiran Patil <[email protected]>

This patch modifies the driver timeout logic by issuing a writeback request via a software interrupt to the hardware the first time the driver detects a hang. The driver was too agrressive in resetting a hung queue, so back that off by removing logic to down the netdevice after too many hangs, and move the function to the service task.

Change-ID Ife100b9d124cd08cbdb81ab659008c1b9abbedea

---
v3: Made i40evf_force_wb static.

 drivers/net/ethernet/intel/i40e/i40e.h        |   1 -
 drivers/net/ethernet/intel/i40e/i40e_main.c   | 274 +++++++++++++++-----------
 drivers/net/ethernet/intel/i40e/i40e_txrx.c   |  75 +------
 drivers/net/ethernet/intel/i40e/i40e_txrx.h   |  10 +-
 drivers/net/ethernet/intel/i40evf/i40e_txrx.c |  96 +--------
 drivers/net/ethernet/intel/i40evf/i40e_txrx.h |   8 -
 6 files changed, 173 insertions(+), 291 deletions(-)
drivers/net/ethernet/intel/i40e/i40e.h +0 -1
index e87d967..0ed6506 100644
@@ -246,7 +246,6 @@ struct i40e_pf {
 	struct pci_dev *pdev;
 	struct i40e_hw hw;
 	unsigned long state;
-	unsigned long link_check_timeout;
 	struct msix_entry *msix_entries;
 	bool fc_autoneg_status;
 
drivers/net/ethernet/intel/i40e/i40e_main.c +161 -113

355 lines not on this page · raw message

drivers/net/ethernet/intel/i40e/i40e_txrx.c +2 -73

105 lines not on this page · raw message

drivers/net/ethernet/intel/i40e/i40e_txrx.h +2 -8
index f67f8cf..c8a7cb4 100644
@@ -201,8 +201,6 @@ struct i40e_rx_queue_stats {
 enum i40e_ring_state_t {
 	__I40E_TX_FDIR_INIT_DONE,
 	__I40E_TX_XPS_INIT_DONE,
-	__I40E_TX_DETECT_HANG,
-	__I40E_HANG_CHECK_ARMED,
 	__I40E_RX_PS_ENABLED,
 	__I40E_RX_16BYTE_DESC_ENABLED,
 };
@@ -213,12 +211,6 @@ enum i40e_ring_state_t {
 	set_bit(__I40E_RX_PS_ENABLED, &(ring)->state)
 #define clear_ring_ps_enabled(ring) \
 	clear_bit(__I40E_RX_PS_ENABLED, &(ring)->state)
-#define check_for_tx_hang(ring) \
-	test_bit(__I40E_TX_DETECT_HANG, &(ring)->state)
-#define set_check_for_tx_hang(ring) \
-	set_bit(__I40E_TX_DETECT_HANG, &(ring)->state)
-#define clear_check_for_tx_hang(ring) \
-	clear_bit(__I40E_TX_DETECT_HANG, &(ring)->state)
 #define ring_is_16byte_desc_enabled(ring) \
 	test_bit(__I40E_RX_16BYTE_DESC_ENABLED, &(ring)->state)
 #define set_ring_16byte_desc_enabled(ring) \
@@ -328,6 +320,8 @@ int i40e_xmit_descriptor_count(struct sk_buff *skb, struct i40e_ring *tx_ring);
 int i40e_tx_prepare_vlan_flags(struct sk_buff *skb,
 			       struct i40e_ring *tx_ring, u32 *flags);
 #endif
+void i40e_force_wb(struct i40e_vsi *vsi, struct i40e_q_vector *q_vector);
+u32 i40e_get_tx_pending(struct i40e_ring *ring);
 
 /**
  * i40e_get_head - Retrieve head from head writeback
drivers/net/ethernet/intel/i40evf/i40e_txrx.c +8 -88

133 lines not on this page · raw message

drivers/net/ethernet/intel/i40evf/i40e_txrx.h +0 -8
index 26fc85f..0c13ece 100644
@@ -200,8 +200,6 @@ struct i40e_rx_queue_stats {
 enum i40e_ring_state_t {
 	__I40E_TX_FDIR_INIT_DONE,
 	__I40E_TX_XPS_INIT_DONE,
-	__I40E_TX_DETECT_HANG,
-	__I40E_HANG_CHECK_ARMED,
 	__I40E_RX_PS_ENABLED,
 	__I40E_RX_16BYTE_DESC_ENABLED,
 };
@@ -212,12 +210,6 @@ enum i40e_ring_state_t {
 	set_bit(__I40E_RX_PS_ENABLED, &(ring)->state)
 #define clear_ring_ps_enabled(ring) \
 	clear_bit(__I40E_RX_PS_ENABLED, &(ring)->state)
-#define check_for_tx_hang(ring) \
-	test_bit(__I40E_TX_DETECT_HANG, &(ring)->state)
-#define set_check_for_tx_hang(ring) \
-	set_bit(__I40E_TX_DETECT_HANG, &(ring)->state)
-#define clear_check_for_tx_hang(ring) \
-	clear_bit(__I40E_TX_DETECT_HANG, &(ring)->state)
 #define ring_is_16byte_desc_enabled(ring) \
 	test_bit(__I40E_RX_16BYTE_DESC_ENABLED, &(ring)->state)
 #define set_ring_16byte_desc_enabled(ring) \
Signature · 1 line
1.9.3
#1 lore raw

PATCH v3 2/2 i40e/i40evf: refactor tx timeout logic

Quoted 23 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Catherine Sullivan
Sent: Thursday, September 24, 2015 3:13 PM
To: intel-wired-lan at lists.osuosl.org
Subject: [Intel-wired-lan] [next PATCH S 2/2 v3] i40e/i40evf: refactor tx timeout logic

From: Kiran Patil <[email protected]>

This patch modifies the driver timeout logic by issuing a writeback request via a software interrupt to the hardware the first time the driver detects a hang. The driver was too agrressive in resetting a hung queue, so back that off by removing logic to down the netdevice after too many hangs, and move the function to the service task.

Change-ID Ife100b9d124cd08cbdb81ab659008c1b9abbedea

---
v3: Made i40evf_force_wb static.

Quoted 8 lines
 drivers/net/ethernet/intel/i40e/i40e.h        |   1 -
 drivers/net/ethernet/intel/i40e/i40e_main.c   | 274 +++++++++++++++------
-----
 drivers/net/ethernet/intel/i40e/i40e_txrx.c   |  75 +------
 drivers/net/ethernet/intel/i40e/i40e_txrx.h   |  10 +-
 drivers/net/ethernet/intel/i40evf/i40e_txrx.c |  96 +--------
 drivers/net/ethernet/intel/i40evf/i40e_txrx.h |   8 -
 6 files changed, 173 insertions(+), 291 deletions(-)

Code changes present in tree, still see occasional tx_timeout with hung queue, however adaptor does not reset and link does not go down.

#2 lore raw

PATCH v3 2/2 i40e/i40evf: refactor tx timeout logic

Quoted 8 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Bowers, AndrewX
Sent: Monday, September 28, 2015 10:37 AM
To: intel-wired-lan at lists.osuosl.org
Subject: Re: [Intel-wired-lan] [next PATCH S 2/2 v3] i40e/i40evf: refactor tx timeout logic

[..]

Quoted 7 lines

Code changes present in tree, still see occasional tx_timeout with hung queue, however adaptor does not reset and link does not go down.
_______________________________________________
Intel-wired-lan mailing list
Intel-wired-lan at lists.osuosl.org
http://lists.osuosl.org/mailman/listinfo/intel-wired-lan

Jeff, please have this also pulled into Dave's net tree, this and the other patch (Fix RS bit update..) need to go into net to fix the tx-timeout problem. The problem exists today in Dave's net tree, if its identical to your net tree, as that's where Andrew is seeing the tx hang.

Andrew, can you test applying this patch and the other patch (Fix RS bit update..) which I'll also add this note to to confirm the solution per Kiran.

Thanks,

Carolyn