* [PATCH net-next 1/4] cxgb3: Add register bit definition for Fatal Parity Error.
2010-09-02 23:04 [PATCH net-next 0/4] cxgb3: new code to deal with adapter parity errors Casey Leedom
@ 2010-09-02 23:07 ` Casey Leedom
2010-09-02 23:07 ` [PATCH net-next 2/4] cxgb3: Set FATALPERREN Casey Leedom
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Casey Leedom @ 2010-09-02 23:07 UTC (permalink / raw)
To: netdev; +Cc: Casey Leedom
Signed-off-by: Casey Leedom <leedom@chelsio.com>
---
drivers/net/cxgb3/regs.h | 4 ++++
1 files changed, 4 insertions(+), 0 deletions(-)
diff --git a/drivers/net/cxgb3/regs.h b/drivers/net/cxgb3/regs.h
index cb42353..6990f6c 100644
--- a/drivers/net/cxgb3/regs.h
+++ b/drivers/net/cxgb3/regs.h
@@ -1997,6 +1997,10 @@
#define A_PL_RST 0x6f0
+#define S_FATALPERREN 4
+#define V_FATALPERREN(x) ((x) << S_FATALPERREN)
+#define F_FATALPERREN V_FATALPERREN(1U)
+
#define S_CRSTWRM 1
#define V_CRSTWRM(x) ((x) << S_CRSTWRM)
#define F_CRSTWRM V_CRSTWRM(1U)
--
1.7.0.4
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH net-next 2/4] cxgb3: Set FATALPERREN.
2010-09-02 23:04 [PATCH net-next 0/4] cxgb3: new code to deal with adapter parity errors Casey Leedom
2010-09-02 23:07 ` [PATCH net-next 1/4] cxgb3: Add register bit definition for Fatal Parity Error Casey Leedom
@ 2010-09-02 23:07 ` Casey Leedom
2010-09-02 23:07 ` [PATCH net-next 3/4] cxgb3: Leave interrupts for fatal errors asserted in common code Casey Leedom
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Casey Leedom @ 2010-09-02 23:07 UTC (permalink / raw)
To: netdev; +Cc: Casey Leedom, Dimitris Michailidis
Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
Signed-off-by: Casey Leedom <leedom@chelsio.com>
---
drivers/net/cxgb3/t3_hw.c | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/net/cxgb3/t3_hw.c b/drivers/net/cxgb3/t3_hw.c
index 427c451..14a5664 100644
--- a/drivers/net/cxgb3/t3_hw.c
+++ b/drivers/net/cxgb3/t3_hw.c
@@ -3569,6 +3569,7 @@ int t3_init_hw(struct adapter *adapter, u32 fw_params)
t3_write_reg(adapter, A_PM1_TX_MODE, 0);
chan_init_hw(adapter, adapter->params.chan_map);
t3_sge_init(adapter, &adapter->params.sge);
+ t3_set_reg_field(adapter, A_PL_RST, 0, F_FATALPERREN);
t3_write_reg(adapter, A_T3DBG_GPIO_ACT_LOW, calc_gpio_intr(adapter));
--
1.7.0.4
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH net-next 3/4] cxgb3: Leave interrupts for fatal errors asserted in common code.
2010-09-02 23:04 [PATCH net-next 0/4] cxgb3: new code to deal with adapter parity errors Casey Leedom
2010-09-02 23:07 ` [PATCH net-next 1/4] cxgb3: Add register bit definition for Fatal Parity Error Casey Leedom
2010-09-02 23:07 ` [PATCH net-next 2/4] cxgb3: Set FATALPERREN Casey Leedom
@ 2010-09-02 23:07 ` Casey Leedom
2010-09-02 23:07 ` [PATCH net-next 4/4] cxgb3: Avoid flush_workqueue() deadlock Casey Leedom
2010-09-03 17:04 ` [PATCH net-next 0/4] cxgb3: new code to deal with adapter parity errors David Miller
4 siblings, 0 replies; 6+ messages in thread
From: Casey Leedom @ 2010-09-02 23:07 UTC (permalink / raw)
To: netdev; +Cc: Casey Leedom, Dimitris Michailidis
Platform code needs to deal with them now.
Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
Signed-off-by: Casey Leedom <leedom@chelsio.com>
---
drivers/net/cxgb3/t3_hw.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/cxgb3/t3_hw.c b/drivers/net/cxgb3/t3_hw.c
index 14a5664..d307c9d 100644
--- a/drivers/net/cxgb3/t3_hw.c
+++ b/drivers/net/cxgb3/t3_hw.c
@@ -1408,6 +1408,7 @@ static int t3_handle_intr_status(struct adapter *adapter, unsigned int reg,
fatal++;
CH_ALERT(adapter, "%s (0x%x)\n",
acts->msg, status & acts->mask);
+ status &= ~acts->mask;
} else if (acts->msg)
CH_WARN(adapter, "%s (0x%x)\n",
acts->msg, status & acts->mask);
@@ -1843,11 +1844,10 @@ static int mac_intr_handler(struct adapter *adap, unsigned int idx)
t3_os_link_fault_handler(adap, idx);
}
- t3_write_reg(adap, A_XGM_INT_CAUSE + mac->offset, cause);
-
if (cause & XGM_INTR_FATAL)
t3_fatal_err(adap);
+ t3_write_reg(adap, A_XGM_INT_CAUSE + mac->offset, cause);
return cause != 0;
}
--
1.7.0.4
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH net-next 4/4] cxgb3: Avoid flush_workqueue() deadlock.
2010-09-02 23:04 [PATCH net-next 0/4] cxgb3: new code to deal with adapter parity errors Casey Leedom
` (2 preceding siblings ...)
2010-09-02 23:07 ` [PATCH net-next 3/4] cxgb3: Leave interrupts for fatal errors asserted in common code Casey Leedom
@ 2010-09-02 23:07 ` Casey Leedom
2010-09-03 17:04 ` [PATCH net-next 0/4] cxgb3: new code to deal with adapter parity errors David Miller
4 siblings, 0 replies; 6+ messages in thread
From: Casey Leedom @ 2010-09-02 23:07 UTC (permalink / raw)
To: netdev; +Cc: Casey Leedom
Don't call flush_workqueue() on the cxgb3 Work Queue in cxgb_down() when
we're being called from the fatal error task ... which is executing on the
cxgb3 Work Queue.
Signed-off-by: Casey Leedom <leedom@chelsio.com>
---
drivers/net/cxgb3/cxgb3_main.c | 24 +++++++++++++++---------
1 files changed, 15 insertions(+), 9 deletions(-)
diff --git a/drivers/net/cxgb3/cxgb3_main.c b/drivers/net/cxgb3/cxgb3_main.c
index ad19585..1ecf53d 100644
--- a/drivers/net/cxgb3/cxgb3_main.c
+++ b/drivers/net/cxgb3/cxgb3_main.c
@@ -1286,7 +1286,7 @@ irq_err:
/*
* Release resources when all the ports and offloading have been stopped.
*/
-static void cxgb_down(struct adapter *adapter)
+static void cxgb_down(struct adapter *adapter, int on_wq)
{
t3_sge_stop(adapter);
spin_lock_irq(&adapter->work_lock); /* sync with PHY intr task */
@@ -1296,7 +1296,8 @@ static void cxgb_down(struct adapter *adapter)
free_irq_resources(adapter);
quiesce_rx(adapter);
t3_sge_stop(adapter);
- flush_workqueue(cxgb3_wq); /* wait for external IRQ handler */
+ if (!on_wq)
+ flush_workqueue(cxgb3_wq);/* wait for external IRQ handler */
}
static void schedule_chk_task(struct adapter *adap)
@@ -1374,7 +1375,7 @@ static int offload_close(struct t3cdev *tdev)
clear_bit(OFFLOAD_DEVMAP_BIT, &adapter->open_device_map);
if (!adapter->open_device_map)
- cxgb_down(adapter);
+ cxgb_down(adapter, 0);
cxgb3_offload_deactivate(adapter);
return 0;
@@ -1409,7 +1410,7 @@ static int cxgb_open(struct net_device *dev)
return 0;
}
-static int cxgb_close(struct net_device *dev)
+static int __cxgb_close(struct net_device *dev, int on_wq)
{
struct port_info *pi = netdev_priv(dev);
struct adapter *adapter = pi->adapter;
@@ -1436,12 +1437,17 @@ static int cxgb_close(struct net_device *dev)
cancel_delayed_work_sync(&adapter->adap_check_task);
if (!adapter->open_device_map)
- cxgb_down(adapter);
+ cxgb_down(adapter, on_wq);
cxgb3_event_notify(&adapter->tdev, OFFLOAD_PORT_DOWN, pi->port_id);
return 0;
}
+static int cxgb_close(struct net_device *dev)
+{
+ return __cxgb_close(dev, 0);
+}
+
static struct net_device_stats *cxgb_get_stats(struct net_device *dev)
{
struct port_info *pi = netdev_priv(dev);
@@ -2862,7 +2868,7 @@ void t3_os_link_fault_handler(struct adapter *adapter, int port_id)
spin_unlock(&adapter->work_lock);
}
-static int t3_adapter_error(struct adapter *adapter, int reset)
+static int t3_adapter_error(struct adapter *adapter, int reset, int on_wq)
{
int i, ret = 0;
@@ -2877,7 +2883,7 @@ static int t3_adapter_error(struct adapter *adapter, int reset)
struct net_device *netdev = adapter->port[i];
if (netif_running(netdev))
- cxgb_close(netdev);
+ __cxgb_close(netdev, on_wq);
}
/* Stop SGE timers */
@@ -2948,7 +2954,7 @@ static void fatal_error_task(struct work_struct *work)
int err = 0;
rtnl_lock();
- err = t3_adapter_error(adapter, 1);
+ err = t3_adapter_error(adapter, 1, 1);
if (!err)
err = t3_reenable_adapter(adapter);
if (!err)
@@ -2998,7 +3004,7 @@ static pci_ers_result_t t3_io_error_detected(struct pci_dev *pdev,
if (state == pci_channel_io_perm_failure)
return PCI_ERS_RESULT_DISCONNECT;
- ret = t3_adapter_error(adapter, 0);
+ ret = t3_adapter_error(adapter, 0, 0);
/* Request a slot reset. */
return PCI_ERS_RESULT_NEED_RESET;
--
1.7.0.4
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH net-next 0/4] cxgb3: new code to deal with adapter parity errors
2010-09-02 23:04 [PATCH net-next 0/4] cxgb3: new code to deal with adapter parity errors Casey Leedom
` (3 preceding siblings ...)
2010-09-02 23:07 ` [PATCH net-next 4/4] cxgb3: Avoid flush_workqueue() deadlock Casey Leedom
@ 2010-09-03 17:04 ` David Miller
4 siblings, 0 replies; 6+ messages in thread
From: David Miller @ 2010-09-03 17:04 UTC (permalink / raw)
To: leedom; +Cc: netdev
From: Casey Leedom <leedom@chelsio.com>
Date: Thu, 2 Sep 2010 16:04:28 -0700
> These patches implement code to detect and recover from adapter parity errors.
>
> Two of the following four patches (2/4 and 3/4) were actually
> implemented by Dimitris Michailidis just before he headed off for
> vacation. I've indicated that via a "Signed-off-by" by-line for
> him as well as one for me which I _think_ is the right way to do
> this. The other two patches (1/4 and 4/4) were implemented by me
> so I've only included a "Signed-off-by" for me on those. If I've
> done anything incorrectly please reject the patches and tell me
> how to correct them. Thanks!
All applied, thanks Casey.
^ permalink raw reply [flat|nested] 6+ messages in thread