diff options
Diffstat (limited to 'sys')
34 files changed, 458 insertions, 166 deletions
diff --git a/sys/cddl/dev/fbt/aarch64/fbt_isa.c b/sys/cddl/dev/fbt/aarch64/fbt_isa.c index fd666770d3a2..b265f6a1e23e 100644 --- a/sys/cddl/dev/fbt/aarch64/fbt_isa.c +++ b/sys/cddl/dev/fbt/aarch64/fbt_isa.c @@ -105,7 +105,7 @@ fbt_provide_module_function(linker_file_t lf, int symindx, */ if (strcmp(name, "handle_el1h_sync") == 0 || strcmp(name, "do_el1h_sync") == 0) - return (1); + return (0); instr = (uint32_t *)(symval->value); limit = (uint32_t *)(symval->value + symval->size); diff --git a/sys/dev/e1000/e1000_osdep.h b/sys/dev/e1000/e1000_osdep.h index 893979025f01..ba1c8a16fad1 100644 --- a/sys/dev/e1000/e1000_osdep.h +++ b/sys/dev/e1000/e1000_osdep.h @@ -152,6 +152,9 @@ struct e1000_osdep { bus_space_tag_t mem_bus_space_tag; bus_space_handle_t mem_bus_space_handle; +#ifdef INVARIANTS + bus_size_t mem_bus_space_size; +#endif bus_space_tag_t io_bus_space_tag; bus_space_handle_t io_bus_space_handle; bus_space_tag_t flash_bus_space_tag; @@ -175,27 +178,44 @@ struct e1000_osdep bus_space_write_4(((struct e1000_osdep *)(hw)->back)->mem_bus_space_tag, \ ((struct e1000_osdep *)(hw)->back)->mem_bus_space_handle, offset, value) +static __inline uint32_t +e1000_rd32(struct e1000_osdep *osdep, uint32_t reg) +{ + + KASSERT(reg < osdep->mem_bus_space_size, + ("e1000: register offset %#jx too large (max is %#jx)", + (uintmax_t)reg, (uintmax_t)osdep->mem_bus_space_size)); + + return (bus_space_read_4(osdep->mem_bus_space_tag, + osdep->mem_bus_space_handle, reg)); +} + + +static __inline void +e1000_wr32(struct e1000_osdep *osdep, uint32_t reg, uint32_t value) +{ + + KASSERT(reg < osdep->mem_bus_space_size, + ("e1000: register offset %#jx too large (max is %#jx)", + (uintmax_t)reg, (uintmax_t)osdep->mem_bus_space_size)); + + bus_space_write_4(osdep->mem_bus_space_tag, + osdep->mem_bus_space_handle, reg, value); +} + /* Register READ/WRITE macros */ -#define E1000_READ_REG(hw, reg) \ - bus_space_read_4(((struct e1000_osdep *)(hw)->back)->mem_bus_space_tag, \ - ((struct e1000_osdep *)(hw)->back)->mem_bus_space_handle, \ - E1000_REGISTER(hw, reg)) +#define E1000_READ_REG(hw, reg) \ + e1000_rd32((hw)->back, E1000_REGISTER(hw, reg)) #define E1000_WRITE_REG(hw, reg, value) \ - bus_space_write_4(((struct e1000_osdep *)(hw)->back)->mem_bus_space_tag, \ - ((struct e1000_osdep *)(hw)->back)->mem_bus_space_handle, \ - E1000_REGISTER(hw, reg), value) + e1000_wr32((hw)->back, E1000_REGISTER(hw, reg), value) #define E1000_READ_REG_ARRAY(hw, reg, index) \ - bus_space_read_4(((struct e1000_osdep *)(hw)->back)->mem_bus_space_tag, \ - ((struct e1000_osdep *)(hw)->back)->mem_bus_space_handle, \ - E1000_REGISTER(hw, reg) + ((index)<< 2)) + e1000_rd32((hw)->back, E1000_REGISTER(hw, reg) + ((index) << 2)) #define E1000_WRITE_REG_ARRAY(hw, reg, index, value) \ - bus_space_write_4(((struct e1000_osdep *)(hw)->back)->mem_bus_space_tag, \ - ((struct e1000_osdep *)(hw)->back)->mem_bus_space_handle, \ - E1000_REGISTER(hw, reg) + ((index)<< 2), value) + e1000_wr32((hw)->back, E1000_REGISTER(hw, reg) + ((index) << 2), value) #define E1000_READ_REG_ARRAY_DWORD E1000_READ_REG_ARRAY #define E1000_WRITE_REG_ARRAY_DWORD E1000_WRITE_REG_ARRAY diff --git a/sys/dev/e1000/if_em.c b/sys/dev/e1000/if_em.c index f0ef6051fab1..9040949b36c7 100644 --- a/sys/dev/e1000/if_em.c +++ b/sys/dev/e1000/if_em.c @@ -1575,7 +1575,7 @@ em_if_init(if_ctx_t ctx) E1000_WRITE_REG(&sc->hw, E1000_VET, ETHERTYPE_VLAN); /* Clear bad data from Rx FIFOs */ - if (sc->hw.mac.type >= igb_mac_min) + if (sc->hw.mac.type >= igb_mac_min && !sc->vf_ifp) e1000_rx_fifo_flush_base(&sc->hw); /* Configure for OS presence */ @@ -1595,7 +1595,9 @@ em_if_init(if_ctx_t ctx) /* Don't lose promiscuous settings */ em_if_set_promisc(ctx, if_getflags(ifp)); - e1000_clear_hw_cntrs_base_generic(&sc->hw); + + if (sc->hw.mac.ops.clear_hw_cntrs != NULL) + sc->hw.mac.ops.clear_hw_cntrs(&sc->hw); /* MSI-X configuration for 82574 */ if (sc->hw.mac.type == e1000_82574) { @@ -2374,7 +2376,7 @@ em_if_stop(if_ctx_t ctx) em_flush_desc_rings(sc); e1000_reset_hw(&sc->hw); - if (sc->hw.mac.type >= e1000_82544) + if (sc->hw.mac.type >= e1000_82544 && !sc->vf_ifp) E1000_WRITE_REG(&sc->hw, E1000_WUFC, 0); e1000_led_off(&sc->hw); @@ -2433,6 +2435,9 @@ em_allocate_pci_resources(if_ctx_t ctx) } sc->osdep.mem_bus_space_tag = rman_get_bustag(sc->memory); sc->osdep.mem_bus_space_handle = rman_get_bushandle(sc->memory); +#ifdef INVARIANTS + sc->osdep.mem_bus_space_size = rman_get_size(sc->memory); +#endif sc->hw.hw_addr = (u8 *)&sc->osdep.mem_bus_space_handle; /* Only older adapters use IO mapping */ @@ -3284,11 +3289,13 @@ em_reset(if_ctx_t ctx) /* Issue a global reset */ e1000_reset_hw(hw); - if (hw->mac.type >= igb_mac_min) { - E1000_WRITE_REG(hw, E1000_WUC, 0); - } else { - E1000_WRITE_REG(hw, E1000_WUFC, 0); - em_disable_aspm(sc); + if (!sc->vf_ifp) { + if (hw->mac.type >= igb_mac_min) { + E1000_WRITE_REG(hw, E1000_WUC, 0); + } else { + E1000_WRITE_REG(hw, E1000_WUFC, 0); + em_disable_aspm(sc); + } } if (sc->flags & IGB_MEDIA_RESET) { e1000_setup_init_funcs(hw, true); @@ -3838,7 +3845,7 @@ em_initialize_receive_unit(if_ctx_t ctx) sc->rx_int_delay.value); } - if (hw->mac.type >= em_mac_min) { + if (hw->mac.type >= em_mac_min && !sc->vf_ifp) { uint32_t rfctl; /* Use extended rx descriptor formats */ rfctl = E1000_READ_REG(hw, E1000_RFCTL); @@ -3858,33 +3865,38 @@ em_initialize_receive_unit(if_ctx_t ctx) E1000_WRITE_REG(hw, E1000_RFCTL, rfctl); } - /* Set up L3 and L4 csum Rx descriptor offloads */ - rxcsum = E1000_READ_REG(hw, E1000_RXCSUM); - if (if_getcapenable(ifp) & IFCAP_RXCSUM) { - rxcsum |= E1000_RXCSUM_TUOFL | E1000_RXCSUM_IPOFL; - if (hw->mac.type > e1000_82575) - rxcsum |= E1000_RXCSUM_CRCOFL; - else if (hw->mac.type < em_mac_min && - if_getcapenable(ifp) & IFCAP_HWCSUM_IPV6) - rxcsum |= E1000_RXCSUM_IPV6OFL; - } else { - rxcsum &= ~(E1000_RXCSUM_IPOFL | E1000_RXCSUM_TUOFL); - if (hw->mac.type > e1000_82575) - rxcsum &= ~E1000_RXCSUM_CRCOFL; - else if (hw->mac.type < em_mac_min) - rxcsum &= ~E1000_RXCSUM_IPV6OFL; - } + /* + * Set up L3 and L4 csum Rx descriptor offloads only on Physical + * Functions. Virtual Functions have no access to this register. + */ + if (!sc->vf_ifp) { + rxcsum = E1000_READ_REG(hw, E1000_RXCSUM); + if (if_getcapenable(ifp) & IFCAP_RXCSUM) { + rxcsum |= E1000_RXCSUM_TUOFL | E1000_RXCSUM_IPOFL; + if (hw->mac.type > e1000_82575) + rxcsum |= E1000_RXCSUM_CRCOFL; + else if (hw->mac.type < em_mac_min && + if_getcapenable(ifp) & IFCAP_HWCSUM_IPV6) + rxcsum |= E1000_RXCSUM_IPV6OFL; + } else { + rxcsum &= ~(E1000_RXCSUM_IPOFL | E1000_RXCSUM_TUOFL); + if (hw->mac.type > e1000_82575) + rxcsum &= ~E1000_RXCSUM_CRCOFL; + else if (hw->mac.type < em_mac_min) + rxcsum &= ~E1000_RXCSUM_IPV6OFL; + } - if (sc->rx_num_queues > 1) { - /* RSS hash needed in the Rx descriptor */ - rxcsum |= E1000_RXCSUM_PCSD; + if (sc->rx_num_queues > 1) { + /* RSS hash needed in the Rx descriptor */ + rxcsum |= E1000_RXCSUM_PCSD; - if (hw->mac.type >= igb_mac_min) - igb_initialize_rss_mapping(sc); - else - em_initialize_rss_mapping(sc); + if (hw->mac.type >= igb_mac_min) + igb_initialize_rss_mapping(sc); + else + em_initialize_rss_mapping(sc); + } + E1000_WRITE_REG(hw, E1000_RXCSUM, rxcsum); } - E1000_WRITE_REG(hw, E1000_RXCSUM, rxcsum); for (i = 0, que = sc->rx_queues; i < sc->rx_num_queues; i++, que++) { struct rx_ring *rxr = &que->rxr; @@ -4392,6 +4404,8 @@ em_get_wakeup(if_ctx_t ctx) switch (sc->hw.mac.type) { case e1000_82542: case e1000_82543: + case e1000_vfadapt: + case e1000_vfadapt_i350: break; case e1000_82544: e1000_read_nvm(&sc->hw, @@ -4437,8 +4451,6 @@ em_get_wakeup(if_ctx_t ctx) case e1000_i354: case e1000_i210: case e1000_i211: - case e1000_vfadapt: - case e1000_vfadapt_i350: apme_mask = E1000_WUC_APME; sc->has_amt = true; eeprom_data = E1000_READ_REG(&sc->hw, E1000_WUC); @@ -4494,7 +4506,6 @@ em_get_wakeup(if_ctx_t ctx) global_quad_port_a = 0; break; } - return; } diff --git a/sys/dev/ice/ice_drv_info.h b/sys/dev/ice/ice_drv_info.h index 46965f4124bc..abb11bdb5fd9 100644 --- a/sys/dev/ice/ice_drv_info.h +++ b/sys/dev/ice/ice_drv_info.h @@ -238,6 +238,9 @@ static const pci_vendor_info_t ice_vendor_info_array[] = { ICE_INTEL_VENDOR_ID, 0x0001, 0, "Intel(R) Ethernet Network Adapter E835-XXV-2 for OCP 3.0"), PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_SFP, + ICE_INTEL_VENDOR_ID, 0x0002, 0, + "Intel(R) Ethernet Network Adapter E835-XXV-4"), + PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_SFP, ICE_INTEL_VENDOR_ID, 0x0003, 0, "Intel(R) Ethernet Network Adapter E835-XXV-2"), PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_SFP, diff --git a/sys/dev/ixl/if_ixl.c b/sys/dev/ixl/if_ixl.c index 43c3af056b67..aa2e69ea0b5a 100644 --- a/sys/dev/ixl/if_ixl.c +++ b/sys/dev/ixl/if_ixl.c @@ -1480,17 +1480,33 @@ ixl_if_multi_set(if_ctx_t ctx) struct ixl_pf *pf = iflib_get_softc(ctx); struct ixl_vsi *vsi = &pf->vsi; struct i40e_hw *hw = vsi->hw; + enum i40e_status_code status; int mcnt; + if_t ifp = iflib_get_ifp(ctx); IOCTL_DEBUGOUT("ixl_if_multi_set: begin"); /* Delete filters for removed multicast addresses */ ixl_del_multi(vsi, false); - mcnt = min(if_llmaddr_count(iflib_get_ifp(ctx)), MAX_MULTICAST_ADDR); + mcnt = min(if_llmaddr_count(ifp), MAX_MULTICAST_ADDR); if (__predict_false(mcnt == MAX_MULTICAST_ADDR)) { - i40e_aq_set_vsi_multicast_promiscuous(hw, + /* Check if promisc mode is already enabled, if yes return */ + if (vsi->flags & IXL_FLAGS_MC_PROMISC) + return; + + status = i40e_aq_set_vsi_multicast_promiscuous(hw, vsi->seid, TRUE, NULL); + if (status != I40E_SUCCESS) + if_printf(ifp, "Failed to enable multicast promiscuous " + "mode, status: %s\n", i40e_stat_str(hw, status)); + else { + if_printf(ifp, "Enabled multicast promiscuous mode\n"); + + /* Set the flag to track promiscuous mode */ + vsi->flags |= IXL_FLAGS_MC_PROMISC; + } + /* Delete all existing MC filters */ ixl_del_multi(vsi, true); return; } @@ -1693,6 +1709,13 @@ ixl_if_promisc_set(if_ctx_t ctx, int flags) return (err); err = i40e_aq_set_vsi_multicast_promiscuous(hw, vsi->seid, multi, NULL); + + /* Update the multicast promiscuous flag based on the new state */ + if (multi) + vsi->flags |= IXL_FLAGS_MC_PROMISC; + else + vsi->flags &= ~IXL_FLAGS_MC_PROMISC; + return (err); } diff --git a/sys/dev/ixl/ixl.h b/sys/dev/ixl/ixl.h index f45354d29300..69925a131b35 100644 --- a/sys/dev/ixl/ixl.h +++ b/sys/dev/ixl/ixl.h @@ -202,6 +202,7 @@ #define IXL_FLAGS_KEEP_TSO6 (1 << 1) #define IXL_FLAGS_USES_MSIX (1 << 2) #define IXL_FLAGS_IS_VF (1 << 3) +#define IXL_FLAGS_MC_PROMISC (1 << 4) #define IXL_VSI_IS_PF(v) ((v->flags & IXL_FLAGS_IS_VF) == 0) #define IXL_VSI_IS_VF(v) ((v->flags & IXL_FLAGS_IS_VF) != 0) diff --git a/sys/dev/ixl/ixl_pf_main.c b/sys/dev/ixl/ixl_pf_main.c index 4f384e7191af..99851af61cfe 100644 --- a/sys/dev/ixl/ixl_pf_main.c +++ b/sys/dev/ixl/ixl_pf_main.c @@ -592,24 +592,29 @@ ixl_add_maddr(void *arg, struct sockaddr_dl *sdl, u_int cnt) * Routines for multicast and vlan filter management. * *********************************************************************/ + +/** + * ixl_add_multi - Add multicast filters to the hardware + * @vsi: The VSI structure + * + * In case number of multicast filters in the IFP exceeds 127 entries, + * multicast promiscuous mode will be enabled and the filters will be removed + * from the hardware + */ void ixl_add_multi(struct ixl_vsi *vsi) { if_t ifp = vsi->ifp; - struct i40e_hw *hw = vsi->hw; int mcnt = 0; struct ixl_add_maddr_arg cb_arg; IOCTL_DEBUGOUT("ixl_add_multi: begin"); - mcnt = if_llmaddr_count(ifp); - if (__predict_false(mcnt >= MAX_MULTICAST_ADDR)) { - i40e_aq_set_vsi_multicast_promiscuous(hw, - vsi->seid, TRUE, NULL); - /* delete all existing MC filters */ - ixl_del_multi(vsi, true); - return; - } + /* + * There is no need to check if the number of multicast addresses + * exceeds the MAX_MULTICAST_ADDR threshold and set promiscuous mode + * here, as all callers already handle this case. + */ cb_arg.vsi = vsi; LIST_INIT(&cb_arg.to_add); @@ -632,30 +637,103 @@ ixl_match_maddr(void *arg, struct sockaddr_dl *sdl, u_int cnt) return (0); } +/** + * ixl_dis_multi_promisc - Disable multicast promiscuous mode + * @vsi: The VSI structure + * @vsi_mcnt: Number of multicast filters in the VSI + * + * Disable multicast promiscuous mode based on number of entries in the IFP + * and the VSI, then re-add multicast filters. + * + */ +static void +ixl_dis_multi_promisc(struct ixl_vsi *vsi, int vsi_mcnt) +{ + struct ifnet *ifp = vsi->ifp; + struct i40e_hw *hw = vsi->hw; + int ifp_mcnt = 0; + enum i40e_status_code status; + + /* + * Check if multicast promiscuous mode was actually enabled. + * If promiscuous mode was not enabled, don't attempt to disable it. + * Also, don't disable if IFF_PROMISC or IFF_ALLMULTI is set. + */ + if (!(vsi->flags & IXL_FLAGS_MC_PROMISC) || + (if_getflags(ifp) & (IFF_PROMISC | IFF_ALLMULTI))) + return; + + ifp_mcnt = if_llmaddr_count(ifp); + /* + * Equal lists or empty ifp list mean the list has not been changed + * and in such case avoid disabling multicast promiscuous mode as it + * was not previously enabled. Case where multicast promiscuous mode has + * been enabled is when vsi_mcnt == 0 && ifp_mcnt > 0. + */ + if (ifp_mcnt == vsi_mcnt || ifp_mcnt == 0 || + ifp_mcnt >= MAX_MULTICAST_ADDR) + return; + + status = i40e_aq_set_vsi_multicast_promiscuous(hw, vsi->seid, + FALSE, NULL); + if (status != I40E_SUCCESS) { + if_printf(ifp, "Failed to disable multicast promiscuous " + "mode, status: %s\n", i40e_stat_str(hw, status)); + + return; + } + + /* Clear the flag since promiscuous mode is now disabled */ + vsi->flags &= ~IXL_FLAGS_MC_PROMISC; + if_printf(ifp, "Disabled multicast promiscuous mode\n"); + + ixl_add_multi(vsi); +} + +/** + * ixl_del_multi - Delete multicast filters from the hardware + * @vsi: The VSI structure + * @all: Bool to determine if all the multicast filters should be removed + * + * In case number of multicast filters in the IFP drops to 127 entries, + * multicast promiscuous mode will be disabled and the filters will be reapplied + * to the hardware. + */ void ixl_del_multi(struct ixl_vsi *vsi, bool all) { - struct ixl_ftl_head to_del; + int to_del_cnt = 0, vsi_mcnt = 0; if_t ifp = vsi->ifp; struct ixl_mac_filter *f, *fn; - int mcnt = 0; + struct ixl_ftl_head to_del; IOCTL_DEBUGOUT("ixl_del_multi: begin"); LIST_INIT(&to_del); /* Search for removed multicast addresses */ LIST_FOREACH_SAFE(f, &vsi->ftl, ftle, fn) { - if ((f->flags & IXL_FILTER_MC) == 0 || - (!all && (if_foreach_llmaddr(ifp, ixl_match_maddr, f) == 0))) + if ((f->flags & IXL_FILTER_MC) == 0) + continue; + + /* Count all the multicast filters in the VSI for comparison */ + vsi_mcnt++; + + if (!all && if_foreach_llmaddr(ifp, ixl_match_maddr, f) != 0) continue; LIST_REMOVE(f, ftle); LIST_INSERT_HEAD(&to_del, f, ftle); - mcnt++; + to_del_cnt++; } - if (mcnt > 0) - ixl_del_hw_filters(vsi, &to_del, mcnt); + if (to_del_cnt > 0) { + ixl_del_hw_filters(vsi, &to_del, to_del_cnt); + return; + } + + ixl_dis_multi_promisc(vsi, vsi_mcnt); + + IOCTL_DEBUGOUT("ixl_del_multi: end"); } void diff --git a/sys/dev/md/md.c b/sys/dev/md/md.c index 1b434eda19a5..3057060d7adb 100644 --- a/sys/dev/md/md.c +++ b/sys/dev/md/md.c @@ -11,9 +11,9 @@ */ /*- - * The following functions are based on the vn(4) driver: mdstart_swap(), - * mdstart_vnode(), mdcreate_swap(), mdcreate_vnode() and mddestroy(), - * and as such under the following copyright: + * The following functions are based on the historical vn(4) driver: + * mdstart_swap(), mdstart_vnode(), mdcreate_swap(), mdcreate_vnode() + * and mddestroy(), and as such under the following copyright: * * Copyright (c) 1988 University of Utah. * Copyright (c) 1990, 1993 diff --git a/sys/dev/nvme/nvme_ctrlr.c b/sys/dev/nvme/nvme_ctrlr.c index 6f5d6ae74add..ce203e2869fd 100644 --- a/sys/dev/nvme/nvme_ctrlr.c +++ b/sys/dev/nvme/nvme_ctrlr.c @@ -1584,9 +1584,14 @@ noadminq: bus_release_resource(ctrlr->dev, SYS_RES_IRQ, rman_get_rid(ctrlr->res), ctrlr->res); - if (ctrlr->bar4_resource != NULL) { + if (ctrlr->msix_table_resource != NULL) { bus_release_resource(dev, SYS_RES_MEMORY, - ctrlr->bar4_resource_id, ctrlr->bar4_resource); + ctrlr->msix_table_resource_id, ctrlr->msix_table_resource); + } + + if (ctrlr->msix_pba_resource != NULL) { + bus_release_resource(dev, SYS_RES_MEMORY, + ctrlr->msix_pba_resource_id, ctrlr->msix_pba_resource); } bus_release_resource(dev, SYS_RES_MEMORY, diff --git a/sys/dev/nvme/nvme_pci.c b/sys/dev/nvme/nvme_pci.c index a78327ba0e8b..9c40c3d9f5c7 100644 --- a/sys/dev/nvme/nvme_pci.c +++ b/sys/dev/nvme/nvme_pci.c @@ -154,11 +154,15 @@ nvme_ctrlr_allocate_bar(struct nvme_controller *ctrlr) { ctrlr->resource_id = PCIR_BAR(0); + ctrlr->msix_table_resource_id = -1; + ctrlr->msix_table_resource = NULL; + ctrlr->msix_pba_resource_id = -1; + ctrlr->msix_pba_resource = NULL; ctrlr->resource = bus_alloc_resource_any(ctrlr->dev, SYS_RES_MEMORY, &ctrlr->resource_id, RF_ACTIVE); - if(ctrlr->resource == NULL) { + if (ctrlr->resource == NULL) { nvme_printf(ctrlr, "unable to allocate pci resource\n"); return (ENOMEM); } @@ -168,15 +172,32 @@ nvme_ctrlr_allocate_bar(struct nvme_controller *ctrlr) ctrlr->regs = (struct nvme_registers *)ctrlr->bus_handle; /* - * The NVMe spec allows for the MSI-X table to be placed behind - * BAR 4/5, separate from the control/doorbell registers. Always - * try to map this bar, because it must be mapped prior to calling - * pci_alloc_msix(). If the table isn't behind BAR 4/5, - * bus_alloc_resource() will just return NULL which is OK. + * The NVMe spec allows for the MSI-X tables to be placed behind + * BAR 4 and/or 5, separate from the control/doorbell registers. */ - ctrlr->bar4_resource_id = PCIR_BAR(4); - ctrlr->bar4_resource = bus_alloc_resource_any(ctrlr->dev, SYS_RES_MEMORY, - &ctrlr->bar4_resource_id, RF_ACTIVE); + + ctrlr->msix_table_resource_id = pci_msix_table_bar(ctrlr->dev); + ctrlr->msix_pba_resource_id = pci_msix_pba_bar(ctrlr->dev); + + if (ctrlr->msix_table_resource_id >= 0 && + ctrlr->msix_table_resource_id != ctrlr->resource_id) { + ctrlr->msix_table_resource = bus_alloc_resource_any(ctrlr->dev, + SYS_RES_MEMORY, &ctrlr->msix_table_resource_id, RF_ACTIVE); + if (ctrlr->msix_table_resource == NULL) { + nvme_printf(ctrlr, "unable to allocate msi-x table resource\n"); + return (ENOMEM); + } + } + if (ctrlr->msix_pba_resource_id >= 0 && + ctrlr->msix_pba_resource_id != ctrlr->resource_id && + ctrlr->msix_pba_resource_id != ctrlr->msix_table_resource_id) { + ctrlr->msix_pba_resource = bus_alloc_resource_any(ctrlr->dev, + SYS_RES_MEMORY, &ctrlr->msix_pba_resource_id, RF_ACTIVE); + if (ctrlr->msix_pba_resource == NULL) { + nvme_printf(ctrlr, "unable to allocate msi-x pba resource\n"); + return (ENOMEM); + } + } return (0); } @@ -202,9 +223,14 @@ bad: ctrlr->resource_id, ctrlr->resource); } - if (ctrlr->bar4_resource != NULL) { + if (ctrlr->msix_table_resource != NULL) { + bus_release_resource(dev, SYS_RES_MEMORY, + ctrlr->msix_table_resource_id, ctrlr->msix_table_resource); + } + + if (ctrlr->msix_pba_resource != NULL) { bus_release_resource(dev, SYS_RES_MEMORY, - ctrlr->bar4_resource_id, ctrlr->bar4_resource); + ctrlr->msix_pba_resource_id, ctrlr->msix_pba_resource); } if (ctrlr->tag) diff --git a/sys/dev/nvme/nvme_private.h b/sys/dev/nvme/nvme_private.h index 36e04ceb7f31..93833672674a 100644 --- a/sys/dev/nvme/nvme_private.h +++ b/sys/dev/nvme/nvme_private.h @@ -233,8 +233,10 @@ struct nvme_controller { * separate from the control registers which are in BAR 0/1. These * members track the mapping of BAR 4/5 for that reason. */ - int bar4_resource_id; - struct resource *bar4_resource; + int msix_table_resource_id; + struct resource *msix_table_resource; + int msix_pba_resource_id; + struct resource *msix_pba_resource; int msi_count; uint32_t enable_aborts; diff --git a/sys/dev/usb/serial/u3g.c b/sys/dev/usb/serial/u3g.c index 28a403a846bd..fad352fc1f63 100644 --- a/sys/dev/usb/serial/u3g.c +++ b/sys/dev/usb/serial/u3g.c @@ -531,6 +531,7 @@ static const STRUCT_USB_HOST_ID u3g_devs[] = { U3G_DEV(QUECTEL, RG520, 0), U3G_DEV(QUECTEL, EC200, 0), U3G_DEV(QUECTEL, EC200S, 0), + U3G_DEV(QUECTEL, EM060K, 0), U3G_DEV(QUECTEL, EC200T, 0), U3G_DEV(QUECTEL, UC200, 0), U3G_DEV(SIERRA, AC402, 0), @@ -600,6 +601,7 @@ static const STRUCT_USB_HOST_ID u3g_devs[] = { U3G_DEV(SIERRA, EM7455_2, 0), U3G_DEV(SIERRA, EM7565, 0), U3G_DEV(SIERRA, EM7565_2, 0), + U3G_DEV(SIERRA, EM7590, 0), U3G_DEV(SILABS, SAEL, U3GINIT_SAEL_M460), U3G_DEV(STELERA, C105, 0), U3G_DEV(STELERA, E1003, 0), diff --git a/sys/dev/usb/usbdevs b/sys/dev/usb/usbdevs index 251d2f483a61..fa201f448fd6 100644 --- a/sys/dev/usb/usbdevs +++ b/sys/dev/usb/usbdevs @@ -4073,6 +4073,7 @@ product QUECTEL RG500 0x0800 Quectel RG500/RM500/RG510/RM510 product QUECTEL RG520 0x0801 Quectel RG520/RM520/SG520 product QUECTEL EC200 0x6000 Quectel EC200/UC200 product QUECTEL EC200S 0x6002 Quectel EC200S +product QUECTEL EM060K 0x6008 Quectel EM060K product QUECTEL EC200T 0x6026 Quectel EC200T product QUECTEL UC200 0x6120 Quectel UC200 @@ -4436,6 +4437,7 @@ product SIERRA EM7455 0x9078 Sierra Wireless EM7455 Qualcomm Snapdragon X7 LTE- product SIERRA EM7455_2 0x9079 Sierra Wireless EM7455 Qualcomm Snapdragon X7 LTE-A product SIERRA EM7565 0x9090 Sierra Wireless EM7565 Qualcomm Snapdragon X7 LTE-A product SIERRA EM7565_2 0x9091 Sierra Wireless EM7565 Qualcomm Snapdragon X7 LTE-A +product SIERRA EM7590 0xc081 Sierra Wireless EM7590 Qualcomm Snapdragon X7 LTE-A /* Sigmatel products */ product SIGMATEL WBT_3052 0x4200 WBT-3052 IrDA/USB Bridge diff --git a/sys/geom/zero/g_zero.c b/sys/geom/zero/g_zero.c index 66cc884fab56..be31cc794cb5 100644 --- a/sys/geom/zero/g_zero.c +++ b/sys/geom/zero/g_zero.c @@ -47,11 +47,11 @@ static SYSCTL_NODE(_kern_geom, OID_AUTO, zero, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, "GEOM_ZERO stuff"); static int g_zero_clear = 1; SYSCTL_PROC(_kern_geom_zero, OID_AUTO, clear, - CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, &g_zero_clear, 0, + CTLTYPE_INT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, &g_zero_clear, 0, g_zero_clear_sysctl, "I", "Clear read data buffer"); static int g_zero_byte = 0; -SYSCTL_INT(_kern_geom_zero, OID_AUTO, byte, CTLFLAG_RW, &g_zero_byte, 0, +SYSCTL_INT(_kern_geom_zero, OID_AUTO, byte, CTLFLAG_RWTUN, &g_zero_byte, 0, "Byte (octet) value to clear the buffers with"); static struct g_provider *gpp; diff --git a/sys/kern/subr_syscall.c b/sys/kern/subr_syscall.c index 25843eec3754..139a4df57f13 100644 --- a/sys/kern/subr_syscall.c +++ b/sys/kern/subr_syscall.c @@ -57,8 +57,8 @@ syscallenter(struct thread *td) struct proc *p; struct syscall_args *sa; struct sysent *se; - int error, traced; - bool sy_thr_static; + int error; + bool sy_thr_static, traced; VM_CNT_INC(v_syscall); p = td->td_proc; @@ -217,7 +217,7 @@ syscallret(struct thread *td) struct proc *p; struct syscall_args *sa; ksiginfo_t ksi; - int traced; + bool traced; KASSERT(td->td_errno != ERELOOKUP, ("ERELOOKUP not consumed syscall %d", td->td_sa.code)); @@ -248,9 +248,9 @@ syscallret(struct thread *td) } #endif - traced = 0; + traced = false; if (__predict_false(p->p_flag & P_TRACED)) { - traced = 1; + traced = true; PROC_LOCK(p); td->td_dbgflags |= TDB_SCX; PROC_UNLOCK(p); diff --git a/sys/kern/vfs_syscalls.c b/sys/kern/vfs_syscalls.c index 9c5c5b9dfa80..99f9e129f4cd 100644 --- a/sys/kern/vfs_syscalls.c +++ b/sys/kern/vfs_syscalls.c @@ -1074,7 +1074,7 @@ flags_to_rights(int flags, cap_rights_t *rightsp) if (flags & O_TRUNC) cap_rights_set_one(rightsp, CAP_FTRUNCATE); - if (flags & (O_SYNC | O_FSYNC)) + if (flags & (O_SYNC | O_FSYNC | O_DSYNC)) cap_rights_set_one(rightsp, CAP_FSYNC); if (flags & (O_EXLOCK | O_SHLOCK)) diff --git a/sys/net/altq/altq_cbq.c b/sys/net/altq/altq_cbq.c index fdf39690160b..2333b9ea8678 100644 --- a/sys/net/altq/altq_cbq.c +++ b/sys/net/altq/altq_cbq.c @@ -173,6 +173,8 @@ cbq_request(struct ifaltq *ifq, int req, void *arg) static void get_class_stats(class_stats_t *statsp, struct rm_class *cl) { + memset(statsp, 0, sizeof(*statsp)); + statsp->xmit_cnt = cl->stats_.xmit_cnt; statsp->drop_cnt = cl->stats_.drop_cnt; statsp->over = cl->stats_.over; diff --git a/sys/net/altq/altq_fairq.c b/sys/net/altq/altq_fairq.c index 6069865101a0..0a00168e547e 100644 --- a/sys/net/altq/altq_fairq.c +++ b/sys/net/altq/altq_fairq.c @@ -857,6 +857,8 @@ get_class_stats(struct fairq_classstats *sp, struct fairq_class *cl) { fairq_bucket_t *b; + memset(sp, 0, sizeof(*sp)); + sp->class_handle = cl->cl_handle; sp->qlimit = cl->cl_qlimit; sp->xmit_cnt = cl->cl_xmitcnt; diff --git a/sys/net/altq/altq_priq.c b/sys/net/altq/altq_priq.c index 026346639b2e..fec488418546 100644 --- a/sys/net/altq/altq_priq.c +++ b/sys/net/altq/altq_priq.c @@ -597,6 +597,8 @@ priq_purgeq(struct priq_class *cl) static void get_class_stats(struct priq_classstats *sp, struct priq_class *cl) { + memset(sp, 0, sizeof(*sp)); + sp->class_handle = cl->cl_handle; sp->qlength = qlen(cl->cl_q); sp->qlimit = qlimit(cl->cl_q); diff --git a/sys/net/if.c b/sys/net/if.c index 607bcdd2aa80..0c7e32e858bc 100644 --- a/sys/net/if.c +++ b/sys/net/if.c @@ -2850,15 +2850,20 @@ ifhwioctl(u_long cmd, struct ifnet *ifp, caddr_t data, struct thread *td) break; case SIOCAIFGROUP: + { + const char *groupname; + error = priv_check(td, PRIV_NET_ADDIFGROUP); if (error) return (error); - error = if_addgroup(ifp, - ((struct ifgroupreq *)data)->ifgr_group); + groupname = ((struct ifgroupreq *)data)->ifgr_group; + if (strnlen(groupname, IFNAMSIZ) == IFNAMSIZ) + return (EINVAL); + error = if_addgroup(ifp, groupname); if (error != 0) return (error); break; - + } case SIOCGIFGROUP: { struct epoch_tracker et; @@ -2870,15 +2875,20 @@ ifhwioctl(u_long cmd, struct ifnet *ifp, caddr_t data, struct thread *td) } case SIOCDIFGROUP: + { + const char *groupname; + error = priv_check(td, PRIV_NET_DELIFGROUP); if (error) return (error); - error = if_delgroup(ifp, - ((struct ifgroupreq *)data)->ifgr_group); + groupname = ((struct ifgroupreq *)data)->ifgr_group; + if (strnlen(groupname, IFNAMSIZ) == IFNAMSIZ) + return (EINVAL); + error = if_delgroup(ifp, groupname); if (error != 0) return (error); break; - + } default: error = ENOIOCTL; break; @@ -3022,9 +3032,17 @@ ifioctl(struct socket *so, u_long cmd, caddr_t data, struct thread *td) goto out_noref; case SIOCGIFGMEMB: - error = if_getgroupmembers((struct ifgroupreq *)data); - goto out_noref; + { + struct ifgroupreq *req; + req = (struct ifgroupreq *)data; + if (strnlen(req->ifgr_name, IFNAMSIZ) == IFNAMSIZ) { + error = EINVAL; + goto out_noref; + } + error = if_getgroupmembers(req); + goto out_noref; + } #if defined(INET) || defined(INET6) case SIOCSVH: case SIOCGVH: diff --git a/sys/net/if_vxlan.c b/sys/net/if_vxlan.c index 71d9223e366e..1f242d86ece5 100644 --- a/sys/net/if_vxlan.c +++ b/sys/net/if_vxlan.c @@ -2534,7 +2534,7 @@ vxlan_encap4(struct vxlan_softc *sc, const union vxlan_sockaddr *fvxlsa, ifp = sc->vxl_ifp; srcaddr = sc->vxl_src_addr.in4.sin_addr; - srcport = vxlan_pick_source_port(sc, m); + srcport = htons(vxlan_pick_source_port(sc, m)); dstaddr = fvxlsa->in4.sin_addr; dstport = fvxlsa->in4.sin_port; @@ -2645,7 +2645,7 @@ vxlan_encap6(struct vxlan_softc *sc, const union vxlan_sockaddr *fvxlsa, ifp = sc->vxl_ifp; srcaddr = &sc->vxl_src_addr.in6.sin6_addr; - srcport = vxlan_pick_source_port(sc, m); + srcport = htons(vxlan_pick_source_port(sc, m)); dstaddr = &fvxlsa->in6.sin6_addr; dstport = fvxlsa->in6.sin6_port; diff --git a/sys/netgraph/ng_nat.c b/sys/netgraph/ng_nat.c index ae083608a199..9c09d3305ef9 100644 --- a/sys/netgraph/ng_nat.c +++ b/sys/netgraph/ng_nat.c @@ -812,7 +812,8 @@ ng_nat_rcvdata(hook_p hook, item_p item ) if (ip->ip_v != IPVERSION) goto send; /* other IP version, let it pass */ - if (m->m_pkthdr.len < ipofs + ntohs(ip->ip_len)) + uint16_t ip_len = ntohs(ip->ip_len); + if (m->m_pkthdr.len < (ipofs + ip_len)) goto send; /* packet too short (i.e. fragmented or broken) */ /* @@ -846,50 +847,68 @@ ng_nat_rcvdata(hook_p hook, item_p item ) if (rval == PKT_ALIAS_RESPOND) m->m_flags |= M_SKIP_FIREWALL; - m->m_pkthdr.len = m->m_len = ntohs(ip->ip_len) + ipofs; - if ((ip->ip_off & htons(IP_OFFMASK)) == 0 && - ip->ip_p == IPPROTO_TCP) { - struct tcphdr *th = (struct tcphdr *)((caddr_t)ip + - (ip->ip_hl << 2)); + /* Re-read just in case it has been updated */ + ip_len = ntohs(ip->ip_len); + int new_m_len = ip_len + ipofs; + if (new_m_len > (m->m_len + M_TRAILINGSPACE(m))) { /* - * Here is our terrible HACK. - * - * Sometimes LibAlias edits contents of TCP packet. - * In this case it needs to recompute full TCP - * checksum. However, the problem is that LibAlias - * doesn't have any idea about checksum offloading - * in kernel. To workaround this, we do not do - * checksumming in LibAlias, but only mark the - * packets in th_x2 field. If we receive a marked - * packet, we calculate correct checksum for it - * aware of offloading. - * - * Why do I do such a terrible hack instead of - * recalculating checksum for each packet? - * Because the previous checksum was not checked! - * Recalculating checksums for EVERY packet will - * hide ALL transmission errors. Yes, marked packets - * still suffer from this problem. But, sigh, natd(8) - * has this problem, too. + * This is just a safety railguard to make sure LibAlias has not + * screwed the IP packet up somehow, should probably be KASSERT() + * at some point. Calling in_delayed_cksum() will parse IP packet + * again and reliably panic if there is less data than the IP + * header declares, there might be some other places too. */ + log(LOG_ERR, "ng_nat_rcvdata: outgoing packet corrupted, " + "not enough data: expected %d, available (%d - %d)\n", + ip_len, m->m_len + (int)M_TRAILINGSPACE(m), ipofs); + NG_FREE_ITEM(item); + return (ENXIO); + } + + m->m_pkthdr.len = m->m_len = new_m_len; - if (th->th_x2) { - uint16_t ip_len = ntohs(ip->ip_len); + if ((ip->ip_off & htons(IP_OFFMASK)) != 0 || ip->ip_p != IPPROTO_TCP) + goto send; - th->th_x2 = 0; - th->th_sum = in_pseudo(ip->ip_src.s_addr, - ip->ip_dst.s_addr, htons(IPPROTO_TCP + - ip_len - (ip->ip_hl << 2))); + uint16_t pl_offset = ip->ip_hl << 2; + struct tcphdr *th = (struct tcphdr *)((caddr_t)ip + pl_offset); - if ((m->m_pkthdr.csum_flags & CSUM_TCP) == 0) { - m->m_pkthdr.csum_data = offsetof(struct tcphdr, - th_sum); - in_delayed_cksum(m); - } - } - } + /* + * Here is our terrible HACK. + * + * Sometimes LibAlias edits contents of TCP packet. + * In this case it needs to recompute full TCP + * checksum. However, the problem is that LibAlias + * doesn't have any idea about checksum offloading + * in kernel. To workaround this, we do not do + * checksumming in LibAlias, but only mark the + * packets in th_x2 field. If we receive a marked + * packet, we calculate correct checksum for it + * aware of offloading. + * + * Why do I do such a terrible hack instead of + * recalculating checksum for each packet? + * Because the previous checksum was not checked! + * Recalculating checksums for EVERY packet will + * hide ALL transmission errors. Yes, marked packets + * still suffer from this problem. But, sigh, natd(8) + * has this problem, too. + */ + + if (!th->th_x2) + goto send; + + th->th_x2 = 0; + th->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, + htons(IPPROTO_TCP + ip_len - pl_offset)); + + if ((m->m_pkthdr.csum_flags & CSUM_TCP) != 0) + goto send; + + m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); + in_delayed_cksum_o(m, ipofs); send: if (hook == priv->in) diff --git a/sys/netinet/in_proto.c b/sys/netinet/in_proto.c index e534fdd77635..129dc1cfe892 100644 --- a/sys/netinet/in_proto.c +++ b/sys/netinet/in_proto.c @@ -110,6 +110,8 @@ SYSCTL_NODE(_net_inet, IPPROTO_ICMP, icmp, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, "ICMP"); SYSCTL_NODE(_net_inet, IPPROTO_UDP, udp, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, "UDP"); +SYSCTL_NODE(_net_inet, IPPROTO_UDPLITE, udplite, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, + "UDP-Lite"); SYSCTL_NODE(_net_inet, IPPROTO_TCP, tcp, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, "TCP"); #if defined(SCTP) || defined(SCTP_SUPPORT) diff --git a/sys/netinet/ip_output.c b/sys/netinet/ip_output.c index 66051c9c711c..e3ef8e2c7dd9 100644 --- a/sys/netinet/ip_output.c +++ b/sys/netinet/ip_output.c @@ -1045,14 +1045,14 @@ done: } void -in_delayed_cksum(struct mbuf *m) +in_delayed_cksum_o(struct mbuf *m, uint16_t iph_offset) { struct ip *ip; struct udphdr *uh; uint16_t cklen, csum, offset; - ip = mtod(m, struct ip *); - offset = ip->ip_hl << 2 ; + ip = (struct ip *)mtodo(m, iph_offset); + offset = iph_offset + (ip->ip_hl << 2); if (m->m_pkthdr.csum_flags & CSUM_UDP) { /* if udp header is not in the first mbuf copy udplen */ @@ -1079,6 +1079,13 @@ in_delayed_cksum(struct mbuf *m) *(u_short *)mtodo(m, offset) = csum; } +void +in_delayed_cksum(struct mbuf *m) +{ + + in_delayed_cksum_o(m, 0); +} + /* * IP socket option processing. */ diff --git a/sys/netinet/ip_var.h b/sys/netinet/ip_var.h index 3220679d749f..a1402f4fa268 100644 --- a/sys/netinet/ip_var.h +++ b/sys/netinet/ip_var.h @@ -259,6 +259,7 @@ VNET_DECLARE(struct pfil_head *, inet_local_pfil_head); #define PFIL_INET_LOCAL_NAME "inet-local" void in_delayed_cksum(struct mbuf *m); +void in_delayed_cksum_o(struct mbuf *m, uint16_t o); /* Hooks for ipfw, dummynet, divert etc. Most are declared in raw_ip.c */ /* diff --git a/sys/netinet/tcp_subr.c b/sys/netinet/tcp_subr.c index 6829d2a743f2..81b378f496c9 100644 --- a/sys/netinet/tcp_subr.c +++ b/sys/netinet/tcp_subr.c @@ -654,7 +654,7 @@ tcp_recv_udp_tunneled_packet(struct mbuf *m, int off, struct inpcb *inp, } } m->m_pkthdr.tcp_tun_port = port = uh->uh_sport; - bcopy(th, uh, m->m_len - off); + bcopy(th, uh, m->m_len - off - sizeof(struct udphdr)); m->m_len -= sizeof(struct udphdr); m->m_pkthdr.len -= sizeof(struct udphdr); /* diff --git a/sys/netinet/tcp_syncache.c b/sys/netinet/tcp_syncache.c index 077c3a5f6ef3..713f6a35ad45 100644 --- a/sys/netinet/tcp_syncache.c +++ b/sys/netinet/tcp_syncache.c @@ -1175,7 +1175,7 @@ syncache_expand(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, /* * If listening socket requested TCP digests, check that * received ACK has signature and it is correct. - * If not, drop the ACK and leave sc entry in th cache, + * If not, drop the ACK and leave sc entry in the cache, * because SYN was received with correct signature. */ if (sc->sc_flags & SCF_SIGNATURE) { @@ -1387,6 +1387,7 @@ syncache_add(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, struct tcpcb *tp; struct socket *rv = NULL; struct syncache *sc = NULL; + struct ucred *cred; struct syncache_head *sch; struct mbuf *ipopts = NULL; u_int ltflags; @@ -1415,6 +1416,7 @@ syncache_add(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, */ KASSERT(SOLISTENING(so), ("%s: %p not listening", __func__, so)); tp = sototcpcb(so); + cred = V_tcp_syncache.see_other ? NULL : crhold(so->so_cred); #ifdef INET6 if (inc->inc_flags & INC_ISIPV6) { @@ -1643,16 +1645,16 @@ syncache_add(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, /* * sc_cred is only used in syncache_pcblist() to list TCP endpoints in * TCPS_SYN_RECEIVED state when V_tcp_syncache.see_other is false. - * Therefore, store the credentials and take a reference count only - * when needed: + * Therefore, store the credentials only when needed: * - sc is allocated from the zone and not using the on stack instance. * - the sysctl variable net.inet.tcp.syncache.see_other is false. * The reference count is decremented when a zone allocated sc is * freed in syncache_free(). */ - if (sc != &scs && !V_tcp_syncache.see_other) - sc->sc_cred = crhold(so->so_cred); - else + if (sc != &scs && !V_tcp_syncache.see_other) { + sc->sc_cred = cred; + cred = NULL; + } else sc->sc_cred = NULL; sc->sc_port = port; sc->sc_ipopts = ipopts; @@ -1790,6 +1792,8 @@ donenoprobe: tcp_fastopen_decrement_counter(tfo_pending); tfo_expanded: + if (cred != NULL) + crfree(cred); if (sc == NULL || sc == &scs) { #ifdef MAC mac_syncache_destroy(&maclabel); diff --git a/sys/netinet/tcp_timer.c b/sys/netinet/tcp_timer.c index 65a9fbc84ff7..1c687e94bb4a 100644 --- a/sys/netinet/tcp_timer.c +++ b/sys/netinet/tcp_timer.c @@ -515,9 +515,12 @@ tcp_timer_persist(struct tcpcb *tp) if (progdrop || (tp->t_rxtshift >= V_tcp_retries && (ticks - tp->t_rcvtime >= tcp_maxpersistidle || ticks - tp->t_rcvtime >= TCP_REXMTVAL(tp) * tcp_totbackoff))) { - if (!progdrop) + if (progdrop) { + tcp_log_end_status(tp, TCP_EI_STATUS_PROGRESS); + } else { TCPSTAT_INC(tcps_persistdrop); - tcp_log_end_status(tp, TCP_EI_STATUS_PERSIST_MAX); + tcp_log_end_status(tp, TCP_EI_STATUS_PERSIST_MAX); + } goto dropit; } /* diff --git a/sys/netinet/udp_usrreq.c b/sys/netinet/udp_usrreq.c index 88df8f8d38fc..a1000dadf583 100644 --- a/sys/netinet/udp_usrreq.c +++ b/sys/netinet/udp_usrreq.c @@ -852,6 +852,11 @@ SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, udp_pcblist, "S,xinpcb", "List of active UDP sockets"); +SYSCTL_PROC(_net_inet_udplite, OID_AUTO, pcblist, + CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, IPPROTO_UDPLITE, + udp_pcblist, "S,xinpcb", + "List of active UDP-Lite sockets"); + #ifdef INET static int udp_getcred(SYSCTL_HANDLER_ARGS) @@ -1141,7 +1146,19 @@ udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, else INP_RLOCK(inp); NET_EPOCH_ENTER(et); +#ifdef INET6 + if ((flags & PRUS_IPV6) != 0) { + if ((inp->in6p_outputopts != NULL) && + (inp->in6p_outputopts->ip6po_tclass != -1)) + tos = (u_char)inp->in6p_outputopts->ip6po_tclass; + else + tos = 0; + } else { + tos = inp->inp_ip_tos; + } +#else tos = inp->inp_ip_tos; +#endif if (control != NULL) { /* * XXX: Currently, we assume all the optional information is @@ -1165,6 +1182,23 @@ udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, error = udp_v4mapped_pktinfo(cm, &src, inp, flags); if (error != 0) break; + if (((flags & PRUS_IPV6) != 0) && + (cm->cmsg_level == IPPROTO_IPV6) && + (cm->cmsg_type == IPV6_TCLASS)) { + int tclass; + + if (cm->cmsg_len != CMSG_LEN(sizeof(int))) { + error = EINVAL; + break; + } + tclass = *(int *)CMSG_DATA(cm); + if (tclass < -1 || tclass > 255) { + error = EINVAL; + break; + } + if (tclass != -1) + tos = (u_char)tclass; + } #endif if (cm->cmsg_level != IPPROTO_IP) continue; diff --git a/sys/netinet/udp_var.h b/sys/netinet/udp_var.h index 51272e7c9349..569b7b8d5240 100644 --- a/sys/netinet/udp_var.h +++ b/sys/netinet/udp_var.h @@ -141,6 +141,7 @@ void kmod_udpstat_inc(int statnum); kmod_udpstat_inc(offsetof(struct udpstat, name) / sizeof(uint64_t)) SYSCTL_DECL(_net_inet_udp); +SYSCTL_DECL(_net_inet_udplite); VNET_DECLARE(struct inpcbinfo, udbinfo); VNET_DECLARE(struct inpcbinfo, ulitecbinfo); diff --git a/sys/netpfil/ipfw/pmod/tcpmod.c b/sys/netpfil/ipfw/pmod/tcpmod.c index d7400f4fd5ed..138fefe22a98 100644 --- a/sys/netpfil/ipfw/pmod/tcpmod.c +++ b/sys/netpfil/ipfw/pmod/tcpmod.c @@ -58,7 +58,8 @@ VNET_DEFINE_STATIC(uint16_t, tcpmod_setmss_eid) = 0; #define V_tcpmod_setmss_eid VNET(tcpmod_setmss_eid) static int -tcpmod_setmss(struct mbuf **mp, struct tcphdr *tcp, int tlen, uint16_t mss) +tcpmod_setmss(struct mbuf **mp, struct tcphdr *tcp, int tlen, uint16_t mss, + int *done) { struct mbuf *m; u_char *cp; @@ -73,8 +74,10 @@ tcpmod_setmss(struct mbuf **mp, struct tcphdr *tcp, int tlen, uint16_t mss) * TCP header with options. */ *mp = m = m_pullup(m, m->m_pkthdr.len); - if (m == NULL) + if (m == NULL) { + *done = 1; return (ret); + } } /* Parse TCP options. */ for (tlen -= sizeof(struct tcphdr), cp = (u_char *)(tcp + 1); @@ -115,7 +118,7 @@ tcpmod_setmss(struct mbuf **mp, struct tcphdr *tcp, int tlen, uint16_t mss) #ifdef INET6 static int -tcpmod_ipv6_setmss(struct mbuf **mp, uint16_t mss) +tcpmod_ipv6_setmss(struct mbuf **mp, uint16_t mss, int *done) { struct ip6_hdr *ip6; struct ip6_hbh *hbh; @@ -143,13 +146,13 @@ tcpmod_ipv6_setmss(struct mbuf **mp, uint16_t mss) /* We must have TCP options and enough data in a packet. */ if (hlen <= sizeof(struct tcphdr) || hlen > plen) return (IP_FW_DENY); - return (tcpmod_setmss(mp, tcp, hlen, mss)); + return (tcpmod_setmss(mp, tcp, hlen, mss, done)); } #endif /* INET6 */ #ifdef INET static int -tcpmod_ipv4_setmss(struct mbuf **mp, uint16_t mss) +tcpmod_ipv4_setmss(struct mbuf **mp, uint16_t mss, int *done) { struct tcphdr *tcp; struct ip *ip; @@ -163,7 +166,7 @@ tcpmod_ipv4_setmss(struct mbuf **mp, uint16_t mss) /* We must have TCP options and enough data in a packet. */ if (hlen <= sizeof(struct tcphdr) || hlen > plen) return (IP_FW_DENY); - return (tcpmod_setmss(mp, tcp, hlen, mss)); + return (tcpmod_setmss(mp, tcp, hlen, mss, done)); } #endif /* INET */ @@ -207,19 +210,23 @@ ipfw_tcpmod(struct ip_fw_chain *chain, struct ip_fw_args *args, switch (args->f_id.addr_type) { #ifdef INET case 4: - ret = tcpmod_ipv4_setmss(&args->m, htons(icmd->arg1)); + ret = tcpmod_ipv4_setmss(&args->m, htons(icmd->arg1), + done); break; #endif #ifdef INET6 case 6: - ret = tcpmod_ipv6_setmss(&args->m, htons(icmd->arg1)); + ret = tcpmod_ipv6_setmss(&args->m, htons(icmd->arg1), + done); break; #endif } /* * We return zero in both @ret and @done on success, and ipfw_chk() * will update rule counters. Otherwise a packet will not be matched - * by rule. + * by rule. We passed @done around above in case we hit a fatal error + * somewhere, we'll return non-zero but signal that rule processing + * cannot succeed. */ return (ret); } diff --git a/sys/netpfil/pf/if_pfsync.c b/sys/netpfil/pf/if_pfsync.c index 8c080b472653..2a5cb0612d36 100644 --- a/sys/netpfil/pf/if_pfsync.c +++ b/sys/netpfil/pf/if_pfsync.c @@ -530,6 +530,9 @@ pfsync_state_import(union pfsync_state_union *sp, int flags, int msg_version) PF_RULES_RASSERT(); + if (strnlen(sp->pfs_1301.ifname, IFNAMSIZ) == IFNAMSIZ) + return (EINVAL); + if (sp->pfs_1301.creatorid == 0) { if (V_pf_status.debug >= PF_DEBUG_MISC) printf("%s: invalid creator id: %08x\n", __func__, diff --git a/sys/netpfil/pf/pf_ioctl.c b/sys/netpfil/pf/pf_ioctl.c index d95f36d06ee3..cf53ea638095 100644 --- a/sys/netpfil/pf/pf_ioctl.c +++ b/sys/netpfil/pf/pf_ioctl.c @@ -4645,6 +4645,17 @@ DIOCCHANGEADDR_error: error = ENODEV; break; } + if (strnlen(io->pfrio_table.pfrt_anchor, MAXPATHLEN) + == MAXPATHLEN) { + error = EINVAL; + goto fail; + } + if (strnlen(io->pfrio_table.pfrt_name, PF_TABLE_NAME_SIZE) + == PF_TABLE_NAME_SIZE) { + error = EINVAL; + goto fail; + } + PF_RULES_WLOCK(); error = pfr_clr_tables(&io->pfrio_table, &io->pfrio_ndel, io->pfrio_flags | PFR_FLAG_USERIOCTL); diff --git a/sys/x86/x86/local_apic.c b/sys/x86/x86/local_apic.c index 6a913883cc5c..bf05855439bc 100644 --- a/sys/x86/x86/local_apic.c +++ b/sys/x86/x86/local_apic.c @@ -371,9 +371,12 @@ lvt_mode_impl(struct lapic *la, struct lvt *lvt, u_int pin, uint32_t value) case APIC_LVT_DM_SMI: case APIC_LVT_DM_INIT: case APIC_LVT_DM_EXTINT: - if (!lvt->lvt_edgetrigger && bootverbose) { - printf("lapic%u: Forcing LINT%u to edge trigger\n", - la->la_id, pin); + if (!lvt->lvt_edgetrigger) { + if (bootverbose) { + printf( + "lapic%u: Forcing LINT%u to edge trigger\n", + la->la_id, pin); + } value &= ~APIC_LVT_TM; } /* Use a vector of 0. */ |