leandrof/linux
1
0
Fork 0
forked from Minki/linux
Code Issues Pull Requests Packages Projects Releases Wiki Activity

sfc: Improve NIC internal error recovery

Browse Source 
Make the error count a per-NIC variable.
Reset this the count after an hour if it has not reached the critical value.
Set the critical value back to 5.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Ben Hutchings 2009-03-04 10:01:57 +00:00 committed by David S. Miller
parent 4720bc6cfe
commit 2c3c3d02f2
1 changed files with 19 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
drivers/net/sfc/falcon.c
View File

@ -39,11 +39,16 @@
* @next_buffer_table: First available buffer table id * @next_buffer_table: First available buffer table id
* @pci_dev2: The secondary PCI device if present * @pci_dev2: The secondary PCI device if present
* @i2c_data: Operations and state for I2C bit-bashing algorithm * @i2c_data: Operations and state for I2C bit-bashing algorithm
* @int_error_count: Number of internal errors seen recently
* @int_error_expire: Time at which error count will be expired
*/ */
struct falcon_nic_data { struct falcon_nic_data {
unsigned next_buffer_table; unsigned next_buffer_table;
struct pci_dev *pci_dev2; struct pci_dev *pci_dev2;
struct i2c_algo_bit_data i2c_data; struct i2c_algo_bit_data i2c_data;
unsigned int_error_count;
unsigned long int_error_expire;
}; };
/************************************************************************** /**************************************************************************
@ -119,8 +124,12 @@ MODULE_PARM_DESC(rx_xon_thresh_bytes, "RX fifo XON threshold");
#define FALCON_EVQ_SIZE 4096 #define FALCON_EVQ_SIZE 4096
#define FALCON_EVQ_MASK (FALCON_EVQ_SIZE - 1) #define FALCON_EVQ_MASK (FALCON_EVQ_SIZE - 1)
/* Max number of internal errors. After this resets will not be performed */ /* If FALCON_MAX_INT_ERRORS internal errors occur within
#define FALCON_MAX_INT_ERRORS 4 * FALCON_INT_ERROR_EXPIRE seconds, we consider the NIC broken and
* disable it.
*/
#define FALCON_INT_ERROR_EXPIRE 3600
#define FALCON_MAX_INT_ERRORS 5
/* We poll for events every FLUSH_INTERVAL ms, and check FLUSH_POLL_COUNT times /* We poll for events every FLUSH_INTERVAL ms, and check FLUSH_POLL_COUNT times
*/ */
@ -1374,7 +1383,6 @@ static irqreturn_t falcon_fatal_interrupt(struct efx_nic *efx)
efx_oword_t *int_ker = efx->irq_status.addr; efx_oword_t *int_ker = efx->irq_status.addr;
efx_oword_t fatal_intr; efx_oword_t fatal_intr;
int error, mem_perr; int error, mem_perr;
static int n_int_errors;
falcon_read(efx, &fatal_intr, FATAL_INTR_REG_KER); falcon_read(efx, &fatal_intr, FATAL_INTR_REG_KER);
error = EFX_OWORD_FIELD(fatal_intr, INT_KER_ERROR); error = EFX_OWORD_FIELD(fatal_intr, INT_KER_ERROR);
@ -1401,7 +1409,14 @@ static irqreturn_t falcon_fatal_interrupt(struct efx_nic *efx)
pci_clear_master(nic_data->pci_dev2); pci_clear_master(nic_data->pci_dev2);
falcon_disable_interrupts(efx); falcon_disable_interrupts(efx);
if (++n_int_errors < FALCON_MAX_INT_ERRORS) { /* Count errors and reset or disable the NIC accordingly */
if (nic_data->int_error_count == 0 ||
time_after(jiffies, nic_data->int_error_expire)) {
nic_data->int_error_count = 0;
nic_data->int_error_expire =
jiffies + FALCON_INT_ERROR_EXPIRE * HZ;
}
if (++nic_data->int_error_count < FALCON_MAX_INT_ERRORS) {
EFX_ERR(efx, "SYSTEM ERROR - reset scheduled\n"); EFX_ERR(efx, "SYSTEM ERROR - reset scheduled\n");
efx_schedule_reset(efx, RESET_TYPE_INT_ERROR); efx_schedule_reset(efx, RESET_TYPE_INT_ERROR);
} else { } else {