From d4ea6f6373ef56d1d795a24f1f5874f4a6019199 Mon Sep 17 00:00:00 2001 From: Marcin Szycik Date: Mon, 9 May 2022 21:01:18 +0200 Subject: ice: ignore protocol field in GTP offload Commit 34a897758efe ("ice: Add support for inner etype in switchdev") added the ability to match on inner ethertype. A side effect of that change is that it is now impossible to add some filters for protocols which do not contain inner ethtype field. tc requires the protocol field to be specified when providing certain other options, e.g. src_ip. This is a problem in case of GTP - when user wants to specify e.g. src_ip, they also need to specify protocol in tc command (otherwise tc fails with: Illegal "src_ip"). Because GTP is a tunnel, the protocol field is treated as inner protocol. GTP does not contain inner ethtype field and the filter cannot be added. To fix this, ignore the ethertype field in case of GTP filters. Fixes: 9a225f81f540 ("ice: Support GTP-U and GTP-C offload in switchdev") Signed-off-by: Marcin Szycik Tested-by: Sandeep Penigalapati Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_tc_lib.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/intel/ice/ice_tc_lib.c b/drivers/net/ethernet/intel/ice/ice_tc_lib.c index 0a0c55fb8699..b677c62a244c 100644 --- a/drivers/net/ethernet/intel/ice/ice_tc_lib.c +++ b/drivers/net/ethernet/intel/ice/ice_tc_lib.c @@ -993,7 +993,9 @@ ice_parse_cls_flower(struct net_device *filter_dev, struct ice_vsi *vsi, n_proto_key = ntohs(match.key->n_proto); n_proto_mask = ntohs(match.mask->n_proto); - if (n_proto_key == ETH_P_ALL || n_proto_key == 0) { + if (n_proto_key == ETH_P_ALL || n_proto_key == 0 || + fltr->tunnel_type == TNL_GTPU || + fltr->tunnel_type == TNL_GTPC) { n_proto_key = 0; n_proto_mask = 0; } else { -- cgit v1.2.3 From 3578dc90013b1fa20da996cdadd8515802716132 Mon Sep 17 00:00:00 2001 From: Wojciech Drewek Date: Mon, 23 May 2022 11:41:55 +0200 Subject: ice: Fix switchdev rules book keeping Adding two filters with same matching criteria ends up with one rule in hardware with act = ICE_FWD_TO_VSI_LIST. In order to remove them properly we have to keep the information about vsi handle which is used in VSI bitmap (ice_adv_fltr_mgmt_list_entry::vsi_list_info::vsi_map). Fixes: 0d08a441fb1a ("ice: ndo_setup_tc implementation for PF") Reported-by: Sridhar Samudrala Signed-off-by: Wojciech Drewek Tested-by: Sandeep Penigalapati Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_tc_lib.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/net/ethernet/intel/ice/ice_tc_lib.c b/drivers/net/ethernet/intel/ice/ice_tc_lib.c index b677c62a244c..b803f2ab3cc7 100644 --- a/drivers/net/ethernet/intel/ice/ice_tc_lib.c +++ b/drivers/net/ethernet/intel/ice/ice_tc_lib.c @@ -524,6 +524,7 @@ ice_eswitch_add_tc_fltr(struct ice_vsi *vsi, struct ice_tc_flower_fltr *fltr) */ fltr->rid = rule_added.rid; fltr->rule_id = rule_added.rule_id; + fltr->dest_id = rule_added.vsi_handle; exit: kfree(list); -- cgit v1.2.3 From c3d184c83ff4b80167e34edfc3d21df424bf27ff Mon Sep 17 00:00:00 2001 From: Anatolii Gerasymenko Date: Mon, 20 Jun 2022 09:47:05 +0200 Subject: ice: ethtool: advertise 1000M speeds properly In current implementation ice_update_phy_type enables all link modes for selected speed. This approach doesn't work for 1000M speeds, because both copper (1000baseT) and optical (1000baseX) standards cannot be enabled at once. Fix this, by adding the function `ice_set_phy_type_from_speed()` for 1000M speeds. Fixes: 48cb27f2fd18 ("ice: Implement handlers for ethtool PHY/link operations") Signed-off-by: Anatolii Gerasymenko Tested-by: Gurucharan (A Contingent worker at Intel) Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_ethtool.c | 39 +++++++++++++++++++++++++++- 1 file changed, 38 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/intel/ice/ice_ethtool.c b/drivers/net/ethernet/intel/ice/ice_ethtool.c index 1e71b70f0e52..8078618ce1b6 100644 --- a/drivers/net/ethernet/intel/ice/ice_ethtool.c +++ b/drivers/net/ethernet/intel/ice/ice_ethtool.c @@ -2189,6 +2189,42 @@ ice_setup_autoneg(struct ice_port_info *p, struct ethtool_link_ksettings *ks, return err; } +/** + * ice_set_phy_type_from_speed - set phy_types based on speeds + * and advertised modes + * @ks: ethtool link ksettings struct + * @phy_type_low: pointer to the lower part of phy_type + * @phy_type_high: pointer to the higher part of phy_type + * @adv_link_speed: targeted link speeds bitmap + */ +static void +ice_set_phy_type_from_speed(const struct ethtool_link_ksettings *ks, + u64 *phy_type_low, u64 *phy_type_high, + u16 adv_link_speed) +{ + /* Handle 1000M speed in a special way because ice_update_phy_type + * enables all link modes, but having mixed copper and optical + * standards is not supported. + */ + adv_link_speed &= ~ICE_AQ_LINK_SPEED_1000MB; + + if (ethtool_link_ksettings_test_link_mode(ks, advertising, + 1000baseT_Full)) + *phy_type_low |= ICE_PHY_TYPE_LOW_1000BASE_T | + ICE_PHY_TYPE_LOW_1G_SGMII; + + if (ethtool_link_ksettings_test_link_mode(ks, advertising, + 1000baseKX_Full)) + *phy_type_low |= ICE_PHY_TYPE_LOW_1000BASE_KX; + + if (ethtool_link_ksettings_test_link_mode(ks, advertising, + 1000baseX_Full)) + *phy_type_low |= ICE_PHY_TYPE_LOW_1000BASE_SX | + ICE_PHY_TYPE_LOW_1000BASE_LX; + + ice_update_phy_type(phy_type_low, phy_type_high, adv_link_speed); +} + /** * ice_set_link_ksettings - Set Speed and Duplex * @netdev: network interface device structure @@ -2320,7 +2356,8 @@ ice_set_link_ksettings(struct net_device *netdev, adv_link_speed = curr_link_speed; /* Convert the advertise link speeds to their corresponded PHY_TYPE */ - ice_update_phy_type(&phy_type_low, &phy_type_high, adv_link_speed); + ice_set_phy_type_from_speed(ks, &phy_type_low, &phy_type_high, + adv_link_speed); if (!autoneg_changed && adv_link_speed == curr_link_speed) { netdev_info(netdev, "Nothing changed, exiting without setting anything.\n"); -- cgit v1.2.3 From a632b2a4c920ce5af29410fb091f7ee6d2e77dc6 Mon Sep 17 00:00:00 2001 From: Anatolii Gerasymenko Date: Mon, 6 Jun 2022 09:01:21 +0200 Subject: ice: ethtool: Prohibit improper channel config for DCB Do not allow setting less channels, than Traffic Classes there are via ethtool. There must be at least one channel per Traffic Class. If you set less channels, than Traffic Classes there are, then during ice_vsi_rebuild there would be allocated only the requested amount of tx/rx rings in ice_vsi_alloc_arrays. But later in ice_vsi_setup_q_map there would be requested at least one channel per Traffic Class. This results in setting num_rxq > alloc_rxq and num_txq > alloc_txq. Later, there would be a NULL pointer dereference in ice_vsi_map_rings_to_vectors, because we go beyond of rx_rings or tx_rings arrays. Change ice_set_channels() to return error if you try to allocate less channels, than Traffic Classes there are. Change ice_vsi_setup_q_map() and ice_vsi_setup_q_map_mqprio() to return status code instead of void. Add error handling for ice_vsi_setup_q_map() and ice_vsi_setup_q_map_mqprio() in ice_vsi_init() and ice_vsi_cfg_tc(). [53753.889983] INFO: Flow control is disabled for this traffic class (0) on this vsi. [53763.984862] BUG: unable to handle kernel NULL pointer dereference at 0000000000000028 [53763.992915] PGD 14b45f5067 P4D 0 [53763.996444] Oops: 0002 [#1] SMP NOPTI [53764.000312] CPU: 12 PID: 30661 Comm: ethtool Kdump: loaded Tainted: GOE --------- - - 4.18.0-240.el8.x86_64 #1 [53764.011825] Hardware name: Intel Corporation WilsonCity/WilsonCity, BIOS WLYDCRB1.SYS.0020.P21.2012150710 12/15/2020 [53764.022584] RIP: 0010:ice_vsi_map_rings_to_vectors+0x7e/0x120 [ice] [53764.029089] Code: 41 0d 0f b7 b7 12 05 00 00 0f b6 d0 44 29 de 44 0f b7 c6 44 01 c2 41 39 d0 7d 2d 4c 8b 47 28 44 0f b7 ce 83 c6 01 4f 8b 04 c8 <49> 89 48 28 4 c 8b 89 b8 01 00 00 4d 89 08 4c 89 81 b8 01 00 00 44 [53764.048379] RSP: 0018:ff550dd88ea47b20 EFLAGS: 00010206 [53764.053884] RAX: 0000000000000002 RBX: 0000000000000004 RCX: ff385ea42fa4a018 [53764.061301] RDX: 0000000000000006 RSI: 0000000000000005 RDI: ff385e9baeedd018 [53764.068717] RBP: 0000000000000010 R08: 0000000000000000 R09: 0000000000000004 [53764.076133] R10: 0000000000000002 R11: 0000000000000004 R12: 0000000000000000 [53764.083553] R13: 0000000000000000 R14: ff385e658fdd9000 R15: ff385e9baeedd018 [53764.090976] FS: 000014872c5b5740(0000) GS:ff385e847f100000(0000) knlGS:0000000000000000 [53764.099362] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [53764.105409] CR2: 0000000000000028 CR3: 0000000a820fa002 CR4: 0000000000761ee0 [53764.112851] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [53764.120301] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [53764.127747] PKRU: 55555554 [53764.130781] Call Trace: [53764.133564] ice_vsi_rebuild+0x611/0x870 [ice] [53764.138341] ice_vsi_recfg_qs+0x94/0x100 [ice] [53764.143116] ice_set_channels+0x1a8/0x3e0 [ice] [53764.147975] ethtool_set_channels+0x14e/0x240 [53764.152667] dev_ethtool+0xd74/0x2a10 [53764.156665] ? __mod_lruvec_state+0x44/0x110 [53764.161280] ? __mod_lruvec_state+0x44/0x110 [53764.165893] ? page_add_file_rmap+0x15/0x170 [53764.170518] ? inet_ioctl+0xd1/0x220 [53764.174445] ? netdev_run_todo+0x5e/0x290 [53764.178808] dev_ioctl+0xb5/0x550 [53764.182485] sock_do_ioctl+0xa0/0x140 [53764.186512] sock_ioctl+0x1a8/0x300 [53764.190367] ? selinux_file_ioctl+0x161/0x200 [53764.195090] do_vfs_ioctl+0xa4/0x640 [53764.199035] ksys_ioctl+0x60/0x90 [53764.202722] __x64_sys_ioctl+0x16/0x20 [53764.206845] do_syscall_64+0x5b/0x1a0 [53764.210887] entry_SYSCALL_64_after_hwframe+0x65/0xca Fixes: 87324e747fde ("ice: Implement ethtool ops for channels") Signed-off-by: Anatolii Gerasymenko Tested-by: Gurucharan (A Contingent worker at Intel) Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_ethtool.c | 10 +++++++ drivers/net/ethernet/intel/ice/ice_lib.c | 42 ++++++++++++++++++++++++---- 2 files changed, 47 insertions(+), 5 deletions(-) diff --git a/drivers/net/ethernet/intel/ice/ice_ethtool.c b/drivers/net/ethernet/intel/ice/ice_ethtool.c index 8078618ce1b6..70335f6e8524 100644 --- a/drivers/net/ethernet/intel/ice/ice_ethtool.c +++ b/drivers/net/ethernet/intel/ice/ice_ethtool.c @@ -3507,6 +3507,16 @@ static int ice_set_channels(struct net_device *dev, struct ethtool_channels *ch) new_rx = ch->combined_count + ch->rx_count; new_tx = ch->combined_count + ch->tx_count; + if (new_rx < vsi->tc_cfg.numtc) { + netdev_err(dev, "Cannot set less Rx channels, than Traffic Classes you have (%u)\n", + vsi->tc_cfg.numtc); + return -EINVAL; + } + if (new_tx < vsi->tc_cfg.numtc) { + netdev_err(dev, "Cannot set less Tx channels, than Traffic Classes you have (%u)\n", + vsi->tc_cfg.numtc); + return -EINVAL; + } if (new_rx > ice_get_max_rxq(pf)) { netdev_err(dev, "Maximum allowed Rx channels is %d\n", ice_get_max_rxq(pf)); diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c b/drivers/net/ethernet/intel/ice/ice_lib.c index 454e01ae09b9..f7f9c973ec54 100644 --- a/drivers/net/ethernet/intel/ice/ice_lib.c +++ b/drivers/net/ethernet/intel/ice/ice_lib.c @@ -909,7 +909,7 @@ static void ice_set_dflt_vsi_ctx(struct ice_hw *hw, struct ice_vsi_ctx *ctxt) * @vsi: the VSI being configured * @ctxt: VSI context structure */ -static void ice_vsi_setup_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt) +static int ice_vsi_setup_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt) { u16 offset = 0, qmap = 0, tx_count = 0, pow = 0; u16 num_txq_per_tc, num_rxq_per_tc; @@ -982,7 +982,18 @@ static void ice_vsi_setup_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt) else vsi->num_rxq = num_rxq_per_tc; + if (vsi->num_rxq > vsi->alloc_rxq) { + dev_err(ice_pf_to_dev(vsi->back), "Trying to use more Rx queues (%u), than were allocated (%u)!\n", + vsi->num_rxq, vsi->alloc_rxq); + return -EINVAL; + } + vsi->num_txq = tx_count; + if (vsi->num_txq > vsi->alloc_txq) { + dev_err(ice_pf_to_dev(vsi->back), "Trying to use more Tx queues (%u), than were allocated (%u)!\n", + vsi->num_txq, vsi->alloc_txq); + return -EINVAL; + } if (vsi->type == ICE_VSI_VF && vsi->num_txq != vsi->num_rxq) { dev_dbg(ice_pf_to_dev(vsi->back), "VF VSI should have same number of Tx and Rx queues. Hence making them equal\n"); @@ -1000,6 +1011,8 @@ static void ice_vsi_setup_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt) */ ctxt->info.q_mapping[0] = cpu_to_le16(vsi->rxq_map[0]); ctxt->info.q_mapping[1] = cpu_to_le16(vsi->num_rxq); + + return 0; } /** @@ -1187,7 +1200,10 @@ static int ice_vsi_init(struct ice_vsi *vsi, bool init_vsi) if (vsi->type == ICE_VSI_CHNL) { ice_chnl_vsi_setup_q_map(vsi, ctxt); } else { - ice_vsi_setup_q_map(vsi, ctxt); + ret = ice_vsi_setup_q_map(vsi, ctxt); + if (ret) + goto out; + if (!init_vsi) /* means VSI being updated */ /* must to indicate which section of VSI context are * being modified @@ -3464,7 +3480,7 @@ void ice_vsi_cfg_netdev_tc(struct ice_vsi *vsi, u8 ena_tc) * * Prepares VSI tc_config to have queue configurations based on MQPRIO options. */ -static void +static int ice_vsi_setup_q_map_mqprio(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt, u8 ena_tc) { @@ -3513,7 +3529,18 @@ ice_vsi_setup_q_map_mqprio(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt, /* Set actual Tx/Rx queue pairs */ vsi->num_txq = offset + qcount_tx; + if (vsi->num_txq > vsi->alloc_txq) { + dev_err(ice_pf_to_dev(vsi->back), "Trying to use more Tx queues (%u), than were allocated (%u)!\n", + vsi->num_txq, vsi->alloc_txq); + return -EINVAL; + } + vsi->num_rxq = offset + qcount_rx; + if (vsi->num_rxq > vsi->alloc_rxq) { + dev_err(ice_pf_to_dev(vsi->back), "Trying to use more Rx queues (%u), than were allocated (%u)!\n", + vsi->num_rxq, vsi->alloc_rxq); + return -EINVAL; + } /* Setup queue TC[0].qmap for given VSI context */ ctxt->info.tc_mapping[0] = cpu_to_le16(qmap); @@ -3531,6 +3558,8 @@ ice_vsi_setup_q_map_mqprio(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt, dev_dbg(ice_pf_to_dev(vsi->back), "vsi->num_rxq = %d\n", vsi->num_rxq); dev_dbg(ice_pf_to_dev(vsi->back), "all_numtc %u, all_enatc: 0x%04x, tc_cfg.numtc %u\n", vsi->all_numtc, vsi->all_enatc, vsi->tc_cfg.numtc); + + return 0; } /** @@ -3580,9 +3609,12 @@ int ice_vsi_cfg_tc(struct ice_vsi *vsi, u8 ena_tc) if (vsi->type == ICE_VSI_PF && test_bit(ICE_FLAG_TC_MQPRIO, pf->flags)) - ice_vsi_setup_q_map_mqprio(vsi, ctx, ena_tc); + ret = ice_vsi_setup_q_map_mqprio(vsi, ctx, ena_tc); else - ice_vsi_setup_q_map(vsi, ctx); + ret = ice_vsi_setup_q_map(vsi, ctx); + + if (ret) + goto out; /* must to indicate which section of VSI context are being modified */ ctx->info.valid_sections = cpu_to_le16(ICE_AQ_VSI_PROP_RXQ_MAP_VALID); -- cgit v1.2.3