summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorJakub Kicinski <kuba@kernel.org>2020-10-26 18:26:38 -0700
committerJakub Kicinski <kuba@kernel.org>2020-10-26 18:26:38 -0700
commit10067b501990cdd1f9d39133d124d140726bd996 (patch)
tree34dc7e52b00833a2547bc53e348af1a0f28637f5
parent19c176eb079834c5cf8574ce18c1d934ec486dd8 (diff)
parent825741b071722f1c8ad692cead562c4b5f5eaa93 (diff)
Merge branch 'bnxt_en-bug-fixes'
Michael Chan says: ==================== bnxt_en: Bug fixes. These 5 bug fixes are all related to the firmware reset or AER recovery. 2 patches fix the cleanup logic for the workqueue used to handle firmware reset and recovery. 1 patch ensures that the chip will have the proper BAR addresses latched after fatal AER recovery. 1 patch fixes the open path to check for firmware reset abort error. The last one sends the fw reset command unconditionally to fix the AER reset logic. ==================== Link: https://lore.kernel.org/r/1603685901-17917-1-git-send-email-michael.chan@broadcom.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
-rw-r--r--drivers/net/ethernet/broadcom/bnxt/bnxt.c49
-rw-r--r--drivers/net/ethernet/broadcom/bnxt/bnxt.h1
2 files changed, 32 insertions, 18 deletions
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index fa147865e33f..7975f59735d6 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -1160,16 +1160,6 @@ static void bnxt_queue_sp_work(struct bnxt *bp)
schedule_work(&bp->sp_task);
}
-static void bnxt_cancel_sp_work(struct bnxt *bp)
-{
- if (BNXT_PF(bp)) {
- flush_workqueue(bnxt_pf_wq);
- } else {
- cancel_work_sync(&bp->sp_task);
- cancel_delayed_work_sync(&bp->fw_reset_task);
- }
-}
-
static void bnxt_sched_reset(struct bnxt *bp, struct bnxt_rx_ring_info *rxr)
{
if (!rxr->bnapi->in_reset) {
@@ -4362,7 +4352,8 @@ static int bnxt_hwrm_do_send_msg(struct bnxt *bp, void *msg, u32 msg_len,
u32 bar_offset = BNXT_GRCPF_REG_CHIMP_COMM;
u16 dst = BNXT_HWRM_CHNL_CHIMP;
- if (BNXT_NO_FW_ACCESS(bp))
+ if (BNXT_NO_FW_ACCESS(bp) &&
+ le16_to_cpu(req->req_type) != HWRM_FUNC_RESET)
return -EBUSY;
if (msg_len > BNXT_HWRM_MAX_REQ_LEN) {
@@ -9789,7 +9780,10 @@ int bnxt_open_nic(struct bnxt *bp, bool irq_re_init, bool link_re_init)
{
int rc = 0;
- rc = __bnxt_open_nic(bp, irq_re_init, link_re_init);
+ if (test_bit(BNXT_STATE_ABORT_ERR, &bp->state))
+ rc = -EIO;
+ if (!rc)
+ rc = __bnxt_open_nic(bp, irq_re_init, link_re_init);
if (rc) {
netdev_err(bp->dev, "nic open fail (rc: %x)\n", rc);
dev_close(bp->dev);
@@ -12108,15 +12102,17 @@ static void bnxt_remove_one(struct pci_dev *pdev)
if (BNXT_PF(bp))
bnxt_sriov_disable(bp);
- clear_bit(BNXT_STATE_IN_FW_RESET, &bp->state);
- bnxt_cancel_sp_work(bp);
- bp->sp_event = 0;
-
- bnxt_dl_fw_reporters_destroy(bp, true);
if (BNXT_PF(bp))
devlink_port_type_clear(&bp->dl_port);
pci_disable_pcie_error_reporting(pdev);
unregister_netdev(dev);
+ clear_bit(BNXT_STATE_IN_FW_RESET, &bp->state);
+ /* Flush any pending tasks */
+ cancel_work_sync(&bp->sp_task);
+ cancel_delayed_work_sync(&bp->fw_reset_task);
+ bp->sp_event = 0;
+
+ bnxt_dl_fw_reporters_destroy(bp, true);
bnxt_dl_unregister(bp);
bnxt_shutdown_tc(bp);
@@ -12860,6 +12856,9 @@ static pci_ers_result_t bnxt_io_error_detected(struct pci_dev *pdev,
return PCI_ERS_RESULT_DISCONNECT;
}
+ if (state == pci_channel_io_frozen)
+ set_bit(BNXT_STATE_PCI_CHANNEL_IO_FROZEN, &bp->state);
+
if (netif_running(netdev))
bnxt_close(netdev);
@@ -12886,7 +12885,7 @@ static pci_ers_result_t bnxt_io_slot_reset(struct pci_dev *pdev)
{
struct net_device *netdev = pci_get_drvdata(pdev);
struct bnxt *bp = netdev_priv(netdev);
- int err = 0;
+ int err = 0, off;
pci_ers_result_t result = PCI_ERS_RESULT_DISCONNECT;
netdev_info(bp->dev, "PCI Slot Reset\n");
@@ -12898,6 +12897,20 @@ static pci_ers_result_t bnxt_io_slot_reset(struct pci_dev *pdev)
"Cannot re-enable PCI device after reset.\n");
} else {
pci_set_master(pdev);
+ /* Upon fatal error, our device internal logic that latches to
+ * BAR value is getting reset and will restore only upon
+ * rewritting the BARs.
+ *
+ * As pci_restore_state() does not re-write the BARs if the
+ * value is same as saved value earlier, driver needs to
+ * write the BARs to 0 to force restore, in case of fatal error.
+ */
+ if (test_and_clear_bit(BNXT_STATE_PCI_CHANNEL_IO_FROZEN,
+ &bp->state)) {
+ for (off = PCI_BASE_ADDRESS_0;
+ off <= PCI_BASE_ADDRESS_5; off += 4)
+ pci_write_config_dword(bp->pdev, off, 0);
+ }
pci_restore_state(pdev);
pci_save_state(pdev);
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.h b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
index 21ef1c21f602..47b3c3127879 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.h
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
@@ -1781,6 +1781,7 @@ struct bnxt {
#define BNXT_STATE_ABORT_ERR 5
#define BNXT_STATE_FW_FATAL_COND 6
#define BNXT_STATE_DRV_REGISTERED 7
+#define BNXT_STATE_PCI_CHANNEL_IO_FROZEN 8
#define BNXT_NO_FW_ACCESS(bp) \
(test_bit(BNXT_STATE_FW_FATAL_COND, &(bp)->state) || \