From ef1c28817bf90aab3a6365ec81c30c09a3b18ece Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 19 Dec 2024 18:52:34 -0800 Subject: [PATCH 01/16] eth: fbnic: don't reset the secondary RSS indir table Secondary RSS indirection table is for additional contexts. It can / should be initialized when such context is created. Since we don't support creating RSS contexts, yet, this change has no user visible effect. Link: https://patch.msgid.link/20241220025241.1522781-4-kuba@kernel.org Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/meta/fbnic/fbnic_rpc.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_rpc.c b/drivers/net/ethernet/meta/fbnic/fbnic_rpc.c index 908c098cd59e..b99c890ac43f 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_rpc.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_rpc.c @@ -13,10 +13,8 @@ void fbnic_reset_indir_tbl(struct fbnic_net *fbn) unsigned int num_rx = fbn->num_rx_queues; unsigned int i; - for (i = 0; i < FBNIC_RPC_RSS_TBL_SIZE; i++) { + for (i = 0; i < FBNIC_RPC_RSS_TBL_SIZE; i++) fbn->indir_tbl[0][i] = ethtool_rxfh_indir_default(i, num_rx); - fbn->indir_tbl[1][i] = ethtool_rxfh_indir_default(i, num_rx); - } } void fbnic_rss_key_fill(u32 *buffer) -- 2.51.0 From 31ab733e999edbc4070d8386c608d9f0b73267c5 Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Thu, 19 Dec 2024 18:52:35 -0800 Subject: [PATCH 02/16] eth: fbnic: support setting RSS configuration Let the user program the RSS indirection table and the RSS key. Straightforward implementation. Track the changes and don't bother poking the HW if user asked for a config identical to what's already programmed. The device only supports Toeplitz hash. Similarly to the GET support - all the real code that does the programming was part of initial driver submission, already. Signed-off-by: Alexander Duyck Link: https://patch.msgid.link/20241220025241.1522781-5-kuba@kernel.org Signed-off-by: Jakub Kicinski --- .../net/ethernet/meta/fbnic/fbnic_ethtool.c | 55 +++++++++++++++++++ 1 file changed, 55 insertions(+) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c b/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c index e71ae6abb0f5..5523803c8edd 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c @@ -201,6 +201,60 @@ fbnic_get_rxfh(struct net_device *netdev, struct ethtool_rxfh_param *rxfh) return 0; } +static unsigned int +fbnic_set_indir(struct fbnic_net *fbn, unsigned int idx, const u32 *indir) +{ + unsigned int i, changes = 0; + + for (i = 0; i < FBNIC_RPC_RSS_TBL_SIZE; i++) { + if (fbn->indir_tbl[idx][i] == indir[i]) + continue; + + fbn->indir_tbl[idx][i] = indir[i]; + changes++; + } + + return changes; +} + +static int +fbnic_set_rxfh(struct net_device *netdev, struct ethtool_rxfh_param *rxfh, + struct netlink_ext_ack *extack) +{ + struct fbnic_net *fbn = netdev_priv(netdev); + unsigned int i, changes = 0; + + if (rxfh->hfunc != ETH_RSS_HASH_NO_CHANGE && + rxfh->hfunc != ETH_RSS_HASH_TOP) + return -EINVAL; + + if (rxfh->key) { + u32 rss_key = 0; + + for (i = FBNIC_RPC_RSS_KEY_BYTE_LEN; i--;) { + rss_key >>= 8; + rss_key |= (u32)(rxfh->key[i]) << 24; + + if (i % 4) + continue; + + if (fbn->rss_key[i / 4] == rss_key) + continue; + + fbn->rss_key[i / 4] = rss_key; + changes++; + } + } + + if (rxfh->indir) + changes += fbnic_set_indir(fbn, 0, rxfh->indir); + + if (changes && netif_running(netdev)) + fbnic_rss_reinit_hw(fbn->fbd, fbn); + + return 0; +} + static int fbnic_get_ts_info(struct net_device *netdev, struct kernel_ethtool_ts_info *tsinfo) @@ -312,6 +366,7 @@ static const struct ethtool_ops fbnic_ethtool_ops = { .get_rxfh_key_size = fbnic_get_rxfh_key_size, .get_rxfh_indir_size = fbnic_get_rxfh_indir_size, .get_rxfh = fbnic_get_rxfh, + .set_rxfh = fbnic_set_rxfh, .get_ts_info = fbnic_get_ts_info, .get_ts_stats = fbnic_get_ts_stats, .get_eth_mac_stats = fbnic_get_eth_mac_stats, -- 2.51.0 From c23a1461bfee0a6f158795a58c768911c49d6cd0 Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Thu, 19 Dec 2024 18:52:36 -0800 Subject: [PATCH 03/16] eth: fbnic: let user control the RSS hash fields Support setting the fields over which RSS computes its hash. Signed-off-by: Alexander Duyck Link: https://patch.msgid.link/20241220025241.1522781-6-kuba@kernel.org Signed-off-by: Jakub Kicinski --- .../net/ethernet/meta/fbnic/fbnic_ethtool.c | 50 +++++++++++++++++++ 1 file changed, 50 insertions(+) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c b/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c index 5523803c8edd..d1be8fc30404 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c @@ -167,6 +167,55 @@ static int fbnic_get_rxnfc(struct net_device *netdev, return ret; } +#define FBNIC_L2_HASH_OPTIONS \ + (RXH_L2DA | RXH_DISCARD) +#define FBNIC_L3_HASH_OPTIONS \ + (FBNIC_L2_HASH_OPTIONS | RXH_IP_SRC | RXH_IP_DST) +#define FBNIC_L4_HASH_OPTIONS \ + (FBNIC_L3_HASH_OPTIONS | RXH_L4_B_0_1 | RXH_L4_B_2_3) + +static int +fbnic_set_rss_hash_opts(struct fbnic_net *fbn, const struct ethtool_rxnfc *cmd) +{ + int hash_opt_idx; + + /* Verify the type requested is correct */ + hash_opt_idx = fbnic_get_rss_hash_idx(cmd->flow_type); + if (hash_opt_idx < 0) + return -EINVAL; + + /* Verify the fields asked for can actually be assigned based on type */ + if (cmd->data & ~FBNIC_L4_HASH_OPTIONS || + (hash_opt_idx > FBNIC_L4_HASH_OPT && + cmd->data & ~FBNIC_L3_HASH_OPTIONS) || + (hash_opt_idx > FBNIC_IP_HASH_OPT && + cmd->data & ~FBNIC_L2_HASH_OPTIONS)) + return -EINVAL; + + fbn->rss_flow_hash[hash_opt_idx] = cmd->data; + + if (netif_running(fbn->netdev)) { + fbnic_rss_reinit(fbn->fbd, fbn); + fbnic_write_rules(fbn->fbd); + } + + return 0; +} + +static int fbnic_set_rxnfc(struct net_device *netdev, struct ethtool_rxnfc *cmd) +{ + struct fbnic_net *fbn = netdev_priv(netdev); + int ret = -EOPNOTSUPP; + + switch (cmd->cmd) { + case ETHTOOL_SRXFH: + ret = fbnic_set_rss_hash_opts(fbn, cmd); + break; + } + + return ret; +} + static u32 fbnic_get_rxfh_key_size(struct net_device *netdev) { return FBNIC_RPC_RSS_KEY_BYTE_LEN; @@ -363,6 +412,7 @@ static const struct ethtool_ops fbnic_ethtool_ops = { .get_ethtool_stats = fbnic_get_ethtool_stats, .get_sset_count = fbnic_get_sset_count, .get_rxnfc = fbnic_get_rxnfc, + .set_rxnfc = fbnic_set_rxnfc, .get_rxfh_key_size = fbnic_get_rxfh_key_size, .get_rxfh_indir_size = fbnic_get_rxfh_indir_size, .get_rxfh = fbnic_get_rxfh, -- 2.51.0 From db7159c400ffbbf3e0df1f3ef6b847b7b62186a3 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 19 Dec 2024 18:52:37 -0800 Subject: [PATCH 04/16] eth: fbnic: store NAPIs in an array instead of the list We will need an array for storing NAPIs in the upcoming IRQ handler reuse rework. Replace the current list we have, so that we are able to reuse it later. In a few places replace i as the iterator with t when we iterate over triads, this seems slightly less confusing than having i, j, k variables. Link: https://patch.msgid.link/20241220025241.1522781-7-kuba@kernel.org Signed-off-by: Jakub Kicinski --- .../net/ethernet/meta/fbnic/fbnic_netdev.c | 1 - .../net/ethernet/meta/fbnic/fbnic_netdev.h | 6 +- drivers/net/ethernet/meta/fbnic/fbnic_txrx.c | 117 ++++++++++-------- drivers/net/ethernet/meta/fbnic/fbnic_txrx.h | 7 +- 4 files changed, 71 insertions(+), 60 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c index fc7d80db5fa6..558644c49a4b 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c @@ -615,7 +615,6 @@ struct net_device *fbnic_netdev_alloc(struct fbnic_dev *fbd) fbn->netdev = netdev; fbn->fbd = fbd; - INIT_LIST_HEAD(&fbn->napis); fbn->txq_size = FBNIC_TXQ_SIZE_DEFAULT; fbn->hpq_size = FBNIC_HPQ_SIZE_DEFAULT; diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.h b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.h index b8417b300778..0986c8f120a8 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.h @@ -11,10 +11,14 @@ #include "fbnic_rpc.h" #include "fbnic_txrx.h" +#define FBNIC_MAX_NAPI_VECTORS 128u + struct fbnic_net { struct fbnic_ring *tx[FBNIC_MAX_TXQS]; struct fbnic_ring *rx[FBNIC_MAX_RXQS]; + struct fbnic_napi_vector *napi[FBNIC_MAX_NAPI_VECTORS]; + struct net_device *netdev; struct fbnic_dev *fbd; @@ -56,8 +60,6 @@ struct fbnic_net { /* Time stampinn filter config */ struct kernel_hwtstamp_config hwtstamp_config; - - struct list_head napis; }; int __fbnic_open(struct fbnic_net *fbn); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c index b5050fabe8fe..87e4eb03d991 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c @@ -1116,16 +1116,17 @@ static void fbnic_free_napi_vector(struct fbnic_net *fbn, fbnic_free_irq(fbd, v_idx, nv); page_pool_destroy(nv->page_pool); netif_napi_del(&nv->napi); - list_del(&nv->napis); + fbn->napi[fbnic_napi_idx(nv)] = NULL; kfree(nv); } void fbnic_free_napi_vectors(struct fbnic_net *fbn) { - struct fbnic_napi_vector *nv, *temp; + int i; - list_for_each_entry_safe(nv, temp, &fbn->napis, napis) - fbnic_free_napi_vector(fbn, nv); + for (i = 0; i < fbn->num_napi; i++) + if (fbn->napi[i]) + fbnic_free_napi_vector(fbn, fbn->napi[i]); } static void fbnic_name_napi_vector(struct fbnic_napi_vector *nv) @@ -1222,7 +1223,7 @@ static int fbnic_alloc_napi_vector(struct fbnic_dev *fbd, struct fbnic_net *fbn, nv->v_idx = v_idx; /* Tie napi to netdev */ - list_add(&nv->napis, &fbn->napis); + fbn->napi[fbnic_napi_idx(nv)] = nv; netif_napi_add(fbn->netdev, &nv->napi, fbnic_poll); /* Record IRQ to NAPI struct */ @@ -1307,7 +1308,7 @@ pp_destroy: page_pool_destroy(nv->page_pool); napi_del: netif_napi_del(&nv->napi); - list_del(&nv->napis); + fbn->napi[fbnic_napi_idx(nv)] = NULL; kfree(nv); return err; } @@ -1612,19 +1613,18 @@ free_resources: void fbnic_free_resources(struct fbnic_net *fbn) { - struct fbnic_napi_vector *nv; + int i; - list_for_each_entry(nv, &fbn->napis, napis) - fbnic_free_nv_resources(fbn, nv); + for (i = 0; i < fbn->num_napi; i++) + fbnic_free_nv_resources(fbn, fbn->napi[i]); } int fbnic_alloc_resources(struct fbnic_net *fbn) { - struct fbnic_napi_vector *nv; - int err = -ENODEV; + int i, err = -ENODEV; - list_for_each_entry(nv, &fbn->napis, napis) { - err = fbnic_alloc_nv_resources(fbn, nv); + for (i = 0; i < fbn->num_napi; i++) { + err = fbnic_alloc_nv_resources(fbn, fbn->napi[i]); if (err) goto free_resources; } @@ -1632,8 +1632,8 @@ int fbnic_alloc_resources(struct fbnic_net *fbn) return 0; free_resources: - list_for_each_entry_continue_reverse(nv, &fbn->napis, napis) - fbnic_free_nv_resources(fbn, nv); + while (i--) + fbnic_free_nv_resources(fbn, fbn->napi[i]); return err; } @@ -1670,33 +1670,34 @@ static void fbnic_disable_rcq(struct fbnic_ring *rxr) void fbnic_napi_disable(struct fbnic_net *fbn) { - struct fbnic_napi_vector *nv; + int i; - list_for_each_entry(nv, &fbn->napis, napis) { - napi_disable(&nv->napi); + for (i = 0; i < fbn->num_napi; i++) { + napi_disable(&fbn->napi[i]->napi); - fbnic_nv_irq_disable(nv); + fbnic_nv_irq_disable(fbn->napi[i]); } } void fbnic_disable(struct fbnic_net *fbn) { struct fbnic_dev *fbd = fbn->fbd; - struct fbnic_napi_vector *nv; - int i, j; + int i, j, t; + + for (i = 0; i < fbn->num_napi; i++) { + struct fbnic_napi_vector *nv = fbn->napi[i]; - list_for_each_entry(nv, &fbn->napis, napis) { /* Disable Tx queue triads */ - for (i = 0; i < nv->txt_count; i++) { - struct fbnic_q_triad *qt = &nv->qt[i]; + for (t = 0; t < nv->txt_count; t++) { + struct fbnic_q_triad *qt = &nv->qt[t]; fbnic_disable_twq0(&qt->sub0); fbnic_disable_tcq(&qt->cmpl); } /* Disable Rx queue triads */ - for (j = 0; j < nv->rxt_count; j++, i++) { - struct fbnic_q_triad *qt = &nv->qt[i]; + for (j = 0; j < nv->rxt_count; j++, t++) { + struct fbnic_q_triad *qt = &nv->qt[t]; fbnic_disable_bdq(&qt->sub0, &qt->sub1); fbnic_disable_rcq(&qt->cmpl); @@ -1792,14 +1793,15 @@ int fbnic_wait_all_queues_idle(struct fbnic_dev *fbd, bool may_fail) void fbnic_flush(struct fbnic_net *fbn) { - struct fbnic_napi_vector *nv; + int i; - list_for_each_entry(nv, &fbn->napis, napis) { - int i, j; + for (i = 0; i < fbn->num_napi; i++) { + struct fbnic_napi_vector *nv = fbn->napi[i]; + int j, t; /* Flush any processed Tx Queue Triads and drop the rest */ - for (i = 0; i < nv->txt_count; i++) { - struct fbnic_q_triad *qt = &nv->qt[i]; + for (t = 0; t < nv->txt_count; t++) { + struct fbnic_q_triad *qt = &nv->qt[t]; struct netdev_queue *tx_queue; /* Clean the work queues of unprocessed work */ @@ -1823,8 +1825,8 @@ void fbnic_flush(struct fbnic_net *fbn) } /* Flush any processed Rx Queue Triads and drop the rest */ - for (j = 0; j < nv->rxt_count; j++, i++) { - struct fbnic_q_triad *qt = &nv->qt[i]; + for (j = 0; j < nv->rxt_count; j++, t++) { + struct fbnic_q_triad *qt = &nv->qt[t]; /* Clean the work queues of unprocessed work */ fbnic_clean_bdq(nv, 0, &qt->sub0, qt->sub0.tail); @@ -1845,14 +1847,15 @@ void fbnic_flush(struct fbnic_net *fbn) void fbnic_fill(struct fbnic_net *fbn) { - struct fbnic_napi_vector *nv; + int i; - list_for_each_entry(nv, &fbn->napis, napis) { - int i, j; + for (i = 0; i < fbn->num_napi; i++) { + struct fbnic_napi_vector *nv = fbn->napi[i]; + int j, t; /* Configure NAPI mapping for Tx */ - for (i = 0; i < nv->txt_count; i++) { - struct fbnic_q_triad *qt = &nv->qt[i]; + for (t = 0; t < nv->txt_count; t++) { + struct fbnic_q_triad *qt = &nv->qt[t]; /* Nothing to do if Tx queue is disabled */ if (qt->sub0.flags & FBNIC_RING_F_DISABLED) @@ -1866,8 +1869,8 @@ void fbnic_fill(struct fbnic_net *fbn) /* Configure NAPI mapping and populate pages * in the BDQ rings to use for Rx */ - for (j = 0; j < nv->rxt_count; j++, i++) { - struct fbnic_q_triad *qt = &nv->qt[i]; + for (j = 0; j < nv->rxt_count; j++, t++) { + struct fbnic_q_triad *qt = &nv->qt[t]; /* Associate Rx queue with NAPI */ netif_queue_set_napi(nv->napi.dev, qt->cmpl.q_idx, @@ -2025,21 +2028,23 @@ static void fbnic_enable_rcq(struct fbnic_napi_vector *nv, void fbnic_enable(struct fbnic_net *fbn) { struct fbnic_dev *fbd = fbn->fbd; - struct fbnic_napi_vector *nv; - int i, j; + int i; + + for (i = 0; i < fbn->num_napi; i++) { + struct fbnic_napi_vector *nv = fbn->napi[i]; + int j, t; - list_for_each_entry(nv, &fbn->napis, napis) { /* Setup Tx Queue Triads */ - for (i = 0; i < nv->txt_count; i++) { - struct fbnic_q_triad *qt = &nv->qt[i]; + for (t = 0; t < nv->txt_count; t++) { + struct fbnic_q_triad *qt = &nv->qt[t]; fbnic_enable_twq0(&qt->sub0); fbnic_enable_tcq(nv, &qt->cmpl); } /* Setup Rx Queue Triads */ - for (j = 0; j < nv->rxt_count; j++, i++) { - struct fbnic_q_triad *qt = &nv->qt[i]; + for (j = 0; j < nv->rxt_count; j++, t++) { + struct fbnic_q_triad *qt = &nv->qt[t]; fbnic_enable_bdq(&qt->sub0, &qt->sub1); fbnic_config_drop_mode_rcq(nv, &qt->cmpl); @@ -2064,10 +2069,11 @@ void fbnic_napi_enable(struct fbnic_net *fbn) { u32 irqs[FBNIC_MAX_MSIX_VECS / 32] = {}; struct fbnic_dev *fbd = fbn->fbd; - struct fbnic_napi_vector *nv; int i; - list_for_each_entry(nv, &fbn->napis, napis) { + for (i = 0; i < fbn->num_napi; i++) { + struct fbnic_napi_vector *nv = fbn->napi[i]; + napi_enable(&nv->napi); fbnic_nv_irq_enable(nv); @@ -2096,17 +2102,18 @@ void fbnic_napi_depletion_check(struct net_device *netdev) struct fbnic_net *fbn = netdev_priv(netdev); u32 irqs[FBNIC_MAX_MSIX_VECS / 32] = {}; struct fbnic_dev *fbd = fbn->fbd; - struct fbnic_napi_vector *nv; - int i, j; + int i, j, t; + + for (i = 0; i < fbn->num_napi; i++) { + struct fbnic_napi_vector *nv = fbn->napi[i]; - list_for_each_entry(nv, &fbn->napis, napis) { /* Find RQs which are completely out of pages */ - for (i = nv->txt_count, j = 0; j < nv->rxt_count; j++, i++) { + for (t = nv->txt_count, j = 0; j < nv->rxt_count; j++, t++) { /* Assume 4 pages is always enough to fit a packet * and therefore generate a completion and an IRQ. */ - if (fbnic_desc_used(&nv->qt[i].sub0) < 4 || - fbnic_desc_used(&nv->qt[i].sub1) < 4) + if (fbnic_desc_used(&nv->qt[t].sub0) < 4 || + fbnic_desc_used(&nv->qt[t].sub1) < 4) irqs[nv->v_idx / 32] |= BIT(nv->v_idx % 32); } } diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h index 8d626287c3f4..1965d1fa38a2 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h @@ -110,8 +110,6 @@ struct fbnic_napi_vector { u8 txt_count; u8 rxt_count; - struct list_head napis; - struct fbnic_q_triad qt[]; }; @@ -137,4 +135,9 @@ void fbnic_fill(struct fbnic_net *fbn); void fbnic_napi_depletion_check(struct net_device *netdev); int fbnic_wait_all_queues_idle(struct fbnic_dev *fbd, bool may_fail); +static inline int fbnic_napi_idx(const struct fbnic_napi_vector *nv) +{ + return nv->v_idx - FBNIC_NON_NAPI_VECTORS; +} + #endif /* _FBNIC_TXRX_H_ */ -- 2.51.0 From 3a856ab347261870d2bb6f3cab95325f27eee104 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 19 Dec 2024 18:52:38 -0800 Subject: [PATCH 05/16] eth: fbnic: add IRQ reuse support Change our method of swapping NAPIs without disturbing existing config. This is primarily needed for "live reconfiguration" such as changing the channel count when interface is already up. Previously we were planning to use a trick of using shared interrupts. We would install a second IRQ handler for the new NAPI, and make it return IRQ_NONE until we were ready for it to take over. This works fine functionally but breaks IRQ naming. The IRQ subsystem uses the IRQ name to create the procfs entry, since both handlers used the same name the second handler wouldn't get a proc directory registered. When first one gets removed on success full ring count change it would remove its directory and we would be left with none. New approach uses a double pointer to the NAPI. The IRQ handler needs to know how to locate the NAPI to schedule. We register a single IRQ handler and give it a pointer to a pointer. We can then change what it points to without re-registering. This may have a tiny perf impact, but really really negligible. Link: https://patch.msgid.link/20241220025241.1522781-8-kuba@kernel.org Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/meta/fbnic/fbnic.h | 14 +++++++ drivers/net/ethernet/meta/fbnic/fbnic_irq.c | 42 +++++++++++++++++++ .../net/ethernet/meta/fbnic/fbnic_netdev.c | 2 + drivers/net/ethernet/meta/fbnic/fbnic_txrx.c | 25 ++--------- drivers/net/ethernet/meta/fbnic/fbnic_txrx.h | 2 +- 5 files changed, 63 insertions(+), 22 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic.h b/drivers/net/ethernet/meta/fbnic/fbnic.h index 706ae6104c8e..ed527209b30c 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic.h @@ -16,6 +16,10 @@ #include "fbnic_mac.h" #include "fbnic_rpc.h" +struct fbnic_napi_vector; + +#define FBNIC_MAX_NAPI_VECTORS 128u + struct fbnic_dev { struct device *dev; struct net_device *netdev; @@ -29,6 +33,11 @@ struct fbnic_dev { unsigned int pcs_msix_vector; unsigned short num_irqs; + struct { + u8 users; + char name[IFNAMSIZ + 9]; + } napi_irq[FBNIC_MAX_NAPI_VECTORS]; + struct delayed_work service_task; struct fbnic_fw_mbx mbx[FBNIC_IPC_MBX_INDICES]; @@ -148,6 +157,11 @@ void fbnic_hwmon_unregister(struct fbnic_dev *fbd); int fbnic_pcs_irq_enable(struct fbnic_dev *fbd); void fbnic_pcs_irq_disable(struct fbnic_dev *fbd); +void fbnic_napi_name_irqs(struct fbnic_dev *fbd); +int fbnic_napi_request_irq(struct fbnic_dev *fbd, + struct fbnic_napi_vector *nv); +void fbnic_napi_free_irq(struct fbnic_dev *fbd, + struct fbnic_napi_vector *nv); int fbnic_request_irq(struct fbnic_dev *dev, int nr, irq_handler_t handler, unsigned long flags, const char *name, void *data); void fbnic_free_irq(struct fbnic_dev *dev, int nr, void *data); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_irq.c b/drivers/net/ethernet/meta/fbnic/fbnic_irq.c index 914362195920..a8ea7b6774a8 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_irq.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_irq.c @@ -169,6 +169,48 @@ void fbnic_free_irq(struct fbnic_dev *fbd, int nr, void *data) free_irq(irq, data); } +void fbnic_napi_name_irqs(struct fbnic_dev *fbd) +{ + unsigned int i; + + for (i = 0; i < ARRAY_SIZE(fbd->napi_irq); i++) + snprintf(fbd->napi_irq[i].name, + sizeof(fbd->napi_irq[i].name), + "%s-TxRx-%u", fbd->netdev->name, i); +} + +int fbnic_napi_request_irq(struct fbnic_dev *fbd, + struct fbnic_napi_vector *nv) +{ + struct fbnic_net *fbn = netdev_priv(fbd->netdev); + int i = fbnic_napi_idx(nv); + int err; + + if (!fbd->napi_irq[i].users) { + err = fbnic_request_irq(fbd, nv->v_idx, + fbnic_msix_clean_rings, 0, + fbd->napi_irq[i].name, + &fbn->napi[i]); + if (err) + return err; + } + + fbd->napi_irq[i].users++; + return 0; +} + +void fbnic_napi_free_irq(struct fbnic_dev *fbd, + struct fbnic_napi_vector *nv) +{ + struct fbnic_net *fbn = netdev_priv(fbd->netdev); + int i = fbnic_napi_idx(nv); + + if (--fbd->napi_irq[i].users) + return; + + fbnic_free_irq(fbd, nv->v_idx, &fbn->napi[i]); +} + void fbnic_free_irqs(struct fbnic_dev *fbd) { struct pci_dev *pdev = to_pci_dev(fbd->dev); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c index 558644c49a4b..2f19144e4410 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c @@ -74,6 +74,8 @@ static int fbnic_open(struct net_device *netdev) struct fbnic_net *fbn = netdev_priv(netdev); int err; + fbnic_napi_name_irqs(fbn->fbd); + err = __fbnic_open(fbn); if (!err) fbnic_up(fbn); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c index 87e4eb03d991..75b491b8e1ca 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c @@ -1036,9 +1036,9 @@ static int fbnic_poll(struct napi_struct *napi, int budget) return 0; } -static irqreturn_t fbnic_msix_clean_rings(int __always_unused irq, void *data) +irqreturn_t fbnic_msix_clean_rings(int __always_unused irq, void *data) { - struct fbnic_napi_vector *nv = data; + struct fbnic_napi_vector *nv = *(void **)data; napi_schedule_irqoff(&nv->napi); @@ -1099,7 +1099,6 @@ static void fbnic_free_napi_vector(struct fbnic_net *fbn, struct fbnic_napi_vector *nv) { struct fbnic_dev *fbd = nv->fbd; - u32 v_idx = nv->v_idx; int i, j; for (i = 0; i < nv->txt_count; i++) { @@ -1113,7 +1112,7 @@ static void fbnic_free_napi_vector(struct fbnic_net *fbn, fbnic_remove_rx_ring(fbn, &nv->qt[i].cmpl); } - fbnic_free_irq(fbd, v_idx, nv); + fbnic_napi_free_irq(fbd, nv); page_pool_destroy(nv->page_pool); netif_napi_del(&nv->napi); fbn->napi[fbnic_napi_idx(nv)] = NULL; @@ -1129,18 +1128,6 @@ void fbnic_free_napi_vectors(struct fbnic_net *fbn) fbnic_free_napi_vector(fbn, fbn->napi[i]); } -static void fbnic_name_napi_vector(struct fbnic_napi_vector *nv) -{ - unsigned char *dev_name = nv->napi.dev->name; - - if (!nv->rxt_count) - snprintf(nv->name, sizeof(nv->name), "%s-Tx-%u", dev_name, - nv->v_idx - FBNIC_NON_NAPI_VECTORS); - else - snprintf(nv->name, sizeof(nv->name), "%s-TxRx-%u", dev_name, - nv->v_idx - FBNIC_NON_NAPI_VECTORS); -} - #define FBNIC_PAGE_POOL_FLAGS \ (PP_FLAG_DMA_MAP | PP_FLAG_DMA_SYNC_DEV) @@ -1240,12 +1227,8 @@ static int fbnic_alloc_napi_vector(struct fbnic_dev *fbd, struct fbnic_net *fbn, goto napi_del; } - /* Initialize vector name */ - fbnic_name_napi_vector(nv); - /* Request the IRQ for napi vector */ - err = fbnic_request_irq(fbd, v_idx, &fbnic_msix_clean_rings, - IRQF_SHARED, nv->name, nv); + err = fbnic_napi_request_irq(fbd, nv); if (err) goto pp_destroy; diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h index 1965d1fa38a2..c8d908860ab0 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h @@ -104,7 +104,6 @@ struct fbnic_napi_vector { struct device *dev; /* Device for DMA unmapping */ struct page_pool *page_pool; struct fbnic_dev *fbd; - char name[IFNAMSIZ + 9]; u16 v_idx; u8 txt_count; @@ -125,6 +124,7 @@ int fbnic_alloc_napi_vectors(struct fbnic_net *fbn); void fbnic_free_napi_vectors(struct fbnic_net *fbn); int fbnic_alloc_resources(struct fbnic_net *fbn); void fbnic_free_resources(struct fbnic_net *fbn); +irqreturn_t fbnic_msix_clean_rings(int irq, void *data); void fbnic_napi_enable(struct fbnic_net *fbn); void fbnic_napi_disable(struct fbnic_net *fbn); void fbnic_enable(struct fbnic_net *fbn); -- 2.51.0 From 557d02238e05eb66b9aba9a1f90f3a2131c6c887 Mon Sep 17 00:00:00 2001 From: Alexander Duyck Date: Thu, 19 Dec 2024 18:52:39 -0800 Subject: [PATCH 06/16] eth: fbnic: centralize the queue count and NAPI<>queue setting To simplify dealing with RTNL_ASSERT() requirements further down the line, move setting queue count and NAPI<>queue association to their own helpers. Signed-off-by: Alexander Duyck Link: https://patch.msgid.link/20241220025241.1522781-9-kuba@kernel.org Signed-off-by: Jakub Kicinski --- .../net/ethernet/meta/fbnic/fbnic_netdev.c | 9 +- drivers/net/ethernet/meta/fbnic/fbnic_txrx.c | 92 +++++++++++++------ drivers/net/ethernet/meta/fbnic/fbnic_txrx.h | 2 + 3 files changed, 70 insertions(+), 33 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c index 2f19144e4410..7a96b6ee773f 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.c @@ -23,13 +23,7 @@ int __fbnic_open(struct fbnic_net *fbn) if (err) goto free_napi_vectors; - err = netif_set_real_num_tx_queues(fbn->netdev, - fbn->num_tx_queues); - if (err) - goto free_resources; - - err = netif_set_real_num_rx_queues(fbn->netdev, - fbn->num_rx_queues); + err = fbnic_set_netif_queues(fbn); if (err) goto free_resources; @@ -93,6 +87,7 @@ static int fbnic_stop(struct net_device *netdev) fbnic_time_stop(fbn); fbnic_fw_xmit_ownership_msg(fbn->fbd, false); + fbnic_reset_netif_queues(fbn); fbnic_free_resources(fbn); fbnic_free_napi_vectors(fbn); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c index 75b491b8e1ca..92fc1ad6ed6f 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c @@ -1621,6 +1621,71 @@ free_resources: return err; } +static void fbnic_set_netif_napi(struct fbnic_napi_vector *nv) +{ + int i, j; + + /* Associate Tx queue with NAPI */ + for (i = 0; i < nv->txt_count; i++) { + struct fbnic_q_triad *qt = &nv->qt[i]; + + netif_queue_set_napi(nv->napi.dev, qt->sub0.q_idx, + NETDEV_QUEUE_TYPE_TX, &nv->napi); + } + + /* Associate Rx queue with NAPI */ + for (j = 0; j < nv->rxt_count; j++, i++) { + struct fbnic_q_triad *qt = &nv->qt[i]; + + netif_queue_set_napi(nv->napi.dev, qt->cmpl.q_idx, + NETDEV_QUEUE_TYPE_RX, &nv->napi); + } +} + +static void fbnic_reset_netif_napi(struct fbnic_napi_vector *nv) +{ + int i, j; + + /* Disassociate Tx queue from NAPI */ + for (i = 0; i < nv->txt_count; i++) { + struct fbnic_q_triad *qt = &nv->qt[i]; + + netif_queue_set_napi(nv->napi.dev, qt->sub0.q_idx, + NETDEV_QUEUE_TYPE_TX, NULL); + } + + /* Disassociate Rx queue from NAPI */ + for (j = 0; j < nv->rxt_count; j++, i++) { + struct fbnic_q_triad *qt = &nv->qt[i]; + + netif_queue_set_napi(nv->napi.dev, qt->cmpl.q_idx, + NETDEV_QUEUE_TYPE_RX, NULL); + } +} + +int fbnic_set_netif_queues(struct fbnic_net *fbn) +{ + int i, err; + + err = netif_set_real_num_queues(fbn->netdev, fbn->num_tx_queues, + fbn->num_rx_queues); + if (err) + return err; + + for (i = 0; i < fbn->num_napi; i++) + fbnic_set_netif_napi(fbn->napi[i]); + + return 0; +} + +void fbnic_reset_netif_queues(struct fbnic_net *fbn) +{ + int i; + + for (i = 0; i < fbn->num_napi; i++) + fbnic_reset_netif_napi(fbn->napi[i]); +} + static void fbnic_disable_twq0(struct fbnic_ring *txr) { u32 twq_ctl = fbnic_ring_rd32(txr, FBNIC_QUEUE_TWQ0_CTL); @@ -1801,10 +1866,6 @@ void fbnic_flush(struct fbnic_net *fbn) tx_queue = netdev_get_tx_queue(nv->napi.dev, qt->sub0.q_idx); netdev_tx_reset_queue(tx_queue); - - /* Disassociate Tx queue from NAPI */ - netif_queue_set_napi(nv->napi.dev, qt->sub0.q_idx, - NETDEV_QUEUE_TYPE_TX, NULL); } /* Flush any processed Rx Queue Triads and drop the rest */ @@ -1820,10 +1881,6 @@ void fbnic_flush(struct fbnic_net *fbn) fbnic_put_pkt_buff(nv, qt->cmpl.pkt, 0); qt->cmpl.pkt->buff.data_hard_start = NULL; - - /* Disassociate Rx queue from NAPI */ - netif_queue_set_napi(nv->napi.dev, qt->cmpl.q_idx, - NETDEV_QUEUE_TYPE_RX, NULL); } } } @@ -1836,29 +1893,12 @@ void fbnic_fill(struct fbnic_net *fbn) struct fbnic_napi_vector *nv = fbn->napi[i]; int j, t; - /* Configure NAPI mapping for Tx */ - for (t = 0; t < nv->txt_count; t++) { - struct fbnic_q_triad *qt = &nv->qt[t]; - - /* Nothing to do if Tx queue is disabled */ - if (qt->sub0.flags & FBNIC_RING_F_DISABLED) - continue; - - /* Associate Tx queue with NAPI */ - netif_queue_set_napi(nv->napi.dev, qt->sub0.q_idx, - NETDEV_QUEUE_TYPE_TX, &nv->napi); - } - /* Configure NAPI mapping and populate pages * in the BDQ rings to use for Rx */ - for (j = 0; j < nv->rxt_count; j++, t++) { + for (j = 0, t = nv->txt_count; j < nv->rxt_count; j++, t++) { struct fbnic_q_triad *qt = &nv->qt[t]; - /* Associate Rx queue with NAPI */ - netif_queue_set_napi(nv->napi.dev, qt->cmpl.q_idx, - NETDEV_QUEUE_TYPE_RX, &nv->napi); - /* Populate the header and payload BDQs */ fbnic_fill_bdq(nv, &qt->sub0); fbnic_fill_bdq(nv, &qt->sub1); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h index c8d908860ab0..92c671135ad7 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h @@ -124,6 +124,8 @@ int fbnic_alloc_napi_vectors(struct fbnic_net *fbn); void fbnic_free_napi_vectors(struct fbnic_net *fbn); int fbnic_alloc_resources(struct fbnic_net *fbn); void fbnic_free_resources(struct fbnic_net *fbn); +int fbnic_set_netif_queues(struct fbnic_net *fbn); +void fbnic_reset_netif_queues(struct fbnic_net *fbn); irqreturn_t fbnic_msix_clean_rings(int irq, void *data); void fbnic_napi_enable(struct fbnic_net *fbn); void fbnic_napi_disable(struct fbnic_net *fbn); -- 2.51.0 From 3a481cc72673b2fbb18271acf2d9b43f6a920ec4 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 19 Dec 2024 18:52:40 -0800 Subject: [PATCH 07/16] eth: fbnic: support ring channel get and set while down Trivial implementation of ethtool channel get and set. Set is only supported when device is closed, next patch will add code for live reconfig. Asymmetric configurations are supported (combined + extra Tx or Rx), so are configurations with independent IRQs for Rx and Tx. Having all 3 NAPI types (combined, Tx, Rx) is not supported. We used to only call fbnic_reset_indir_tbl() during init. Now that we call it after device had been register must be careful not to override user config. Link: https://patch.msgid.link/20241220025241.1522781-10-kuba@kernel.org Signed-off-by: Jakub Kicinski --- .../net/ethernet/meta/fbnic/fbnic_ethtool.c | 64 +++++++++++++++++++ drivers/net/ethernet/meta/fbnic/fbnic_rpc.c | 3 + 2 files changed, 67 insertions(+) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c b/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c index d1be8fc30404..d2fe97ae6a71 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c @@ -304,6 +304,68 @@ fbnic_set_rxfh(struct net_device *netdev, struct ethtool_rxfh_param *rxfh, return 0; } +static void fbnic_get_channels(struct net_device *netdev, + struct ethtool_channels *ch) +{ + struct fbnic_net *fbn = netdev_priv(netdev); + struct fbnic_dev *fbd = fbn->fbd; + + ch->max_rx = fbd->max_num_queues; + ch->max_tx = fbd->max_num_queues; + ch->max_combined = min(ch->max_rx, ch->max_tx); + ch->max_other = FBNIC_NON_NAPI_VECTORS; + + if (fbn->num_rx_queues > fbn->num_napi || + fbn->num_tx_queues > fbn->num_napi) + ch->combined_count = min(fbn->num_rx_queues, + fbn->num_tx_queues); + else + ch->combined_count = + fbn->num_rx_queues + fbn->num_tx_queues - fbn->num_napi; + ch->rx_count = fbn->num_rx_queues - ch->combined_count; + ch->tx_count = fbn->num_tx_queues - ch->combined_count; + ch->other_count = FBNIC_NON_NAPI_VECTORS; +} + +static void fbnic_set_queues(struct fbnic_net *fbn, struct ethtool_channels *ch, + unsigned int max_napis) +{ + fbn->num_rx_queues = ch->rx_count + ch->combined_count; + fbn->num_tx_queues = ch->tx_count + ch->combined_count; + fbn->num_napi = min(ch->rx_count + ch->tx_count + ch->combined_count, + max_napis); +} + +static int fbnic_set_channels(struct net_device *netdev, + struct ethtool_channels *ch) +{ + struct fbnic_net *fbn = netdev_priv(netdev); + unsigned int max_napis, standalone; + struct fbnic_dev *fbd = fbn->fbd; + + max_napis = fbd->num_irqs - FBNIC_NON_NAPI_VECTORS; + standalone = ch->rx_count + ch->tx_count; + + /* Limits for standalone queues: + * - each queue has it's own NAPI (num_napi >= rx + tx + combined) + * - combining queues (combined not 0, rx or tx must be 0) + */ + if ((ch->rx_count && ch->tx_count && ch->combined_count) || + (standalone && standalone + ch->combined_count > max_napis) || + ch->rx_count + ch->combined_count > fbd->max_num_queues || + ch->tx_count + ch->combined_count > fbd->max_num_queues || + ch->other_count != FBNIC_NON_NAPI_VECTORS) + return -EINVAL; + + if (!netif_running(netdev)) { + fbnic_set_queues(fbn, ch, max_napis); + fbnic_reset_indir_tbl(fbn); + return 0; + } + + return -EBUSY; +} + static int fbnic_get_ts_info(struct net_device *netdev, struct kernel_ethtool_ts_info *tsinfo) @@ -417,6 +479,8 @@ static const struct ethtool_ops fbnic_ethtool_ops = { .get_rxfh_indir_size = fbnic_get_rxfh_indir_size, .get_rxfh = fbnic_get_rxfh, .set_rxfh = fbnic_set_rxfh, + .get_channels = fbnic_get_channels, + .set_channels = fbnic_set_channels, .get_ts_info = fbnic_get_ts_info, .get_ts_stats = fbnic_get_ts_stats, .get_eth_mac_stats = fbnic_get_eth_mac_stats, diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_rpc.c b/drivers/net/ethernet/meta/fbnic/fbnic_rpc.c index b99c890ac43f..c25bd300b902 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_rpc.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_rpc.c @@ -13,6 +13,9 @@ void fbnic_reset_indir_tbl(struct fbnic_net *fbn) unsigned int num_rx = fbn->num_rx_queues; unsigned int i; + if (netif_is_rxfh_configured(fbn->netdev)) + return; + for (i = 0; i < FBNIC_RPC_RSS_TBL_SIZE; i++) fbn->indir_tbl[0][i] = ethtool_rxfh_indir_default(i, num_rx); } -- 2.51.0 From 52dc722db0d98bcdf40927dd1719468f7d08bd59 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 19 Dec 2024 18:52:41 -0800 Subject: [PATCH 08/16] eth: fbnic: support ring channel set while up Implement the channel count changes. Copy the netdev priv, allocate new channels using it. Stop, swap, start. Then free the copy of the priv along with the channels it holds, which are now the channels that used to be on the real priv. Reviewed-by: Przemek Kitszel Link: https://patch.msgid.link/20241220025241.1522781-11-kuba@kernel.org Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/meta/fbnic/fbnic.h | 1 + .../net/ethernet/meta/fbnic/fbnic_ethtool.c | 121 +++++++++++++++++- drivers/net/ethernet/meta/fbnic/fbnic_irq.c | 11 ++ .../net/ethernet/meta/fbnic/fbnic_netdev.h | 1 + drivers/net/ethernet/meta/fbnic/fbnic_pci.c | 2 +- drivers/net/ethernet/meta/fbnic/fbnic_txrx.c | 8 +- drivers/net/ethernet/meta/fbnic/fbnic_txrx.h | 5 + 7 files changed, 143 insertions(+), 6 deletions(-) diff --git a/drivers/net/ethernet/meta/fbnic/fbnic.h b/drivers/net/ethernet/meta/fbnic/fbnic.h index ed527209b30c..14751f16e125 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic.h @@ -162,6 +162,7 @@ int fbnic_napi_request_irq(struct fbnic_dev *fbd, struct fbnic_napi_vector *nv); void fbnic_napi_free_irq(struct fbnic_dev *fbd, struct fbnic_napi_vector *nv); +void fbnic_synchronize_irq(struct fbnic_dev *fbd, int nr); int fbnic_request_irq(struct fbnic_dev *dev, int nr, irq_handler_t handler, unsigned long flags, const char *name, void *data); void fbnic_free_irq(struct fbnic_dev *dev, int nr, void *data); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c b/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c index d2fe97ae6a71..20cd9f5f89e2 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_ethtool.c @@ -65,6 +65,76 @@ static void fbnic_get_regs(struct net_device *netdev, fbnic_csr_get_regs(fbn->fbd, data, ®s->version); } +static struct fbnic_net *fbnic_clone_create(struct fbnic_net *orig) +{ + struct fbnic_net *clone; + + clone = kmemdup(orig, sizeof(*orig), GFP_KERNEL); + if (!clone) + return NULL; + + memset(clone->tx, 0, sizeof(clone->tx)); + memset(clone->rx, 0, sizeof(clone->rx)); + memset(clone->napi, 0, sizeof(clone->napi)); + return clone; +} + +static void fbnic_clone_swap_cfg(struct fbnic_net *orig, + struct fbnic_net *clone) +{ + swap(clone->rcq_size, orig->rcq_size); + swap(clone->hpq_size, orig->hpq_size); + swap(clone->ppq_size, orig->ppq_size); + swap(clone->txq_size, orig->txq_size); + swap(clone->num_rx_queues, orig->num_rx_queues); + swap(clone->num_tx_queues, orig->num_tx_queues); + swap(clone->num_napi, orig->num_napi); +} + +static void fbnic_aggregate_vector_counters(struct fbnic_net *fbn, + struct fbnic_napi_vector *nv) +{ + int i, j; + + for (i = 0; i < nv->txt_count; i++) { + fbnic_aggregate_ring_tx_counters(fbn, &nv->qt[i].sub0); + fbnic_aggregate_ring_tx_counters(fbn, &nv->qt[i].sub1); + fbnic_aggregate_ring_tx_counters(fbn, &nv->qt[i].cmpl); + } + + for (j = 0; j < nv->rxt_count; j++, i++) { + fbnic_aggregate_ring_rx_counters(fbn, &nv->qt[i].sub0); + fbnic_aggregate_ring_rx_counters(fbn, &nv->qt[i].sub1); + fbnic_aggregate_ring_rx_counters(fbn, &nv->qt[i].cmpl); + } +} + +static void fbnic_clone_swap(struct fbnic_net *orig, + struct fbnic_net *clone) +{ + struct fbnic_dev *fbd = orig->fbd; + unsigned int i; + + for (i = 0; i < max(clone->num_napi, orig->num_napi); i++) + fbnic_synchronize_irq(fbd, FBNIC_NON_NAPI_VECTORS + i); + for (i = 0; i < orig->num_napi; i++) + fbnic_aggregate_vector_counters(orig, orig->napi[i]); + + fbnic_clone_swap_cfg(orig, clone); + + for (i = 0; i < ARRAY_SIZE(orig->napi); i++) + swap(clone->napi[i], orig->napi[i]); + for (i = 0; i < ARRAY_SIZE(orig->tx); i++) + swap(clone->tx[i], orig->tx[i]); + for (i = 0; i < ARRAY_SIZE(orig->rx); i++) + swap(clone->rx[i], orig->rx[i]); +} + +static void fbnic_clone_free(struct fbnic_net *clone) +{ + kfree(clone); +} + static void fbnic_get_strings(struct net_device *dev, u32 sset, u8 *data) { int i; @@ -342,6 +412,8 @@ static int fbnic_set_channels(struct net_device *netdev, struct fbnic_net *fbn = netdev_priv(netdev); unsigned int max_napis, standalone; struct fbnic_dev *fbd = fbn->fbd; + struct fbnic_net *clone; + int err; max_napis = fbd->num_irqs - FBNIC_NON_NAPI_VECTORS; standalone = ch->rx_count + ch->tx_count; @@ -363,7 +435,54 @@ static int fbnic_set_channels(struct net_device *netdev, return 0; } - return -EBUSY; + clone = fbnic_clone_create(fbn); + if (!clone) + return -ENOMEM; + + fbnic_set_queues(clone, ch, max_napis); + + err = fbnic_alloc_napi_vectors(clone); + if (err) + goto err_free_clone; + + err = fbnic_alloc_resources(clone); + if (err) + goto err_free_napis; + + fbnic_down_noidle(fbn); + err = fbnic_wait_all_queues_idle(fbn->fbd, true); + if (err) + goto err_start_stack; + + err = fbnic_set_netif_queues(clone); + if (err) + goto err_start_stack; + + /* Nothing can fail past this point */ + fbnic_flush(fbn); + + fbnic_clone_swap(fbn, clone); + + /* Reset RSS indirection table */ + fbnic_reset_indir_tbl(fbn); + + fbnic_up(fbn); + + fbnic_free_resources(clone); + fbnic_free_napi_vectors(clone); + fbnic_clone_free(clone); + + return 0; + +err_start_stack: + fbnic_flush(fbn); + fbnic_up(fbn); + fbnic_free_resources(clone); +err_free_napis: + fbnic_free_napi_vectors(clone); +err_free_clone: + fbnic_clone_free(clone); + return err; } static int diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_irq.c b/drivers/net/ethernet/meta/fbnic/fbnic_irq.c index a8ea7b6774a8..1bbc0e56f3a0 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_irq.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_irq.c @@ -146,6 +146,17 @@ void fbnic_pcs_irq_disable(struct fbnic_dev *fbd) free_irq(fbd->pcs_msix_vector, fbd); } +void fbnic_synchronize_irq(struct fbnic_dev *fbd, int nr) +{ + struct pci_dev *pdev = to_pci_dev(fbd->dev); + int irq = pci_irq_vector(pdev, nr); + + if (irq < 0) + return; + + synchronize_irq(irq); +} + int fbnic_request_irq(struct fbnic_dev *fbd, int nr, irq_handler_t handler, unsigned long flags, const char *name, void *data) { diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.h b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.h index 0986c8f120a8..a392ac1cc4f2 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_netdev.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic_netdev.h @@ -65,6 +65,7 @@ struct fbnic_net { int __fbnic_open(struct fbnic_net *fbn); void fbnic_up(struct fbnic_net *fbn); void fbnic_down(struct fbnic_net *fbn); +void fbnic_down_noidle(struct fbnic_net *fbn); struct net_device *fbnic_netdev_alloc(struct fbnic_dev *fbd); void fbnic_netdev_free(struct fbnic_dev *fbd); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_pci.c b/drivers/net/ethernet/meta/fbnic/fbnic_pci.c index 32702dc4a066..6cbbc2ee3e1f 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_pci.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_pci.c @@ -145,7 +145,7 @@ void fbnic_up(struct fbnic_net *fbn) fbnic_service_task_start(fbn); } -static void fbnic_down_noidle(struct fbnic_net *fbn) +void fbnic_down_noidle(struct fbnic_net *fbn) { fbnic_service_task_stop(fbn); diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c index 92fc1ad6ed6f..bb54ce5f5787 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c +++ b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.c @@ -1045,8 +1045,8 @@ irqreturn_t fbnic_msix_clean_rings(int __always_unused irq, void *data) return IRQ_HANDLED; } -static void fbnic_aggregate_ring_rx_counters(struct fbnic_net *fbn, - struct fbnic_ring *rxr) +void fbnic_aggregate_ring_rx_counters(struct fbnic_net *fbn, + struct fbnic_ring *rxr) { struct fbnic_queue_stats *stats = &rxr->stats; @@ -1056,8 +1056,8 @@ static void fbnic_aggregate_ring_rx_counters(struct fbnic_net *fbn, fbn->rx_stats.dropped += stats->dropped; } -static void fbnic_aggregate_ring_tx_counters(struct fbnic_net *fbn, - struct fbnic_ring *txr) +void fbnic_aggregate_ring_tx_counters(struct fbnic_net *fbn, + struct fbnic_ring *txr) { struct fbnic_queue_stats *stats = &txr->stats; diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h index 92c671135ad7..c2a94f31f71b 100644 --- a/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h +++ b/drivers/net/ethernet/meta/fbnic/fbnic_txrx.h @@ -120,6 +120,11 @@ netdev_features_t fbnic_features_check(struct sk_buff *skb, struct net_device *dev, netdev_features_t features); +void fbnic_aggregate_ring_rx_counters(struct fbnic_net *fbn, + struct fbnic_ring *rxr); +void fbnic_aggregate_ring_tx_counters(struct fbnic_net *fbn, + struct fbnic_ring *txr); + int fbnic_alloc_napi_vectors(struct fbnic_net *fbn); void fbnic_free_napi_vectors(struct fbnic_net *fbn); int fbnic_alloc_resources(struct fbnic_net *fbn); -- 2.51.0 From 4c61d809cf608842112c77880f50810a564cd9cb Mon Sep 17 00:00:00 2001 From: Kory Maincent Date: Fri, 20 Dec 2024 09:37:40 +0100 Subject: [PATCH 09/16] net: ethtool: Fix suspicious rcu_dereference usage The __ethtool_get_ts_info function can be called with or without the rtnl lock held. When the rtnl lock is not held, using rtnl_dereference() triggers a warning due to the lack of lock context. Add an rcu_read_lock() to ensure the lock is acquired and to maintain synchronization. Reported-by: syzbot+a344326c05c98ba19682@syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/676147f8.050a0220.37aaf.0154.GAE@google.com/ Fixes: b9e3f7dc9ed9 ("net: ethtool: tsinfo: Enhance tsinfo to support several hwtstamp by net topology") Signed-off-by: Kory Maincent Reviewed-by: Eric Dumazet Link: https://patch.msgid.link/20241220083741.175329-1-kory.maincent@bootlin.com Signed-off-by: Jakub Kicinski --- net/ethtool/common.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/net/ethtool/common.c b/net/ethtool/common.c index 02f941f667dd..2607aea1fbfb 100644 --- a/net/ethtool/common.c +++ b/net/ethtool/common.c @@ -870,7 +870,8 @@ int __ethtool_get_ts_info(struct net_device *dev, { struct hwtstamp_provider *hwprov; - hwprov = rtnl_dereference(dev->hwprov); + rcu_read_lock(); + hwprov = rcu_dereference(dev->hwprov); /* No provider specified, use default behavior */ if (!hwprov) { const struct ethtool_ops *ops = dev->ethtool_ops; @@ -887,9 +888,11 @@ int __ethtool_get_ts_info(struct net_device *dev, info->so_timestamping |= SOF_TIMESTAMPING_RX_SOFTWARE | SOF_TIMESTAMPING_SOFTWARE; + rcu_read_unlock(); return err; } + rcu_read_unlock(); return ethtool_get_ts_info_by_phc(dev, info, &hwprov->desc); } -- 2.51.0 From c71b59690aa12daf3edbb4dd02b8821490dc727e Mon Sep 17 00:00:00 2001 From: Daniel Machon Date: Fri, 20 Dec 2024 14:48:40 +0100 Subject: [PATCH 10/16] net: sparx5: do some preparation work The sparx5_port_init() does initial configuration of a variety of different features and options for each port. Some are shared for all types of devices, some are not. As it is now, common configuration is done after configuration of low-speed devices. This will not work when adding RGMII support in a subsequent patch. In preparation for lan969x RGMII support, move a block of code, that configures 2g5 devices, down. This ensures that the configuration common to all devices is done before configuration of 2g5, 5g, 10g and 25g devices. Reviewed-by: Steen Hegelund Reviewed-by: Horatiu Vultur Tested-by: Robert Marko Signed-off-by: Daniel Machon Link: https://patch.msgid.link/20241220-sparx5-lan969x-switch-driver-4-v5-1-fa8ba5dff732@microchip.com Signed-off-by: Jakub Kicinski --- .../ethernet/microchip/sparx5/sparx5_port.c | 36 +++++++++---------- 1 file changed, 18 insertions(+), 18 deletions(-) diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_port.c b/drivers/net/ethernet/microchip/sparx5/sparx5_port.c index f9d1a6bb9bff..f39bf4878e11 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_port.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_port.c @@ -1067,24 +1067,6 @@ int sparx5_port_init(struct sparx5 *sparx5, if (err) return err; - /* Configure MAC vlan awareness */ - err = sparx5_port_max_tags_set(sparx5, port); - if (err) - return err; - - /* Set Max Length */ - spx5_rmw(DEV2G5_MAC_MAXLEN_CFG_MAX_LEN_SET(ETH_MAXLEN), - DEV2G5_MAC_MAXLEN_CFG_MAX_LEN, - sparx5, - DEV2G5_MAC_MAXLEN_CFG(port->portno)); - - /* 1G/2G5: Signal Detect configuration */ - spx5_wr(DEV2G5_PCS1G_SD_CFG_SD_POL_SET(sd_pol) | - DEV2G5_PCS1G_SD_CFG_SD_SEL_SET(sd_sel) | - DEV2G5_PCS1G_SD_CFG_SD_ENA_SET(sd_ena), - sparx5, - DEV2G5_PCS1G_SD_CFG(port->portno)); - /* Set Pause WM hysteresis */ spx5_rmw(QSYS_PAUSE_CFG_PAUSE_START_SET(pause_start) | QSYS_PAUSE_CFG_PAUSE_STOP_SET(pause_stop) | @@ -1108,6 +1090,24 @@ int sparx5_port_init(struct sparx5 *sparx5, ANA_CL_FILTER_CTRL_FILTER_SMAC_MC_DIS, sparx5, ANA_CL_FILTER_CTRL(port->portno)); + /* Configure MAC vlan awareness */ + err = sparx5_port_max_tags_set(sparx5, port); + if (err) + return err; + + /* Set Max Length */ + spx5_rmw(DEV2G5_MAC_MAXLEN_CFG_MAX_LEN_SET(ETH_MAXLEN), + DEV2G5_MAC_MAXLEN_CFG_MAX_LEN, + sparx5, + DEV2G5_MAC_MAXLEN_CFG(port->portno)); + + /* 1G/2G5: Signal Detect configuration */ + spx5_wr(DEV2G5_PCS1G_SD_CFG_SD_POL_SET(sd_pol) | + DEV2G5_PCS1G_SD_CFG_SD_SEL_SET(sd_sel) | + DEV2G5_PCS1G_SD_CFG_SD_ENA_SET(sd_ena), + sparx5, + DEV2G5_PCS1G_SD_CFG(port->portno)); + if (conf->portmode == PHY_INTERFACE_MODE_QSGMII || conf->portmode == PHY_INTERFACE_MODE_SGMII) { err = sparx5_serdes_set(sparx5, port, conf); -- 2.51.0 From dd2baee1084034b8666290cbcc02cb32fe5a8666 Mon Sep 17 00:00:00 2001 From: Daniel Machon Date: Fri, 20 Dec 2024 14:48:41 +0100 Subject: [PATCH 11/16] net: sparx5: add function for RGMII port check The lan969x device contains two RGMII port interfaces, sitting at port 28 and 29. Add function: is_port_rgmii() to the match data ops, that checks if a given port is an RGMII port or not. For Sparx5, this function always returns false. Reviewed-by: Steen Hegelund Reviewed-by: Horatiu Vultur Tested-by: Robert Marko Signed-off-by: Daniel Machon Link: https://patch.msgid.link/20241220-sparx5-lan969x-switch-driver-4-v5-2-fa8ba5dff732@microchip.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.c | 1 + drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.h | 5 +++++ drivers/net/ethernet/microchip/sparx5/sparx5_main.c | 1 + drivers/net/ethernet/microchip/sparx5/sparx5_main.h | 1 + drivers/net/ethernet/microchip/sparx5/sparx5_port.h | 5 +++++ 5 files changed, 13 insertions(+) diff --git a/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.c b/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.c index c2afa2176b08..76f0c8635eb9 100644 --- a/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.c +++ b/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.c @@ -329,6 +329,7 @@ static const struct sparx5_ops lan969x_ops = { .is_port_5g = &lan969x_port_is_5g, .is_port_10g = &lan969x_port_is_10g, .is_port_25g = &lan969x_port_is_25g, + .is_port_rgmii = &lan969x_port_is_rgmii, .get_port_dev_index = &lan969x_port_dev_mapping, .get_port_dev_bit = &lan969x_get_dev_mode_bit, .get_hsch_max_group_rate = &lan969x_get_hsch_max_group_rate, diff --git a/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.h b/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.h index 2489d0d32dfd..4b91c47d6d21 100644 --- a/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.h +++ b/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.h @@ -59,6 +59,11 @@ static inline bool lan969x_port_is_25g(int portno) return false; } +static inline bool lan969x_port_is_rgmii(int portno) +{ + return portno == 28 || portno == 29; +} + /* lan969x_calendar.c */ int lan969x_dsm_calendar_calc(struct sparx5 *sparx5, u32 taxi, struct sparx5_calendar_data *data); diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_main.c b/drivers/net/ethernet/microchip/sparx5/sparx5_main.c index f61aa15beab7..4be717ba7d37 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_main.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_main.c @@ -1072,6 +1072,7 @@ static const struct sparx5_ops sparx5_ops = { .is_port_5g = &sparx5_port_is_5g, .is_port_10g = &sparx5_port_is_10g, .is_port_25g = &sparx5_port_is_25g, + .is_port_rgmii = &sparx5_port_is_rgmii, .get_port_dev_index = &sparx5_port_dev_mapping, .get_port_dev_bit = &sparx5_port_dev_mapping, .get_hsch_max_group_rate = &sparx5_get_hsch_max_group_rate, diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_main.h b/drivers/net/ethernet/microchip/sparx5/sparx5_main.h index d5dd953b0a71..c58d7841638e 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_main.h +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_main.h @@ -313,6 +313,7 @@ struct sparx5_ops { bool (*is_port_5g)(int portno); bool (*is_port_10g)(int portno); bool (*is_port_25g)(int portno); + bool (*is_port_rgmii)(int portno); u32 (*get_port_dev_index)(struct sparx5 *sparx5, int port); u32 (*get_port_dev_bit)(struct sparx5 *sparx5, int port); u32 (*get_hsch_max_group_rate)(int grp); diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_port.h b/drivers/net/ethernet/microchip/sparx5/sparx5_port.h index 9b9bcc6834bc..c8a37468a3d1 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_port.h +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_port.h @@ -40,6 +40,11 @@ static inline bool sparx5_port_is_25g(int portno) return portno >= 56 && portno <= 63; } +static inline bool sparx5_port_is_rgmii(int portno) +{ + return false; +} + static inline u32 sparx5_to_high_dev(struct sparx5 *sparx5, int port) { const struct sparx5_ops *ops = sparx5->data->ops; -- 2.51.0 From 05bda8a1bdedd5dfc21522d732c3bf9413d70eb3 Mon Sep 17 00:00:00 2001 From: Daniel Machon Date: Fri, 20 Dec 2024 14:48:42 +0100 Subject: [PATCH 12/16] net: sparx5: use is_port_rgmii() throughout Now that we can check if a given port is an RGMII port, use it in the following cases: - To set RGMII PHY modes for RGMII port devices. - To avoid checking for a SerDes node in the devicetree, when the port is an RGMII port. - To bail out of sparx5_port_init() when the common configuration is done. Reviewed-by: Steen Hegelund Reviewed-by: Horatiu Vultur Tested-by: Robert Marko Signed-off-by: Daniel Machon Link: https://patch.msgid.link/20241220-sparx5-lan969x-switch-driver-4-v5-3-fa8ba5dff732@microchip.com Signed-off-by: Jakub Kicinski --- .../ethernet/microchip/sparx5/sparx5_main.c | 28 +++++++++++++------ .../ethernet/microchip/sparx5/sparx5_port.c | 3 ++ 2 files changed, 23 insertions(+), 8 deletions(-) diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_main.c b/drivers/net/ethernet/microchip/sparx5/sparx5_main.c index 4be717ba7d37..e68277c38adc 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_main.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_main.c @@ -313,10 +313,13 @@ static int sparx5_create_port(struct sparx5 *sparx5, struct initial_port_config *config) { struct sparx5_port *spx5_port; + const struct sparx5_ops *ops; struct net_device *ndev; struct phylink *phylink; int err; + ops = sparx5->data->ops; + ndev = sparx5_create_netdev(sparx5, config->portno); if (IS_ERR(ndev)) { dev_err(sparx5->dev, "Could not create net device: %02u\n", @@ -357,6 +360,9 @@ static int sparx5_create_port(struct sparx5 *sparx5, MAC_SYM_PAUSE | MAC_10 | MAC_100 | MAC_1000FD | MAC_2500FD | MAC_5000FD | MAC_10000FD | MAC_25000FD; + if (ops->is_port_rgmii(spx5_port->portno)) + phy_interface_set_rgmii(spx5_port->phylink_config.supported_interfaces); + __set_bit(PHY_INTERFACE_MODE_SGMII, spx5_port->phylink_config.supported_interfaces); __set_bit(PHY_INTERFACE_MODE_QSGMII, @@ -830,6 +836,7 @@ static int mchp_sparx5_probe(struct platform_device *pdev) struct initial_port_config *configs, *config; struct device_node *np = pdev->dev.of_node; struct device_node *ports, *portnp; + const struct sparx5_ops *ops; struct reset_control *reset; struct sparx5 *sparx5; int idx = 0, err = 0; @@ -851,6 +858,7 @@ static int mchp_sparx5_probe(struct platform_device *pdev) return -EINVAL; regs = sparx5->data->regs; + ops = sparx5->data->ops; /* Do switch core reset if available */ reset = devm_reset_control_get_optional_shared(&pdev->dev, "switch"); @@ -880,7 +888,7 @@ static int mchp_sparx5_probe(struct platform_device *pdev) for_each_available_child_of_node(ports, portnp) { struct sparx5_port_config *conf; - struct phy *serdes; + struct phy *serdes = NULL; u32 portno; err = of_property_read_u32(portnp, "reg", &portno); @@ -910,13 +918,17 @@ static int mchp_sparx5_probe(struct platform_device *pdev) conf->sd_sgpio = ~0; else sparx5->sd_sgpio_remapping = true; - serdes = devm_of_phy_get(sparx5->dev, portnp, NULL); - if (IS_ERR(serdes)) { - err = dev_err_probe(sparx5->dev, PTR_ERR(serdes), - "port %u: missing serdes\n", - portno); - of_node_put(portnp); - goto cleanup_config; + /* There is no SerDes node for RGMII ports. */ + if (!ops->is_port_rgmii(portno)) { + serdes = devm_of_phy_get(sparx5->dev, portnp, NULL); + if (IS_ERR(serdes)) { + err = dev_err_probe(sparx5->dev, + PTR_ERR(serdes), + "port %u: missing serdes\n", + portno); + of_node_put(portnp); + goto cleanup_config; + } } config->portno = portno; config->node = portnp; diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_port.c b/drivers/net/ethernet/microchip/sparx5/sparx5_port.c index f39bf4878e11..996dc4343019 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_port.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_port.c @@ -1090,6 +1090,9 @@ int sparx5_port_init(struct sparx5 *sparx5, ANA_CL_FILTER_CTRL_FILTER_SMAC_MC_DIS, sparx5, ANA_CL_FILTER_CTRL(port->portno)); + if (ops->is_port_rgmii(port->portno)) + return 0; /* RGMII device - nothing more to configure */ + /* Configure MAC vlan awareness */ err = sparx5_port_max_tags_set(sparx5, port); if (err) -- 2.51.0 From d9450934f915a97b09f035866acd5da302f0dc12 Mon Sep 17 00:00:00 2001 From: Daniel Machon Date: Fri, 20 Dec 2024 14:48:43 +0100 Subject: [PATCH 13/16] net: sparx5: skip low-speed configuration when port is RGMII When doing a port config, we configure low-speed port devices, among other things. We have a check to ensure, that the device is indeed a low-speed device, an not a high-speed device. Add an additional check, to ensure that the device is not an RGMII device. Reviewed-by: Steen Hegelund Reviewed-by: Horatiu Vultur Tested-by: Robert Marko Signed-off-by: Daniel Machon Link: https://patch.msgid.link/20241220-sparx5-lan969x-switch-driver-4-v5-4-fa8ba5dff732@microchip.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/microchip/sparx5/sparx5_port.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_port.c b/drivers/net/ethernet/microchip/sparx5/sparx5_port.c index 996dc4343019..0a1374422ccb 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_port.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_port.c @@ -994,6 +994,7 @@ int sparx5_port_config(struct sparx5 *sparx5, struct sparx5_port *port, struct sparx5_port_config *conf) { + bool rgmii = phy_interface_mode_is_rgmii(conf->phy_mode); bool high_speed_dev = sparx5_is_baser(conf->portmode); const struct sparx5_ops *ops = sparx5->data->ops; int err, urgency, stop_wm; @@ -1003,7 +1004,7 @@ int sparx5_port_config(struct sparx5 *sparx5, return err; /* high speed device is already configured */ - if (!high_speed_dev) + if (!rgmii && !high_speed_dev) sparx5_port_config_low_set(sparx5, port, conf); /* Configure flow control */ -- 2.51.0 From 9b8d70ecfef7abcabe265be4faeb07e552383520 Mon Sep 17 00:00:00 2001 From: Daniel Machon Date: Fri, 20 Dec 2024 14:48:44 +0100 Subject: [PATCH 14/16] net: sparx5: only return PCS for modes that require it The RGMII ports have no PCS to configure. Make sure we only return the PCS for port modes that require it. Reviewed-by: Russell King (Oracle) Tested-by: Robert Marko Signed-off-by: Daniel Machon Link: https://patch.msgid.link/20241220-sparx5-lan969x-switch-driver-4-v5-5-fa8ba5dff732@microchip.com Signed-off-by: Jakub Kicinski --- .../net/ethernet/microchip/sparx5/sparx5_phylink.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_phylink.c b/drivers/net/ethernet/microchip/sparx5/sparx5_phylink.c index f8562c1a894d..035d2f1bea0d 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_phylink.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_phylink.c @@ -32,7 +32,19 @@ sparx5_phylink_mac_select_pcs(struct phylink_config *config, { struct sparx5_port *port = netdev_priv(to_net_dev(config->dev)); - return &port->phylink_pcs; + /* Return the PCS for all the modes that require it. */ + switch (interface) { + case PHY_INTERFACE_MODE_SGMII: + case PHY_INTERFACE_MODE_QSGMII: + case PHY_INTERFACE_MODE_1000BASEX: + case PHY_INTERFACE_MODE_2500BASEX: + case PHY_INTERFACE_MODE_5GBASER: + case PHY_INTERFACE_MODE_10GBASER: + case PHY_INTERFACE_MODE_25GBASER: + return &port->phylink_pcs; + default: + return NULL; + } } static void sparx5_phylink_mac_config(struct phylink_config *config, -- 2.51.0 From 95e467b85e6930d34093b7770c7ed964113589b0 Mon Sep 17 00:00:00 2001 From: Daniel Machon Date: Fri, 20 Dec 2024 14:48:45 +0100 Subject: [PATCH 15/16] net: sparx5: verify RGMII speeds When doing a port config, we verify the port speed against the PHY mode and supported speeds of that PHY mode. Add checks for the four RGMII phy modes: RGMII, RGMII_ID, RGMII_TXID and RGMII_RXID. Reviewed-by: Steen Hegelund Reviewed-by: Horatiu Vultur Tested-by: Robert Marko Signed-off-by: Daniel Machon Link: https://patch.msgid.link/20241220-sparx5-lan969x-switch-driver-4-v5-6-fa8ba5dff732@microchip.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/microchip/sparx5/sparx5_port.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_port.c b/drivers/net/ethernet/microchip/sparx5/sparx5_port.c index 0a1374422ccb..86d6c9e9ec7c 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_port.c +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_port.c @@ -257,6 +257,15 @@ static int sparx5_port_verify_speed(struct sparx5 *sparx5, conf->speed != SPEED_25000)) return sparx5_port_error(port, conf, SPX5_PERR_SPEED); break; + case PHY_INTERFACE_MODE_RGMII: + case PHY_INTERFACE_MODE_RGMII_ID: + case PHY_INTERFACE_MODE_RGMII_TXID: + case PHY_INTERFACE_MODE_RGMII_RXID: + if (conf->speed != SPEED_1000 && + conf->speed != SPEED_100 && + conf->speed != SPEED_10) + return sparx5_port_error(port, conf, SPX5_PERR_SPEED); + break; default: return sparx5_port_error(port, conf, SPX5_PERR_IFTYPE); } -- 2.51.0 From fb6ac1829bb5865768e75517aefb416a3a19569e Mon Sep 17 00:00:00 2001 From: Daniel Machon Date: Fri, 20 Dec 2024 14:48:46 +0100 Subject: [PATCH 16/16] net: lan969x: add RGMII registers Configuration of RGMII is done by configuring the GPIO and clock settings in the HSIOWRAP target, and configuring the RGMII port devices in the DEVRGMII target. Both targets contain registers replicated for the number of RGMII port devices, which is two. Add said targets and register macros required to configure RGMII. Reviewed-by: Steen Hegelund Reviewed-by: Horatiu Vultur Tested-by: Robert Marko Signed-off-by: Daniel Machon Link: https://patch.msgid.link/20241220-sparx5-lan969x-switch-driver-4-v5-7-fa8ba5dff732@microchip.com Signed-off-by: Jakub Kicinski --- .../microchip/sparx5/lan969x/lan969x.c | 3 + .../microchip/sparx5/sparx5_main_regs.h | 145 ++++++++++++++++++ 2 files changed, 148 insertions(+) diff --git a/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.c b/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.c index 76f0c8635eb9..be49a99556fe 100644 --- a/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.c +++ b/drivers/net/ethernet/microchip/sparx5/lan969x/lan969x.c @@ -90,9 +90,12 @@ static const struct sparx5_main_io_resource lan969x_main_iomap[] = { { TARGET_DEV2G5 + 27, 0x30d8000, 1 }, /* 0xe30d8000 */ { TARGET_DEV10G + 9, 0x30dc000, 1 }, /* 0xe30dc000 */ { TARGET_PCS10G_BR + 9, 0x30e0000, 1 }, /* 0xe30e0000 */ + { TARGET_DEVRGMII, 0x30e4000, 1 }, /* 0xe30e4000 */ + { TARGET_DEVRGMII + 1, 0x30e8000, 1 }, /* 0xe30e8000 */ { TARGET_DSM, 0x30ec000, 1 }, /* 0xe30ec000 */ { TARGET_PORT_CONF, 0x30f0000, 1 }, /* 0xe30f0000 */ { TARGET_ASM, 0x3200000, 1 }, /* 0xe3200000 */ + { TARGET_HSIO_WRAP, 0x3408000, 1 }, /* 0xe3408000 */ }; static struct sparx5_sdlb_group lan969x_sdlb_groups[LAN969X_SDLB_GRP_CNT] = { diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_main_regs.h b/drivers/net/ethernet/microchip/sparx5/sparx5_main_regs.h index 561344f19062..d9ef4ef137b8 100644 --- a/drivers/net/ethernet/microchip/sparx5/sparx5_main_regs.h +++ b/drivers/net/ethernet/microchip/sparx5/sparx5_main_regs.h @@ -37,6 +37,7 @@ enum sparx5_target { TARGET_FDMA = 117, TARGET_GCB = 118, TARGET_HSCH = 119, + TARGET_HSIO_WRAP = 120, TARGET_LRN = 122, TARGET_PCEP = 129, TARGET_PCS10G_BR = 132, @@ -54,6 +55,7 @@ enum sparx5_target { TARGET_VCAP_SUPER = 326, TARGET_VOP = 327, TARGET_XQS = 331, + TARGET_DEVRGMII = 392, NUM_TARGETS = 517 }; @@ -5367,6 +5369,69 @@ extern const struct sparx5_regs *regs; #define HSCH_TAS_STATEMACHINE_CFG_REVISIT_DLY_GET(x)\ FIELD_GET(HSCH_TAS_STATEMACHINE_CFG_REVISIT_DLY, x) +/* LAN969X ONLY */ +/* HSIOWRAP:XMII_CFG:XMII_CFG */ +#define HSIO_WRAP_XMII_CFG(g) \ + __REG(TARGET_HSIO_WRAP, 0, 1, 116, g, 2, 20, 0, 0, 1, 4) + +#define HSIO_WRAP_XMII_CFG_GPIO_XMII_CFG GENMASK(2, 1) +#define HSIO_WRAP_XMII_CFG_GPIO_XMII_CFG_SET(x)\ + FIELD_PREP(HSIO_WRAP_XMII_CFG_GPIO_XMII_CFG, x) +#define HSIO_WRAP_XMII_CFG_GPIO_XMII_CFG_GET(x)\ + FIELD_GET(HSIO_WRAP_XMII_CFG_GPIO_XMII_CFG, x) + +/* LAN969X ONLY */ +/* HSIOWRAP:XMII_CFG:RGMII_CFG */ +#define HSIO_WRAP_RGMII_CFG(g) \ + __REG(TARGET_HSIO_WRAP, 0, 1, 116, g, 2, 20, 4, 0, 1, 4) + +#define HSIO_WRAP_RGMII_CFG_TX_CLK_CFG GENMASK(4, 2) +#define HSIO_WRAP_RGMII_CFG_TX_CLK_CFG_SET(x)\ + FIELD_PREP(HSIO_WRAP_RGMII_CFG_TX_CLK_CFG, x) +#define HSIO_WRAP_RGMII_CFG_TX_CLK_CFG_GET(x)\ + FIELD_GET(HSIO_WRAP_RGMII_CFG_TX_CLK_CFG, x) + +#define HSIO_WRAP_RGMII_CFG_RGMII_TX_RST BIT(1) +#define HSIO_WRAP_RGMII_CFG_RGMII_TX_RST_SET(x)\ + FIELD_PREP(HSIO_WRAP_RGMII_CFG_RGMII_TX_RST, x) +#define HSIO_WRAP_RGMII_CFG_RGMII_TX_RST_GET(x)\ + FIELD_GET(HSIO_WRAP_RGMII_CFG_RGMII_TX_RST, x) + +#define HSIO_WRAP_RGMII_CFG_RGMII_RX_RST BIT(0) +#define HSIO_WRAP_RGMII_CFG_RGMII_RX_RST_SET(x)\ + FIELD_PREP(HSIO_WRAP_RGMII_CFG_RGMII_RX_RST, x) +#define HSIO_WRAP_RGMII_CFG_RGMII_RX_RST_GET(x)\ + FIELD_GET(HSIO_WRAP_RGMII_CFG_RGMII_RX_RST, x) + +/* LAN969X ONLY */ +/* HSIOWRAP:XMII_CFG:DLL_CFG */ +#define HSIO_WRAP_DLL_CFG(g, r) \ + __REG(TARGET_HSIO_WRAP, 0, 1, 116, g, 2, 20, 12, r, 2, 4) + +#define HSIO_WRAP_DLL_CFG_DLL_ENA BIT(19) +#define HSIO_WRAP_DLL_CFG_DLL_ENA_SET(x)\ + FIELD_PREP(HSIO_WRAP_DLL_CFG_DLL_ENA, x) +#define HSIO_WRAP_DLL_CFG_DLL_ENA_GET(x)\ + FIELD_GET(HSIO_WRAP_DLL_CFG_DLL_ENA, x) + +#define HSIO_WRAP_DLL_CFG_DLL_CLK_ENA BIT(18) +#define HSIO_WRAP_DLL_CFG_DLL_CLK_ENA_SET(x)\ + FIELD_PREP(HSIO_WRAP_DLL_CFG_DLL_CLK_ENA, x) +#define HSIO_WRAP_DLL_CFG_DLL_CLK_ENA_GET(x)\ + FIELD_GET(HSIO_WRAP_DLL_CFG_DLL_CLK_ENA, x) + +#define HSIO_WRAP_DLL_CFG_DLL_CLK_SEL GENMASK(17, 15) +#define HSIO_WRAP_DLL_CFG_DLL_CLK_SEL_SET(x)\ + FIELD_PREP(HSIO_WRAP_DLL_CFG_DLL_CLK_SEL, x) +#define HSIO_WRAP_DLL_CFG_DLL_CLK_SEL_GET(x)\ + FIELD_GET(HSIO_WRAP_DLL_CFG_DLL_CLK_SEL, x) + +#define HSIO_WRAP_DLL_CFG_DLL_RST BIT(0) +#define HSIO_WRAP_DLL_CFG_DLL_RST_SET(x)\ + FIELD_PREP(HSIO_WRAP_DLL_CFG_DLL_RST, x) +#define HSIO_WRAP_DLL_CFG_DLL_RST_GET(x)\ + FIELD_GET(HSIO_WRAP_DLL_CFG_DLL_RST, x) + /* LRN:COMMON:COMMON_ACCESS_CTRL */ #define LRN_COMMON_ACCESS_CTRL \ __REG(TARGET_LRN, 0, 1, 0, 0, 1, 72, 0, 0, 1, 4) @@ -8110,4 +8175,84 @@ extern const struct sparx5_regs *regs; #define XQS_CNT(g) \ __REG(TARGET_XQS, 0, 1, 0, g, 1024, 4, 0, 0, 1, 4) +/* LAN969X ONLY */ +/* DEV1G:DEV_CFG_STATUS:DEV_RST_CTRL */ +#define DEVRGMII_DEV_RST_CTRL(t) \ + __REG(TARGET_DEVRGMII, t, 2, 0, 0, 1, 36, 0, 0, 1, 4) + +#define DEVRGMII_DEV_RST_CTRL_SPEED_SEL GENMASK(22, 20) +#define DEVRGMII_DEV_RST_CTRL_SPEED_SEL_SET(x)\ + FIELD_PREP(DEVRGMII_DEV_RST_CTRL_SPEED_SEL, x) +#define DEVRGMII_DEV_RST_CTRL_SPEED_SEL_GET(x)\ + FIELD_GET(DEVRGMII_DEV_RST_CTRL_SPEED_SEL, x) + +/* LAN969X ONLY */ +/* DEV1G:MAC_CFG_STATUS:MAC_ENA_CFG */ +#define DEVRGMII_MAC_ENA_CFG(t) \ + __REG(TARGET_DEVRGMII, t, 2, 36, 0, 1, 36, 0, 0, 1, 4) + +#define DEVRGMII_MAC_ENA_CFG_RX_ENA BIT(4) +#define DEVRGMII_MAC_ENA_CFG_RX_ENA_SET(x)\ + FIELD_PREP(DEVRGMII_MAC_ENA_CFG_RX_ENA, x) +#define DEVRGMII_MAC_ENA_CFG_RX_ENA_GET(x)\ + FIELD_GET(DEVRGMII_MAC_ENA_CFG_RX_ENA, x) + +#define DEVRGMII_MAC_ENA_CFG_TX_ENA BIT(0) +#define DEVRGMII_MAC_ENA_CFG_TX_ENA_SET(x)\ + FIELD_PREP(DEVRGMII_MAC_ENA_CFG_TX_ENA, x) +#define DEVRGMII_MAC_ENA_CFG_TX_ENA_GET(x)\ + FIELD_GET(DEVRGMII_MAC_ENA_CFG_TX_ENA, x) + +/* LAN969X ONLY */ +/* DEV1G:MAC_CFG_STATUS:MAC_TAGS_CFG */ +#define DEVRGMII_MAC_TAGS_CFG(t) \ + __REG(TARGET_DEVRGMII, t, 2, 36, 0, 1, 36, 12, 0, 1, 4) + +#define DEVRGMII_MAC_TAGS_CFG_TAG_ID GENMASK(31, 16) +#define DEVRGMII_MAC_TAGS_CFG_TAG_ID_SET(x)\ + FIELD_PREP(DEVRGMII_MAC_TAGS_CFG_TAG_ID, x) +#define DEVRGMII_MAC_TAGS_CFG_TAG_ID_GET(x)\ + FIELD_GET(DEVRGMII_MAC_TAGS_CFG_TAG_ID, x) + +#define DEVRGMII_MAC_TAGS_CFG_VLAN_LEN_AWR_ENA BIT(3) +#define DEVRGMII_MAC_TAGS_CFG_VLAN_LEN_AWR_ENA_SET(x)\ + FIELD_PREP(DEVRGMII_MAC_TAGS_CFG_VLAN_LEN_AWR_ENA, x) +#define DEVRGMII_MAC_TAGS_CFG_VLAN_LEN_AWR_ENA_GET(x)\ + FIELD_GET(DEVRGMII_MAC_TAGS_CFG_VLAN_LEN_AWR_ENA, x) + +#define DEVRGMII_MAC_TAGS_CFG_PB_ENA GENMASK(2, 1) +#define DEVRGMII_MAC_TAGS_CFG_PB_ENA_SET(x)\ + FIELD_PREP(DEVRGMII_MAC_TAGS_CFG_PB_ENA, x) +#define DEVRGMII_MAC_TAGS_CFG_PB_ENA_GET(x)\ + FIELD_GET(DEVRGMII_MAC_TAGS_CFG_PB_ENA, x) + +#define DEVRGMII_MAC_TAGS_CFG_VLAN_AWR_ENA BIT(0) +#define DEVRGMII_MAC_TAGS_CFG_VLAN_AWR_ENA_SET(x)\ + FIELD_PREP(DEVRGMII_MAC_TAGS_CFG_VLAN_AWR_ENA, x) +#define DEVRGMII_MAC_TAGS_CFG_VLAN_AWR_ENA_GET(x)\ + FIELD_GET(DEVRGMII_MAC_TAGS_CFG_VLAN_AWR_ENA, x) + +/* LAN969X ONLY */ +/* DEV1G:MAC_CFG_STATUS:MAC_IFG_CFG */ +#define DEVRGMII_MAC_IFG_CFG(t) \ + __REG(TARGET_DEVRGMII, t, 2, 36, 0, 1, 36, 24, 0, 1, 4) + +#define DEVRGMII_MAC_IFG_CFG_TX_IFG GENMASK(12, 8) +#define DEVRGMII_MAC_IFG_CFG_TX_IFG_SET(x)\ + FIELD_PREP(DEVRGMII_MAC_IFG_CFG_TX_IFG, x) +#define DEVRGMII_MAC_IFG_CFG_TX_IFG_GET(x)\ + FIELD_GET(DEVRGMII_MAC_IFG_CFG_TX_IFG, x) + +#define DEVRGMII_MAC_IFG_CFG_RX_IFG2 GENMASK(7, 4) +#define DEVRGMII_MAC_IFG_CFG_RX_IFG2_SET(x)\ + FIELD_PREP(DEVRGMII_MAC_IFG_CFG_RX_IFG2, x) +#define DEVRGMII_MAC_IFG_CFG_RX_IFG2_GET(x)\ + FIELD_GET(DEVRGMII_MAC_IFG_CFG_RX_IFG2, x) + +#define DEVRGMII_MAC_IFG_CFG_RX_IFG1 GENMASK(3, 0) +#define DEVRGMII_MAC_IFG_CFG_RX_IFG1_SET(x)\ + FIELD_PREP(DEVRGMII_MAC_IFG_CFG_RX_IFG1, x) +#define DEVRGMII_MAC_IFG_CFG_RX_IFG1_GET(x)\ + FIELD_GET(DEVRGMII_MAC_IFG_CFG_RX_IFG1, x) + #endif /* _SPARX5_MAIN_REGS_H_ */ -- 2.51.0