Root cause: When kernel crashes, bfa IOC state machine and FW doesn't get
a notification and hence are not cleanly shutdown. So registers holding
driver/IOC state information are not reset back to valid disabled/parking
values. This causes subsequent driver initialization to hang during kdump
kernel boot.
Fix description: during the initialization of first PCI function, reset
corresponding register when unclean shutown is detect by reading chip
registers. This will make sure that ioc/fw gets clean re-initialization.
Signed-off-by: Jing Huang <huangj@xxxxxxxxxxx>
---
drivers/scsi/bfa/bfa_ioc.c | 4 +++-
drivers/scsi/bfa/bfa_ioc.h | 1 +
drivers/scsi/bfa/bfa_ioc_cb.c | 11 +++++++++++
drivers/scsi/bfa/bfa_ioc_ct.c | 26 ++++++++++++++++++++++++++
4 files changed, 41 insertions(+), 1 deletions(-)
diff --git a/drivers/scsi/bfa/bfa_ioc.c b/drivers/scsi/bfa/bfa_ioc.c
index c1f72c4..6c7e033 100644
--- a/drivers/scsi/bfa/bfa_ioc.c
+++ b/drivers/scsi/bfa/bfa_ioc.c
@@ -56,6 +56,8 @@ BFA_TRC_FILE(CNA, IOC);
#define bfa_ioc_map_port(__ioc) ((__ioc)->ioc_hwif->ioc_map_port(__ioc))
#define bfa_ioc_notify_fail(__ioc) \
((__ioc)->ioc_hwif->ioc_notify_fail(__ioc))
+#define bfa_ioc_sync_start(__ioc) \
+ ((__ioc)->ioc_hwif->ioc_sync_start(__ioc))
#define bfa_ioc_sync_join(__ioc) \
((__ioc)->ioc_hwif->ioc_sync_join(__ioc))
#define bfa_ioc_sync_leave(__ioc) \
@@ -647,7 +649,7 @@ bfa_iocpf_sm_fwcheck(struct bfa_iocpf_s *iocpf, enum iocpf_event event)
switch (event) {
case IOCPF_E_SEMLOCKED:
if (bfa_ioc_firmware_lock(ioc)) {
- if (bfa_ioc_sync_complete(ioc)) {
+ if (bfa_ioc_sync_start(ioc)) {
iocpf->retry_count = 0;
bfa_ioc_sync_join(ioc);
bfa_fsm_set_state(iocpf, bfa_iocpf_sm_hwinit);
diff --git a/drivers/scsi/bfa/bfa_ioc.h b/drivers/scsi/bfa/bfa_ioc.h
index ec9cf08..c85182a 100644
--- a/drivers/scsi/bfa/bfa_ioc.h
+++ b/drivers/scsi/bfa/bfa_ioc.h
@@ -263,6 +263,7 @@ struct bfa_ioc_hwif_s {
bfa_boolean_t msix);
void (*ioc_notify_fail) (struct bfa_ioc_s *ioc);
void (*ioc_ownership_reset) (struct bfa_ioc_s *ioc);
+ bfa_boolean_t (*ioc_sync_start) (struct bfa_ioc_s *ioc);
void (*ioc_sync_join) (struct bfa_ioc_s *ioc);
void (*ioc_sync_leave) (struct bfa_ioc_s *ioc);
void (*ioc_sync_ack) (struct bfa_ioc_s *ioc);
diff --git a/drivers/scsi/bfa/bfa_ioc_cb.c b/drivers/scsi/bfa/bfa_ioc_cb.c
index e4a0713..89ae4c8 100644
--- a/drivers/scsi/bfa/bfa_ioc_cb.c
+++ b/drivers/scsi/bfa/bfa_ioc_cb.c
@@ -32,6 +32,7 @@ static void bfa_ioc_cb_map_port(struct bfa_ioc_s *ioc);
static void bfa_ioc_cb_isr_mode_set(struct bfa_ioc_s *ioc, bfa_boolean_t msix);
static void bfa_ioc_cb_notify_fail(struct bfa_ioc_s *ioc);
static void bfa_ioc_cb_ownership_reset(struct bfa_ioc_s *ioc);
+static bfa_boolean_t bfa_ioc_cb_sync_start(struct bfa_ioc_s *ioc);
static void bfa_ioc_cb_sync_join(struct bfa_ioc_s *ioc);
static void bfa_ioc_cb_sync_leave(struct bfa_ioc_s *ioc);
static void bfa_ioc_cb_sync_ack(struct bfa_ioc_s *ioc);
@@ -53,6 +54,7 @@ bfa_ioc_set_cb_hwif(struct bfa_ioc_s *ioc)
hwif_cb.ioc_isr_mode_set = bfa_ioc_cb_isr_mode_set;
hwif_cb.ioc_notify_fail = bfa_ioc_cb_notify_fail;
hwif_cb.ioc_ownership_reset = bfa_ioc_cb_ownership_reset;
+ hwif_cb.ioc_sync_start = bfa_ioc_cb_sync_start;
hwif_cb.ioc_sync_join = bfa_ioc_cb_sync_join;
hwif_cb.ioc_sync_leave = bfa_ioc_cb_sync_leave;
hwif_cb.ioc_sync_ack = bfa_ioc_cb_sync_ack;
@@ -195,6 +197,15 @@ bfa_ioc_cb_isr_mode_set(struct bfa_ioc_s *ioc, bfa_boolean_t msix)
}
/*
+ * Synchronized IOC failure processing routines
+ */
+static bfa_boolean_t
+bfa_ioc_cb_sync_start(struct bfa_ioc_s *ioc)
+{
+ return bfa_ioc_cb_sync_complete(ioc);
+}
+
+/*
* Cleanup hw semaphore and usecnt registers
*/
static void
diff --git a/drivers/scsi/bfa/bfa_ioc_ct.c b/drivers/scsi/bfa/bfa_ioc_ct.c
index 008d129..9361252 100644
--- a/drivers/scsi/bfa/bfa_ioc_ct.c
+++ b/drivers/scsi/bfa/bfa_ioc_ct.c
@@ -41,6 +41,7 @@ static void bfa_ioc_ct_map_port(struct bfa_ioc_s *ioc);
static void bfa_ioc_ct_isr_mode_set(struct bfa_ioc_s *ioc, bfa_boolean_t msix);
static void bfa_ioc_ct_notify_fail(struct bfa_ioc_s *ioc);
static void bfa_ioc_ct_ownership_reset(struct bfa_ioc_s *ioc);
+static bfa_boolean_t bfa_ioc_ct_sync_start(struct bfa_ioc_s *ioc);
static void bfa_ioc_ct_sync_join(struct bfa_ioc_s *ioc);
static void bfa_ioc_ct_sync_leave(struct bfa_ioc_s *ioc);
static void bfa_ioc_ct_sync_ack(struct bfa_ioc_s *ioc);
@@ -62,6 +63,7 @@ bfa_ioc_set_ct_hwif(struct bfa_ioc_s *ioc)
hwif_ct.ioc_isr_mode_set = bfa_ioc_ct_isr_mode_set;
hwif_ct.ioc_notify_fail = bfa_ioc_ct_notify_fail;
hwif_ct.ioc_ownership_reset = bfa_ioc_ct_ownership_reset;
+ hwif_ct.ioc_sync_start = bfa_ioc_ct_sync_start;
hwif_ct.ioc_sync_join = bfa_ioc_ct_sync_join;
hwif_ct.ioc_sync_leave = bfa_ioc_ct_sync_leave;
hwif_ct.ioc_sync_ack = bfa_ioc_ct_sync_ack;
@@ -351,6 +353,30 @@ bfa_ioc_ct_ownership_reset(struct bfa_ioc_s *ioc)
writel(1, ioc->ioc_regs.ioc_sem_reg);
}
+static bfa_boolean_t
+bfa_ioc_ct_sync_start(struct bfa_ioc_s *ioc)
+{
+ uint32_t r32 = readl(ioc->ioc_regs.ioc_fail_sync);
+ uint32_t sync_reqd = bfa_ioc_ct_get_sync_reqd(r32);
+
+ /*
+ * Driver load time. If the sync required bit for this PCI fn
+ * is set, it is due to an unclean exit by the driver for this
+ * PCI fn in the previous incarnation. Whoever comes here first
+ * should clean it up, no matter which PCI fn.
+ */
+
+ if (sync_reqd & bfa_ioc_ct_sync_pos(ioc)) {
+ writel(0, ioc->ioc_regs.ioc_fail_sync);
+ writel(1, ioc->ioc_regs.ioc_usage_reg);
+ writel(BFI_IOC_UNINIT, ioc->ioc_regs.ioc_fwstate);
+ writel(BFI_IOC_UNINIT, ioc->ioc_regs.alt_ioc_fwstate);
+ return BFA_TRUE;
+ }
+
+ return bfa_ioc_ct_sync_complete(ioc);
+}
+
/*
* Synchronized IOC failure processing routines
*/
--
1.6.5.2
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
