bnxt_en: Bug fixes
Michael Chan
maintainercover bnxt_en: Bug fixes
This patch series includes 3 fixes:
1. Fix an occasional VF open failure.
2. Fix a PTP spinlock usage before initialization
3. Fix unnecesary RX packet drops under high TX traffic load.
Michael Chan (2):
bnxt_en: Initiallize bp->ptp_lock first before using it
bnxt_en: Fix unnecessary dropping of RX packets
Somnath Kotur (1):
bnxt_en: Fix possible bnxt_open() failure caused by wrong RFS flag
drivers/net/ethernet/broadcom/bnxt/bnxt.c | 13 ++++++++----- drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c | 15 +++++++-------- 2 files changed, 15 insertions(+), 13 deletions(-)
Signature · 1 line
2.18.1
Michael Chan
maintainerPATCH 1/3 bnxt_en: Fix possible bnxt_open() failure caused by wrong RFS flag
From: Somnath Kotur <[email protected]>
bnxt_open() can fail in this code path, especially on a VF when it fails to reserve default rings:
bnxt_open()
__bnxt_open_nic()
bnxt_clear_int_mode()
bnxt_init_dflt_ring_mode()
RX rings would be set to 0 when we hit this error path.
It is possible for a subsequent bnxt_open() call to potentially succeed with a code path like this:
bnxt_open()
bnxt_hwrm_if_change()
bnxt_fw_init_one()
bnxt_fw_init_one_p3()
bnxt_set_dflt_rfs()
bnxt_rfs_capable()
bnxt_hwrm_reserve_rings()On older chips, RFS is capable if we can reserve the number of vnics that is equal to RX rings + 1. But since RX rings is still set to 0 in this code path, we may mistakenly think that RFS is supported for 0 RX rings.
Later, when the default RX rings are reserved and we try to enable RFS, it would fail and cause bnxt_open() to fail unnecessarily.
We fix this in 2 places. bnxt_rfs_capable() will always return false if RX rings is not yet set. bnxt_init_dflt_ring_mode() will call bnxt_set_dflt_rfs() which will always clear the RFS flags if RFS is not supported.
--- drivers/net/ethernet/broadcom/bnxt/bnxt.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +4 -5
@@ -10983,7 +10983,7 @@ static bool bnxt_rfs_capable(struct bnxt *bp)if (bp->flags &BNXT_FLAG_CHIP_P5 )return bnxt_rfs_supported (bp); -if (!(bp->flags &BNXT_FLAG_MSIX_CAP ) || !bnxt_can_reserve_rings (bp)) +if (!(bp->flags &BNXT_FLAG_MSIX_CAP ) || !bnxt_can_reserve_rings (bp) || !bp->rx_nr_rings)return false; vnics =1 + bp->rx_nr_rings;
@@ -13234,10 +13234,9 @@ static int bnxt_init_dflt_ring_mode(struct bnxt *bp) goto init_dflt_ring_err; bp->tx_nr_rings_per_tc = bp->tx_nr_rings; -if (bnxt_rfs_supported (bp) &&bnxt_rfs_capable (bp)) { - bp->flags |=BNXT_FLAG_RFS ; - bp->dev->features |=NETIF_F_NTUPLE ; - } + +bnxt_set_dflt_rfs (bp); + init_dflt_ring_err:bnxt_ulp_irq_restart (bp, rc);return rc;
Signature · 1 line
2.18.1
Michael Chan
maintainerPATCH 2/3 bnxt_en: Initiallize bp->ptp_lock first before using it
In reply to Michael Chanbnxt_ptp_init() calls bnxt_ptp_init_rtc() which will acquire the ptp_lock spinlock. The spinlock is not initialized until later. Move the bnxt_ptp_init_rtc() call after the spinlock is initialized.
--- drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c +7 -8
@@ -846,13 +846,6 @@ int bnxt_ptp_init(struct bnxt *bp, bool phc_cfg)if (rc)return rc; -if (bp->fw_cap &BNXT_FW_CAP_PTP_RTC ) { -bnxt_ptp_timecounter_init (bp, false); - rc =bnxt_ptp_init_rtc (bp, phc_cfg); -if (rc) - goto out; - } -if (ptp->ptp_clock &&bnxt_pps_config_ok (bp))return 0 ;
@@ -861,8 +854,14 @@ int bnxt_ptp_init(struct bnxt *bp, bool phc_cfg)atomic_set (&ptp->tx_avail,BNXT_MAX_TX_TS );spin_lock_init (&ptp->ptp_lock); -if (!(bp->fw_cap &BNXT_FW_CAP_PTP_RTC )) +if (bp->fw_cap &BNXT_FW_CAP_PTP_RTC ) { +bnxt_ptp_timecounter_init (bp, false); + rc =bnxt_ptp_init_rtc (bp, phc_cfg); +if (rc) + goto out; + }else {bnxt_ptp_timecounter_init (bp, true); + } ptp->ptp_info = bnxt_ptp_caps;if ((bp->fw_cap &BNXT_FW_CAP_PTP_PPS )) {
Signature · 1 line
2.18.1
Michael Chan
maintainerPATCH 3/3 bnxt_en: Fix unnecessary dropping of RX packets
In reply to Michael ChanIn bnxt_poll_p5(), we first check cpr->has_more_work. If it is true, we are in NAPI polling mode and we will call __bnxt_poll_cqs() to continue polling. It is possible to exhanust the budget again when __bnxt_poll_cqs() returns.
We then enter the main while loop to check for new entries in the NQ. If we had previously exhausted the NAPI budget, we may call __bnxt_poll_work() to process an RX entry with zero budget. This will cause packets to be dropped unnecessarily, thinking that we are in the netpoll path. Fix it by breaking out of the while loop if we need to process an RX NQ entry with no budget left. We will then exit NAPI and stay in polling mode.
--- drivers/net/ethernet/broadcom/bnxt/bnxt.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +4 -0
@@ -2707,6 +2707,10 @@ static int bnxt_poll_p5(struct napi_struct *napi, int budget)u32 idx =le32_to_cpu (nqcmp->cq_handle_low);struct bnxt_cp_ring_info *cpr2; +/* No more budget for RX work */ +if (budget && work_done >= budget && idx ==BNXT_RX_HDL ) +break ; + cpr2 = cpr->cp_ring_arr[idx]; work_done +=__bnxt_poll_work (bp, cpr2, budget - work_done);
Signature · 1 line
2.18.1
This series was applied to netdev/net.git (master) by Jakub Kicinski <[email protected]>:
Hello:
This series was applied to netdev/net.git (master) by Jakub Kicinski <[email protected]>:
Quoted 11 lines · On Mon, 2 May 2022 21:13:09 -0400 you wrote:
This patch series includes 3 fixes:
1. Fix an occasional VF open failure.
2. Fix a PTP spinlock usage before initialization
3. Fix unnecesary RX packet drops under high TX traffic load.Michael Chan (2):
bnxt_en: Initiallizebp->ptp_lockfirst before using it
bnxt_en: Fix unnecessary dropping of RX packets[...]
Here is the summary with links:
- [net,1/3] bnxt_en: Fix possible bnxt_open() failure caused by wrong RFS flag
https://git.kernel.org/netdev/net/c/13ba794397e4 - [net,2/3] bnxt_en: Initiallize bp->ptp_lock first before using it https://git.kernel.org/netdev/net/c/2b156fb57d8f - [net,3/3] bnxt_en: Fix unnecessary dropping of RX packets https://git.kernel.org/netdev/net/c/195af57914d1
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html