From 744d197162c2070a6045a71e2666ed93a57cc65d Mon Sep 17 00:00:00 2001 From: Larysa Zaremba Date: Mon, 3 Jun 2024 14:42:33 -0700 Subject: ice: add flag to distinguish reset from .ndo_bpf in XDP rings config Commit 6624e780a577 ("ice: split ice_vsi_setup into smaller functions") has placed ice_vsi_free_q_vectors() after ice_destroy_xdp_rings() in the rebuild process. The behaviour of the XDP rings config functions is context-dependent, so the change of order has led to ice_destroy_xdp_rings() doing additional work and removing XDP prog, when it was supposed to be preserved. Also, dependency on the PF state reset flags creates an additional, fortunately less common problem: * PFR is requested e.g. by tx_timeout handler * .ndo_bpf() is asked to delete the program, calls ice_destroy_xdp_rings(), but reset flag is set, so rings are destroyed without deleting the program * ice_vsi_rebuild tries to delete non-existent XDP rings, because the program is still on the VSI * system crashes With a similar race, when requested to attach a program, ice_prepare_xdp_rings() can actually skip setting the program in the VSI and nevertheless report success. Instead of reverting to the old order of function calls, add an enum argument to both ice_prepare_xdp_rings() and ice_destroy_xdp_rings() in order to distinguish between calls from rebuild and .ndo_bpf(). Fixes: efc2214b6047 ("ice: Add support for XDP") Reviewed-by: Igor Bagnucki Signed-off-by: Larysa Zaremba Reviewed-by: Simon Horman Tested-by: Chandan Kumar Rout Signed-off-by: Jacob Keller Link: https://lore.kernel.org/r/20240603-net-2024-05-30-intel-net-fixes-v2-4-e3563aa89b0c@intel.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/intel/ice/ice_main.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) (limited to 'drivers/net/ethernet/intel/ice/ice_main.c') diff --git a/drivers/net/ethernet/intel/ice/ice_main.c b/drivers/net/ethernet/intel/ice/ice_main.c index f60c022f7960..2a270aacd24a 100644 --- a/drivers/net/ethernet/intel/ice/ice_main.c +++ b/drivers/net/ethernet/intel/ice/ice_main.c @@ -2711,10 +2711,12 @@ static void ice_vsi_assign_bpf_prog(struct ice_vsi *vsi, struct bpf_prog *prog) * ice_prepare_xdp_rings - Allocate, configure and setup Tx rings for XDP * @vsi: VSI to bring up Tx rings used by XDP * @prog: bpf program that will be assigned to VSI + * @cfg_type: create from scratch or restore the existing configuration * * Return 0 on success and negative value on error */ -int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog) +int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog, + enum ice_xdp_cfg cfg_type) { u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 }; int xdp_rings_rem = vsi->num_xdp_txq; @@ -2790,7 +2792,7 @@ int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog) * taken into account at the end of ice_vsi_rebuild, where * ice_cfg_vsi_lan is being called */ - if (ice_is_reset_in_progress(pf->state)) + if (cfg_type == ICE_XDP_CFG_PART) return 0; /* tell the Tx scheduler that right now we have @@ -2842,22 +2844,21 @@ err_map_xdp: /** * ice_destroy_xdp_rings - undo the configuration made by ice_prepare_xdp_rings * @vsi: VSI to remove XDP rings + * @cfg_type: disable XDP permanently or allow it to be restored later * * Detach XDP rings from irq vectors, clean up the PF bitmap and free * resources */ -int ice_destroy_xdp_rings(struct ice_vsi *vsi) +int ice_destroy_xdp_rings(struct ice_vsi *vsi, enum ice_xdp_cfg cfg_type) { u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 }; struct ice_pf *pf = vsi->back; int i, v_idx; /* q_vectors are freed in reset path so there's no point in detaching - * rings; in case of rebuild being triggered not from reset bits - * in pf->state won't be set, so additionally check first q_vector - * against NULL + * rings */ - if (ice_is_reset_in_progress(pf->state) || !vsi->q_vectors[0]) + if (cfg_type == ICE_XDP_CFG_PART) goto free_qmap; ice_for_each_q_vector(vsi, v_idx) { @@ -2898,7 +2899,7 @@ free_qmap: if (static_key_enabled(&ice_xdp_locking_key)) static_branch_dec(&ice_xdp_locking_key); - if (ice_is_reset_in_progress(pf->state) || !vsi->q_vectors[0]) + if (cfg_type == ICE_XDP_CFG_PART) return 0; ice_vsi_assign_bpf_prog(vsi, NULL); @@ -3009,7 +3010,8 @@ ice_xdp_setup_prog(struct ice_vsi *vsi, struct bpf_prog *prog, if (xdp_ring_err) { NL_SET_ERR_MSG_MOD(extack, "Not enough Tx resources for XDP"); } else { - xdp_ring_err = ice_prepare_xdp_rings(vsi, prog); + xdp_ring_err = ice_prepare_xdp_rings(vsi, prog, + ICE_XDP_CFG_FULL); if (xdp_ring_err) NL_SET_ERR_MSG_MOD(extack, "Setting up XDP Tx resources failed"); } @@ -3020,7 +3022,7 @@ ice_xdp_setup_prog(struct ice_vsi *vsi, struct bpf_prog *prog, NL_SET_ERR_MSG_MOD(extack, "Setting up XDP Rx resources failed"); } else if (ice_is_xdp_ena_vsi(vsi) && !prog) { xdp_features_clear_redirect_target(vsi->netdev); - xdp_ring_err = ice_destroy_xdp_rings(vsi); + xdp_ring_err = ice_destroy_xdp_rings(vsi, ICE_XDP_CFG_FULL); if (xdp_ring_err) NL_SET_ERR_MSG_MOD(extack, "Freeing XDP Tx resources failed"); /* reallocate Rx queues that were used for zero-copy */ -- cgit From f3df4044254c98128890b512bf19cc05588f1fe5 Mon Sep 17 00:00:00 2001 From: Larysa Zaremba Date: Mon, 3 Jun 2024 14:42:34 -0700 Subject: ice: map XDP queues to vectors in ice_vsi_map_rings_to_vectors() ice_pf_dcb_recfg() re-maps queues to vectors with ice_vsi_map_rings_to_vectors(), which does not restore the previous state for XDP queues. This leads to no AF_XDP traffic after rebuild. Map XDP queues to vectors in ice_vsi_map_rings_to_vectors(). Also, move the code around, so XDP queues are mapped independently only through .ndo_bpf(). Fixes: 6624e780a577 ("ice: split ice_vsi_setup into smaller functions") Reviewed-by: Przemek Kitszel Signed-off-by: Larysa Zaremba Reviewed-by: Simon Horman Tested-by: Chandan Kumar Rout Signed-off-by: Jacob Keller Link: https://lore.kernel.org/r/20240603-net-2024-05-30-intel-net-fixes-v2-5-e3563aa89b0c@intel.com Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/intel/ice/ice_main.c | 96 ++++++++++++++++++------------- 1 file changed, 57 insertions(+), 39 deletions(-) (limited to 'drivers/net/ethernet/intel/ice/ice_main.c') diff --git a/drivers/net/ethernet/intel/ice/ice_main.c b/drivers/net/ethernet/intel/ice/ice_main.c index 2a270aacd24a..1b61ca3a6eb6 100644 --- a/drivers/net/ethernet/intel/ice/ice_main.c +++ b/drivers/net/ethernet/intel/ice/ice_main.c @@ -2707,6 +2707,60 @@ static void ice_vsi_assign_bpf_prog(struct ice_vsi *vsi, struct bpf_prog *prog) bpf_prog_put(old_prog); } +static struct ice_tx_ring *ice_xdp_ring_from_qid(struct ice_vsi *vsi, int qid) +{ + struct ice_q_vector *q_vector; + struct ice_tx_ring *ring; + + if (static_key_enabled(&ice_xdp_locking_key)) + return vsi->xdp_rings[qid % vsi->num_xdp_txq]; + + q_vector = vsi->rx_rings[qid]->q_vector; + ice_for_each_tx_ring(ring, q_vector->tx) + if (ice_ring_is_xdp(ring)) + return ring; + + return NULL; +} + +/** + * ice_map_xdp_rings - Map XDP rings to interrupt vectors + * @vsi: the VSI with XDP rings being configured + * + * Map XDP rings to interrupt vectors and perform the configuration steps + * dependent on the mapping. + */ +void ice_map_xdp_rings(struct ice_vsi *vsi) +{ + int xdp_rings_rem = vsi->num_xdp_txq; + int v_idx, q_idx; + + /* follow the logic from ice_vsi_map_rings_to_vectors */ + ice_for_each_q_vector(vsi, v_idx) { + struct ice_q_vector *q_vector = vsi->q_vectors[v_idx]; + int xdp_rings_per_v, q_id, q_base; + + xdp_rings_per_v = DIV_ROUND_UP(xdp_rings_rem, + vsi->num_q_vectors - v_idx); + q_base = vsi->num_xdp_txq - xdp_rings_rem; + + for (q_id = q_base; q_id < (q_base + xdp_rings_per_v); q_id++) { + struct ice_tx_ring *xdp_ring = vsi->xdp_rings[q_id]; + + xdp_ring->q_vector = q_vector; + xdp_ring->next = q_vector->tx.tx_ring; + q_vector->tx.tx_ring = xdp_ring; + } + xdp_rings_rem -= xdp_rings_per_v; + } + + ice_for_each_rxq(vsi, q_idx) { + vsi->rx_rings[q_idx]->xdp_ring = ice_xdp_ring_from_qid(vsi, + q_idx); + ice_tx_xsk_pool(vsi, q_idx); + } +} + /** * ice_prepare_xdp_rings - Allocate, configure and setup Tx rings for XDP * @vsi: VSI to bring up Tx rings used by XDP @@ -2719,7 +2773,6 @@ int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog, enum ice_xdp_cfg cfg_type) { u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 }; - int xdp_rings_rem = vsi->num_xdp_txq; struct ice_pf *pf = vsi->back; struct ice_qs_cfg xdp_qs_cfg = { .qs_mutex = &pf->avail_q_mutex, @@ -2732,8 +2785,7 @@ int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog, .mapping_mode = ICE_VSI_MAP_CONTIG }; struct device *dev; - int i, v_idx; - int status; + int status, i; dev = ice_pf_to_dev(pf); vsi->xdp_rings = devm_kcalloc(dev, vsi->num_xdp_txq, @@ -2752,42 +2804,6 @@ int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog, if (ice_xdp_alloc_setup_rings(vsi)) goto clear_xdp_rings; - /* follow the logic from ice_vsi_map_rings_to_vectors */ - ice_for_each_q_vector(vsi, v_idx) { - struct ice_q_vector *q_vector = vsi->q_vectors[v_idx]; - int xdp_rings_per_v, q_id, q_base; - - xdp_rings_per_v = DIV_ROUND_UP(xdp_rings_rem, - vsi->num_q_vectors - v_idx); - q_base = vsi->num_xdp_txq - xdp_rings_rem; - - for (q_id = q_base; q_id < (q_base + xdp_rings_per_v); q_id++) { - struct ice_tx_ring *xdp_ring = vsi->xdp_rings[q_id]; - - xdp_ring->q_vector = q_vector; - xdp_ring->next = q_vector->tx.tx_ring; - q_vector->tx.tx_ring = xdp_ring; - } - xdp_rings_rem -= xdp_rings_per_v; - } - - ice_for_each_rxq(vsi, i) { - if (static_key_enabled(&ice_xdp_locking_key)) { - vsi->rx_rings[i]->xdp_ring = vsi->xdp_rings[i % vsi->num_xdp_txq]; - } else { - struct ice_q_vector *q_vector = vsi->rx_rings[i]->q_vector; - struct ice_tx_ring *ring; - - ice_for_each_tx_ring(ring, q_vector->tx) { - if (ice_ring_is_xdp(ring)) { - vsi->rx_rings[i]->xdp_ring = ring; - break; - } - } - } - ice_tx_xsk_pool(vsi, i); - } - /* omit the scheduler update if in reset path; XDP queues will be * taken into account at the end of ice_vsi_rebuild, where * ice_cfg_vsi_lan is being called @@ -2795,6 +2811,8 @@ int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog, if (cfg_type == ICE_XDP_CFG_PART) return 0; + ice_map_xdp_rings(vsi); + /* tell the Tx scheduler that right now we have * additional queues */ -- cgit From bc69ad74867dba1377abe14356c94a946d9837a3 Mon Sep 17 00:00:00 2001 From: En-Wei Wu Date: Thu, 30 May 2024 22:21:31 +0800 Subject: ice: avoid IRQ collision to fix init failure on ACPI S3 resume A bug in https://bugzilla.kernel.org/show_bug.cgi?id=218906 describes that irdma would break and report hardware initialization failed after suspend/resume with Intel E810 NIC (tested on 6.9.0-rc5). The problem is caused due to the collision between the irq numbers requested in irdma and the irq numbers requested in other drivers after suspend/resume. The irq numbers used by irdma are derived from ice's ice_pf->msix_entries which stores mappings between MSI-X index and Linux interrupt number. It's supposed to be cleaned up when suspend and rebuilt in resume but it's not, causing irdma using the old irq numbers stored in the old ice_pf->msix_entries to request_irq() when resume. And eventually collide with other drivers. This patch fixes this problem. On suspend, we call ice_deinit_rdma() to clean up the ice_pf->msix_entries (and free the MSI-X vectors used by irdma if we've dynamically allocated them). On resume, we call ice_init_rdma() to rebuild the ice_pf->msix_entries (and allocate the MSI-X vectors if we would like to dynamically allocate them). Fixes: f9f5301e7e2d ("ice: Register auxiliary device to provide RDMA") Tested-by: Cyrus Lien Signed-off-by: En-Wei Wu Reviewed-by: Wojciech Drewek Tested-by: Pucha Himasekhar Reddy (A Contingent worker at Intel) Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_main.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'drivers/net/ethernet/intel/ice/ice_main.c') diff --git a/drivers/net/ethernet/intel/ice/ice_main.c b/drivers/net/ethernet/intel/ice/ice_main.c index 1b61ca3a6eb6..45d850514f4c 100644 --- a/drivers/net/ethernet/intel/ice/ice_main.c +++ b/drivers/net/ethernet/intel/ice/ice_main.c @@ -5564,7 +5564,7 @@ static int ice_suspend(struct device *dev) */ disabled = ice_service_task_stop(pf); - ice_unplug_aux_dev(pf); + ice_deinit_rdma(pf); /* Already suspended?, then there is nothing to do */ if (test_and_set_bit(ICE_SUSPENDED, pf->state)) { @@ -5644,6 +5644,11 @@ static int ice_resume(struct device *dev) if (ret) dev_err(dev, "Cannot restore interrupt scheme: %d\n", ret); + ret = ice_init_rdma(pf); + if (ret) + dev_err(dev, "Reinitialize RDMA during resume failed: %d\n", + ret); + clear_bit(ICE_DOWN, pf->state); /* Now perform PF reset and rebuild */ reset_type = ICE_RESET_PFR; -- cgit From aeccadb24d9dacdde673a0f68f0a9135c6be4993 Mon Sep 17 00:00:00 2001 From: Paul Greenwalt Date: Thu, 30 May 2024 13:06:17 -0400 Subject: ice: fix 200G link speed message log Commit 24407a01e57c ("ice: Add 200G speed/phy type use") added support for 200G PHY speeds, but did not include 200G link speed message support. As a result the driver incorrectly reports Unknown for 200G link speed. Fix this by adding 200G support to ice_print_link_msg(). Fixes: 24407a01e57c ("ice: Add 200G speed/phy type use") Reviewed-by: Michal Swiatkowski Signed-off-by: Paul Greenwalt Reviewed-by: Jesse Brandeburg Tested-by: Pucha Himasekhar Reddy (A Contingent worker at Intel) Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_main.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'drivers/net/ethernet/intel/ice/ice_main.c') diff --git a/drivers/net/ethernet/intel/ice/ice_main.c b/drivers/net/ethernet/intel/ice/ice_main.c index 45d850514f4c..1766230abfff 100644 --- a/drivers/net/ethernet/intel/ice/ice_main.c +++ b/drivers/net/ethernet/intel/ice/ice_main.c @@ -805,6 +805,9 @@ void ice_print_link_msg(struct ice_vsi *vsi, bool isup) } switch (vsi->port_info->phy.link_info.link_speed) { + case ICE_AQ_LINK_SPEED_200GB: + speed = "200 G"; + break; case ICE_AQ_LINK_SPEED_100GB: speed = "100 G"; break; -- cgit From f4b91c1d17c676b8ad4c6bd674da874f3f7d5701 Mon Sep 17 00:00:00 2001 From: Jan Sokolowski Date: Fri, 21 Jun 2024 10:54:19 -0700 Subject: ice: Rebuild TC queues on VSI queue reconfiguration TC queues needs to be correctly updated when the number of queues on a VSI is reconfigured, so netdev's queue and TC settings will be dynamically adjusted and could accurately represent the underlying hardware state after changes to the VSI queue counts. Fixes: 0754d65bd4be ("ice: Add infrastructure for mqprio support via ndo_setup_tc") Reviewed-by: Wojciech Drewek Signed-off-by: Jan Sokolowski Signed-off-by: Karen Ostrowska Tested-by: Pucha Himasekhar Reddy (A Contingent worker at Intel) Signed-off-by: Tony Nguyen Signed-off-by: David S. Miller --- drivers/net/ethernet/intel/ice/ice_main.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'drivers/net/ethernet/intel/ice/ice_main.c') diff --git a/drivers/net/ethernet/intel/ice/ice_main.c b/drivers/net/ethernet/intel/ice/ice_main.c index 1766230abfff..55a42aad92a5 100644 --- a/drivers/net/ethernet/intel/ice/ice_main.c +++ b/drivers/net/ethernet/intel/ice/ice_main.c @@ -4139,7 +4139,7 @@ bool ice_is_wol_supported(struct ice_hw *hw) int ice_vsi_recfg_qs(struct ice_vsi *vsi, int new_rx, int new_tx, bool locked) { struct ice_pf *pf = vsi->back; - int err = 0, timeout = 50; + int i, err = 0, timeout = 50; if (!new_rx && !new_tx) return -EINVAL; @@ -4165,6 +4165,14 @@ int ice_vsi_recfg_qs(struct ice_vsi *vsi, int new_rx, int new_tx, bool locked) ice_vsi_close(vsi); ice_vsi_rebuild(vsi, ICE_VSI_FLAG_NO_INIT); + + ice_for_each_traffic_class(i) { + if (vsi->tc_cfg.ena_tc & BIT(i)) + netdev_set_tc_queue(vsi->netdev, + vsi->tc_cfg.tc_info[i].netdev_tc, + vsi->tc_cfg.tc_info[i].qcount_tx, + vsi->tc_cfg.tc_info[i].qoffset); + } ice_pf_dcb_recfg(pf, locked); ice_vsi_open(vsi); done: -- cgit