diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2021-12-30 11:12:12 -0800 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2021-12-30 11:12:12 -0800 |
commit | 74c78b4291b4466b44a57b3b7c3b98ad02628686 (patch) | |
tree | 56ce20b0f82e52645e42b8afc84f61bea6d63bd1 /drivers/net/ethernet/mellanox/mlx5/core/en | |
parent | Merge tag 'char-misc-5.16' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc (diff) | |
parent | fsl/fman: Fix missing put_device() call in fman_port_probe (diff) | |
download | linux-dev-74c78b4291b4466b44a57b3b7c3b98ad02628686.tar.xz linux-dev-74c78b4291b4466b44a57b3b7c3b98ad02628686.zip |
Merge tag 'net-5.16-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Pull networking fixes from Jakub Kicinski:
"Including fixes from.. Santa?
No regressions on our radar at this point. The igc problem fixed here
was the last one I was tracking but it was broken in previous
releases, anyway. Mostly driver fixes and a couple of largish SMC
fixes.
Current release - regressions:
- xsk: initialise xskb free_list_node, fixup for a -rc7 fix
Current release - new code bugs:
- mlx5: handful of minor fixes:
- use first online CPU instead of hard coded CPU
- fix some error handling paths in 'mlx5e_tc_add_fdb_flow()'
- fix skb memory leak when TC classifier action offloads are disabled
- fix memory leak with rules with internal OvS port
Previous releases - regressions:
- igc: do not enable crosstimestamping for i225-V models
Previous releases - always broken:
- udp: use datalen to cap ipv6 udp max gso segments
- fix use-after-free in tw_timer_handler due to early free of stats
- smc: fix kernel panic caused by race of smc_sock
- smc: don't send CDC/LLC message if link not ready, avoid timeouts
- sctp: use call_rcu to free endpoint, avoid UAF in sock diag
- bridge: mcast: add and enforce query interval minimum
- usb: pegasus: do not drop long Ethernet frames
- mlx5e: fix ICOSQ recovery flow for XSK
- nfc: uapi: use kernel size_t to fix user-space builds"
* tag 'net-5.16-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (47 commits)
fsl/fman: Fix missing put_device() call in fman_port_probe
selftests: net: using ping6 for IPv6 in udpgro_fwd.sh
Documentation: fix outdated interpretation of ip_no_pmtu_disc
net/ncsi: check for error return from call to nla_put_u32
net: bridge: mcast: fix br_multicast_ctx_vlan_global_disabled helper
net: fix use-after-free in tw_timer_handler
selftests: net: Fix a typo in udpgro_fwd.sh
selftests/net: udpgso_bench_tx: fix dst ip argument
net: bridge: mcast: add and enforce startup query interval minimum
net: bridge: mcast: add and enforce query interval minimum
ipv6: raw: check passed optlen before reading
xsk: Initialise xskb free_list_node
net/mlx5e: Fix wrong features assignment in case of error
net/mlx5e: TC, Fix memory leak with rules with internal port
ionic: Initialize the 'lif->dbid_inuse' bitmap
igc: Fix TX timestamp support for non-MSI-X platforms
igc: Do not enable crosstimestamping for i225-V models
net/smc: fix kernel panic caused by race of smc_sock
net/smc: don't send CDC/LLC message if link not ready
NFC: st21nfca: Fix memory leak in device probe and remove
...
Diffstat (limited to 'drivers/net/ethernet/mellanox/mlx5/core/en')
5 files changed, 61 insertions, 4 deletions
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/health.h b/drivers/net/ethernet/mellanox/mlx5/core/en/health.h index d5b7110a4265..0107e4e73bb0 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/health.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/health.h @@ -30,6 +30,8 @@ void mlx5e_reporter_rx_destroy(struct mlx5e_priv *priv); void mlx5e_reporter_icosq_cqe_err(struct mlx5e_icosq *icosq); void mlx5e_reporter_rq_cqe_err(struct mlx5e_rq *rq); void mlx5e_reporter_rx_timeout(struct mlx5e_rq *rq); +void mlx5e_reporter_icosq_suspend_recovery(struct mlx5e_channel *c); +void mlx5e_reporter_icosq_resume_recovery(struct mlx5e_channel *c); #define MLX5E_REPORTER_PER_Q_MAX_LEN 256 diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.h b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.h index d6c7c81690eb..7c9dd3a75f8a 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.h @@ -66,7 +66,7 @@ mlx5e_rep_setup_tc(struct net_device *dev, enum tc_setup_type type, static inline void mlx5e_rep_tc_receive(struct mlx5_cqe64 *cqe, struct mlx5e_rq *rq, - struct sk_buff *skb) {} + struct sk_buff *skb) { napi_gro_receive(rq->cq.napi, skb); } #endif /* CONFIG_MLX5_CLS_ACT */ diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c index 74086eb556ae..2684e9da9f41 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c @@ -62,6 +62,7 @@ static void mlx5e_reset_icosq_cc_pc(struct mlx5e_icosq *icosq) static int mlx5e_rx_reporter_err_icosq_cqe_recover(void *ctx) { + struct mlx5e_rq *xskrq = NULL; struct mlx5_core_dev *mdev; struct mlx5e_icosq *icosq; struct net_device *dev; @@ -70,7 +71,13 @@ static int mlx5e_rx_reporter_err_icosq_cqe_recover(void *ctx) int err; icosq = ctx; + + mutex_lock(&icosq->channel->icosq_recovery_lock); + + /* mlx5e_close_rq cancels this work before RQ and ICOSQ are killed. */ rq = &icosq->channel->rq; + if (test_bit(MLX5E_RQ_STATE_ENABLED, &icosq->channel->xskrq.state)) + xskrq = &icosq->channel->xskrq; mdev = icosq->channel->mdev; dev = icosq->channel->netdev; err = mlx5_core_query_sq_state(mdev, icosq->sqn, &state); @@ -84,6 +91,9 @@ static int mlx5e_rx_reporter_err_icosq_cqe_recover(void *ctx) goto out; mlx5e_deactivate_rq(rq); + if (xskrq) + mlx5e_deactivate_rq(xskrq); + err = mlx5e_wait_for_icosq_flush(icosq); if (err) goto out; @@ -97,15 +107,28 @@ static int mlx5e_rx_reporter_err_icosq_cqe_recover(void *ctx) goto out; mlx5e_reset_icosq_cc_pc(icosq); + mlx5e_free_rx_in_progress_descs(rq); + if (xskrq) + mlx5e_free_rx_in_progress_descs(xskrq); + clear_bit(MLX5E_SQ_STATE_RECOVERING, &icosq->state); mlx5e_activate_icosq(icosq); - mlx5e_activate_rq(rq); + mlx5e_activate_rq(rq); rq->stats->recover++; + + if (xskrq) { + mlx5e_activate_rq(xskrq); + xskrq->stats->recover++; + } + + mutex_unlock(&icosq->channel->icosq_recovery_lock); + return 0; out: clear_bit(MLX5E_SQ_STATE_RECOVERING, &icosq->state); + mutex_unlock(&icosq->channel->icosq_recovery_lock); return err; } @@ -706,6 +729,16 @@ void mlx5e_reporter_icosq_cqe_err(struct mlx5e_icosq *icosq) mlx5e_health_report(priv, priv->rx_reporter, err_str, &err_ctx); } +void mlx5e_reporter_icosq_suspend_recovery(struct mlx5e_channel *c) +{ + mutex_lock(&c->icosq_recovery_lock); +} + +void mlx5e_reporter_icosq_resume_recovery(struct mlx5e_channel *c) +{ + mutex_unlock(&c->icosq_recovery_lock); +} + static const struct devlink_health_reporter_ops mlx5_rx_reporter_ops = { .name = "rx", .recover = mlx5e_rx_reporter_recover, diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c index 4f4bc8726ec4..614cd9477600 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c @@ -466,6 +466,14 @@ static int mlx5e_tx_reporter_dump_sq(struct mlx5e_priv *priv, struct devlink_fms return mlx5e_health_fmsg_named_obj_nest_end(fmsg); } +static int mlx5e_tx_reporter_timeout_dump(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg, + void *ctx) +{ + struct mlx5e_tx_timeout_ctx *to_ctx = ctx; + + return mlx5e_tx_reporter_dump_sq(priv, fmsg, to_ctx->sq); +} + static int mlx5e_tx_reporter_dump_all_sqs(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg) { @@ -561,7 +569,7 @@ int mlx5e_reporter_tx_timeout(struct mlx5e_txqsq *sq) to_ctx.sq = sq; err_ctx.ctx = &to_ctx; err_ctx.recover = mlx5e_tx_reporter_timeout_recover; - err_ctx.dump = mlx5e_tx_reporter_dump_sq; + err_ctx.dump = mlx5e_tx_reporter_timeout_dump; snprintf(err_str, sizeof(err_str), "TX timeout on queue: %d, SQ: 0x%x, CQ: 0x%x, SQ Cons: 0x%x SQ Prod: 0x%x, usecs since last trans: %u", sq->ch_ix, sq->sqn, sq->cq.mcq.cqn, sq->cc, sq->pc, diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c index 538bc2419bd8..8526a5fbbf0b 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c @@ -4,6 +4,7 @@ #include "setup.h" #include "en/params.h" #include "en/txrx.h" +#include "en/health.h" /* It matches XDP_UMEM_MIN_CHUNK_SIZE, but as this constant is private and may * change unexpectedly, and mlx5e has a minimum valid stride size for striding @@ -170,7 +171,13 @@ void mlx5e_close_xsk(struct mlx5e_channel *c) void mlx5e_activate_xsk(struct mlx5e_channel *c) { + /* ICOSQ recovery deactivates RQs. Suspend the recovery to avoid + * activating XSKRQ in the middle of recovery. + */ + mlx5e_reporter_icosq_suspend_recovery(c); set_bit(MLX5E_RQ_STATE_ENABLED, &c->xskrq.state); + mlx5e_reporter_icosq_resume_recovery(c); + /* TX queue is created active. */ spin_lock_bh(&c->async_icosq_lock); @@ -180,6 +187,13 @@ void mlx5e_activate_xsk(struct mlx5e_channel *c) void mlx5e_deactivate_xsk(struct mlx5e_channel *c) { - mlx5e_deactivate_rq(&c->xskrq); + /* ICOSQ recovery may reactivate XSKRQ if clear_bit is called in the + * middle of recovery. Suspend the recovery to avoid it. + */ + mlx5e_reporter_icosq_suspend_recovery(c); + clear_bit(MLX5E_RQ_STATE_ENABLED, &c->xskrq.state); + mlx5e_reporter_icosq_resume_recovery(c); + synchronize_net(); /* Sync with NAPI to prevent mlx5e_post_rx_wqes. */ + /* TX queue is disabled on close. */ } |