bnxt_en: Bug fixes.
Michael Chan
maintainercover bnxt_en: Bug fixes.
This set of driver patches include bug fixes for ethtool get channels, ethtool statistics, ethtool NVRAM, AER recovery, a firmware reset issue that could potentially crash, hwmon temperature reporting issue on VF, and 2 fixes for regressions introduced by the recent user-defined RSS map feature.
Please queue patches 1 to 6 for -stable. Thanks.
Edwin Peer (2):
bnxt_en: fix HWRM error when querying VF temperature
bnxt_en: init RSS table for Minimal-Static VF reservation
Michael Chan (3):
bnxt_en: Fix ethtool -S statitics with XDP or TCs enabled.
bnxt_en: Fix possible crash in bnxt_fw_reset_task().
bnxt_en: Setup default RSS map in all scenarios.
Pavan Chebbi (1):
bnxt_en: Don't query FW when netif_running() is false.
Vasundhara Volam (2):
bnxt_en: Check for zero dir entries in NVRAM.
bnxt_en: Fix PCI AER error recovery flow
drivers/net/ethernet/broadcom/bnxt/bnxt.c | 78 ++++++++++++++++------- drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c | 16 ++--- 2 files changed, 61 insertions(+), 33 deletions(-)
Signature · 1 line
1.8.3.1
Michael Chan
maintainerPATCH 1/8 bnxt_en: Don't query FW when netif_running() is false.
From: Pavan Chebbi <[email protected]>
In rare conditions like two stage OS installation, the ethtool's get_channels function may be called when the device is in D3 state, leading to uncorrectable PCI error. Check netif_running() first before making any query to FW which involves writing to BAR.
--- drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c +1 -1
@@ -806,7 +806,7 @@ static void bnxt_get_channels(struct net_device *dev,int max_tx_sch_inputs;/* Get the most up-to-date max_tx_sch_inputs. */ -if (BNXT_NEW_RM (bp)) +if (netif_running (dev) &&BNXT_NEW_RM (bp))bnxt_hwrm_func_resc_qcaps (bp, false); max_tx_sch_inputs = hw_resc->max_tx_sch_inputs;
Signature · 1 line
1.8.3.1
Michael Chan
maintainerPATCH 2/8 bnxt_en: Check for zero dir entries in NVRAM.
In reply to Michael ChanFrom: Vasundhara Volam <[email protected]>
If firmware goes into unstable state, HWRM_NVM_GET_DIR_INFO firmware command may return zero dir entries. Return error in such case to avoid zero length dma buffer request.
--- drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c +3 -0
@@ -2323,6 +2323,9 @@ static int bnxt_get_nvram_directory(struct net_device *dev, u32 len, u8 *data)if (rc !=0 )return rc; +if (!dir_entries || !entry_length) +return -EIO ; +/* Insert 2 bytes of directory info (count and size of entries) */ if (len <2 )return -EINVAL ;
Signature · 1 line
1.8.3.1
Michael Chan
maintainerPATCH 3/8 bnxt_en: Fix ethtool -S statitics with XDP or TCs enabled.
In reply to Michael ChanWe are returning the wrong count for ETH_SS_STATS in get_sset_count() when XDP or TCs are enabled. In a recent commit, we got rid of irrelevant counters when the ring is RX only or TX only, but we did not make the proper adjustments for the count. As a result, when we have XDP or TCs enabled, we are returning an excess count because some of the rings are TX only. This causes ethtool -S to display extra counters with no counter names.
Fix bnxt_get_num_ring_stats() by not assuming that all rings will always have RX and TX counters in combined mode.
--- drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c | 11 ++--------- 1 file changed, 2 insertions(+), 9 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c +2 -9
@@ -472,20 +472,13 @@ static int bnxt_get_num_tpa_ring_stats(struct bnxt *bp)static int bnxt_get_num_ring_stats (struct bnxt *bp) {int rx, tx, cmn; -bool sh = false; - -if (bp->flags &BNXT_FLAG_SHARED_RINGS ) - sh = true; rx =NUM_RING_RX_HW_STATS +NUM_RING_RX_SW_STATS +bnxt_get_num_tpa_ring_stats (bp); tx =NUM_RING_TX_HW_STATS ; cmn =NUM_RING_CMN_SW_STATS ; -if (sh) -return (rx + tx + cmn) * bp->cp_nr_rings; -else -return rx * bp->rx_nr_rings + tx * bp->tx_nr_rings + - cmn * bp->cp_nr_rings; +return rx * bp->rx_nr_rings + tx * bp->tx_nr_rings + + cmn * bp->cp_nr_rings; }static int bnxt_get_num_stats (struct bnxt *bp)
Signature · 1 line
1.8.3.1
Michael Chan
maintainerPATCH 4/8 bnxt_en: Fix PCI AER error recovery flow
In reply to Michael ChanFrom: Vasundhara Volam <[email protected]>
When a PCI error is detected the PCI state could be corrupt, save the PCI state after initialization and restore it after the slot reset.
--- drivers/net/ethernet/broadcom/bnxt/bnxt.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +3 -0
@@ -12339,6 +12339,7 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent) (long)pci_resource_start(pdev, 0), dev->dev_addr);pcie_print_link_status (pdev); +pci_save_state (pdev);return 0 ; init_err_cleanup:
@@ -12536,6 +12537,8 @@ 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); +pci_restore_state (pdev); +pci_save_state (pdev); err =bnxt_hwrm_func_reset (bp);if (!err) {
Signature · 1 line
1.8.3.1
Michael Chan
maintainerPATCH 5/8 bnxt_en: Fix possible crash in bnxt_fw_reset_task().
In reply to Michael Chanbnxt_fw_reset_task() is run from a delayed workqueue. The current code is not cancelling the workqueue in the driver's .remove() method and it can potentially crash if the device is removed with the workqueue still pending.
The fix is to clear the BNXT_STATE_IN_FW_RESET flag and then cancel the delayed workqueue in bnxt_remove_one(). bnxt_queue_fw_reset_work() also needs to check that this flag is set before scheduling. This will guarantee that no rescheduling will be done after it is cancelled.
--- drivers/net/ethernet/broadcom/bnxt/bnxt.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +8 -2
@@ -1141,6 +1141,9 @@ static int bnxt_discard_rx(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,static void bnxt_queue_fw_reset_work (struct bnxt *bp,unsigned long delay) { +if (!(test_bit (BNXT_STATE_IN_FW_RESET , &bp->state))) +return ; +if (BNXT_PF (bp))queue_delayed_work (bnxt_pf_wq, &bp->fw_reset_task, delay);else
@@ -1157,10 +1160,12 @@ static void bnxt_queue_sp_work(struct bnxt *bp)static void bnxt_cancel_sp_work (struct bnxt *bp) { -if (BNXT_PF (bp)) +if (BNXT_PF (bp)) {flush_workqueue (bnxt_pf_wq); -else + }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)
@@ -11761,6 +11766,7 @@ static void bnxt_remove_one(struct pci_dev *pdev)unregister_netdev (dev);bnxt_dl_unregister (bp);bnxt_shutdown_tc (bp); +clear_bit (BNXT_STATE_IN_FW_RESET , &bp->state);bnxt_cancel_sp_work (bp); bp->sp_event =0 ;
Signature · 1 line
1.8.3.1
Michael Chan
maintainerPATCH 6/8 bnxt_en: fix HWRM error when querying VF temperature
In reply to Michael ChanFrom: Edwin Peer <[email protected]>
Firmware returns RESOURCE_ACCESS_DENIED for HWRM_TEMP_MONITORY_QUERY for VFs. This produces unpleasing error messages in the log when temp1_input is queried via the hwmon sysfs interface from a VF.
The error is harmless and expected, so silence it and return unknown as the value. Since the device temperature is not particularly sensitive information, provide flexibility to change this policy in future by silencing the error rather than avoiding the HWRM call entirely for VFs.
--- drivers/net/ethernet/broadcom/bnxt/bnxt.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +7 -4
@@ -9289,16 +9289,19 @@ static ssize_t bnxt_show_temp(struct device *dev,struct hwrm_temp_monitor_query_input req = {0 };struct hwrm_temp_monitor_query_output *resp;struct bnxt *bp =dev_get_drvdata (dev); -u32 temp =0 ; +u32 len =0 ; resp = bp->hwrm_cmd_resp_addr;bnxt_hwrm_cmd_hdr_init (bp, &req,HWRM_TEMP_MONITOR_QUERY ,-1 ,-1 );mutex_lock (&bp->hwrm_cmd_lock); -if (!_hwrm_send_message (bp, &req,sizeof (req),HWRM_CMD_TIMEOUT )) - temp = resp->temp *1000 ;/* display millidegree */ +if (!_hwrm_send_message_silent (bp, &req,sizeof (req),HWRM_CMD_TIMEOUT )) + len =sprintf (buf,"%u\n" , resp->temp *1000 );/* display millidegree */ mutex_unlock (&bp->hwrm_cmd_lock); -return sprintf (buf,"%u\n" , temp); +if (len) +return len; + +return sprintf (buf,"unknown\n" ); } static SENSOR_DEVICE_ATTR(temp1_input, 0444, bnxt_show_temp, NULL, 0);
Signature · 1 line
1.8.3.1
Michael Chan
maintainerPATCH 7/8 bnxt_en: init RSS table for Minimal-Static VF reservation
In reply to Michael ChanFrom: Edwin Peer <[email protected]>
There are no VF rings available during probe when the device is configured using the Minimal-Static reservation strategy. In this case, the RSS indirection table can only be initialized later, during bnxt_open_nic(). However, this was not happening because the rings will already have been reserved via bnxt_init_dflt_ring_mode(), causing bnxt_need_reserve_rings() to return false in bnxt_reserve_rings() and bypass the RSS table init.
Solve this by pushing the call to bnxt_set_dflt_rss_indir_tbl() into __bnxt_reserve_rings(), which is common to both paths and is called whenever ring configuration is changed. After doing this, the RSS table init that must be called from bnxt_init_one() happens implicitly via bnxt_set_default_rings(), necessitating doing the allocation earlier in order to avoid a null pointer dereference.
--- drivers/net/ethernet/broadcom/bnxt/bnxt.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +7 -8
@@ -6219,6 +6219,9 @@ static int __bnxt_reserve_rings(struct bnxt *bp)if (!tx || !rx || !cp || !grp || !vnic || !stat)return -ENOMEM ; +if (!netif_is_rxfh_configured (bp->dev)) +bnxt_set_dflt_rss_indir_tbl (bp); +return rc; }
@@ -8500,9 +8503,6 @@ int bnxt_reserve_rings(struct bnxt *bp, bool irq_re_init) rc =bnxt_init_int_mode (bp);bnxt_ulp_irq_restart (bp, rc); } -if (!netif_is_rxfh_configured (bp->dev)) -bnxt_set_dflt_rss_indir_tbl (bp); -if (rc) {netdev_err (bp->dev,"ring reservation/IRQ init failure rc: %d\n" , rc);return rc;
@@ -12209,6 +12209,10 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)if (BNXT_CHIP_P5 (bp)) bp->flags |=BNXT_FLAG_CHIP_P5 ; + rc =bnxt_alloc_rss_indir_tbl (bp); +if (rc) + goto init_err_pci_clean; + rc =bnxt_fw_init_one_p2 (bp);if (rc) goto init_err_pci_clean;
@@ -12313,11 +12317,6 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)*/ bp->tx_nr_rings_per_tc = bp->tx_nr_rings; - rc =bnxt_alloc_rss_indir_tbl (bp); -if (rc) - goto init_err_pci_clean; -bnxt_set_dflt_rss_indir_tbl (bp); -if (BNXT_PF (bp)) {if (!bnxt_pf_wq) { bnxt_pf_wq =
Signature · 1 line
1.8.3.1
Michael Chan
maintainerPATCH 8/8 bnxt_en: Setup default RSS map in all scenarios.
In reply to Michael ChanThe recent changes to support user-defined RSS map assume that RX rings are always reserved and the default RSS map is set after the RX rings are successfully reserved. If the firmware spec is older than 1.6.1, no ring reservations are required and the default RSS map is not setup at all. In another scenario where the fw Resource Manager is older, RX rings are not reserved and we also end up with no valid RSS map.
Fix both issues in bnxt_need_reserve_rings(). In both scenarios described above, we don't need to reserve RX rings so we need to call this new function bnxt_check_rss_map_no_rmgr() to setup the default RSS map when needed.
Without valid RSS map, the NIC won't receive packets properly.
--- drivers/net/ethernet/broadcom/bnxt/bnxt.c | 39 ++++++++++++++++++++++++------- 1 file changed, 30 insertions(+), 9 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +30 -9
@@ -6107,6 +6107,21 @@ static int bnxt_get_func_stat_ctxs(struct bnxt *bp)return cp + ulp_stat; } +/* Check if a default RSS map needs to be setup. This function is only +* used on older firmware that does not require reserving RX rings. +*/ +static void bnxt_check_rss_tbl_no_rmgr (struct bnxt *bp) +{ +struct bnxt_hw_resc *hw_resc = &bp->hw_resc; + +/* The RSS map is valid for RX rings set to resv_rx_rings */ +if (hw_resc->resv_rx_rings != bp->rx_nr_rings) { + hw_resc->resv_rx_rings = bp->rx_nr_rings; +if (!netif_is_rxfh_configured (bp->dev)) +bnxt_set_dflt_rss_indir_tbl (bp); + } +} +static bool bnxt_need_reserve_rings (struct bnxt *bp) {struct bnxt_hw_resc *hw_resc = &bp->hw_resc;
@@ -6115,22 +6130,28 @@ static bool bnxt_need_reserve_rings(struct bnxt *bp)int rx = bp->rx_nr_rings, stat;int vnic =1 , grp = rx; -if (bp->hwrm_spec_code <0x10601 ) -return false; - -if (hw_resc->resv_tx_rings != bp->tx_nr_rings) +if (hw_resc->resv_tx_rings != bp->tx_nr_rings && + bp->hwrm_spec_code >=0x10601 )return true; +/* Old firmware does not need RX ring reservations but we still +* need to setup a default RSS map when needed. With new firmware +* we go through RX ring reservations first and then set up the +* RSS map for the successfully reserved RX rings when needed. +*/ +if (!BNXT_NEW_RM (bp)) { +bnxt_check_rss_tbl_no_rmgr (bp); +return false; + }if ((bp->flags &BNXT_FLAG_RFS ) && !(bp->flags &BNXT_FLAG_CHIP_P5 )) vnic = rx +1 ;if (bp->flags &BNXT_FLAG_AGG_RINGS ) rx <<=1 ; stat =bnxt_get_func_stat_ctxs (bp); -if (BNXT_NEW_RM (bp) && - (hw_resc->resv_rx_rings != rx || hw_resc->resv_cp_rings != cp || - hw_resc->resv_vnics != vnic || hw_resc->resv_stat_ctxs != stat || - (hw_resc->resv_hw_ring_grps != grp && - !(bp->flags &BNXT_FLAG_CHIP_P5 )))) +if (hw_resc->resv_rx_rings != rx || hw_resc->resv_cp_rings != cp || + hw_resc->resv_vnics != vnic || hw_resc->resv_stat_ctxs != stat || + (hw_resc->resv_hw_ring_grps != grp && + !(bp->flags &BNXT_FLAG_CHIP_P5 )))return true;if ((bp->flags &BNXT_FLAG_CHIP_P5 ) &&BNXT_PF (bp) && hw_resc->resv_irqs != nq)
Signature · 1 line
1.8.3.1