diff options
author | Sebastian Huber <sebastian.huber@embedded-brains.de> | 2018-12-20 11:12:40 +0100 |
---|---|---|
committer | Sebastian Huber <sebastian.huber@embedded-brains.de> | 2018-12-20 13:36:34 +0100 |
commit | 2b2563da953978f63e3e707f758fd600dcd19a32 (patch) | |
tree | a207b096c10788192b56025e8187f14d1b5a978d /freebsd/sys | |
parent | freebsd/if_cpsw: Port. (diff) | |
download | rtems-libbsd-2b2563da953978f63e3e707f758fd600dcd19a32.tar.bz2 |
Update to FreeBSD head 2018-12-20
Git mirror commit 19a6ceb89dbacf74697d493e48c388767126d418.
It includes an update of wpa_supplicant to version 2.7.
It includes an update of the OpenSSL baseline to version 1.1.1a.
Update #3472.
Diffstat (limited to '')
106 files changed, 1286 insertions, 5489 deletions
diff --git a/freebsd/sys/cam/scsi/scsi_all.c b/freebsd/sys/cam/scsi/scsi_all.c index 1a469f32..0be7e692 100644 --- a/freebsd/sys/cam/scsi/scsi_all.c +++ b/freebsd/sys/cam/scsi/scsi_all.c @@ -1168,7 +1168,7 @@ static struct asc_table_entry asc_table[] = { { SST(0x04, 0x1B, SS_RDEF, /* XXX TBD */ "Logical unit not ready, sanitize in progress") }, /* DT MAEB */ - { SST(0x04, 0x1C, SS_RDEF, /* XXX TBD */ + { SST(0x04, 0x1C, SS_START | SSQ_DECREMENT_COUNT | ENXIO, "Logical unit not ready, additional power use not yet granted") }, /* D */ { SST(0x04, 0x1D, SS_RDEF, /* XXX TBD */ @@ -8414,6 +8414,38 @@ scsi_ata_read_log(struct ccb_scsiio *csio, uint32_t retries, return (retval); } +int scsi_ata_setfeatures(struct ccb_scsiio *csio, uint32_t retries, + void (*cbfcnp)(struct cam_periph *, union ccb *), + uint8_t tag_action, uint8_t feature, + uint64_t lba, uint32_t count, + uint8_t sense_len, uint32_t timeout) +{ + return (scsi_ata_pass(csio, + retries, + cbfcnp, + /*flags*/CAM_DIR_NONE, + tag_action, + /*protocol*/AP_PROTO_PIO_IN, + /*ata_flags*/AP_FLAG_TDIR_FROM_DEV | + AP_FLAG_BYT_BLOK_BYTES | + AP_FLAG_TLEN_SECT_CNT, + /*features*/feature, + /*sector_count*/count, + /*lba*/lba, + /*command*/ATA_SETFEATURES, + /*device*/ 0, + /*icc*/ 0, + /*auxiliary*/0, + /*control*/0, + /*data_ptr*/NULL, + /*dxfer_len*/0, + /*cdb_storage*/NULL, + /*cdb_storage_len*/0, + /*minimum_cmd_size*/0, + sense_len, + timeout)); +} + /* * Note! This is an unusual CDB building function because it can return * an error in the event that the command in question requires a variable diff --git a/freebsd/sys/cam/scsi/scsi_all.h b/freebsd/sys/cam/scsi/scsi_all.h index 80f8a221..b5c45bc0 100644 --- a/freebsd/sys/cam/scsi/scsi_all.h +++ b/freebsd/sys/cam/scsi/scsi_all.h @@ -4176,6 +4176,12 @@ int scsi_ata_read_log(struct ccb_scsiio *csio, uint32_t retries, uint8_t protocol, uint8_t *data_ptr, uint32_t dxfer_len, uint8_t sense_len, uint32_t timeout); +int scsi_ata_setfeatures(struct ccb_scsiio *csio, uint32_t retries, + void (*cbfcnp)(struct cam_periph *, union ccb *), + uint8_t tag_action, uint8_t feature, + uint64_t lba, uint32_t count, + uint8_t sense_len, uint32_t timeout); + int scsi_ata_pass(struct ccb_scsiio *csio, uint32_t retries, void (*cbfcnp)(struct cam_periph *, union ccb *), uint32_t flags, uint8_t tag_action, diff --git a/freebsd/sys/contrib/ck/include/gcc/sparcv9/ck_pr.h b/freebsd/sys/contrib/ck/include/gcc/sparcv9/ck_pr.h index 7dc71725..b60e199d 100644 --- a/freebsd/sys/contrib/ck/include/gcc/sparcv9/ck_pr.h +++ b/freebsd/sys/contrib/ck/include/gcc/sparcv9/ck_pr.h @@ -136,11 +136,26 @@ CK_PR_STORE_S(int, int, "stsw") #undef CK_PR_STORE_S #undef CK_PR_STORE +/* Use the appropriate address space for atomics within the FreeBSD kernel. */ +#if defined(__FreeBSD__) && defined(_KERNEL) +#include <sys/cdefs.h> +#include <machine/atomic.h> +#define CK_PR_INS_CAS "casa" +#define CK_PR_INS_CASX "casxa" +#define CK_PR_INS_SWAP "swapa" +#define CK_PR_ASI_ATOMIC __XSTRING(__ASI_ATOMIC) +#else +#define CK_PR_INS_CAS "cas" +#define CK_PR_INS_CASX "casx" +#define CK_PR_INS_SWAP "swap" +#define CK_PR_ASI_ATOMIC "" +#endif + CK_CC_INLINE static bool ck_pr_cas_64_value(uint64_t *target, uint64_t compare, uint64_t set, uint64_t *value) { - __asm__ __volatile__("casx [%1], %2, %0" + __asm__ __volatile__(CK_PR_INS_CASX " [%1] " CK_PR_ASI_ATOMIC ", %2, %0" : "+&r" (set) : "r" (target), "r" (compare) @@ -154,7 +169,7 @@ CK_CC_INLINE static bool ck_pr_cas_64(uint64_t *target, uint64_t compare, uint64_t set) { - __asm__ __volatile__("casx [%1], %2, %0" + __asm__ __volatile__(CK_PR_INS_CASX " [%1] " CK_PR_ASI_ATOMIC ", %2, %0" : "+&r" (set) : "r" (target), "r" (compare) @@ -181,7 +196,7 @@ ck_pr_cas_ptr_value(void *target, void *compare, void *set, void *previous) CK_CC_INLINE static bool \ ck_pr_cas_##N##_value(T *target, T compare, T set, T *value) \ { \ - __asm__ __volatile__("cas [%1], %2, %0" \ + __asm__ __volatile__(CK_PR_INS_CAS " [%1] " CK_PR_ASI_ATOMIC ", %2, %0" \ : "+&r" (set) \ : "r" (target), \ "r" (compare) \ @@ -192,7 +207,7 @@ ck_pr_cas_ptr_value(void *target, void *compare, void *set, void *previous) CK_CC_INLINE static bool \ ck_pr_cas_##N(T *target, T compare, T set) \ { \ - __asm__ __volatile__("cas [%1], %2, %0" \ + __asm__ __volatile__(CK_PR_INS_CAS " [%1] " CK_PR_ASI_ATOMIC ", %2, %0" \ : "+&r" (set) \ : "r" (target), \ "r" (compare) \ @@ -211,7 +226,7 @@ CK_PR_CAS(int, int) ck_pr_fas_##N(T *target, T update) \ { \ \ - __asm__ __volatile__("swap [%1], %0" \ + __asm__ __volatile__(CK_PR_INS_SWAP " [%1] " CK_PR_ASI_ATOMIC ", %0" \ : "+&r" (update) \ : "r" (target) \ : "memory"); \ @@ -224,5 +239,10 @@ CK_PR_FAS(32, uint32_t) #undef CK_PR_FAS +#undef CK_PR_INS_CAS +#undef CK_PR_INS_CASX +#undef CK_PR_INS_SWAP +#undef CK_PR_ASI_ATOMIC + #endif /* CK_PR_SPARCV9_H */ diff --git a/freebsd/sys/dev/evdev/evdev_utils.c b/freebsd/sys/dev/evdev/evdev_utils.c index 17a815fb..9bdc5a23 100644 --- a/freebsd/sys/dev/evdev/evdev_utils.c +++ b/freebsd/sys/dev/evdev/evdev_utils.c @@ -252,12 +252,15 @@ evdev_scancode2key(int *state, int scancode) */ *state = 0; if ((scancode & 0x7f) == 0x1D) - *state = 0x1D; + *state = scancode; return (NONE); /* NOT REACHED */ case 0x1D: /* pause / break */ + case 0x9D: + if ((*state ^ scancode) & 0x80) + return (NONE); *state = 0; - if (scancode != 0x45) + if ((scancode & 0x7f) != 0x45) return (NONE); keycode = KEY_PAUSE; break; diff --git a/freebsd/sys/dev/fdt/simplebus.c b/freebsd/sys/dev/fdt/simplebus.c index 54b17074..583590d4 100644 --- a/freebsd/sys/dev/fdt/simplebus.c +++ b/freebsd/sys/dev/fdt/simplebus.c @@ -64,13 +64,6 @@ static const struct ofw_bus_devinfo *simplebus_get_devinfo(device_t bus, device_t child); /* - * local methods - */ - -static int simplebus_fill_ranges(phandle_t node, - struct simplebus_softc *sc); - -/* * Driver methods. */ static device_method_t simplebus_methods[] = { @@ -186,7 +179,7 @@ simplebus_init(device_t dev, phandle_t node) OF_getencprop(node, "#size-cells", &sc->scells, sizeof(sc->scells)); } -static int +int simplebus_fill_ranges(phandle_t node, struct simplebus_softc *sc) { int host_address_cells; diff --git a/freebsd/sys/dev/fdt/simplebus.h b/freebsd/sys/dev/fdt/simplebus.h index caccf162..bc695578 100644 --- a/freebsd/sys/dev/fdt/simplebus.h +++ b/freebsd/sys/dev/fdt/simplebus.h @@ -61,4 +61,6 @@ device_t simplebus_add_device(device_t dev, phandle_t node, u_int order, const char *name, int unit, struct simplebus_devinfo *di); struct simplebus_devinfo *simplebus_setup_dinfo(device_t dev, phandle_t node, struct simplebus_devinfo *di); +int simplebus_fill_ranges(phandle_t node, + struct simplebus_softc *sc); #endif /* _FDT_SIMPLEBUS_H */ diff --git a/freebsd/sys/dev/fxp/if_fxp.c b/freebsd/sys/dev/fxp/if_fxp.c index e9c8721e..b781905b 100644 --- a/freebsd/sys/dev/fxp/if_fxp.c +++ b/freebsd/sys/dev/fxp/if_fxp.c @@ -1629,7 +1629,7 @@ fxp_encap(struct fxp_softc *sc, struct mbuf **m_head) cbp->tbd_number = nseg; /* Configure TSO. */ if (m->m_pkthdr.csum_flags & CSUM_TSO) { - cbp->tbd[-1].tb_size = htole32(m->m_pkthdr.tso_segsz << 16); + cbp->tbdtso.tb_size = htole32(m->m_pkthdr.tso_segsz << 16); cbp->tbd[1].tb_size |= htole32(tcp_payload << 16); cbp->ipcb_ip_schedule |= FXP_IPCB_LARGESEND_ENABLE | FXP_IPCB_IP_CHECKSUM_ENABLE | diff --git a/freebsd/sys/dev/fxp/if_fxpreg.h b/freebsd/sys/dev/fxp/if_fxpreg.h index d1e6a45a..6b5f3d8b 100644 --- a/freebsd/sys/dev/fxp/if_fxpreg.h +++ b/freebsd/sys/dev/fxp/if_fxpreg.h @@ -281,10 +281,15 @@ struct fxp_cb_tx { uint16_t cb_status; uint16_t cb_command; uint32_t link_addr; - uint32_t tbd_array_addr; - uint16_t byte_count; - uint8_t tx_threshold; - uint8_t tbd_number; + union { + struct { + uint32_t tbd_array_addr; + uint16_t byte_count; + uint8_t tx_threshold; + uint8_t tbd_number; + }; + struct fxp_tbd tbdtso; + }; /* * The following structure isn't actually part of the TxCB, diff --git a/freebsd/sys/dev/mmc/mmc.c b/freebsd/sys/dev/mmc/mmc.c index cfbce2a5..4c8aefcf 100644 --- a/freebsd/sys/dev/mmc/mmc.c +++ b/freebsd/sys/dev/mmc/mmc.c @@ -844,9 +844,14 @@ mmc_set_power_class(struct mmc_softc *sc, struct mmc_ivars *ivar) const uint8_t *ext_csd; uint32_t clock; uint8_t value; + enum mmc_bus_timing timing; + enum mmc_bus_width bus_width; dev = sc->dev; - if (mmcbr_get_mode(dev) != mode_mmc || ivar->csd.spec_vers < 4) + timing = mmcbr_get_timing(dev); + bus_width = ivar->bus_width; + if (mmcbr_get_mode(dev) != mode_mmc || ivar->csd.spec_vers < 4 || + timing == bus_timing_normal || bus_width == bus_width_1) return (MMC_ERR_NONE); value = 0; @@ -857,8 +862,8 @@ mmc_set_power_class(struct mmc_softc *sc, struct mmc_ivars *ivar) if (clock <= MMC_TYPE_HS_26_MAX) value = ext_csd[EXT_CSD_PWR_CL_26_195]; else if (clock <= MMC_TYPE_HS_52_MAX) { - if (mmcbr_get_timing(dev) >= bus_timing_mmc_ddr52 && - ivar->bus_width >= bus_width_4) + if (timing >= bus_timing_mmc_ddr52 && + bus_width >= bus_width_4) value = ext_csd[EXT_CSD_PWR_CL_52_195_DDR]; else value = ext_csd[EXT_CSD_PWR_CL_52_195]; @@ -877,13 +882,13 @@ mmc_set_power_class(struct mmc_softc *sc, struct mmc_ivars *ivar) if (clock <= MMC_TYPE_HS_26_MAX) value = ext_csd[EXT_CSD_PWR_CL_26_360]; else if (clock <= MMC_TYPE_HS_52_MAX) { - if (mmcbr_get_timing(dev) == bus_timing_mmc_ddr52 && - ivar->bus_width >= bus_width_4) + if (timing == bus_timing_mmc_ddr52 && + bus_width >= bus_width_4) value = ext_csd[EXT_CSD_PWR_CL_52_360_DDR]; else value = ext_csd[EXT_CSD_PWR_CL_52_360]; } else if (clock <= MMC_TYPE_HS200_HS400ES_MAX) { - if (ivar->bus_width == bus_width_8) + if (bus_width == bus_width_8) value = ext_csd[EXT_CSD_PWR_CL_200_360_DDR]; else value = ext_csd[EXT_CSD_PWR_CL_200_360]; @@ -895,7 +900,7 @@ mmc_set_power_class(struct mmc_softc *sc, struct mmc_ivars *ivar) return (MMC_ERR_INVALID); } - if (ivar->bus_width == bus_width_8) + if (bus_width == bus_width_8) value = (value & EXT_CSD_POWER_CLASS_8BIT_MASK) >> EXT_CSD_POWER_CLASS_8BIT_SHIFT; else @@ -2178,7 +2183,7 @@ mmc_calculate_clock(struct mmc_softc *sc) for (i = 0; i < sc->child_count; i++) { ivar = device_get_ivars(sc->child_list[i]); if ((ivar->timings & ~(1 << bus_timing_normal)) == 0) - continue; + goto clock; rca = ivar->rca; if (mmc_select_card(sc, rca) != MMC_ERR_NONE) { @@ -2244,6 +2249,7 @@ mmc_calculate_clock(struct mmc_softc *sc) } } +clock: /* Set clock (must be done before initial tuning). */ mmcbr_set_clock(dev, max_dtr); mmcbr_update_ios(dev); diff --git a/freebsd/sys/dev/nvme/nvme.h b/freebsd/sys/dev/nvme/nvme.h index 747767ce..845ba75b 100644 --- a/freebsd/sys/dev/nvme/nvme.h +++ b/freebsd/sys/dev/nvme/nvme.h @@ -1259,6 +1259,13 @@ void nvme_unregister_consumer(struct nvme_consumer *consumer); device_t nvme_ctrlr_get_device(struct nvme_controller *ctrlr); const struct nvme_controller_data * nvme_ctrlr_get_data(struct nvme_controller *ctrlr); +static inline bool +nvme_ctrlr_has_dataset_mgmt(const struct nvme_controller_data *cd) +{ + /* Assumes cd was byte swapped by nvme_controller_data_swapbytes() */ + return ((cd->oncs >> NVME_CTRLR_DATA_ONCS_DSM_SHIFT) & + NVME_CTRLR_DATA_ONCS_DSM_MASK); +} /* Namespace helper functions */ uint32_t nvme_ns_get_max_io_xfer_size(struct nvme_namespace *ns); diff --git a/freebsd/sys/dev/pci/pci.c b/freebsd/sys/dev/pci/pci.c index 512e8636..2c977b85 100644 --- a/freebsd/sys/dev/pci/pci.c +++ b/freebsd/sys/dev/pci/pci.c @@ -216,7 +216,8 @@ static device_method_t pci_methods[] = { DEFINE_CLASS_0(pci, pci_driver, pci_methods, sizeof(struct pci_softc)); static devclass_t pci_devclass; -DRIVER_MODULE(pci, pcib, pci_driver, pci_devclass, pci_modevent, NULL); +EARLY_DRIVER_MODULE(pci, pcib, pci_driver, pci_devclass, pci_modevent, NULL, + BUS_PASS_BUS); MODULE_VERSION(pci, 1); static char *pci_vendordata; @@ -231,6 +232,7 @@ struct pci_quirk { #define PCI_QUIRK_UNMAP_REG 4 /* Ignore PCI map register */ #define PCI_QUIRK_DISABLE_MSIX 5 /* MSI-X doesn't work */ #define PCI_QUIRK_MSI_INTX_BUG 6 /* PCIM_CMD_INTxDIS disables MSI */ +#define PCI_QUIRK_REALLOC_BAR 7 /* Can't allocate memory at the default address */ int arg1; int arg2; }; @@ -312,6 +314,12 @@ static const struct pci_quirk pci_quirks[] = { { 0x167814e4, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, /* BCM5715 */ { 0x167914e4, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, /* BCM5715S */ + /* + * HPE Gen 10 VGA has a memory range that can't be allocated in the + * expected place. + */ + { 0x98741002, PCI_QUIRK_REALLOC_BAR, 0, 0 }, + { 0 } }; @@ -3286,7 +3294,9 @@ pci_add_map(device_t bus, device_t dev, int reg, struct resource_list *rl, */ res = resource_list_reserve(rl, bus, dev, type, ®, start, end, count, flags); - if (pci_do_realloc_bars && res == NULL && (start != 0 || end != ~0)) { + if ((pci_do_realloc_bars + || pci_has_quirk(pci_get_devid(dev), PCI_QUIRK_REALLOC_BAR)) + && res == NULL && (start != 0 || end != ~0)) { /* * If the allocation fails, try to allocate a resource for * this BAR using any available range. The firmware felt @@ -4467,6 +4477,7 @@ int pci_suspend_child(device_t dev, device_t child) { struct pci_devinfo *dinfo; + struct resource_list_entry *rle; int error; dinfo = device_get_ivars(child); @@ -4483,8 +4494,20 @@ pci_suspend_child(device_t dev, device_t child) if (error) return (error); - if (pci_do_power_suspend) + if (pci_do_power_suspend) { + /* + * Make sure this device's interrupt handler is not invoked + * in the case the device uses a shared interrupt that can + * be raised by some other device. + * This is applicable only to regular (legacy) PCI interrupts + * as MSI/MSI-X interrupts are never shared. + */ + rle = resource_list_find(&dinfo->resources, + SYS_RES_IRQ, 0); + if (rle != NULL && rle->res != NULL) + (void)bus_suspend_intr(child, rle->res); pci_set_power_child(dev, child, PCI_POWERSTATE_D3); + } return (0); } @@ -4493,6 +4516,7 @@ int pci_resume_child(device_t dev, device_t child) { struct pci_devinfo *dinfo; + struct resource_list_entry *rle; if (pci_do_power_resume) pci_set_power_child(dev, child, PCI_POWERSTATE_D0); @@ -4504,6 +4528,16 @@ pci_resume_child(device_t dev, device_t child) bus_generic_resume_child(dev, child); + /* + * Allow interrupts only after fully resuming the driver and hardware. + */ + if (pci_do_power_suspend) { + /* See pci_suspend_child for details. */ + rle = resource_list_find(&dinfo->resources, SYS_RES_IRQ, 0); + if (rle != NULL && rle->res != NULL) + (void)bus_resume_intr(child, rle->res); + } + return (0); } diff --git a/freebsd/sys/dev/pci/pci_pci.c b/freebsd/sys/dev/pci/pci_pci.c index ec52e1fb..607a0614 100644 --- a/freebsd/sys/dev/pci/pci_pci.c +++ b/freebsd/sys/dev/pci/pci_pci.c @@ -133,7 +133,8 @@ static device_method_t pcib_methods[] = { static devclass_t pcib_devclass; DEFINE_CLASS_0(pcib, pcib_driver, pcib_methods, sizeof(struct pcib_softc)); -DRIVER_MODULE(pcib, pci, pcib_driver, pcib_devclass, NULL, NULL); +EARLY_DRIVER_MODULE(pcib, pci, pcib_driver, pcib_devclass, NULL, NULL, + BUS_PASS_BUS); #if defined(NEW_PCIB) || defined(PCI_HP) SYSCTL_DECL(_hw_pci); diff --git a/freebsd/sys/dev/pci/pci_user.c b/freebsd/sys/dev/pci/pci_user.c index 27fb3475..3e2c3c7e 100644 --- a/freebsd/sys/dev/pci/pci_user.c +++ b/freebsd/sys/dev/pci/pci_user.c @@ -951,6 +951,9 @@ pci_ioctl(struct cdev *dev, u_long cmd, caddr_t data, int flag, struct thread *t if (!(flag & FWRITE)) { switch (cmd) { case PCIOCGETCONF: +#ifdef COMPAT_FREEBSD32 + case PCIOCGETCONF32: +#endif #ifdef PRE7_COMPAT case PCIOCGETCONF_OLD: #ifdef COMPAT_FREEBSD32 @@ -968,6 +971,9 @@ pci_ioctl(struct cdev *dev, u_long cmd, caddr_t data, int flag, struct thread *t switch (cmd) { case PCIOCGETCONF: +#ifdef COMPAT_FREEBSD32 + case PCIOCGETCONF32: +#endif #ifdef PRE7_COMPAT case PCIOCGETCONF_OLD: #ifdef COMPAT_FREEBSD32 diff --git a/freebsd/sys/dev/rtwn/usb/rtwn_usb_attach.h b/freebsd/sys/dev/rtwn/usb/rtwn_usb_attach.h index e30ae6be..432d7e97 100644 --- a/freebsd/sys/dev/rtwn/usb/rtwn_usb_attach.h +++ b/freebsd/sys/dev/rtwn/usb/rtwn_usb_attach.h @@ -118,6 +118,7 @@ static const STRUCT_USB_HOST_ID rtwn_devs[] = { RTWN_RTL8188EU_DEV(DLINK, DWA123D1), RTWN_RTL8188EU_DEV(DLINK, DWA125D1), RTWN_RTL8188EU_DEV(ELECOM, WDC150SU2M), + RTWN_RTL8188EU_DEV(TPLINK, WN722NV2), RTWN_RTL8188EU_DEV(REALTEK, RTL8188ETV), RTWN_RTL8188EU_DEV(REALTEK, RTL8188EU), #undef RTWN_RTL8188EU_DEV @@ -139,6 +140,7 @@ static const STRUCT_USB_HOST_ID rtwn_devs[] = { RTWN_RTL8812AU_DEV(SITECOMEU, WLA7100), RTWN_RTL8812AU_DEV(TPLINK, T4U), RTWN_RTL8812AU_DEV(TPLINK, T4UV2), + RTWN_RTL8812AU_DEV(TPLINK, T4UHV1), RTWN_RTL8812AU_DEV(TPLINK, T4UHV2), RTWN_RTL8812AU_DEV(TRENDNET, TEW805UB), RTWN_RTL8812AU_DEV(ZYXEL, NWD6605), diff --git a/freebsd/sys/dev/sdhci/sdhci.c b/freebsd/sys/dev/sdhci/sdhci.c index ccbf5d85..0bb9edc7 100644 --- a/freebsd/sys/dev/sdhci/sdhci.c +++ b/freebsd/sys/dev/sdhci/sdhci.c @@ -900,6 +900,9 @@ sdhci_init_slot(device_t dev, struct sdhci_slot *slot, int num) if (slot->quirks & SDHCI_QUIRK_CAPS_BIT63_FOR_MMC_HS400 && caps2 & SDHCI_CAN_MMC_HS400) host_caps |= MMC_CAP_MMC_HS400; + if (slot->quirks & SDHCI_QUIRK_MMC_HS400_IF_CAN_SDR104 && + caps2 & SDHCI_CAN_SDR104) + host_caps |= MMC_CAP_MMC_HS400; /* * Disable UHS-I and eMMC modes if the set_uhs_timing method is the diff --git a/freebsd/sys/dev/sdhci/sdhci.h b/freebsd/sys/dev/sdhci/sdhci.h index f2a7a9b1..a22e0235 100644 --- a/freebsd/sys/dev/sdhci/sdhci.h +++ b/freebsd/sys/dev/sdhci/sdhci.h @@ -93,6 +93,8 @@ #define SDHCI_QUIRK_PRESET_VALUE_BROKEN (1 << 27) /* Controller does not support or the support for ACMD12 is broken. */ #define SDHCI_QUIRK_BROKEN_AUTO_STOP (1 << 28) +/* Controller supports eMMC HS400 mode if SDHCI_CAN_SDR104 is set. */ +#define SDHCI_QUIRK_MMC_HS400_IF_CAN_SDR104 (1 << 29) /* * Controller registers diff --git a/freebsd/sys/dev/usb/wlan/if_rsu.c b/freebsd/sys/dev/usb/wlan/if_rsu.c index 1cb504cc..b730ce59 100644 --- a/freebsd/sys/dev/usb/wlan/if_rsu.c +++ b/freebsd/sys/dev/usb/wlan/if_rsu.c @@ -119,6 +119,7 @@ static const STRUCT_USB_HOST_ID rsu_devs[] = { RSU_HT_NOT_SUPPORTED) } RSU_DEV(ASUS, RTL8192SU), RSU_DEV(AZUREWAVE, RTL8192SU_4), + RSU_DEV(SITECOMEU, WLA1000), RSU_DEV_HT(ACCTON, RTL8192SU), RSU_DEV_HT(ASUS, USBN10), RSU_DEV_HT(AZUREWAVE, RTL8192SU_1), diff --git a/freebsd/sys/kern/init_main.c b/freebsd/sys/kern/init_main.c index 8fcc314b..c6a9e310 100644 --- a/freebsd/sys/kern/init_main.c +++ b/freebsd/sys/kern/init_main.c @@ -58,6 +58,7 @@ __FBSDID("$FreeBSD$"); #include <sys/exec.h> #include <sys/file.h> #include <sys/filedesc.h> +#include <sys/imgact.h> #include <sys/jail.h> #include <sys/ktr.h> #include <sys/lock.h> @@ -428,7 +429,6 @@ null_set_syscall_retval(struct thread *td __unused, int error __unused) struct sysentvec null_sysvec = { .sv_size = 0, .sv_table = NULL, - .sv_mask = 0, .sv_errsize = 0, .sv_errtbl = NULL, .sv_transtrap = NULL, @@ -744,15 +744,13 @@ SYSCTL_INT(_kern, OID_AUTO, init_shutdown_timeout, static void start_init(void *dummy) { - vm_offset_t addr; - struct execve_args args; - int options, error; - size_t pathlen; + struct image_args args; + int error; char *var, *path; char *free_init_path, *tmp_init_path; - char *ucp, **uap, *arg0, *arg1; struct thread *td; struct proc *p; + struct vmspace *oldvmspace; TSENTER(); /* Here so we don't overlap with mi_startup. */ @@ -764,16 +762,6 @@ start_init(void *dummy) /* Wipe GELI passphrase from the environment. */ kern_unsetenv("kern.geom.eli.passphrase"); - /* - * Need just enough stack to hold the faked-up "execve()" arguments. - */ - addr = p->p_sysent->sv_usrstack - PAGE_SIZE; - if (vm_map_find(&p->p_vmspace->vm_map, NULL, 0, &addr, PAGE_SIZE, 0, - VMFS_NO_SPACE, VM_PROT_ALL, VM_PROT_ALL, 0) != 0) - panic("init: couldn't allocate argument space"); - p->p_vmspace->vm_maxsaddr = (caddr_t)addr; - p->p_vmspace->vm_ssize = 1; - if ((var = kern_getenv("init_path")) != NULL) { strlcpy(init_path, var, sizeof(init_path)); freeenv(var); @@ -781,58 +769,25 @@ start_init(void *dummy) free_init_path = tmp_init_path = strdup(init_path, M_TEMP); while ((path = strsep(&tmp_init_path, ":")) != NULL) { - pathlen = strlen(path) + 1; if (bootverbose) printf("start_init: trying %s\n", path); - /* - * Move out the boot flag argument. - */ - options = 0; - ucp = (char *)p->p_sysent->sv_usrstack; - (void)subyte(--ucp, 0); /* trailing zero */ - if (boothowto & RB_SINGLE) { - (void)subyte(--ucp, 's'); - options = 1; - } -#ifdef notyet - if (boothowto & RB_FASTBOOT) { - (void)subyte(--ucp, 'f'); - options = 1; - } -#endif - -#ifdef BOOTCDROM - (void)subyte(--ucp, 'C'); - options = 1; -#endif - - if (options == 0) - (void)subyte(--ucp, '-'); - (void)subyte(--ucp, '-'); /* leading hyphen */ - arg1 = ucp; - - /* - * Move out the file name (also arg 0). - */ - ucp -= pathlen; - copyout(path, ucp, pathlen); - arg0 = ucp; - - /* - * Move out the arg pointers. - */ - uap = (char **)rounddown2((intptr_t)ucp, sizeof(intptr_t)); - (void)suword((caddr_t)--uap, (long)0); /* terminator */ - (void)suword((caddr_t)--uap, (long)(intptr_t)arg1); - (void)suword((caddr_t)--uap, (long)(intptr_t)arg0); - - /* - * Point at the arguments. - */ - args.fname = arg0; - args.argv = uap; - args.envv = NULL; + memset(&args, 0, sizeof(args)); + error = exec_alloc_args(&args); + if (error != 0) + panic("%s: Can't allocate space for init arguments %d", + __func__, error); + + error = exec_args_add_fname(&args, path, UIO_SYSSPACE); + if (error != 0) + panic("%s: Can't add fname %d", __func__, error); + error = exec_args_add_arg(&args, path, UIO_SYSSPACE); + if (error != 0) + panic("%s: Can't add argv[0] %d", __func__, error); + if (boothowto & RB_SINGLE) + error = exec_args_add_arg(&args, "-s", UIO_SYSSPACE); + if (error != 0) + panic("%s: Can't add argv[0] %d", __func__, error); /* * Now try to exec the program. If can't for any reason @@ -841,7 +796,19 @@ start_init(void *dummy) * Otherwise, return via fork_trampoline() all the way * to user mode as init! */ - if ((error = sys_execve(td, &args)) == EJUSTRETURN) { + KASSERT((td->td_pflags & TDP_EXECVMSPC) == 0, + ("nested execve")); + oldvmspace = td->td_proc->p_vmspace; + error = kern_execve(td, &args, NULL); + KASSERT(error != 0, + ("kern_execve returned success, not EJUSTRETURN")); + if (error == EJUSTRETURN) { + if ((td->td_pflags & TDP_EXECVMSPC) != 0) { + KASSERT(p->p_vmspace != oldvmspace, + ("oldvmspace still used")); + vmspace_free(oldvmspace); + td->td_pflags &= ~TDP_EXECVMSPC; + } free(free_init_path, M_TEMP); TSEXIT(); return; diff --git a/freebsd/sys/kern/kern_event.c b/freebsd/sys/kern/kern_event.c index 2fae2d89..5c75c657 100644 --- a/freebsd/sys/kern/kern_event.c +++ b/freebsd/sys/kern/kern_event.c @@ -110,13 +110,13 @@ TASKQUEUE_DEFINE_THREAD(kqueue_ctx); static int kevent_copyout(void *arg, struct kevent *kevp, int count); static int kevent_copyin(void *arg, struct kevent *kevp, int count); static int kqueue_register(struct kqueue *kq, struct kevent *kev, - struct thread *td, int waitok); + struct thread *td, int mflag); static int kqueue_acquire(struct file *fp, struct kqueue **kqp); static void kqueue_release(struct kqueue *kq, int locked); static void kqueue_destroy(struct kqueue *kq); static void kqueue_drain(struct kqueue *kq, struct thread *td); static int kqueue_expand(struct kqueue *kq, struct filterops *fops, - uintptr_t ident, int waitok); + uintptr_t ident, int mflag); static void kqueue_task(void *arg, int pending); static int kqueue_scan(struct kqueue *kq, int maxevents, struct kevent_copyops *k_ops, @@ -165,7 +165,7 @@ static void knote_drop_detached(struct knote *kn, struct thread *td); static void knote_enqueue(struct knote *kn); static void knote_dequeue(struct knote *kn); static void knote_init(void); -static struct knote *knote_alloc(int waitok); +static struct knote *knote_alloc(int mflag); static void knote_free(struct knote *kn); static void filt_kqdetach(struct knote *kn); @@ -571,10 +571,12 @@ knote_fork(struct knlist *list, int pid) struct kevent kev; int error; - if (list == NULL) + MPASS(list != NULL); + KNL_ASSERT_LOCKED(list); + if (SLIST_EMPTY(&list->kl_list)) return; - list->kl_lock(list->kl_lockarg); + memset(&kev, 0, sizeof(kev)); SLIST_FOREACH(kn, &list->kl_list, kn_selnext) { kq = kn->kn_kq; KQ_LOCK(kq); @@ -587,10 +589,8 @@ knote_fork(struct knlist *list, int pid) * The same as knote(), activate the event. */ if ((kn->kn_sfflags & NOTE_TRACK) == 0) { - kn->kn_status |= KN_HASKQLOCK; if (kn->kn_fop->f_event(kn, NOTE_FORK)) KNOTE_ACTIVATE(kn, 1); - kn->kn_status &= ~KN_HASKQLOCK; KQ_UNLOCK(kq); continue; } @@ -621,7 +621,7 @@ knote_fork(struct knlist *list, int pid) kev.fflags = kn->kn_sfflags; kev.data = kn->kn_id; /* parent */ kev.udata = kn->kn_kevent.udata;/* preserve udata */ - error = kqueue_register(kq, &kev, NULL, 0); + error = kqueue_register(kq, &kev, NULL, M_NOWAIT); if (error) kn->kn_fflags |= NOTE_TRACKERR; @@ -635,17 +635,16 @@ knote_fork(struct knlist *list, int pid) kev.fflags = kn->kn_sfflags; kev.data = kn->kn_id; /* parent */ kev.udata = kn->kn_kevent.udata;/* preserve udata */ - error = kqueue_register(kq, &kev, NULL, 0); + error = kqueue_register(kq, &kev, NULL, M_NOWAIT); if (error) kn->kn_fflags |= NOTE_TRACKERR; if (kn->kn_fop->f_event(kn, NOTE_FORK)) KNOTE_ACTIVATE(kn, 0); + list->kl_lock(list->kl_lockarg); KQ_LOCK(kq); kn_leave_flux(kn); KQ_UNLOCK_FLUX(kq); - list->kl_lock(list->kl_lockarg); } - list->kl_unlock(list->kl_lockarg); } #endif /* __rtems__ */ @@ -1352,7 +1351,7 @@ kqueue_kevent(struct kqueue *kq, struct thread *td, int nchanges, int nevents, if (!kevp->filter) continue; kevp->flags &= ~EV_SYSFLAGS; - error = kqueue_register(kq, kevp, td, 1); + error = kqueue_register(kq, kevp, td, M_WAITOK); if (error || (kevp->flags & EV_RECEIPT)) { if (nevents == 0) return (error); @@ -1495,12 +1494,11 @@ kqueue_fo_release(int filt) } /* - * A ref to kq (obtained via kqueue_acquire) must be held. waitok will - * influence if memory allocation should wait. Make sure it is 0 if you - * hold any mutexes. + * A ref to kq (obtained via kqueue_acquire) must be held. */ static int -kqueue_register(struct kqueue *kq, struct kevent *kev, struct thread *td, int waitok) +kqueue_register(struct kqueue *kq, struct kevent *kev, struct thread *td, + int mflag) { struct filterops *fops; struct file *fp; @@ -1530,7 +1528,7 @@ kqueue_register(struct kqueue *kq, struct kevent *kev, struct thread *td, int wa * allocation failures are handled in the loop, only * if the spare knote appears to be actually required. */ - tkn = knote_alloc(waitok); + tkn = knote_alloc(mflag); } else { tkn = NULL; } @@ -1546,11 +1544,11 @@ findkn: goto done; if ((kev->flags & EV_ADD) == EV_ADD && kqueue_expand(kq, fops, - kev->ident, 0) != 0) { + kev->ident, M_NOWAIT) != 0) { /* try again */ fdrop(fp, td); fp = NULL; - error = kqueue_expand(kq, fops, kev->ident, waitok); + error = kqueue_expand(kq, fops, kev->ident, mflag); if (error) goto done; goto findkn; @@ -1590,8 +1588,11 @@ findkn: break; } } else { - if ((kev->flags & EV_ADD) == EV_ADD) - kqueue_expand(kq, fops, kev->ident, waitok); + if ((kev->flags & EV_ADD) == EV_ADD) { + error = kqueue_expand(kq, fops, kev->ident, mflag); + if (error != 0) + goto done; + } KQ_LOCK(kq); @@ -1663,6 +1664,8 @@ findkn: kn->kn_kevent.flags &= ~(EV_ADD | EV_DELETE | EV_ENABLE | EV_DISABLE | EV_FORCEONESHOT); kn->kn_status = KN_DETACHED; + if ((kev->flags & EV_DISABLE) != 0) + kn->kn_status |= KN_DISABLED; kn_enter_flux(kn); error = knote_attach(kn, kq); @@ -1698,6 +1701,11 @@ findkn: KNOTE_ACTIVATE(kn, 1); } + if ((kev->flags & EV_ENABLE) != 0) + kn->kn_status &= ~KN_DISABLED; + else if ((kev->flags & EV_DISABLE) != 0) + kn->kn_status |= KN_DISABLED; + /* * The user may change some filter values after the initial EV_ADD, * but doing so will not reset any filter which has already been @@ -1715,19 +1723,17 @@ findkn: kn->kn_sdata = kev->data; } +done_ev_add: /* * We can get here with kn->kn_knlist == NULL. This can happen when * the initial attach event decides that the event is "completed" - * already. i.e. filt_procattach is called on a zombie process. It - * will call filt_proc which will remove it from the list, and NULL + * already, e.g., filt_procattach() is called on a zombie process. It + * will call filt_proc() which will remove it from the list, and NULL * kn_knlist. + * + * KN_DISABLED will be stable while the knote is in flux, so the + * unlocked read will not race with an update. */ -done_ev_add: - if ((kev->flags & EV_ENABLE) != 0) - kn->kn_status &= ~KN_DISABLED; - else if ((kev->flags & EV_DISABLE) != 0) - kn->kn_status |= KN_DISABLED; - if ((kn->kn_status & KN_DISABLED) == 0) event = kn->kn_fop->f_event(kn, 0); else @@ -1815,23 +1821,18 @@ kqueue_schedtask(struct kqueue *kq) * Expand the kq to make sure we have storage for fops/ident pair. * * Return 0 on success (or no work necessary), return errno on failure. - * - * Not calling hashinit w/ waitok (proper malloc flag) should be safe. - * If kqueue_register is called from a non-fd context, there usually/should - * be no locks held. */ static int kqueue_expand(struct kqueue *kq, struct filterops *fops, uintptr_t ident, - int waitok) + int mflag) { struct klist *list, *tmp_knhash, *to_free; u_long tmp_knhashmask; - int size; - int fd; - int mflag = waitok ? M_WAITOK : M_NOWAIT; + int error, fd, size; KQ_NOTOWNED(kq); + error = 0; to_free = NULL; if (fops->f_isfd) { fd = ident; @@ -1843,9 +1844,11 @@ kqueue_expand(struct kqueue *kq, struct filterops *fops, uintptr_t ident, if (list == NULL) return ENOMEM; KQ_LOCK(kq); - if (kq->kq_knlistsize > fd) { + if ((kq->kq_state & KQ_CLOSING) != 0) { + to_free = list; + error = EBADF; + } else if (kq->kq_knlistsize > fd) { to_free = list; - list = NULL; } else { if (kq->kq_knlist != NULL) { bcopy(kq->kq_knlist, list, @@ -1863,12 +1866,16 @@ kqueue_expand(struct kqueue *kq, struct filterops *fops, uintptr_t ident, } } else { if (kq->kq_knhashmask == 0) { - tmp_knhash = hashinit(KN_HASHSIZE, M_KQUEUE, - &tmp_knhashmask); + tmp_knhash = hashinit_flags(KN_HASHSIZE, M_KQUEUE, + &tmp_knhashmask, (mflag & M_WAITOK) != 0 ? + HASH_WAITOK : HASH_NOWAIT); if (tmp_knhash == NULL) - return ENOMEM; + return (ENOMEM); KQ_LOCK(kq); - if (kq->kq_knhashmask == 0) { + if ((kq->kq_state & KQ_CLOSING) != 0) { + to_free = tmp_knhash; + error = EBADF; + } else if (kq->kq_knhashmask == 0) { kq->kq_knhash = tmp_knhash; kq->kq_knhashmask = tmp_knhashmask; } else { @@ -1880,7 +1887,7 @@ kqueue_expand(struct kqueue *kq, struct filterops *fops, uintptr_t ident, free(to_free, M_KQUEUE); KQ_NOTOWNED(kq); - return 0; + return (error); } static void @@ -1950,7 +1957,7 @@ kqueue_scan(struct kqueue *kq, int maxevents, struct kevent_copyops *k_ops, asbt = -1; } else asbt = 0; - marker = knote_alloc(1); + marker = knote_alloc(M_WAITOK); marker->kn_status = KN_MARKER; KQ_LOCK(kq); @@ -2463,10 +2470,8 @@ knote(struct knlist *list, long hint, int lockflags) KNOTE_ACTIVATE(kn, 1); KQ_UNLOCK_FLUX(kq); } else { - kn->kn_status |= KN_HASKQLOCK; if (kn->kn_fop->f_event(kn, hint)) KNOTE_ACTIVATE(kn, 1); - kn->kn_status &= ~KN_HASKQLOCK; KQ_UNLOCK(kq); } } @@ -2807,6 +2812,8 @@ knote_attach(struct knote *kn, struct kqueue *kq) KASSERT(kn_in_flux(kn), ("knote %p not marked influx", kn)); KQ_OWNED(kq); + if ((kq->kq_state & KQ_CLOSING) != 0) + return (EBADF); if (kn->kn_fop->f_isfd) { if (kn->kn_id >= kq->kq_knlistsize) return (ENOMEM); @@ -2902,11 +2909,10 @@ knote_init(void) SYSINIT(knote, SI_SUB_PSEUDO, SI_ORDER_ANY, knote_init, NULL); static struct knote * -knote_alloc(int waitok) +knote_alloc(int mflag) { - return (uma_zalloc(knote_zone, (waitok ? M_WAITOK : M_NOWAIT) | - M_ZERO)); + return (uma_zalloc(knote_zone, mflag | M_ZERO)); } static void @@ -2920,7 +2926,7 @@ knote_free(struct knote *kn) * Register the kev w/ the kq specified by fd. */ int -kqfd_register(int fd, struct kevent *kev, struct thread *td, int waitok) +kqfd_register(int fd, struct kevent *kev, struct thread *td, int mflag) { struct kqueue *kq; struct file *fp; @@ -2933,7 +2939,7 @@ kqfd_register(int fd, struct kevent *kev, struct thread *td, int waitok) if ((error = kqueue_acquire(fp, &kq)) != 0) goto noacquire; - error = kqueue_register(kq, kev, td, waitok); + error = kqueue_register(kq, kev, td, mflag); kqueue_release(kq, 0); noacquire: diff --git a/freebsd/sys/kern/kern_intr.c b/freebsd/sys/kern/kern_intr.c index 04914e93..4e311cb8 100644 --- a/freebsd/sys/kern/kern_intr.c +++ b/freebsd/sys/kern/kern_intr.c @@ -756,6 +756,28 @@ intr_event_barrier(struct intr_event *ie) atomic_thread_fence_acq(); } +static void +intr_handler_barrier(struct intr_handler *handler) +{ + struct intr_event *ie; + + ie = handler->ih_event; + mtx_assert(&ie->ie_lock, MA_OWNED); + KASSERT((handler->ih_flags & IH_DEAD) == 0, + ("update for a removed handler")); + + if (ie->ie_thread == NULL) { + intr_event_barrier(ie); + return; + } + if ((handler->ih_flags & IH_CHANGED) == 0) { + handler->ih_flags |= IH_CHANGED; + intr_event_schedule_thread(ie); + } + while ((handler->ih_flags & IH_CHANGED) != 0) + msleep(handler, &ie->ie_lock, 0, "ih_barr", 0); +} + /* * Sleep until an ithread finishes executing an interrupt handler. * @@ -879,7 +901,50 @@ intr_event_remove_handler(void *cookie) return (0); } +int +intr_event_suspend_handler(void *cookie) +{ + struct intr_handler *handler = (struct intr_handler *)cookie; + struct intr_event *ie; + + if (handler == NULL) + return (EINVAL); + ie = handler->ih_event; + KASSERT(ie != NULL, + ("interrupt handler \"%s\" has a NULL interrupt event", + handler->ih_name)); + mtx_lock(&ie->ie_lock); + handler->ih_flags |= IH_SUSP; + intr_handler_barrier(handler); + mtx_unlock(&ie->ie_lock); + return (0); +} + +int +intr_event_resume_handler(void *cookie) +{ + struct intr_handler *handler = (struct intr_handler *)cookie; + struct intr_event *ie; + + if (handler == NULL) + return (EINVAL); + ie = handler->ih_event; + KASSERT(ie != NULL, + ("interrupt handler \"%s\" has a NULL interrupt event", + handler->ih_name)); + + /* + * intr_handler_barrier() acts not only as a barrier, + * it also allows to check for any pending interrupts. + */ + mtx_lock(&ie->ie_lock); + handler->ih_flags &= ~IH_SUSP; + intr_handler_barrier(handler); + mtx_unlock(&ie->ie_lock); + return (0); +} #endif /* __rtems__ */ + static int intr_event_schedule_thread(struct intr_event *ie) { @@ -1068,10 +1133,21 @@ intr_event_execute_handlers(struct proc *p, struct intr_event *ie) */ ihp = ih; + if ((ih->ih_flags & IH_CHANGED) != 0) { + mtx_lock(&ie->ie_lock); + ih->ih_flags &= ~IH_CHANGED; + wakeup(ih); + mtx_unlock(&ie->ie_lock); + } + /* Skip filter only handlers */ if (ih->ih_handler == NULL) continue; + /* Skip suspended handlers */ + if ((ih->ih_flags & IH_SUSP) != 0) + continue; + /* * For software interrupt threads, we only execute * handlers that have their need flag set. Hardware @@ -1255,8 +1331,9 @@ intr_event_handle(struct intr_event *ie, struct trapframe *frame) struct intr_handler *ih; struct trapframe *oldframe; struct thread *td; - int ret, thread; int phase; + int ret; + bool filter, thread; td = curthread; @@ -1275,7 +1352,8 @@ intr_event_handle(struct intr_event *ie, struct trapframe *frame) * a trapframe as its argument. */ td->td_intr_nesting_level++; - thread = 0; + filter = false; + thread = false; ret = 0; critical_enter(); oldframe = td->td_intr_frame; @@ -1291,8 +1369,10 @@ intr_event_handle(struct intr_event *ie, struct trapframe *frame) atomic_thread_fence_seq_cst(); CK_SLIST_FOREACH(ih, &ie->ie_handlers, ih_next) { + if ((ih->ih_flags & IH_SUSP) != 0) + continue; if (ih->ih_filter == NULL) { - thread = 1; + thread = true; continue; } CTR4(KTR_INTR, "%s: exec %p(%p) for %s", __func__, @@ -1307,24 +1387,25 @@ intr_event_handle(struct intr_event *ie, struct trapframe *frame) (ret & ~(FILTER_SCHEDULE_THREAD | FILTER_HANDLED)) == 0), ("%s: incorrect return value %#x from %s", __func__, ret, ih->ih_name)); + filter = filter || ret == FILTER_HANDLED; - /* + /* * Wrapper handler special handling: * - * in some particular cases (like pccard and pccbb), + * in some particular cases (like pccard and pccbb), * the _real_ device handler is wrapped in a couple of * functions - a filter wrapper and an ithread wrapper. - * In this case (and just in this case), the filter wrapper + * In this case (and just in this case), the filter wrapper * could ask the system to schedule the ithread and mask * the interrupt source if the wrapped handler is composed * of just an ithread handler. * - * TODO: write a generic wrapper to avoid people rolling - * their own + * TODO: write a generic wrapper to avoid people rolling + * their own. */ if (!thread) { if (ret == FILTER_SCHEDULE_THREAD) - thread = 1; + thread = true; } } atomic_add_rel_int(&ie->ie_active[phase], -1); @@ -1348,6 +1429,11 @@ intr_event_handle(struct intr_event *ie, struct trapframe *frame) } critical_exit(); td->td_intr_nesting_level--; +#ifdef notyet + /* The interrupt is not aknowledged by any filter and has no ithread. */ + if (!thread && !filter) + return (EINVAL); +#endif return (0); } diff --git a/freebsd/sys/kern/kern_synch.c b/freebsd/sys/kern/kern_synch.c index 9c0d1206..2597f91d 100644 --- a/freebsd/sys/kern/kern_synch.c +++ b/freebsd/sys/kern/kern_synch.c @@ -466,7 +466,7 @@ mi_switch(int flags, struct thread *newtd) CTR4(KTR_PROC, "mi_switch: old thread %ld (td_sched %p, pid %ld, %s)", td->td_tid, td_get_sched(td), td->td_proc->p_pid, td->td_name); #ifdef KDTRACE_HOOKS - if (__predict_false(sdt_probes_enabled) && + if (SDT_PROBES_ENABLED() && ((flags & SW_PREEMPT) != 0 || ((flags & SW_INVOL) != 0 && (flags & SW_TYPE_MASK) == SWT_NEEDRESCHED))) SDT_PROBE0(sched, , , preempt); diff --git a/freebsd/sys/kern/kern_sysctl.c b/freebsd/sys/kern/kern_sysctl.c index ea9c1821..dc7c4c72 100644 --- a/freebsd/sys/kern/kern_sysctl.c +++ b/freebsd/sys/kern/kern_sysctl.c @@ -552,10 +552,10 @@ sysctl_unregister_oid(struct sysctl_oid *oidp) int error; SYSCTL_ASSERT_WLOCKED(); - error = ENOENT; if (oidp->oid_number == OID_AUTO) { error = EINVAL; } else { + error = ENOENT; SLIST_FOREACH(p, oidp->oid_parent, oid_link) { if (p == oidp) { SLIST_REMOVE(oidp->oid_parent, oidp, @@ -571,8 +571,10 @@ sysctl_unregister_oid(struct sysctl_oid *oidp) * being unloaded afterwards. It should not be a panic() * for normal use. */ - if (error) - printf("%s: failed to unregister sysctl\n", __func__); + if (error) { + printf("%s: failed(%d) to unregister sysctl(%s)\n", + __func__, error, oidp->oid_name); + } } /* Initialize a new context to keep track of dynamically added sysctls. */ @@ -1714,13 +1716,13 @@ sysctl_usec_to_sbintime(SYSCTL_HANDLER_ARGS) sbintime_t sb; tt = *(int64_t *)arg1; - sb = ustosbt(tt); + sb = sbttous(tt); error = sysctl_handle_64(oidp, &sb, 0, req); if (error || !req->newptr) return (error); - tt = sbttous(sb); + tt = ustosbt(sb); *(int64_t *)arg1 = tt; return (0); @@ -1737,13 +1739,13 @@ sysctl_msec_to_sbintime(SYSCTL_HANDLER_ARGS) sbintime_t sb; tt = *(int64_t *)arg1; - sb = mstosbt(tt); + sb = sbttoms(tt); error = sysctl_handle_64(oidp, &sb, 0, req); if (error || !req->newptr) return (error); - tt = sbttoms(sb); + tt = mstosbt(sb); *(int64_t *)arg1 = tt; return (0); @@ -1782,7 +1784,7 @@ sysctl_new_kernel(struct sysctl_req *req, void *p, size_t l) return (0); if (req->newlen - req->newidx < l) return (EINVAL); - bcopy((char *)req->newptr + req->newidx, p, l); + bcopy((const char *)req->newptr + req->newidx, p, l); req->newidx += l; return (0); } @@ -1919,7 +1921,7 @@ sysctl_new_user(struct sysctl_req *req, void *p, size_t l) return (EINVAL); WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL, "sysctl_new_user()"); - error = copyin((char *)req->newptr + req->newidx, p, l); + error = copyin((const char *)req->newptr + req->newidx, p, l); req->newidx += l; return (error); } @@ -2153,8 +2155,8 @@ sys___sysctl(struct thread *td, struct sysctl_args *uap) */ int userland_sysctl(struct thread *td, int *name, u_int namelen, void *old, - size_t *oldlenp, int inkernel, void *new, size_t newlen, size_t *retval, - int flags) + size_t *oldlenp, int inkernel, const void *new, size_t newlen, + size_t *retval, int flags) { int error = 0, memlocked; struct sysctl_req req; diff --git a/freebsd/sys/kern/subr_blist.c b/freebsd/sys/kern/subr_blist.c index 79a5a7b4..807a7f3c 100644 --- a/freebsd/sys/kern/subr_blist.c +++ b/freebsd/sys/kern/subr_blist.c @@ -297,9 +297,9 @@ blist_alloc(blist_t bl, daddr_t count) * This loop iterates at most twice. An allocation failure in the * first iteration leads to a second iteration only if the cursor was * non-zero. When the cursor is zero, an allocation failure will - * reduce the hint, stopping further iterations. + * stop further iterations. */ - while (count <= bl->bl_root->bm_bighint) { + for (;;) { blk = blst_meta_alloc(bl->bl_root, bl->bl_cursor, count, bl->bl_radix); if (blk != SWAPBLK_NONE) { @@ -308,10 +308,10 @@ blist_alloc(blist_t bl, daddr_t count) if (bl->bl_cursor == bl->bl_blocks) bl->bl_cursor = 0; return (blk); - } + } else if (bl->bl_cursor == 0) + return (SWAPBLK_NONE); bl->bl_cursor = 0; } - return (SWAPBLK_NONE); } /* @@ -646,14 +646,14 @@ blst_next_leaf_alloc(blmeta_t *scan, daddr_t blk, int count) /* * BLST_LEAF_ALLOC() - allocate at a leaf in the radix tree (a bitmap). * - * This is the core of the allocator and is optimized for the - * BLIST_BMAP_RADIX block allocation case. Otherwise, execution - * time is proportional to log2(count) + bitpos time. + * This function is the core of the allocator. Its execution time is + * proportional to log(count), plus height of the tree if the allocation + * crosses a leaf boundary. */ static daddr_t blst_leaf_alloc(blmeta_t *scan, daddr_t blk, int count) { - u_daddr_t mask; + u_daddr_t cursor_mask, mask; int count1, hi, lo, num_shifts, range1, range_ext; range1 = 0; @@ -663,14 +663,14 @@ blst_leaf_alloc(blmeta_t *scan, daddr_t blk, int count) while ((-mask & ~mask) != 0 && num_shifts > 0) { /* * If bit i is set in mask, then bits in [i, i+range1] are set - * in scan->bm_bitmap. The value of range1 is equal to - * count1 >> num_shifts. Grow range and reduce num_shifts to 0, - * while preserving these invariants. The updates to mask leave - * fewer bits set, but each bit that remains set represents a - * longer string of consecutive bits set in scan->bm_bitmap. - * If more updates to mask cannot clear more bits, because mask - * is partitioned with all 0 bits preceding all 1 bits, the loop - * terminates immediately. + * in scan->bm_bitmap. The value of range1 is equal to count1 + * >> num_shifts. Grow range1 and reduce num_shifts to 0, + * while preserving these invariants. The updates to mask + * leave fewer bits set, but each bit that remains set + * represents a longer string of consecutive bits set in + * scan->bm_bitmap. If more updates to mask cannot clear more + * bits, because mask is partitioned with all 0 bits preceding + * all 1 bits, the loop terminates immediately. */ num_shifts--; range_ext = range1 + ((count1 >> num_shifts) & 1); @@ -693,9 +693,22 @@ blst_leaf_alloc(blmeta_t *scan, daddr_t blk, int count) } /* Discard any candidates that appear before blk. */ - mask &= (u_daddr_t)-1 << (blk & BLIST_BMAP_MASK); - if (mask == 0) - return (SWAPBLK_NONE); + if ((blk & BLIST_BMAP_MASK) != 0) { + cursor_mask = mask & bitrange(0, blk & BLIST_BMAP_MASK); + if (cursor_mask != 0) { + mask ^= cursor_mask; + if (mask == 0) + return (SWAPBLK_NONE); + + /* + * Bighint change for last block allocation cannot + * assume that any other blocks are allocated, so the + * bighint cannot be reduced much. + */ + range1 = BLIST_MAX_ALLOC - 1; + } + blk &= ~BLIST_BMAP_MASK; + } /* * The least significant set bit in mask marks the start of the first @@ -736,7 +749,7 @@ blst_leaf_alloc(blmeta_t *scan, daddr_t blk, int count) } /* Clear the allocated bits from this leaf. */ scan->bm_bitmap &= ~mask; - return ((blk & ~BLIST_BMAP_MASK) + lo); + return (blk + lo); } /* @@ -766,6 +779,8 @@ blst_meta_alloc(blmeta_t *scan, daddr_t cursor, daddr_t count, u_daddr_t radix) /* Discard any candidates that appear before cursor. */ digit = (cursor / radix) & BLIST_META_MASK; mask &= (u_daddr_t)-1 << digit; + if (mask == 0) + return (SWAPBLK_NONE); /* * If the first try is for a block that includes the cursor, pre-undo diff --git a/freebsd/sys/kern/subr_bus.c b/freebsd/sys/kern/subr_bus.c index 391b2ed6..a87c02a5 100644 --- a/freebsd/sys/kern/subr_bus.c +++ b/freebsd/sys/kern/subr_bus.c @@ -2821,6 +2821,16 @@ device_set_devclass_fixed(device_t dev, const char *classname) } /** + * @brief Query the device to determine if it's of a fixed devclass + * @see device_set_devclass_fixed() + */ +bool +device_is_devclass_fixed(device_t dev) +{ + return ((dev->flags & DF_FIXEDCLASS) != 0); +} + +/** * @brief Set the driver of a device * * @retval 0 success @@ -4101,6 +4111,36 @@ bus_generic_teardown_intr(device_t dev, device_t child, struct resource *irq, } /** + * @brief Helper function for implementing BUS_SUSPEND_INTR(). + * + * This simple implementation of BUS_SUSPEND_INTR() simply calls the + * BUS_SUSPEND_INTR() method of the parent of @p dev. + */ +int +bus_generic_suspend_intr(device_t dev, device_t child, struct resource *irq) +{ + /* Propagate up the bus hierarchy until someone handles it. */ + if (dev->parent) + return (BUS_SUSPEND_INTR(dev->parent, child, irq)); + return (EINVAL); +} + +/** + * @brief Helper function for implementing BUS_RESUME_INTR(). + * + * This simple implementation of BUS_RESUME_INTR() simply calls the + * BUS_RESUME_INTR() method of the parent of @p dev. + */ +int +bus_generic_resume_intr(device_t dev, device_t child, struct resource *irq) +{ + /* Propagate up the bus hierarchy until someone handles it. */ + if (dev->parent) + return (BUS_RESUME_INTR(dev->parent, child, irq)); + return (EINVAL); +} + +/** * @brief Helper function for implementing BUS_ADJUST_RESOURCE(). * * This simple implementation of BUS_ADJUST_RESOURCE() simply calls the @@ -4668,6 +4708,34 @@ bus_teardown_intr(device_t dev, struct resource *r, void *cookie) } /** + * @brief Wrapper function for BUS_SUSPEND_INTR(). + * + * This function simply calls the BUS_SUSPEND_INTR() method of the + * parent of @p dev. + */ +int +bus_suspend_intr(device_t dev, struct resource *r) +{ + if (dev->parent == NULL) + return (EINVAL); + return (BUS_SUSPEND_INTR(dev->parent, dev, r)); +} + +/** + * @brief Wrapper function for BUS_RESUME_INTR(). + * + * This function simply calls the BUS_RESUME_INTR() method of the + * parent of @p dev. + */ +int +bus_resume_intr(device_t dev, struct resource *r) +{ + if (dev->parent == NULL) + return (EINVAL); + return (BUS_RESUME_INTR(dev->parent, dev, r)); +} + +/** * @brief Wrapper function for BUS_BIND_INTR(). * * This function simply calls the BUS_BIND_INTR() method of the diff --git a/freebsd/sys/kern/subr_rman.c b/freebsd/sys/kern/subr_rman.c index 75f22dd5..e307df46 100644 --- a/freebsd/sys/kern/subr_rman.c +++ b/freebsd/sys/kern/subr_rman.c @@ -96,6 +96,7 @@ struct resource_i { rman_res_t r_end; /* index of the last entry (inclusive) */ u_int r_flags; void *r_virtual; /* virtual address of this resource */ + void *r_irq_cookie; /* interrupt cookie for this (interrupt) resource */ device_t r_dev; /* device which has allocated this resource */ struct rman *r_rm; /* resource manager from whence this came */ int r_rid; /* optional rid for this resource. */ @@ -871,6 +872,20 @@ rman_get_virtual(struct resource *r) } void +rman_set_irq_cookie(struct resource *r, void *c) +{ + + r->__r_i->r_irq_cookie = c; +} + +void * +rman_get_irq_cookie(struct resource *r) +{ + + return (r->__r_i->r_irq_cookie); +} + +void rman_set_bustag(struct resource *r, bus_space_tag_t t) { diff --git a/freebsd/sys/kern/subr_taskqueue.c b/freebsd/sys/kern/subr_taskqueue.c index 60057f78..39d9f939 100644 --- a/freebsd/sys/kern/subr_taskqueue.c +++ b/freebsd/sys/kern/subr_taskqueue.c @@ -375,13 +375,13 @@ taskqueue_task_nop_fn(void *context, int pending) * have begun execution. Tasks queued during execution of * this function are ignored. */ -static void +static int taskqueue_drain_tq_queue(struct taskqueue *queue) { struct task t_barrier; if (STAILQ_EMPTY(&queue->tq_queue)) - return; + return (0); /* * Enqueue our barrier after all current tasks, but with @@ -401,6 +401,7 @@ taskqueue_drain_tq_queue(struct taskqueue *queue) */ while (t_barrier.ta_pending != 0) TQ_SLEEP(queue, &t_barrier, &queue->tq_mutex, PWAIT, "-", 0); + return (1); } /* @@ -408,13 +409,13 @@ taskqueue_drain_tq_queue(struct taskqueue *queue) * complete. Tasks that begin execution during the execution * of this function are ignored. */ -static void +static int taskqueue_drain_tq_active(struct taskqueue *queue) { struct taskqueue_busy tb_marker, *tb_first; if (TAILQ_EMPTY(&queue->tq_active)) - return; + return (0); /* Block taskq_terminate().*/ queue->tq_callouts++; @@ -441,6 +442,7 @@ taskqueue_drain_tq_active(struct taskqueue *queue) queue->tq_callouts--; if ((queue->tq_flags & TQ_FLAGS_ACTIVE) == 0) wakeup_one(queue->tq_threads); + return (1); } void @@ -615,8 +617,8 @@ taskqueue_drain_all(struct taskqueue *queue) #endif /* __rtems__ */ TQ_LOCK(queue); - taskqueue_drain_tq_queue(queue); - taskqueue_drain_tq_active(queue); + (void)taskqueue_drain_tq_queue(queue); + (void)taskqueue_drain_tq_active(queue); TQ_UNLOCK(queue); } @@ -645,6 +647,20 @@ taskqueue_drain_timeout(struct taskqueue *queue, TQ_UNLOCK(queue); } +void +taskqueue_quiesce(struct taskqueue *queue) +{ + int ret; + + TQ_LOCK(queue); + do { + ret = taskqueue_drain_tq_queue(queue); + if (ret == 0) + ret = taskqueue_drain_tq_active(queue); + } while (ret != 0); + TQ_UNLOCK(queue); +} + static void taskqueue_swi_enqueue(void *context) { diff --git a/freebsd/sys/kern/subr_unit.c b/freebsd/sys/kern/subr_unit.c index 426253dc..c4bdea34 100644 --- a/freebsd/sys/kern/subr_unit.c +++ b/freebsd/sys/kern/subr_unit.c @@ -100,6 +100,19 @@ static struct mtx unitmtx; MTX_SYSINIT(unit, &unitmtx, "unit# allocation", MTX_DEF); +#ifdef UNR64_LOCKED +uint64_t +alloc_unr64(struct unrhdr64 *unr64) +{ + uint64_t item; + + mtx_lock(&unitmtx); + item = unr64->counter++; + mtx_unlock(&unitmtx); + return (item); +} +#endif + #else /* ...USERLAND */ #include <bitstring.h> diff --git a/freebsd/sys/kern/sys_pipe.c b/freebsd/sys/kern/sys_pipe.c index b6616271..050d63a4 100755 --- a/freebsd/sys/kern/sys_pipe.c +++ b/freebsd/sys/kern/sys_pipe.c @@ -294,7 +294,7 @@ static int pipe_zone_init(void *mem, int size, int flags); static void pipe_zone_fini(void *mem, int size); static uma_zone_t pipe_zone; -static struct unrhdr *pipeino_unr; +static struct unrhdr64 pipeino_unr; static dev_t pipedev_ino; SYSINIT(vfs, SI_SUB_VFS, SI_ORDER_ANY, pipeinit, NULL); @@ -307,8 +307,7 @@ pipeinit(void *dummy __unused) pipe_zone_ctor, NULL, pipe_zone_init, pipe_zone_fini, UMA_ALIGN_PTR, 0); KASSERT(pipe_zone != NULL, ("pipe_zone not initialized")); - pipeino_unr = new_unrhdr(1, INT32_MAX, NULL); - KASSERT(pipeino_unr != NULL, ("pipe fake inodes not initialized")); + new_unrhdr64(&pipeino_unr, 1); pipedev_ino = devfs_alloc_cdp_inode(); KASSERT(pipedev_ino > 0, ("pipe dev inode not initialized")); } @@ -444,8 +443,6 @@ pipe_dtor(struct pipe *dpipe) funsetown(&peer->pipe_sigio); pipeclose(peer); } - if (ino != 0 && ino != (ino_t)-1) - free_unr(pipeino_unr, ino); } /* @@ -730,7 +727,7 @@ pipe_create(struct pipe *pipe, int backing) (void)pipespace_new(pipe, PIPE_SIZE); } - pipe->pipe_ino = -1; + pipe->pipe_ino = alloc_unr64(&pipeino_unr); } /* ARGSUSED */ @@ -1762,7 +1759,6 @@ static int pipe_stat(struct pipe *pipe, struct stat *ub) { #endif /* __rtems__ */ - int new_unr; #ifdef MAC int error; #endif @@ -1789,23 +1785,6 @@ pipe_stat(struct pipe *pipe, struct stat *ub) #endif /* __rtems__ */ } - /* - * Lazily allocate an inode number for the pipe. Most pipe - * users do not call fstat(2) on the pipe, which means that - * postponing the inode allocation until it is must be - * returned to userland is useful. If alloc_unr failed, - * assign st_ino zero instead of returning an error. - * Special pipe_ino values: - * -1 - not yet initialized; - * 0 - alloc_unr failed, return 0 as st_ino forever. - */ - if (pipe->pipe_ino == (ino_t)-1) { - new_unr = alloc_unr(pipeino_unr); - if (new_unr != -1) - pipe->pipe_ino = new_unr; - else - pipe->pipe_ino = 0; - } PIPE_UNLOCK(pipe); #ifndef __rtems__ diff --git a/freebsd/sys/kern/uipc_sockbuf.c b/freebsd/sys/kern/uipc_sockbuf.c index cf99c615..0830206a 100644 --- a/freebsd/sys/kern/uipc_sockbuf.c +++ b/freebsd/sys/kern/uipc_sockbuf.c @@ -1230,53 +1230,6 @@ sbdrop(struct sockbuf *sb, int len) m_freem(mfree); } -/* - * Maintain a pointer and offset pair into the socket buffer mbuf chain to - * avoid traversal of the entire socket buffer for larger offsets. - */ -struct mbuf * -sbsndptr(struct sockbuf *sb, u_int off, u_int len, u_int *moff) -{ - struct mbuf *m, *ret; - - KASSERT(sb->sb_mb != NULL, ("%s: sb_mb is NULL", __func__)); - KASSERT(off + len <= sb->sb_acc, ("%s: beyond sb", __func__)); - KASSERT(sb->sb_sndptroff <= sb->sb_acc, ("%s: sndptroff broken", __func__)); - - /* - * Is off below stored offset? Happens on retransmits. - * Just return, we can't help here. - */ - if (sb->sb_sndptroff > off) { - *moff = off; - return (sb->sb_mb); - } - - /* Return closest mbuf in chain for current offset. */ - *moff = off - sb->sb_sndptroff; - m = ret = sb->sb_sndptr ? sb->sb_sndptr : sb->sb_mb; - if (*moff == m->m_len) { - *moff = 0; - sb->sb_sndptroff += m->m_len; - m = ret = m->m_next; - KASSERT(ret->m_len > 0, - ("mbuf %p in sockbuf %p chain has no valid data", ret, sb)); - } - - /* Advance by len to be as close as possible for the next transmit. */ - for (off = off - sb->sb_sndptroff + len - 1; - off > 0 && m != NULL && off >= m->m_len; - m = m->m_next) { - sb->sb_sndptroff += m->m_len; - off -= m->m_len; - } - if (off > 0 && m == NULL) - panic("%s: sockbuf %p and mbuf %p clashing", __func__, sb, ret); - sb->sb_sndptr = m; - - return (ret); -} - struct mbuf * #ifndef __rtems__ sbsndptr_noadv(struct sockbuf *sb, uint32_t off, uint32_t *moff) diff --git a/freebsd/sys/kern/uipc_socket.c b/freebsd/sys/kern/uipc_socket.c index aa045cd9..380c97dd 100644 --- a/freebsd/sys/kern/uipc_socket.c +++ b/freebsd/sys/kern/uipc_socket.c @@ -4063,6 +4063,7 @@ void sotoxsocket(struct socket *so, struct xsocket *xso) { + bzero(xso, sizeof(*xso)); xso->xso_len = sizeof *xso; xso->xso_so = (uintptr_t)so; xso->so_type = so->so_type; @@ -4085,8 +4086,6 @@ sotoxsocket(struct socket *so, struct xsocket *xso) xso->so_incqlen = so->sol_incqlen; xso->so_qlimit = so->sol_qlimit; xso->so_oobmark = 0; - bzero(&xso->so_snd, sizeof(xso->so_snd)); - bzero(&xso->so_rcv, sizeof(xso->so_rcv)); } else { xso->so_state |= so->so_qstate; xso->so_qlen = xso->so_incqlen = xso->so_qlimit = 0; diff --git a/freebsd/sys/kern/uipc_syscalls.c b/freebsd/sys/kern/uipc_syscalls.c index f338629e..529268a9 100644 --- a/freebsd/sys/kern/uipc_syscalls.c +++ b/freebsd/sys/kern/uipc_syscalls.c @@ -643,9 +643,7 @@ sys_accept4(td, uap) #ifdef COMPAT_OLDSOCK int -oaccept(td, uap) - struct thread *td; - struct accept_args *uap; +oaccept(struct thread *td, struct oaccept_args *uap) { return (accept1(td, uap->s, uap->name, uap->anamelen, diff --git a/freebsd/sys/kern/uipc_usrreq.c b/freebsd/sys/kern/uipc_usrreq.c index c1885ed6..6b34dcb8 100644 --- a/freebsd/sys/kern/uipc_usrreq.c +++ b/freebsd/sys/kern/uipc_usrreq.c @@ -551,6 +551,7 @@ uipc_attach(struct socket *so, int proto, struct thread *td) UNP_LINK_WLOCK(); unp->unp_gencnt = ++unp_gencnt; + unp->unp_ino = ++unp_ino; unp_count++; switch (so->so_type) { case SOCK_STREAM: @@ -1434,12 +1435,8 @@ uipc_sense(struct socket *so, struct stat *sb) KASSERT(unp != NULL, ("uipc_sense: unp == NULL")); sb->st_blksize = so->so_snd.sb_hiwat; - UNP_PCB_LOCK(unp); sb->st_dev = NODEV; - if (unp->unp_ino == 0) - unp->unp_ino = (++unp_ino == 0) ? ++unp_ino : unp_ino; sb->st_ino = unp->unp_ino; - UNP_PCB_UNLOCK(unp); return (0); } @@ -1993,7 +1990,7 @@ unp_pcblist(SYSCTL_HANDLER_ARGS) /* * OK, now we're committed to doing something. */ - xug = malloc(sizeof(*xug), M_TEMP, M_WAITOK); + xug = malloc(sizeof(*xug), M_TEMP, M_WAITOK | M_ZERO); UNP_LINK_RLOCK(); gencnt = unp_gencnt; n = unp_count; diff --git a/freebsd/sys/net/altq/altq.h b/freebsd/sys/net/altq/altq.h index 35024461..731fead4 100644 --- a/freebsd/sys/net/altq/altq.h +++ b/freebsd/sys/net/altq/altq.h @@ -38,16 +38,6 @@ #define ALTQ3_CLFIER_COMPAT /* for compatibility with altq-3 classifier */ #endif -#ifdef ALTQ3_COMPAT -#include <sys/param.h> -#include <sys/ioccom.h> -#include <sys/queue.h> -#include <netinet/in.h> - -#ifndef IFNAMSIZ -#define IFNAMSIZ 16 -#endif -#endif /* ALTQ3_COMPAT */ /* altq discipline type */ #define ALTQT_NONE 0 /* reserved */ @@ -67,12 +57,6 @@ #define ALTQT_CODEL 14 /* CoDel */ #define ALTQT_MAX 15 /* should be max discipline type + 1 */ -#ifdef ALTQ3_COMPAT -struct altqreq { - char ifname[IFNAMSIZ]; /* if name, e.g. "en0" */ - u_long arg; /* request-specific argument */ -}; -#endif /* simple token backet meter profile */ struct tb_profile { @@ -80,85 +64,6 @@ struct tb_profile { u_int32_t depth; /* depth in bytes */ }; -#ifdef ALTQ3_COMPAT -struct tbrreq { - char ifname[IFNAMSIZ]; /* if name, e.g. "en0" */ - struct tb_profile tb_prof; /* token bucket profile */ -}; - -#ifdef ALTQ3_CLFIER_COMPAT -/* - * common network flow info structure - */ -struct flowinfo { - u_char fi_len; /* total length */ - u_char fi_family; /* address family */ - u_int8_t fi_data[46]; /* actually longer; address family - specific flow info. */ -}; - -/* - * flow info structure for internet protocol family. - * (currently this is the only protocol family supported) - */ -struct flowinfo_in { - u_char fi_len; /* sizeof(struct flowinfo_in) */ - u_char fi_family; /* AF_INET */ - u_int8_t fi_proto; /* IPPROTO_XXX */ - u_int8_t fi_tos; /* type-of-service */ - struct in_addr fi_dst; /* dest address */ - struct in_addr fi_src; /* src address */ - u_int16_t fi_dport; /* dest port */ - u_int16_t fi_sport; /* src port */ - u_int32_t fi_gpi; /* generalized port id for ipsec */ - u_int8_t _pad[28]; /* make the size equal to - flowinfo_in6 */ -}; - -#ifdef SIN6_LEN -struct flowinfo_in6 { - u_char fi6_len; /* sizeof(struct flowinfo_in6) */ - u_char fi6_family; /* AF_INET6 */ - u_int8_t fi6_proto; /* IPPROTO_XXX */ - u_int8_t fi6_tclass; /* traffic class */ - u_int32_t fi6_flowlabel; /* ipv6 flowlabel */ - u_int16_t fi6_dport; /* dest port */ - u_int16_t fi6_sport; /* src port */ - u_int32_t fi6_gpi; /* generalized port id */ - struct in6_addr fi6_dst; /* dest address */ - struct in6_addr fi6_src; /* src address */ -}; -#endif /* INET6 */ - -/* - * flow filters for AF_INET and AF_INET6 - */ -struct flow_filter { - int ff_ruleno; - struct flowinfo_in ff_flow; - struct { - struct in_addr mask_dst; - struct in_addr mask_src; - u_int8_t mask_tos; - u_int8_t _pad[3]; - } ff_mask; - u_int8_t _pad2[24]; /* make the size equal to flow_filter6 */ -}; - -#ifdef SIN6_LEN -struct flow_filter6 { - int ff_ruleno; - struct flowinfo_in6 ff_flow6; - struct { - struct in6_addr mask6_dst; - struct in6_addr mask6_src; - u_int8_t mask6_tclass; - u_int8_t _pad[3]; - } ff_mask6; -}; -#endif /* INET6 */ -#endif /* ALTQ3_CLFIER_COMPAT */ -#endif /* ALTQ3_COMPAT */ /* * generic packet counter @@ -171,33 +76,6 @@ struct pktcntr { #define PKTCNTR_ADD(cntr, len) \ do { (cntr)->packets++; (cntr)->bytes += len; } while (/*CONSTCOND*/ 0) -#ifdef ALTQ3_COMPAT -/* - * altq related ioctls - */ -#define ALTQGTYPE _IOWR('q', 0, struct altqreq) /* get queue type */ -#if 0 -/* - * these ioctls are currently discipline-specific but could be shared - * in the future. - */ -#define ALTQATTACH _IOW('q', 1, struct altqreq) /* attach discipline */ -#define ALTQDETACH _IOW('q', 2, struct altqreq) /* detach discipline */ -#define ALTQENABLE _IOW('q', 3, struct altqreq) /* enable discipline */ -#define ALTQDISABLE _IOW('q', 4, struct altqreq) /* disable discipline*/ -#define ALTQCLEAR _IOW('q', 5, struct altqreq) /* (re)initialize */ -#define ALTQCONFIG _IOWR('q', 6, struct altqreq) /* set config params */ -#define ALTQADDCLASS _IOWR('q', 7, struct altqreq) /* add a class */ -#define ALTQMODCLASS _IOWR('q', 8, struct altqreq) /* modify a class */ -#define ALTQDELCLASS _IOWR('q', 9, struct altqreq) /* delete a class */ -#define ALTQADDFILTER _IOWR('q', 10, struct altqreq) /* add a filter */ -#define ALTQDELFILTER _IOWR('q', 11, struct altqreq) /* delete a filter */ -#define ALTQGETSTATS _IOWR('q', 12, struct altqreq) /* get statistics */ -#define ALTQGETCNTR _IOWR('q', 13, struct altqreq) /* get a pkt counter */ -#endif /* 0 */ -#define ALTQTBRSET _IOW('q', 14, struct tbrreq) /* set tb regulator */ -#define ALTQTBRGET _IOWR('q', 15, struct tbrreq) /* get tb regulator */ -#endif /* ALTQ3_COMPAT */ #ifdef _KERNEL #include <net/altq/altq_var.h> diff --git a/freebsd/sys/net/altq/altq_cbq.c b/freebsd/sys/net/altq/altq_cbq.c index ac108bd1..015e35bf 100644 --- a/freebsd/sys/net/altq/altq_cbq.c +++ b/freebsd/sys/net/altq/altq_cbq.c @@ -46,10 +46,6 @@ #include <sys/proc.h> #include <sys/errno.h> #include <sys/time.h> -#ifdef ALTQ3_COMPAT -#include <sys/uio.h> -#include <sys/kernel.h> -#endif #include <net/if.h> #include <net/if_var.h> @@ -60,16 +56,7 @@ #include <netpfil/pf/pf_mtag.h> #include <net/altq/altq.h> #include <net/altq/altq_cbq.h> -#ifdef ALTQ3_COMPAT -#include <net/altq/altq_conf.h> -#endif -#ifdef ALTQ3_COMPAT -/* - * Local Data structures. - */ -static cbq_state_t *cbq_list = NULL; -#endif /* * Forward Declarations. @@ -84,21 +71,6 @@ static struct mbuf *cbq_dequeue(struct ifaltq *, int); static void cbqrestart(struct ifaltq *); static void get_class_stats(class_stats_t *, struct rm_class *); static void cbq_purge(cbq_state_t *); -#ifdef ALTQ3_COMPAT -static int cbq_add_class(struct cbq_add_class *); -static int cbq_delete_class(struct cbq_delete_class *); -static int cbq_modify_class(struct cbq_modify_class *); -static int cbq_class_create(cbq_state_t *, struct cbq_add_class *, - struct rm_class *, struct rm_class *); -static int cbq_clear_hierarchy(struct cbq_interface *); -static int cbq_set_enable(struct cbq_interface *, int); -static int cbq_ifattach(struct cbq_interface *); -static int cbq_ifdetach(struct cbq_interface *); -static int cbq_getstats(struct cbq_getstats *); - -static int cbq_add_filter(struct cbq_add_filter *); -static int cbq_delete_filter(struct cbq_delete_filter *); -#endif /* ALTQ3_COMPAT */ /* * int @@ -125,10 +97,6 @@ cbq_class_destroy(cbq_state_t *cbqp, struct rm_class *cl) cbqp->ifnp.root_ = NULL; if (cl == cbqp->ifnp.default_) cbqp->ifnp.default_ = NULL; -#ifdef ALTQ3_COMPAT - if (cl == cbqp->ifnp.ctl_) - cbqp->ifnp.ctl_ = NULL; -#endif return (0); } @@ -181,10 +149,6 @@ cbq_clear_interface(cbq_state_t *cbqp) cbqp->ifnp.root_ = NULL; if (cl == cbqp->ifnp.default_) cbqp->ifnp.default_ = NULL; -#ifdef ALTQ3_COMPAT - if (cl == cbqp->ifnp.ctl_) - cbqp->ifnp.ctl_ = NULL; -#endif } } } @@ -514,10 +478,6 @@ cbq_enqueue(struct ifaltq *ifq, struct mbuf *m, struct altq_pktattr *pktattr) cl = NULL; if ((t = pf_find_mtag(m)) != NULL) cl = clh_to_clp(cbqp, t->qid); -#ifdef ALTQ3_COMPAT - else if ((ifq->altq_flags & ALTQF_CLASSIFY) && pktattr != NULL) - cl = pktattr->pattr_class; -#endif if (cl == NULL) { cl = cbqp->ifnp.default_; if (cl == NULL) { @@ -525,12 +485,7 @@ cbq_enqueue(struct ifaltq *ifq, struct mbuf *m, struct altq_pktattr *pktattr) return (ENOBUFS); } } -#ifdef ALTQ3_COMPAT - if (pktattr != NULL) - cl->pktattr_ = pktattr; /* save proto hdr used by ECN */ - else -#endif - cl->pktattr_ = NULL; + cl->pktattr_ = NULL; len = m_pktlen(m); if (rmc_queue_packet(cl, m) != 0) { /* drop occurred. some mbuf was freed in rmc_queue_packet. */ @@ -608,564 +563,5 @@ static void cbq_purge(cbq_state_t *cbqp) if (ALTQ_IS_ENABLED(cbqp->ifnp.ifq_)) cbqp->ifnp.ifq_->ifq_len = 0; } -#ifdef ALTQ3_COMPAT - -static int -cbq_add_class(acp) - struct cbq_add_class *acp; -{ - char *ifacename; - struct rm_class *borrow, *parent; - cbq_state_t *cbqp; - - ifacename = acp->cbq_iface.cbq_ifacename; - if ((cbqp = altq_lookup(ifacename, ALTQT_CBQ)) == NULL) - return (EBADF); - - /* check parameters */ - if (acp->cbq_class.priority >= CBQ_MAXPRI || - acp->cbq_class.maxq > CBQ_MAXQSIZE) - return (EINVAL); - - /* Get pointers to parent and borrow classes. */ - parent = clh_to_clp(cbqp, acp->cbq_class.parent_class_handle); - borrow = clh_to_clp(cbqp, acp->cbq_class.borrow_class_handle); - - /* - * A class must borrow from it's parent or it can not - * borrow at all. Hence, borrow can be null. - */ - if (parent == NULL && (acp->cbq_class.flags & CBQCLF_ROOTCLASS) == 0) { - printf("cbq_add_class: no parent class!\n"); - return (EINVAL); - } - - if ((borrow != parent) && (borrow != NULL)) { - printf("cbq_add_class: borrow class != parent\n"); - return (EINVAL); - } - - return cbq_class_create(cbqp, acp, parent, borrow); -} - -static int -cbq_delete_class(dcp) - struct cbq_delete_class *dcp; -{ - char *ifacename; - struct rm_class *cl; - cbq_state_t *cbqp; - - ifacename = dcp->cbq_iface.cbq_ifacename; - if ((cbqp = altq_lookup(ifacename, ALTQT_CBQ)) == NULL) - return (EBADF); - - if ((cl = clh_to_clp(cbqp, dcp->cbq_class_handle)) == NULL) - return (EINVAL); - - /* if we are a parent class, then return an error. */ - if (is_a_parent_class(cl)) - return (EINVAL); - - /* if a filter has a reference to this class delete the filter */ - acc_discard_filters(&cbqp->cbq_classifier, cl, 0); - - return cbq_class_destroy(cbqp, cl); -} - -static int -cbq_modify_class(acp) - struct cbq_modify_class *acp; -{ - char *ifacename; - struct rm_class *cl; - cbq_state_t *cbqp; - - ifacename = acp->cbq_iface.cbq_ifacename; - if ((cbqp = altq_lookup(ifacename, ALTQT_CBQ)) == NULL) - return (EBADF); - - /* Get pointer to this class */ - if ((cl = clh_to_clp(cbqp, acp->cbq_class_handle)) == NULL) - return (EINVAL); - - if (rmc_modclass(cl, acp->cbq_class.nano_sec_per_byte, - acp->cbq_class.maxq, acp->cbq_class.maxidle, - acp->cbq_class.minidle, acp->cbq_class.offtime, - acp->cbq_class.pktsize) < 0) - return (EINVAL); - return (0); -} - -/* - * struct rm_class * - * cbq_class_create(cbq_mod_state_t *cbqp, struct cbq_add_class *acp, - * struct rm_class *parent, struct rm_class *borrow) - * - * This function create a new traffic class in the CBQ class hierarchy of - * given parameters. The class that created is either the root, default, - * or a new dynamic class. If CBQ is not initilaized, the root class - * will be created. - */ -static int -cbq_class_create(cbqp, acp, parent, borrow) - cbq_state_t *cbqp; - struct cbq_add_class *acp; - struct rm_class *parent, *borrow; -{ - struct rm_class *cl; - cbq_class_spec_t *spec = &acp->cbq_class; - u_int32_t chandle; - int i; - - /* - * allocate class handle - */ - for (i = 1; i < CBQ_MAX_CLASSES; i++) - if (cbqp->cbq_class_tbl[i] == NULL) - break; - if (i == CBQ_MAX_CLASSES) - return (EINVAL); - chandle = i; /* use the slot number as class handle */ - - /* - * create a class. if this is a root class, initialize the - * interface. - */ - if ((spec->flags & CBQCLF_CLASSMASK) == CBQCLF_ROOTCLASS) { - rmc_init(cbqp->ifnp.ifq_, &cbqp->ifnp, spec->nano_sec_per_byte, - cbqrestart, spec->maxq, RM_MAXQUEUED, - spec->maxidle, spec->minidle, spec->offtime, - spec->flags); - cl = cbqp->ifnp.root_; - } else { - cl = rmc_newclass(spec->priority, - &cbqp->ifnp, spec->nano_sec_per_byte, - rmc_delay_action, spec->maxq, parent, borrow, - spec->maxidle, spec->minidle, spec->offtime, - spec->pktsize, spec->flags); - } - if (cl == NULL) - return (ENOMEM); - - /* return handle to user space. */ - acp->cbq_class_handle = chandle; - - cl->stats_.handle = chandle; - cl->stats_.depth = cl->depth_; - - /* save the allocated class */ - cbqp->cbq_class_tbl[i] = cl; - - if ((spec->flags & CBQCLF_CLASSMASK) == CBQCLF_DEFCLASS) - cbqp->ifnp.default_ = cl; - if ((spec->flags & CBQCLF_CLASSMASK) == CBQCLF_CTLCLASS) - cbqp->ifnp.ctl_ = cl; - - return (0); -} - -static int -cbq_add_filter(afp) - struct cbq_add_filter *afp; -{ - char *ifacename; - cbq_state_t *cbqp; - struct rm_class *cl; - - ifacename = afp->cbq_iface.cbq_ifacename; - if ((cbqp = altq_lookup(ifacename, ALTQT_CBQ)) == NULL) - return (EBADF); - - /* Get the pointer to class. */ - if ((cl = clh_to_clp(cbqp, afp->cbq_class_handle)) == NULL) - return (EINVAL); - - return acc_add_filter(&cbqp->cbq_classifier, &afp->cbq_filter, - cl, &afp->cbq_filter_handle); -} - -static int -cbq_delete_filter(dfp) - struct cbq_delete_filter *dfp; -{ - char *ifacename; - cbq_state_t *cbqp; - - ifacename = dfp->cbq_iface.cbq_ifacename; - if ((cbqp = altq_lookup(ifacename, ALTQT_CBQ)) == NULL) - return (EBADF); - - return acc_delete_filter(&cbqp->cbq_classifier, - dfp->cbq_filter_handle); -} - -/* - * cbq_clear_hierarchy deletes all classes and their filters on the - * given interface. - */ -static int -cbq_clear_hierarchy(ifacep) - struct cbq_interface *ifacep; -{ - char *ifacename; - cbq_state_t *cbqp; - - ifacename = ifacep->cbq_ifacename; - if ((cbqp = altq_lookup(ifacename, ALTQT_CBQ)) == NULL) - return (EBADF); - - return cbq_clear_interface(cbqp); -} - -/* - * static int - * cbq_set_enable(struct cbq_enable *ep) - this function processed the - * ioctl request to enable class based queueing. It searches the list - * of interfaces for the specified interface and then enables CBQ on - * that interface. - * - * Returns: 0, for no error. - * EBADF, for specified inteface not found. - */ - -static int -cbq_set_enable(ep, enable) - struct cbq_interface *ep; - int enable; -{ - int error = 0; - cbq_state_t *cbqp; - char *ifacename; - - ifacename = ep->cbq_ifacename; - if ((cbqp = altq_lookup(ifacename, ALTQT_CBQ)) == NULL) - return (EBADF); - - switch (enable) { - case ENABLE: - if (cbqp->ifnp.root_ == NULL || cbqp->ifnp.default_ == NULL || - cbqp->ifnp.ctl_ == NULL) { - if (cbqp->ifnp.root_ == NULL) - printf("No Root Class for %s\n", ifacename); - if (cbqp->ifnp.default_ == NULL) - printf("No Default Class for %s\n", ifacename); - if (cbqp->ifnp.ctl_ == NULL) - printf("No Control Class for %s\n", ifacename); - error = EINVAL; - } else if ((error = altq_enable(cbqp->ifnp.ifq_)) == 0) { - cbqp->cbq_qlen = 0; - } - break; - - case DISABLE: - error = altq_disable(cbqp->ifnp.ifq_); - break; - } - return (error); -} - -static int -cbq_getstats(gsp) - struct cbq_getstats *gsp; -{ - char *ifacename; - int i, n, nclasses; - cbq_state_t *cbqp; - struct rm_class *cl; - class_stats_t stats, *usp; - int error = 0; - - ifacename = gsp->iface.cbq_ifacename; - nclasses = gsp->nclasses; - usp = gsp->stats; - - if ((cbqp = altq_lookup(ifacename, ALTQT_CBQ)) == NULL) - return (EBADF); - if (nclasses <= 0) - return (EINVAL); - - for (n = 0, i = 0; n < nclasses && i < CBQ_MAX_CLASSES; n++, i++) { - while ((cl = cbqp->cbq_class_tbl[i]) == NULL) - if (++i >= CBQ_MAX_CLASSES) - goto out; - - get_class_stats(&stats, cl); - stats.handle = cl->stats_.handle; - - if ((error = copyout((caddr_t)&stats, (caddr_t)usp++, - sizeof(stats))) != 0) - return (error); - } - - out: - gsp->nclasses = n; - return (error); -} - -static int -cbq_ifattach(ifacep) - struct cbq_interface *ifacep; -{ - int error = 0; - char *ifacename; - cbq_state_t *new_cbqp; - struct ifnet *ifp; - - ifacename = ifacep->cbq_ifacename; - if ((ifp = ifunit(ifacename)) == NULL) - return (ENXIO); - if (!ALTQ_IS_READY(&ifp->if_snd)) - return (ENXIO); - - /* allocate and initialize cbq_state_t */ - new_cbqp = malloc(sizeof(cbq_state_t), M_DEVBUF, M_WAITOK); - if (new_cbqp == NULL) - return (ENOMEM); - bzero(new_cbqp, sizeof(cbq_state_t)); - CALLOUT_INIT(&new_cbqp->cbq_callout); - - new_cbqp->cbq_qlen = 0; - new_cbqp->ifnp.ifq_ = &ifp->if_snd; /* keep the ifq */ - - /* - * set CBQ to this ifnet structure. - */ - error = altq_attach(&ifp->if_snd, ALTQT_CBQ, new_cbqp, - cbq_enqueue, cbq_dequeue, cbq_request, - &new_cbqp->cbq_classifier, acc_classify); - if (error) { - free(new_cbqp, M_DEVBUF); - return (error); - } - - /* prepend to the list of cbq_state_t's. */ - new_cbqp->cbq_next = cbq_list; - cbq_list = new_cbqp; - - return (0); -} - -static int -cbq_ifdetach(ifacep) - struct cbq_interface *ifacep; -{ - char *ifacename; - cbq_state_t *cbqp; - - ifacename = ifacep->cbq_ifacename; - if ((cbqp = altq_lookup(ifacename, ALTQT_CBQ)) == NULL) - return (EBADF); - - (void)cbq_set_enable(ifacep, DISABLE); - - cbq_clear_interface(cbqp); - - /* remove CBQ from the ifnet structure. */ - (void)altq_detach(cbqp->ifnp.ifq_); - - /* remove from the list of cbq_state_t's. */ - if (cbq_list == cbqp) - cbq_list = cbqp->cbq_next; - else { - cbq_state_t *cp; - - for (cp = cbq_list; cp != NULL; cp = cp->cbq_next) - if (cp->cbq_next == cbqp) { - cp->cbq_next = cbqp->cbq_next; - break; - } - ASSERT(cp != NULL); - } - - /* deallocate cbq_state_t */ - free(cbqp, M_DEVBUF); - - return (0); -} - -/* - * cbq device interface - */ - -altqdev_decl(cbq); - -int -cbqopen(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - return (0); -} - -int -cbqclose(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - struct ifnet *ifp; - struct cbq_interface iface; - int err, error = 0; - - while (cbq_list) { - ifp = cbq_list->ifnp.ifq_->altq_ifp; - sprintf(iface.cbq_ifacename, "%s", ifp->if_xname); - err = cbq_ifdetach(&iface); - if (err != 0 && error == 0) - error = err; - } - - return (error); -} - -int -cbqioctl(dev, cmd, addr, flag, p) - dev_t dev; - ioctlcmd_t cmd; - caddr_t addr; - int flag; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - int error = 0; - - /* check cmd for superuser only */ - switch (cmd) { - case CBQ_GETSTATS: - /* currently only command that an ordinary user can call */ - break; - default: -#if (__FreeBSD_version > 700000) - error = priv_check(p, PRIV_ALTQ_MANAGE); -#elsif (__FreeBSD_version > 400000) - error = suser(p); -#else - error = suser(p->p_ucred, &p->p_acflag); -#endif - if (error) - return (error); - break; - } - - switch (cmd) { - - case CBQ_ENABLE: - error = cbq_set_enable((struct cbq_interface *)addr, ENABLE); - break; - - case CBQ_DISABLE: - error = cbq_set_enable((struct cbq_interface *)addr, DISABLE); - break; - - case CBQ_ADD_FILTER: - error = cbq_add_filter((struct cbq_add_filter *)addr); - break; - - case CBQ_DEL_FILTER: - error = cbq_delete_filter((struct cbq_delete_filter *)addr); - break; - - case CBQ_ADD_CLASS: - error = cbq_add_class((struct cbq_add_class *)addr); - break; - - case CBQ_DEL_CLASS: - error = cbq_delete_class((struct cbq_delete_class *)addr); - break; - - case CBQ_MODIFY_CLASS: - error = cbq_modify_class((struct cbq_modify_class *)addr); - break; - - case CBQ_CLEAR_HIERARCHY: - error = cbq_clear_hierarchy((struct cbq_interface *)addr); - break; - - case CBQ_IF_ATTACH: - error = cbq_ifattach((struct cbq_interface *)addr); - break; - - case CBQ_IF_DETACH: - error = cbq_ifdetach((struct cbq_interface *)addr); - break; - - case CBQ_GETSTATS: - error = cbq_getstats((struct cbq_getstats *)addr); - break; - - default: - error = EINVAL; - break; - } - - return error; -} - -#if 0 -/* for debug */ -static void cbq_class_dump(int); - -static void cbq_class_dump(i) - int i; -{ - struct rm_class *cl; - rm_class_stats_t *s; - struct _class_queue_ *q; - - if (cbq_list == NULL) { - printf("cbq_class_dump: no cbq_state found\n"); - return; - } - cl = cbq_list->cbq_class_tbl[i]; - - printf("class %d cl=%p\n", i, cl); - if (cl != NULL) { - s = &cl->stats_; - q = cl->q_; - - printf("pri=%d, depth=%d, maxrate=%d, allotment=%d\n", - cl->pri_, cl->depth_, cl->maxrate_, cl->allotment_); - printf("w_allotment=%d, bytes_alloc=%d, avgidle=%d, maxidle=%d\n", - cl->w_allotment_, cl->bytes_alloc_, cl->avgidle_, - cl->maxidle_); - printf("minidle=%d, offtime=%d, sleeping=%d, leaf=%d\n", - cl->minidle_, cl->offtime_, cl->sleeping_, cl->leaf_); - printf("handle=%d, depth=%d, packets=%d, bytes=%d\n", - s->handle, s->depth, - (int)s->xmit_cnt.packets, (int)s->xmit_cnt.bytes); - printf("over=%d\n, borrows=%d, drops=%d, overactions=%d, delays=%d\n", - s->over, s->borrows, (int)s->drop_cnt.packets, - s->overactions, s->delays); - printf("tail=%p, head=%p, qlen=%d, qlim=%d, qthresh=%d,qtype=%d\n", - q->tail_, q->head_, q->qlen_, q->qlim_, - q->qthresh_, q->qtype_); - } -} -#endif /* 0 */ - -#ifdef KLD_MODULE - -static struct altqsw cbq_sw = - {"cbq", cbqopen, cbqclose, cbqioctl}; - -ALTQ_MODULE(altq_cbq, ALTQT_CBQ, &cbq_sw); -MODULE_DEPEND(altq_cbq, altq_red, 1, 1, 1); -MODULE_DEPEND(altq_cbq, altq_rio, 1, 1, 1); - -#endif /* KLD_MODULE */ -#endif /* ALTQ3_COMPAT */ #endif /* ALTQ_CBQ */ diff --git a/freebsd/sys/net/altq/altq_cbq.h b/freebsd/sys/net/altq/altq_cbq.h index 04bcab1a..70974715 100644 --- a/freebsd/sys/net/altq/altq_cbq.h +++ b/freebsd/sys/net/altq/altq_cbq.h @@ -46,7 +46,7 @@ extern "C" { #define NULL_CLASS_HANDLE 0 -/* class flags should be same as class flags in rm_class.h */ +/* class flags must be same as class flags in altq_rmclass.h */ #define CBQCLF_RED 0x0001 /* use RED */ #define CBQCLF_ECN 0x0002 /* use RED/ECN */ #define CBQCLF_RIO 0x0004 /* use RIO */ @@ -55,6 +55,15 @@ extern "C" { #define CBQCLF_BORROW 0x0020 /* borrow from parent */ #define CBQCLF_CODEL 0x0040 /* use CoDel */ +#ifdef _KERNEL +CTASSERT(CBQCLF_RED == RMCF_RED); +CTASSERT(CBQCLF_ECN == RMCF_ECN); +CTASSERT(CBQCLF_RIO == RMCF_RIO); +CTASSERT(CBQCLF_FLOWVALVE == RMCF_FLOWVALVE); +CTASSERT(CBQCLF_CLEARDSCP == RMCF_CLEARDSCP); +CTASSERT(CBQCLF_CODEL == RMCF_CODEL); +#endif + /* class flags only for root class */ #define CBQCLF_WRR 0x0100 /* weighted-round robin */ #define CBQCLF_EFFICIENT 0x0200 /* work-conserving */ @@ -62,9 +71,6 @@ extern "C" { /* class flags for special classes */ #define CBQCLF_ROOTCLASS 0x1000 /* root class */ #define CBQCLF_DEFCLASS 0x2000 /* default class */ -#ifdef ALTQ3_COMPAT -#define CBQCLF_CTLCLASS 0x4000 /* control class */ -#endif #define CBQCLF_CLASSMASK 0xf000 /* class mask */ #define CBQ_MAXQSIZE 200 @@ -105,88 +111,6 @@ typedef struct _cbq_class_stats_ { * header. */ -#ifdef ALTQ3_COMPAT -/* - * Define structures associated with IOCTLS for cbq. - */ - -/* - * Define the CBQ interface structure. This must be included in all - * IOCTL's such that the CBQ driver may find the appropriate CBQ module - * associated with the network interface to be affected. - */ -struct cbq_interface { - char cbq_ifacename[IFNAMSIZ]; -}; - -typedef struct cbq_class_spec { - u_int priority; - u_int nano_sec_per_byte; - u_int maxq; - u_int maxidle; - int minidle; - u_int offtime; - u_int32_t parent_class_handle; - u_int32_t borrow_class_handle; - - u_int pktsize; - int flags; -} cbq_class_spec_t; - -struct cbq_add_class { - struct cbq_interface cbq_iface; - - cbq_class_spec_t cbq_class; - u_int32_t cbq_class_handle; -}; - -struct cbq_delete_class { - struct cbq_interface cbq_iface; - u_int32_t cbq_class_handle; -}; - -struct cbq_modify_class { - struct cbq_interface cbq_iface; - - cbq_class_spec_t cbq_class; - u_int32_t cbq_class_handle; -}; - -struct cbq_add_filter { - struct cbq_interface cbq_iface; - u_int32_t cbq_class_handle; - struct flow_filter cbq_filter; - - u_long cbq_filter_handle; -}; - -struct cbq_delete_filter { - struct cbq_interface cbq_iface; - u_long cbq_filter_handle; -}; - -/* number of classes are returned in nclasses field */ -struct cbq_getstats { - struct cbq_interface iface; - int nclasses; - class_stats_t *stats; -}; - -/* - * Define IOCTLs for CBQ. - */ -#define CBQ_IF_ATTACH _IOW('Q', 1, struct cbq_interface) -#define CBQ_IF_DETACH _IOW('Q', 2, struct cbq_interface) -#define CBQ_ENABLE _IOW('Q', 3, struct cbq_interface) -#define CBQ_DISABLE _IOW('Q', 4, struct cbq_interface) -#define CBQ_CLEAR_HIERARCHY _IOW('Q', 5, struct cbq_interface) -#define CBQ_ADD_CLASS _IOWR('Q', 7, struct cbq_add_class) -#define CBQ_DEL_CLASS _IOW('Q', 8, struct cbq_delete_class) -#define CBQ_MODIFY_CLASS _IOWR('Q', 9, struct cbq_modify_class) -#define CBQ_ADD_FILTER _IOWR('Q', 10, struct cbq_add_filter) -#define CBQ_DEL_FILTER _IOW('Q', 11, struct cbq_delete_filter) -#define CBQ_GETSTATS _IOWR('Q', 12, struct cbq_getstats) -#endif /* ALTQ3_COMPAT */ #ifdef _KERNEL /* @@ -198,20 +122,11 @@ struct cbq_getstats { #define CBQ_MAX_CLASSES 256 -#ifdef ALTQ3_COMPAT -#define CBQ_MAX_FILTERS 256 - -#define DISABLE 0x00 -#define ENABLE 0x01 -#endif /* ALTQ3_COMPAT */ /* * Define State structures. */ typedef struct cbqstate { -#ifdef ALTQ3_COMPAT - struct cbqstate *cbq_next; -#endif int cbq_qlen; /* # of packets in cbq */ struct rm_class *cbq_class_tbl[CBQ_MAX_CLASSES]; diff --git a/freebsd/sys/net/altq/altq_cdnr.c b/freebsd/sys/net/altq/altq_cdnr.c deleted file mode 100644 index 0f4eeec1..00000000 --- a/freebsd/sys/net/altq/altq_cdnr.c +++ /dev/null @@ -1,1384 +0,0 @@ -#include <machine/rtems-bsd-kernel-space.h> - -/*- - * Copyright (C) 1999-2002 - * Sony Computer Science Laboratories Inc. All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY SONY CSL AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL SONY CSL OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * $KAME: altq_cdnr.c,v 1.15 2005/04/13 03:44:24 suz Exp $ - * $FreeBSD$ - */ - -#include <rtems/bsd/local/opt_altq.h> -#include <rtems/bsd/local/opt_inet.h> -#include <rtems/bsd/local/opt_inet6.h> - -#include <sys/param.h> -#include <sys/malloc.h> -#include <sys/mbuf.h> -#include <sys/socket.h> -#include <sys/sockio.h> -#include <sys/systm.h> -#include <sys/proc.h> -#include <sys/errno.h> -#include <sys/kernel.h> -#include <sys/queue.h> - -#include <net/if.h> -#include <net/if_types.h> -#include <netinet/in.h> -#include <netinet/in_systm.h> -#include <netinet/ip.h> -#ifdef INET6 -#include <netinet/ip6.h> -#endif - -#include <net/altq/if_altq.h> -#include <net/altq/altq.h> -#ifdef ALTQ3_COMPAT -#include <net/altq/altq_conf.h> -#endif -#include <net/altq/altq_cdnr.h> - -#ifdef ALTQ3_COMPAT -/* - * diffserv traffic conditioning module - */ - -int altq_cdnr_enabled = 0; - -/* traffic conditioner is enabled by ALTQ_CDNR option in opt_altq.h */ -#ifdef ALTQ_CDNR - -/* cdnr_list keeps all cdnr's allocated. */ -static LIST_HEAD(, top_cdnr) tcb_list; - -static int altq_cdnr_input(struct mbuf *, int); -static struct top_cdnr *tcb_lookup(char *ifname); -static struct cdnr_block *cdnr_handle2cb(u_long); -static u_long cdnr_cb2handle(struct cdnr_block *); -static void *cdnr_cballoc(struct top_cdnr *, int, - struct tc_action *(*)(struct cdnr_block *, struct cdnr_pktinfo *)); -static void cdnr_cbdestroy(void *); -static int tca_verify_action(struct tc_action *); -static void tca_import_action(struct tc_action *, struct tc_action *); -static void tca_invalidate_action(struct tc_action *); - -static int generic_element_destroy(struct cdnr_block *); -static struct top_cdnr *top_create(struct ifaltq *); -static int top_destroy(struct top_cdnr *); -static struct cdnr_block *element_create(struct top_cdnr *, struct tc_action *); -static int element_destroy(struct cdnr_block *); -static void tb_import_profile(struct tbe *, struct tb_profile *); -static struct tbmeter *tbm_create(struct top_cdnr *, struct tb_profile *, - struct tc_action *, struct tc_action *); -static int tbm_destroy(struct tbmeter *); -static struct tc_action *tbm_input(struct cdnr_block *, struct cdnr_pktinfo *); -static struct trtcm *trtcm_create(struct top_cdnr *, - struct tb_profile *, struct tb_profile *, - struct tc_action *, struct tc_action *, struct tc_action *, - int); -static int trtcm_destroy(struct trtcm *); -static struct tc_action *trtcm_input(struct cdnr_block *, struct cdnr_pktinfo *); -static struct tswtcm *tswtcm_create(struct top_cdnr *, - u_int32_t, u_int32_t, u_int32_t, - struct tc_action *, struct tc_action *, struct tc_action *); -static int tswtcm_destroy(struct tswtcm *); -static struct tc_action *tswtcm_input(struct cdnr_block *, struct cdnr_pktinfo *); - -static int cdnrcmd_if_attach(char *); -static int cdnrcmd_if_detach(char *); -static int cdnrcmd_add_element(struct cdnr_add_element *); -static int cdnrcmd_delete_element(struct cdnr_delete_element *); -static int cdnrcmd_add_filter(struct cdnr_add_filter *); -static int cdnrcmd_delete_filter(struct cdnr_delete_filter *); -static int cdnrcmd_add_tbm(struct cdnr_add_tbmeter *); -static int cdnrcmd_modify_tbm(struct cdnr_modify_tbmeter *); -static int cdnrcmd_tbm_stats(struct cdnr_tbmeter_stats *); -static int cdnrcmd_add_trtcm(struct cdnr_add_trtcm *); -static int cdnrcmd_modify_trtcm(struct cdnr_modify_trtcm *); -static int cdnrcmd_tcm_stats(struct cdnr_tcm_stats *); -static int cdnrcmd_add_tswtcm(struct cdnr_add_tswtcm *); -static int cdnrcmd_modify_tswtcm(struct cdnr_modify_tswtcm *); -static int cdnrcmd_get_stats(struct cdnr_get_stats *); - -altqdev_decl(cdnr); - -/* - * top level input function called from ip_input. - * should be called before converting header fields to host-byte-order. - */ -int -altq_cdnr_input(m, af) - struct mbuf *m; - int af; /* address family */ -{ - struct ifnet *ifp; - struct ip *ip; - struct top_cdnr *top; - struct tc_action *tca; - struct cdnr_block *cb; - struct cdnr_pktinfo pktinfo; - - ifp = m->m_pkthdr.rcvif; - if (!ALTQ_IS_CNDTNING(&ifp->if_snd)) - /* traffic conditioner is not enabled on this interface */ - return (1); - - top = ifp->if_snd.altq_cdnr; - - ip = mtod(m, struct ip *); -#ifdef INET6 - if (af == AF_INET6) { - u_int32_t flowlabel; - - flowlabel = ((struct ip6_hdr *)ip)->ip6_flow; - pktinfo.pkt_dscp = (ntohl(flowlabel) >> 20) & DSCP_MASK; - } else -#endif - pktinfo.pkt_dscp = ip->ip_tos & DSCP_MASK; - pktinfo.pkt_len = m_pktlen(m); - - tca = NULL; - - cb = acc_classify(&top->tc_classifier, m, af); - if (cb != NULL) - tca = &cb->cb_action; - - if (tca == NULL) - tca = &top->tc_block.cb_action; - - while (1) { - PKTCNTR_ADD(&top->tc_cnts[tca->tca_code], pktinfo.pkt_len); - - switch (tca->tca_code) { - case TCACODE_PASS: - return (1); - case TCACODE_DROP: - m_freem(m); - return (0); - case TCACODE_RETURN: - return (0); - case TCACODE_MARK: -#ifdef INET6 - if (af == AF_INET6) { - struct ip6_hdr *ip6 = (struct ip6_hdr *)ip; - u_int32_t flowlabel; - - flowlabel = ntohl(ip6->ip6_flow); - flowlabel = (tca->tca_dscp << 20) | - (flowlabel & ~(DSCP_MASK << 20)); - ip6->ip6_flow = htonl(flowlabel); - } else -#endif - ip->ip_tos = tca->tca_dscp | - (ip->ip_tos & DSCP_CUMASK); - return (1); - case TCACODE_NEXT: - cb = tca->tca_next; - tca = (*cb->cb_input)(cb, &pktinfo); - break; - case TCACODE_NONE: - default: - return (1); - } - } -} - -static struct top_cdnr * -tcb_lookup(ifname) - char *ifname; -{ - struct top_cdnr *top; - struct ifnet *ifp; - - if ((ifp = ifunit(ifname)) != NULL) - LIST_FOREACH(top, &tcb_list, tc_next) - if (top->tc_ifq->altq_ifp == ifp) - return (top); - return (NULL); -} - -static struct cdnr_block * -cdnr_handle2cb(handle) - u_long handle; -{ - struct cdnr_block *cb; - - cb = (struct cdnr_block *)handle; - if (handle != ALIGN(cb)) - return (NULL); - - if (cb == NULL || cb->cb_handle != handle) - return (NULL); - return (cb); -} - -static u_long -cdnr_cb2handle(cb) - struct cdnr_block *cb; -{ - return (cb->cb_handle); -} - -static void * -cdnr_cballoc(top, type, input_func) - struct top_cdnr *top; - int type; - struct tc_action *(*input_func)(struct cdnr_block *, - struct cdnr_pktinfo *); -{ - struct cdnr_block *cb; - int size; - - switch (type) { - case TCETYPE_TOP: - size = sizeof(struct top_cdnr); - break; - case TCETYPE_ELEMENT: - size = sizeof(struct cdnr_block); - break; - case TCETYPE_TBMETER: - size = sizeof(struct tbmeter); - break; - case TCETYPE_TRTCM: - size = sizeof(struct trtcm); - break; - case TCETYPE_TSWTCM: - size = sizeof(struct tswtcm); - break; - default: - return (NULL); - } - - cb = malloc(size, M_DEVBUF, M_WAITOK); - if (cb == NULL) - return (NULL); - bzero(cb, size); - - cb->cb_len = size; - cb->cb_type = type; - cb->cb_ref = 0; - cb->cb_handle = (u_long)cb; - if (top == NULL) - cb->cb_top = (struct top_cdnr *)cb; - else - cb->cb_top = top; - - if (input_func != NULL) { - /* - * if this cdnr has an action function, - * make tc_action to call itself. - */ - cb->cb_action.tca_code = TCACODE_NEXT; - cb->cb_action.tca_next = cb; - cb->cb_input = input_func; - } else - cb->cb_action.tca_code = TCACODE_NONE; - - /* if this isn't top, register the element to the top level cdnr */ - if (top != NULL) - LIST_INSERT_HEAD(&top->tc_elements, cb, cb_next); - - return ((void *)cb); -} - -static void -cdnr_cbdestroy(cblock) - void *cblock; -{ - struct cdnr_block *cb = cblock; - - /* delete filters belonging to this cdnr */ - acc_discard_filters(&cb->cb_top->tc_classifier, cb, 0); - - /* remove from the top level cdnr */ - if (cb->cb_top != cblock) - LIST_REMOVE(cb, cb_next); - - free(cb, M_DEVBUF); -} - -/* - * conditioner common destroy routine - */ -static int -generic_element_destroy(cb) - struct cdnr_block *cb; -{ - int error = 0; - - switch (cb->cb_type) { - case TCETYPE_TOP: - error = top_destroy((struct top_cdnr *)cb); - break; - case TCETYPE_ELEMENT: - error = element_destroy(cb); - break; - case TCETYPE_TBMETER: - error = tbm_destroy((struct tbmeter *)cb); - break; - case TCETYPE_TRTCM: - error = trtcm_destroy((struct trtcm *)cb); - break; - case TCETYPE_TSWTCM: - error = tswtcm_destroy((struct tswtcm *)cb); - break; - default: - error = EINVAL; - } - return (error); -} - -static int -tca_verify_action(utca) - struct tc_action *utca; -{ - switch (utca->tca_code) { - case TCACODE_PASS: - case TCACODE_DROP: - case TCACODE_MARK: - /* these are ok */ - break; - - case TCACODE_HANDLE: - /* verify handle value */ - if (cdnr_handle2cb(utca->tca_handle) == NULL) - return (-1); - break; - - case TCACODE_NONE: - case TCACODE_RETURN: - case TCACODE_NEXT: - default: - /* should not be passed from a user */ - return (-1); - } - return (0); -} - -static void -tca_import_action(ktca, utca) - struct tc_action *ktca, *utca; -{ - struct cdnr_block *cb; - - *ktca = *utca; - if (ktca->tca_code == TCACODE_HANDLE) { - cb = cdnr_handle2cb(ktca->tca_handle); - if (cb == NULL) { - ktca->tca_code = TCACODE_NONE; - return; - } - ktca->tca_code = TCACODE_NEXT; - ktca->tca_next = cb; - cb->cb_ref++; - } else if (ktca->tca_code == TCACODE_MARK) { - ktca->tca_dscp &= DSCP_MASK; - } - return; -} - -static void -tca_invalidate_action(tca) - struct tc_action *tca; -{ - struct cdnr_block *cb; - - if (tca->tca_code == TCACODE_NEXT) { - cb = tca->tca_next; - if (cb == NULL) - return; - cb->cb_ref--; - } - tca->tca_code = TCACODE_NONE; -} - -/* - * top level traffic conditioner - */ -static struct top_cdnr * -top_create(ifq) - struct ifaltq *ifq; -{ - struct top_cdnr *top; - - if ((top = cdnr_cballoc(NULL, TCETYPE_TOP, NULL)) == NULL) - return (NULL); - - top->tc_ifq = ifq; - /* set default action for the top level conditioner */ - top->tc_block.cb_action.tca_code = TCACODE_PASS; - - LIST_INSERT_HEAD(&tcb_list, top, tc_next); - - ifq->altq_cdnr = top; - - return (top); -} - -static int -top_destroy(top) - struct top_cdnr *top; -{ - struct cdnr_block *cb; - - if (ALTQ_IS_CNDTNING(top->tc_ifq)) - ALTQ_CLEAR_CNDTNING(top->tc_ifq); - top->tc_ifq->altq_cdnr = NULL; - - /* - * destroy all the conditioner elements belonging to this interface - */ - while ((cb = LIST_FIRST(&top->tc_elements)) != NULL) { - while (cb != NULL && cb->cb_ref > 0) - cb = LIST_NEXT(cb, cb_next); - if (cb != NULL) - generic_element_destroy(cb); - } - - LIST_REMOVE(top, tc_next); - - cdnr_cbdestroy(top); - - /* if there is no active conditioner, remove the input hook */ - if (altq_input != NULL) { - LIST_FOREACH(top, &tcb_list, tc_next) - if (ALTQ_IS_CNDTNING(top->tc_ifq)) - break; - if (top == NULL) - altq_input = NULL; - } - - return (0); -} - -/* - * simple tc elements without input function (e.g., dropper and makers). - */ -static struct cdnr_block * -element_create(top, action) - struct top_cdnr *top; - struct tc_action *action; -{ - struct cdnr_block *cb; - - if (tca_verify_action(action) < 0) - return (NULL); - - if ((cb = cdnr_cballoc(top, TCETYPE_ELEMENT, NULL)) == NULL) - return (NULL); - - tca_import_action(&cb->cb_action, action); - - return (cb); -} - -static int -element_destroy(cb) - struct cdnr_block *cb; -{ - if (cb->cb_ref > 0) - return (EBUSY); - - tca_invalidate_action(&cb->cb_action); - - cdnr_cbdestroy(cb); - return (0); -} - -/* - * internal representation of token bucket parameters - * rate: byte_per_unittime << 32 - * (((bits_per_sec) / 8) << 32) / machclk_freq - * depth: byte << 32 - * - */ -#define TB_SHIFT 32 -#define TB_SCALE(x) ((u_int64_t)(x) << TB_SHIFT) -#define TB_UNSCALE(x) ((x) >> TB_SHIFT) - -static void -tb_import_profile(tb, profile) - struct tbe *tb; - struct tb_profile *profile; -{ - tb->rate = TB_SCALE(profile->rate / 8) / machclk_freq; - tb->depth = TB_SCALE(profile->depth); - if (tb->rate > 0) - tb->filluptime = tb->depth / tb->rate; - else - tb->filluptime = 0xffffffffffffffffLL; - tb->token = tb->depth; - tb->last = read_machclk(); -} - -/* - * simple token bucket meter - */ -static struct tbmeter * -tbm_create(top, profile, in_action, out_action) - struct top_cdnr *top; - struct tb_profile *profile; - struct tc_action *in_action, *out_action; -{ - struct tbmeter *tbm = NULL; - - if (tca_verify_action(in_action) < 0 - || tca_verify_action(out_action) < 0) - return (NULL); - - if ((tbm = cdnr_cballoc(top, TCETYPE_TBMETER, - tbm_input)) == NULL) - return (NULL); - - tb_import_profile(&tbm->tb, profile); - - tca_import_action(&tbm->in_action, in_action); - tca_import_action(&tbm->out_action, out_action); - - return (tbm); -} - -static int -tbm_destroy(tbm) - struct tbmeter *tbm; -{ - if (tbm->cdnrblk.cb_ref > 0) - return (EBUSY); - - tca_invalidate_action(&tbm->in_action); - tca_invalidate_action(&tbm->out_action); - - cdnr_cbdestroy(tbm); - return (0); -} - -static struct tc_action * -tbm_input(cb, pktinfo) - struct cdnr_block *cb; - struct cdnr_pktinfo *pktinfo; -{ - struct tbmeter *tbm = (struct tbmeter *)cb; - u_int64_t len; - u_int64_t interval, now; - - len = TB_SCALE(pktinfo->pkt_len); - - if (tbm->tb.token < len) { - now = read_machclk(); - interval = now - tbm->tb.last; - if (interval >= tbm->tb.filluptime) - tbm->tb.token = tbm->tb.depth; - else { - tbm->tb.token += interval * tbm->tb.rate; - if (tbm->tb.token > tbm->tb.depth) - tbm->tb.token = tbm->tb.depth; - } - tbm->tb.last = now; - } - - if (tbm->tb.token < len) { - PKTCNTR_ADD(&tbm->out_cnt, pktinfo->pkt_len); - return (&tbm->out_action); - } - - tbm->tb.token -= len; - PKTCNTR_ADD(&tbm->in_cnt, pktinfo->pkt_len); - return (&tbm->in_action); -} - -/* - * two rate three color marker - * as described in draft-heinanen-diffserv-trtcm-01.txt - */ -static struct trtcm * -trtcm_create(top, cmtd_profile, peak_profile, - green_action, yellow_action, red_action, coloraware) - struct top_cdnr *top; - struct tb_profile *cmtd_profile, *peak_profile; - struct tc_action *green_action, *yellow_action, *red_action; - int coloraware; -{ - struct trtcm *tcm = NULL; - - if (tca_verify_action(green_action) < 0 - || tca_verify_action(yellow_action) < 0 - || tca_verify_action(red_action) < 0) - return (NULL); - - if ((tcm = cdnr_cballoc(top, TCETYPE_TRTCM, - trtcm_input)) == NULL) - return (NULL); - - tb_import_profile(&tcm->cmtd_tb, cmtd_profile); - tb_import_profile(&tcm->peak_tb, peak_profile); - - tca_import_action(&tcm->green_action, green_action); - tca_import_action(&tcm->yellow_action, yellow_action); - tca_import_action(&tcm->red_action, red_action); - - /* set dscps to use */ - if (tcm->green_action.tca_code == TCACODE_MARK) - tcm->green_dscp = tcm->green_action.tca_dscp & DSCP_MASK; - else - tcm->green_dscp = DSCP_AF11; - if (tcm->yellow_action.tca_code == TCACODE_MARK) - tcm->yellow_dscp = tcm->yellow_action.tca_dscp & DSCP_MASK; - else - tcm->yellow_dscp = DSCP_AF12; - if (tcm->red_action.tca_code == TCACODE_MARK) - tcm->red_dscp = tcm->red_action.tca_dscp & DSCP_MASK; - else - tcm->red_dscp = DSCP_AF13; - - tcm->coloraware = coloraware; - - return (tcm); -} - -static int -trtcm_destroy(tcm) - struct trtcm *tcm; -{ - if (tcm->cdnrblk.cb_ref > 0) - return (EBUSY); - - tca_invalidate_action(&tcm->green_action); - tca_invalidate_action(&tcm->yellow_action); - tca_invalidate_action(&tcm->red_action); - - cdnr_cbdestroy(tcm); - return (0); -} - -static struct tc_action * -trtcm_input(cb, pktinfo) - struct cdnr_block *cb; - struct cdnr_pktinfo *pktinfo; -{ - struct trtcm *tcm = (struct trtcm *)cb; - u_int64_t len; - u_int64_t interval, now; - u_int8_t color; - - len = TB_SCALE(pktinfo->pkt_len); - if (tcm->coloraware) { - color = pktinfo->pkt_dscp; - if (color != tcm->yellow_dscp && color != tcm->red_dscp) - color = tcm->green_dscp; - } else { - /* if color-blind, precolor it as green */ - color = tcm->green_dscp; - } - - now = read_machclk(); - if (tcm->cmtd_tb.token < len) { - interval = now - tcm->cmtd_tb.last; - if (interval >= tcm->cmtd_tb.filluptime) - tcm->cmtd_tb.token = tcm->cmtd_tb.depth; - else { - tcm->cmtd_tb.token += interval * tcm->cmtd_tb.rate; - if (tcm->cmtd_tb.token > tcm->cmtd_tb.depth) - tcm->cmtd_tb.token = tcm->cmtd_tb.depth; - } - tcm->cmtd_tb.last = now; - } - if (tcm->peak_tb.token < len) { - interval = now - tcm->peak_tb.last; - if (interval >= tcm->peak_tb.filluptime) - tcm->peak_tb.token = tcm->peak_tb.depth; - else { - tcm->peak_tb.token += interval * tcm->peak_tb.rate; - if (tcm->peak_tb.token > tcm->peak_tb.depth) - tcm->peak_tb.token = tcm->peak_tb.depth; - } - tcm->peak_tb.last = now; - } - - if (color == tcm->red_dscp || tcm->peak_tb.token < len) { - pktinfo->pkt_dscp = tcm->red_dscp; - PKTCNTR_ADD(&tcm->red_cnt, pktinfo->pkt_len); - return (&tcm->red_action); - } - - if (color == tcm->yellow_dscp || tcm->cmtd_tb.token < len) { - pktinfo->pkt_dscp = tcm->yellow_dscp; - tcm->peak_tb.token -= len; - PKTCNTR_ADD(&tcm->yellow_cnt, pktinfo->pkt_len); - return (&tcm->yellow_action); - } - - pktinfo->pkt_dscp = tcm->green_dscp; - tcm->cmtd_tb.token -= len; - tcm->peak_tb.token -= len; - PKTCNTR_ADD(&tcm->green_cnt, pktinfo->pkt_len); - return (&tcm->green_action); -} - -/* - * time sliding window three color marker - * as described in draft-fang-diffserv-tc-tswtcm-00.txt - */ -static struct tswtcm * -tswtcm_create(top, cmtd_rate, peak_rate, avg_interval, - green_action, yellow_action, red_action) - struct top_cdnr *top; - u_int32_t cmtd_rate, peak_rate, avg_interval; - struct tc_action *green_action, *yellow_action, *red_action; -{ - struct tswtcm *tsw; - - if (tca_verify_action(green_action) < 0 - || tca_verify_action(yellow_action) < 0 - || tca_verify_action(red_action) < 0) - return (NULL); - - if ((tsw = cdnr_cballoc(top, TCETYPE_TSWTCM, - tswtcm_input)) == NULL) - return (NULL); - - tca_import_action(&tsw->green_action, green_action); - tca_import_action(&tsw->yellow_action, yellow_action); - tca_import_action(&tsw->red_action, red_action); - - /* set dscps to use */ - if (tsw->green_action.tca_code == TCACODE_MARK) - tsw->green_dscp = tsw->green_action.tca_dscp & DSCP_MASK; - else - tsw->green_dscp = DSCP_AF11; - if (tsw->yellow_action.tca_code == TCACODE_MARK) - tsw->yellow_dscp = tsw->yellow_action.tca_dscp & DSCP_MASK; - else - tsw->yellow_dscp = DSCP_AF12; - if (tsw->red_action.tca_code == TCACODE_MARK) - tsw->red_dscp = tsw->red_action.tca_dscp & DSCP_MASK; - else - tsw->red_dscp = DSCP_AF13; - - /* convert rates from bits/sec to bytes/sec */ - tsw->cmtd_rate = cmtd_rate / 8; - tsw->peak_rate = peak_rate / 8; - tsw->avg_rate = 0; - - /* timewin is converted from msec to machine clock unit */ - tsw->timewin = (u_int64_t)machclk_freq * avg_interval / 1000; - - return (tsw); -} - -static int -tswtcm_destroy(tsw) - struct tswtcm *tsw; -{ - if (tsw->cdnrblk.cb_ref > 0) - return (EBUSY); - - tca_invalidate_action(&tsw->green_action); - tca_invalidate_action(&tsw->yellow_action); - tca_invalidate_action(&tsw->red_action); - - cdnr_cbdestroy(tsw); - return (0); -} - -static struct tc_action * -tswtcm_input(cb, pktinfo) - struct cdnr_block *cb; - struct cdnr_pktinfo *pktinfo; -{ - struct tswtcm *tsw = (struct tswtcm *)cb; - int len; - u_int32_t avg_rate; - u_int64_t interval, now, tmp; - - /* - * rate estimator - */ - len = pktinfo->pkt_len; - now = read_machclk(); - - interval = now - tsw->t_front; - /* - * calculate average rate: - * avg = (avg * timewin + pkt_len)/(timewin + interval) - * pkt_len needs to be multiplied by machclk_freq in order to - * get (bytes/sec). - * note: when avg_rate (bytes/sec) and timewin (machclk unit) are - * less than 32 bits, the following 64-bit operation has enough - * precision. - */ - tmp = ((u_int64_t)tsw->avg_rate * tsw->timewin - + (u_int64_t)len * machclk_freq) / (tsw->timewin + interval); - tsw->avg_rate = avg_rate = (u_int32_t)tmp; - tsw->t_front = now; - - /* - * marker - */ - if (avg_rate > tsw->cmtd_rate) { - u_int32_t randval = arc4random() % avg_rate; - - if (avg_rate > tsw->peak_rate) { - if (randval < avg_rate - tsw->peak_rate) { - /* mark red */ - pktinfo->pkt_dscp = tsw->red_dscp; - PKTCNTR_ADD(&tsw->red_cnt, len); - return (&tsw->red_action); - } else if (randval < avg_rate - tsw->cmtd_rate) - goto mark_yellow; - } else { - /* peak_rate >= avg_rate > cmtd_rate */ - if (randval < avg_rate - tsw->cmtd_rate) { - mark_yellow: - pktinfo->pkt_dscp = tsw->yellow_dscp; - PKTCNTR_ADD(&tsw->yellow_cnt, len); - return (&tsw->yellow_action); - } - } - } - - /* mark green */ - pktinfo->pkt_dscp = tsw->green_dscp; - PKTCNTR_ADD(&tsw->green_cnt, len); - return (&tsw->green_action); -} - -/* - * ioctl requests - */ -static int -cdnrcmd_if_attach(ifname) - char *ifname; -{ - struct ifnet *ifp; - struct top_cdnr *top; - - if ((ifp = ifunit(ifname)) == NULL) - return (EBADF); - - if (ifp->if_snd.altq_cdnr != NULL) - return (EBUSY); - - if ((top = top_create(&ifp->if_snd)) == NULL) - return (ENOMEM); - return (0); -} - -static int -cdnrcmd_if_detach(ifname) - char *ifname; -{ - struct top_cdnr *top; - - if ((top = tcb_lookup(ifname)) == NULL) - return (EBADF); - - return top_destroy(top); -} - -static int -cdnrcmd_add_element(ap) - struct cdnr_add_element *ap; -{ - struct top_cdnr *top; - struct cdnr_block *cb; - - if ((top = tcb_lookup(ap->iface.cdnr_ifname)) == NULL) - return (EBADF); - - cb = element_create(top, &ap->action); - if (cb == NULL) - return (EINVAL); - /* return a class handle to the user */ - ap->cdnr_handle = cdnr_cb2handle(cb); - return (0); -} - -static int -cdnrcmd_delete_element(ap) - struct cdnr_delete_element *ap; -{ - struct top_cdnr *top; - struct cdnr_block *cb; - - if ((top = tcb_lookup(ap->iface.cdnr_ifname)) == NULL) - return (EBADF); - - if ((cb = cdnr_handle2cb(ap->cdnr_handle)) == NULL) - return (EINVAL); - - if (cb->cb_type != TCETYPE_ELEMENT) - return generic_element_destroy(cb); - - return element_destroy(cb); -} - -static int -cdnrcmd_add_filter(ap) - struct cdnr_add_filter *ap; -{ - struct top_cdnr *top; - struct cdnr_block *cb; - - if ((top = tcb_lookup(ap->iface.cdnr_ifname)) == NULL) - return (EBADF); - - if ((cb = cdnr_handle2cb(ap->cdnr_handle)) == NULL) - return (EINVAL); - - return acc_add_filter(&top->tc_classifier, &ap->filter, - cb, &ap->filter_handle); -} - -static int -cdnrcmd_delete_filter(ap) - struct cdnr_delete_filter *ap; -{ - struct top_cdnr *top; - - if ((top = tcb_lookup(ap->iface.cdnr_ifname)) == NULL) - return (EBADF); - - return acc_delete_filter(&top->tc_classifier, ap->filter_handle); -} - -static int -cdnrcmd_add_tbm(ap) - struct cdnr_add_tbmeter *ap; -{ - struct top_cdnr *top; - struct tbmeter *tbm; - - if ((top = tcb_lookup(ap->iface.cdnr_ifname)) == NULL) - return (EBADF); - - tbm = tbm_create(top, &ap->profile, &ap->in_action, &ap->out_action); - if (tbm == NULL) - return (EINVAL); - /* return a class handle to the user */ - ap->cdnr_handle = cdnr_cb2handle(&tbm->cdnrblk); - return (0); -} - -static int -cdnrcmd_modify_tbm(ap) - struct cdnr_modify_tbmeter *ap; -{ - struct tbmeter *tbm; - - if ((tbm = (struct tbmeter *)cdnr_handle2cb(ap->cdnr_handle)) == NULL) - return (EINVAL); - - tb_import_profile(&tbm->tb, &ap->profile); - - return (0); -} - -static int -cdnrcmd_tbm_stats(ap) - struct cdnr_tbmeter_stats *ap; -{ - struct tbmeter *tbm; - - if ((tbm = (struct tbmeter *)cdnr_handle2cb(ap->cdnr_handle)) == NULL) - return (EINVAL); - - ap->in_cnt = tbm->in_cnt; - ap->out_cnt = tbm->out_cnt; - - return (0); -} - -static int -cdnrcmd_add_trtcm(ap) - struct cdnr_add_trtcm *ap; -{ - struct top_cdnr *top; - struct trtcm *tcm; - - if ((top = tcb_lookup(ap->iface.cdnr_ifname)) == NULL) - return (EBADF); - - tcm = trtcm_create(top, &ap->cmtd_profile, &ap->peak_profile, - &ap->green_action, &ap->yellow_action, - &ap->red_action, ap->coloraware); - if (tcm == NULL) - return (EINVAL); - - /* return a class handle to the user */ - ap->cdnr_handle = cdnr_cb2handle(&tcm->cdnrblk); - return (0); -} - -static int -cdnrcmd_modify_trtcm(ap) - struct cdnr_modify_trtcm *ap; -{ - struct trtcm *tcm; - - if ((tcm = (struct trtcm *)cdnr_handle2cb(ap->cdnr_handle)) == NULL) - return (EINVAL); - - tb_import_profile(&tcm->cmtd_tb, &ap->cmtd_profile); - tb_import_profile(&tcm->peak_tb, &ap->peak_profile); - - return (0); -} - -static int -cdnrcmd_tcm_stats(ap) - struct cdnr_tcm_stats *ap; -{ - struct cdnr_block *cb; - - if ((cb = cdnr_handle2cb(ap->cdnr_handle)) == NULL) - return (EINVAL); - - if (cb->cb_type == TCETYPE_TRTCM) { - struct trtcm *tcm = (struct trtcm *)cb; - - ap->green_cnt = tcm->green_cnt; - ap->yellow_cnt = tcm->yellow_cnt; - ap->red_cnt = tcm->red_cnt; - } else if (cb->cb_type == TCETYPE_TSWTCM) { - struct tswtcm *tsw = (struct tswtcm *)cb; - - ap->green_cnt = tsw->green_cnt; - ap->yellow_cnt = tsw->yellow_cnt; - ap->red_cnt = tsw->red_cnt; - } else - return (EINVAL); - - return (0); -} - -static int -cdnrcmd_add_tswtcm(ap) - struct cdnr_add_tswtcm *ap; -{ - struct top_cdnr *top; - struct tswtcm *tsw; - - if ((top = tcb_lookup(ap->iface.cdnr_ifname)) == NULL) - return (EBADF); - - if (ap->cmtd_rate > ap->peak_rate) - return (EINVAL); - - tsw = tswtcm_create(top, ap->cmtd_rate, ap->peak_rate, - ap->avg_interval, &ap->green_action, - &ap->yellow_action, &ap->red_action); - if (tsw == NULL) - return (EINVAL); - - /* return a class handle to the user */ - ap->cdnr_handle = cdnr_cb2handle(&tsw->cdnrblk); - return (0); -} - -static int -cdnrcmd_modify_tswtcm(ap) - struct cdnr_modify_tswtcm *ap; -{ - struct tswtcm *tsw; - - if ((tsw = (struct tswtcm *)cdnr_handle2cb(ap->cdnr_handle)) == NULL) - return (EINVAL); - - if (ap->cmtd_rate > ap->peak_rate) - return (EINVAL); - - /* convert rates from bits/sec to bytes/sec */ - tsw->cmtd_rate = ap->cmtd_rate / 8; - tsw->peak_rate = ap->peak_rate / 8; - tsw->avg_rate = 0; - - /* timewin is converted from msec to machine clock unit */ - tsw->timewin = (u_int64_t)machclk_freq * ap->avg_interval / 1000; - - return (0); -} - -static int -cdnrcmd_get_stats(ap) - struct cdnr_get_stats *ap; -{ - struct top_cdnr *top; - struct cdnr_block *cb; - struct tbmeter *tbm; - struct trtcm *tcm; - struct tswtcm *tsw; - struct tce_stats tce, *usp; - int error, n, nskip, nelements; - - if ((top = tcb_lookup(ap->iface.cdnr_ifname)) == NULL) - return (EBADF); - - /* copy action stats */ - bcopy(top->tc_cnts, ap->cnts, sizeof(ap->cnts)); - - /* stats for each element */ - nelements = ap->nelements; - usp = ap->tce_stats; - if (nelements <= 0 || usp == NULL) - return (0); - - nskip = ap->nskip; - n = 0; - LIST_FOREACH(cb, &top->tc_elements, cb_next) { - if (nskip > 0) { - nskip--; - continue; - } - - bzero(&tce, sizeof(tce)); - tce.tce_handle = cb->cb_handle; - tce.tce_type = cb->cb_type; - switch (cb->cb_type) { - case TCETYPE_TBMETER: - tbm = (struct tbmeter *)cb; - tce.tce_cnts[0] = tbm->in_cnt; - tce.tce_cnts[1] = tbm->out_cnt; - break; - case TCETYPE_TRTCM: - tcm = (struct trtcm *)cb; - tce.tce_cnts[0] = tcm->green_cnt; - tce.tce_cnts[1] = tcm->yellow_cnt; - tce.tce_cnts[2] = tcm->red_cnt; - break; - case TCETYPE_TSWTCM: - tsw = (struct tswtcm *)cb; - tce.tce_cnts[0] = tsw->green_cnt; - tce.tce_cnts[1] = tsw->yellow_cnt; - tce.tce_cnts[2] = tsw->red_cnt; - break; - default: - continue; - } - - if ((error = copyout((caddr_t)&tce, (caddr_t)usp++, - sizeof(tce))) != 0) - return (error); - - if (++n == nelements) - break; - } - ap->nelements = n; - - return (0); -} - -/* - * conditioner device interface - */ -int -cdnropen(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - if (machclk_freq == 0) - init_machclk(); - - if (machclk_freq == 0) { - printf("cdnr: no cpu clock available!\n"); - return (ENXIO); - } - - /* everything will be done when the queueing scheme is attached. */ - return 0; -} - -int -cdnrclose(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - struct top_cdnr *top; - int err, error = 0; - - while ((top = LIST_FIRST(&tcb_list)) != NULL) { - /* destroy all */ - err = top_destroy(top); - if (err != 0 && error == 0) - error = err; - } - altq_input = NULL; - - return (error); -} - -int -cdnrioctl(dev, cmd, addr, flag, p) - dev_t dev; - ioctlcmd_t cmd; - caddr_t addr; - int flag; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - struct top_cdnr *top; - struct cdnr_interface *ifacep; - int s, error = 0; - - /* check super-user privilege */ - switch (cmd) { - case CDNR_GETSTATS: - break; - default: -#if (__FreeBSD_version > 700000) - if ((error = priv_check(p, PRIV_ALTQ_MANAGE)) != 0) -#elsif (__FreeBSD_version > 400000) - if ((error = suser(p)) != 0) -#else - if ((error = suser(p->p_ucred, &p->p_acflag)) != 0) -#endif - return (error); - break; - } - - s = splnet(); - switch (cmd) { - - case CDNR_IF_ATTACH: - ifacep = (struct cdnr_interface *)addr; - error = cdnrcmd_if_attach(ifacep->cdnr_ifname); - break; - - case CDNR_IF_DETACH: - ifacep = (struct cdnr_interface *)addr; - error = cdnrcmd_if_detach(ifacep->cdnr_ifname); - break; - - case CDNR_ENABLE: - case CDNR_DISABLE: - ifacep = (struct cdnr_interface *)addr; - if ((top = tcb_lookup(ifacep->cdnr_ifname)) == NULL) { - error = EBADF; - break; - } - - switch (cmd) { - - case CDNR_ENABLE: - ALTQ_SET_CNDTNING(top->tc_ifq); - if (altq_input == NULL) - altq_input = altq_cdnr_input; - break; - - case CDNR_DISABLE: - ALTQ_CLEAR_CNDTNING(top->tc_ifq); - LIST_FOREACH(top, &tcb_list, tc_next) - if (ALTQ_IS_CNDTNING(top->tc_ifq)) - break; - if (top == NULL) - altq_input = NULL; - break; - } - break; - - case CDNR_ADD_ELEM: - error = cdnrcmd_add_element((struct cdnr_add_element *)addr); - break; - - case CDNR_DEL_ELEM: - error = cdnrcmd_delete_element((struct cdnr_delete_element *)addr); - break; - - case CDNR_ADD_TBM: - error = cdnrcmd_add_tbm((struct cdnr_add_tbmeter *)addr); - break; - - case CDNR_MOD_TBM: - error = cdnrcmd_modify_tbm((struct cdnr_modify_tbmeter *)addr); - break; - - case CDNR_TBM_STATS: - error = cdnrcmd_tbm_stats((struct cdnr_tbmeter_stats *)addr); - break; - - case CDNR_ADD_TCM: - error = cdnrcmd_add_trtcm((struct cdnr_add_trtcm *)addr); - break; - - case CDNR_MOD_TCM: - error = cdnrcmd_modify_trtcm((struct cdnr_modify_trtcm *)addr); - break; - - case CDNR_TCM_STATS: - error = cdnrcmd_tcm_stats((struct cdnr_tcm_stats *)addr); - break; - - case CDNR_ADD_FILTER: - error = cdnrcmd_add_filter((struct cdnr_add_filter *)addr); - break; - - case CDNR_DEL_FILTER: - error = cdnrcmd_delete_filter((struct cdnr_delete_filter *)addr); - break; - - case CDNR_GETSTATS: - error = cdnrcmd_get_stats((struct cdnr_get_stats *)addr); - break; - - case CDNR_ADD_TSW: - error = cdnrcmd_add_tswtcm((struct cdnr_add_tswtcm *)addr); - break; - - case CDNR_MOD_TSW: - error = cdnrcmd_modify_tswtcm((struct cdnr_modify_tswtcm *)addr); - break; - - default: - error = EINVAL; - break; - } - splx(s); - - return error; -} - -#ifdef KLD_MODULE - -static struct altqsw cdnr_sw = - {"cdnr", cdnropen, cdnrclose, cdnrioctl}; - -ALTQ_MODULE(altq_cdnr, ALTQT_CDNR, &cdnr_sw); - -#endif /* KLD_MODULE */ - -#endif /* ALTQ3_COMPAT */ -#endif /* ALTQ_CDNR */ diff --git a/freebsd/sys/net/altq/altq_hfsc.c b/freebsd/sys/net/altq/altq_hfsc.c index 8d8fdfdc..202915a8 100644 --- a/freebsd/sys/net/altq/altq_hfsc.c +++ b/freebsd/sys/net/altq/altq_hfsc.c @@ -72,9 +72,6 @@ #include <netpfil/pf/pf_mtag.h> #include <net/altq/altq.h> #include <net/altq/altq_hfsc.h> -#ifdef ALTQ3_COMPAT -#include <net/altq/altq_conf.h> -#endif /* * function prototypes @@ -139,23 +136,6 @@ static void get_class_stats_v1(struct hfsc_classstats_v1 *, static struct hfsc_class *clh_to_clp(struct hfsc_if *, u_int32_t); -#ifdef ALTQ3_COMPAT -static struct hfsc_if *hfsc_attach(struct ifaltq *, u_int); -static int hfsc_detach(struct hfsc_if *); -static int hfsc_class_modify(struct hfsc_class *, struct service_curve *, - struct service_curve *, struct service_curve *); - -static int hfsccmd_if_attach(struct hfsc_attach *); -static int hfsccmd_if_detach(struct hfsc_interface *); -static int hfsccmd_add_class(struct hfsc_add_class *); -static int hfsccmd_delete_class(struct hfsc_delete_class *); -static int hfsccmd_modify_class(struct hfsc_modify_class *); -static int hfsccmd_add_filter(struct hfsc_add_filter *); -static int hfsccmd_delete_filter(struct hfsc_delete_filter *); -static int hfsccmd_class_stats(struct hfsc_class_stats *); - -altqdev_decl(hfsc); -#endif /* ALTQ3_COMPAT */ /* * macros @@ -164,10 +144,6 @@ altqdev_decl(hfsc); #define HT_INFINITY 0xffffffffffffffffULL /* infinite time value */ -#ifdef ALTQ3_COMPAT -/* hif_list keeps all hfsc_if's allocated. */ -static struct hfsc_if *hif_list = NULL; -#endif /* ALTQ3_COMPAT */ int hfsc_pfattach(struct pf_altq *a) @@ -334,10 +310,6 @@ hfsc_clear_interface(struct hfsc_if *hif) { struct hfsc_class *cl; -#ifdef ALTQ3_COMPAT - /* free the filters for this interface */ - acc_discard_filters(&hif->hif_classifier, NULL, 1); -#endif /* clear out the classes */ while (hif->hif_rootclass != NULL && @@ -599,10 +571,6 @@ hfsc_class_destroy(struct hfsc_class *cl) s = splnet(); IFQ_LOCK(cl->cl_hif->hif_ifq); -#ifdef ALTQ3_COMPAT - /* delete filters referencing to this class */ - acc_discard_filters(&cl->cl_hif->hif_classifier, cl, 0); -#endif /* ALTQ3_COMPAT */ if (!qempty(cl->cl_q)) hfsc_purgeq(cl); @@ -716,10 +684,6 @@ hfsc_enqueue(struct ifaltq *ifq, struct mbuf *m, struct altq_pktattr *pktattr) cl = NULL; if ((t = pf_find_mtag(m)) != NULL) cl = clh_to_clp(hif, t->qid); -#ifdef ALTQ3_COMPAT - else if ((ifq->altq_flags & ALTQF_CLASSIFY) && pktattr != NULL) - cl = pktattr->pattr_class; -#endif if (cl == NULL || is_a_parent_class(cl)) { cl = hif->hif_defaultclass; if (cl == NULL) { @@ -727,12 +691,7 @@ hfsc_enqueue(struct ifaltq *ifq, struct mbuf *m, struct altq_pktattr *pktattr) return (ENOBUFS); } } -#ifdef ALTQ3_COMPAT - if (pktattr != NULL) - cl->cl_pktattr = pktattr; /* save proto hdr used by ECN */ - else -#endif - cl->cl_pktattr = NULL; + cl->cl_pktattr = NULL; len = m_pktlen(m); if (hfsc_addq(cl, m) != 0) { /* drop occurred. mbuf was freed in hfsc_addq. */ @@ -1790,542 +1749,5 @@ clh_to_clp(struct hfsc_if *hif, u_int32_t chandle) return (NULL); } -#ifdef ALTQ3_COMPAT -static struct hfsc_if * -hfsc_attach(ifq, bandwidth) - struct ifaltq *ifq; - u_int bandwidth; -{ - struct hfsc_if *hif; - - hif = malloc(sizeof(struct hfsc_if), M_DEVBUF, M_WAITOK); - if (hif == NULL) - return (NULL); - bzero(hif, sizeof(struct hfsc_if)); - - hif->hif_eligible = ellist_alloc(); - if (hif->hif_eligible == NULL) { - free(hif, M_DEVBUF); - return NULL; - } - - hif->hif_ifq = ifq; - - /* add this state to the hfsc list */ - hif->hif_next = hif_list; - hif_list = hif; - - return (hif); -} - -static int -hfsc_detach(hif) - struct hfsc_if *hif; -{ - (void)hfsc_clear_interface(hif); - (void)hfsc_class_destroy(hif->hif_rootclass); - - /* remove this interface from the hif list */ - if (hif_list == hif) - hif_list = hif->hif_next; - else { - struct hfsc_if *h; - - for (h = hif_list; h != NULL; h = h->hif_next) - if (h->hif_next == hif) { - h->hif_next = hif->hif_next; - break; - } - ASSERT(h != NULL); - } - - ellist_destroy(hif->hif_eligible); - - free(hif, M_DEVBUF); - - return (0); -} - -static int -hfsc_class_modify(cl, rsc, fsc, usc) - struct hfsc_class *cl; - struct service_curve *rsc, *fsc, *usc; -{ - struct internal_sc *rsc_tmp, *fsc_tmp, *usc_tmp; - u_int64_t cur_time; - int s; - - rsc_tmp = fsc_tmp = usc_tmp = NULL; - if (rsc != NULL && (rsc->m1 != 0 || rsc->m2 != 0) && - cl->cl_rsc == NULL) { - rsc_tmp = malloc(sizeof(struct internal_sc), - M_DEVBUF, M_WAITOK); - if (rsc_tmp == NULL) - return (ENOMEM); - } - if (fsc != NULL && (fsc->m1 != 0 || fsc->m2 != 0) && - cl->cl_fsc == NULL) { - fsc_tmp = malloc(sizeof(struct internal_sc), - M_DEVBUF, M_WAITOK); - if (fsc_tmp == NULL) { - free(rsc_tmp); - return (ENOMEM); - } - } - if (usc != NULL && (usc->m1 != 0 || usc->m2 != 0) && - cl->cl_usc == NULL) { - usc_tmp = malloc(sizeof(struct internal_sc), - M_DEVBUF, M_WAITOK); - if (usc_tmp == NULL) { - free(rsc_tmp); - free(fsc_tmp); - return (ENOMEM); - } - } - - cur_time = read_machclk(); - s = splnet(); - IFQ_LOCK(cl->cl_hif->hif_ifq); - - if (rsc != NULL) { - if (rsc->m1 == 0 && rsc->m2 == 0) { - if (cl->cl_rsc != NULL) { - if (!qempty(cl->cl_q)) - hfsc_purgeq(cl); - free(cl->cl_rsc, M_DEVBUF); - cl->cl_rsc = NULL; - } - } else { - if (cl->cl_rsc == NULL) - cl->cl_rsc = rsc_tmp; - sc2isc(rsc, cl->cl_rsc); - rtsc_init(&cl->cl_deadline, cl->cl_rsc, cur_time, - cl->cl_cumul); - cl->cl_eligible = cl->cl_deadline; - if (cl->cl_rsc->sm1 <= cl->cl_rsc->sm2) { - cl->cl_eligible.dx = 0; - cl->cl_eligible.dy = 0; - } - } - } - - if (fsc != NULL) { - if (fsc->m1 == 0 && fsc->m2 == 0) { - if (cl->cl_fsc != NULL) { - if (!qempty(cl->cl_q)) - hfsc_purgeq(cl); - free(cl->cl_fsc, M_DEVBUF); - cl->cl_fsc = NULL; - } - } else { - if (cl->cl_fsc == NULL) - cl->cl_fsc = fsc_tmp; - sc2isc(fsc, cl->cl_fsc); - rtsc_init(&cl->cl_virtual, cl->cl_fsc, cl->cl_vt, - cl->cl_total); - } - } - - if (usc != NULL) { - if (usc->m1 == 0 && usc->m2 == 0) { - if (cl->cl_usc != NULL) { - free(cl->cl_usc, M_DEVBUF); - cl->cl_usc = NULL; - cl->cl_myf = 0; - } - } else { - if (cl->cl_usc == NULL) - cl->cl_usc = usc_tmp; - sc2isc(usc, cl->cl_usc); - rtsc_init(&cl->cl_ulimit, cl->cl_usc, cur_time, - cl->cl_total); - } - } - - if (!qempty(cl->cl_q)) { - if (cl->cl_rsc != NULL) - update_ed(cl, m_pktlen(qhead(cl->cl_q))); - if (cl->cl_fsc != NULL) - update_vf(cl, 0, cur_time); - /* is this enough? */ - } - - IFQ_UNLOCK(cl->cl_hif->hif_ifq); - splx(s); - - return (0); -} - -/* - * hfsc device interface - */ -int -hfscopen(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - if (machclk_freq == 0) - init_machclk(); - - if (machclk_freq == 0) { - printf("hfsc: no cpu clock available!\n"); - return (ENXIO); - } - - /* everything will be done when the queueing scheme is attached. */ - return 0; -} - -int -hfscclose(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - struct hfsc_if *hif; - int err, error = 0; - - while ((hif = hif_list) != NULL) { - /* destroy all */ - if (ALTQ_IS_ENABLED(hif->hif_ifq)) - altq_disable(hif->hif_ifq); - - err = altq_detach(hif->hif_ifq); - if (err == 0) - err = hfsc_detach(hif); - if (err != 0 && error == 0) - error = err; - } - - return error; -} - -int -hfscioctl(dev, cmd, addr, flag, p) - dev_t dev; - ioctlcmd_t cmd; - caddr_t addr; - int flag; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - struct hfsc_if *hif; - struct hfsc_interface *ifacep; - int error = 0; - - /* check super-user privilege */ - switch (cmd) { - case HFSC_GETSTATS: - break; - default: -#if (__FreeBSD_version > 700000) - if ((error = priv_check(p, PRIV_ALTQ_MANAGE)) != 0) - return (error); -#elsif (__FreeBSD_version > 400000) - if ((error = suser(p)) != 0) - return (error); -#else - if ((error = suser(p->p_ucred, &p->p_acflag)) != 0) - return (error); -#endif - break; - } - - switch (cmd) { - - case HFSC_IF_ATTACH: - error = hfsccmd_if_attach((struct hfsc_attach *)addr); - break; - - case HFSC_IF_DETACH: - error = hfsccmd_if_detach((struct hfsc_interface *)addr); - break; - - case HFSC_ENABLE: - case HFSC_DISABLE: - case HFSC_CLEAR_HIERARCHY: - ifacep = (struct hfsc_interface *)addr; - if ((hif = altq_lookup(ifacep->hfsc_ifname, - ALTQT_HFSC)) == NULL) { - error = EBADF; - break; - } - - switch (cmd) { - - case HFSC_ENABLE: - if (hif->hif_defaultclass == NULL) { -#ifdef ALTQ_DEBUG - printf("hfsc: no default class\n"); -#endif - error = EINVAL; - break; - } - error = altq_enable(hif->hif_ifq); - break; - - case HFSC_DISABLE: - error = altq_disable(hif->hif_ifq); - break; - - case HFSC_CLEAR_HIERARCHY: - hfsc_clear_interface(hif); - break; - } - break; - - case HFSC_ADD_CLASS: - error = hfsccmd_add_class((struct hfsc_add_class *)addr); - break; - - case HFSC_DEL_CLASS: - error = hfsccmd_delete_class((struct hfsc_delete_class *)addr); - break; - - case HFSC_MOD_CLASS: - error = hfsccmd_modify_class((struct hfsc_modify_class *)addr); - break; - - case HFSC_ADD_FILTER: - error = hfsccmd_add_filter((struct hfsc_add_filter *)addr); - break; - - case HFSC_DEL_FILTER: - error = hfsccmd_delete_filter((struct hfsc_delete_filter *)addr); - break; - - case HFSC_GETSTATS: - error = hfsccmd_class_stats((struct hfsc_class_stats *)addr); - break; - - default: - error = EINVAL; - break; - } - return error; -} - -static int -hfsccmd_if_attach(ap) - struct hfsc_attach *ap; -{ - struct hfsc_if *hif; - struct ifnet *ifp; - int error; - - if ((ifp = ifunit(ap->iface.hfsc_ifname)) == NULL) - return (ENXIO); - - if ((hif = hfsc_attach(&ifp->if_snd, ap->bandwidth)) == NULL) - return (ENOMEM); - - /* - * set HFSC to this ifnet structure. - */ - if ((error = altq_attach(&ifp->if_snd, ALTQT_HFSC, hif, - hfsc_enqueue, hfsc_dequeue, hfsc_request, - &hif->hif_classifier, acc_classify)) != 0) - (void)hfsc_detach(hif); - - return (error); -} - -static int -hfsccmd_if_detach(ap) - struct hfsc_interface *ap; -{ - struct hfsc_if *hif; - int error; - - if ((hif = altq_lookup(ap->hfsc_ifname, ALTQT_HFSC)) == NULL) - return (EBADF); - - if (ALTQ_IS_ENABLED(hif->hif_ifq)) - altq_disable(hif->hif_ifq); - - if ((error = altq_detach(hif->hif_ifq))) - return (error); - - return hfsc_detach(hif); -} - -static int -hfsccmd_add_class(ap) - struct hfsc_add_class *ap; -{ - struct hfsc_if *hif; - struct hfsc_class *cl, *parent; - int i; - - if ((hif = altq_lookup(ap->iface.hfsc_ifname, ALTQT_HFSC)) == NULL) - return (EBADF); - - if (ap->parent_handle == HFSC_NULLCLASS_HANDLE && - hif->hif_rootclass == NULL) - parent = NULL; - else if ((parent = clh_to_clp(hif, ap->parent_handle)) == NULL) - return (EINVAL); - - /* assign a class handle (use a free slot number for now) */ - for (i = 1; i < HFSC_MAX_CLASSES; i++) - if (hif->hif_class_tbl[i] == NULL) - break; - if (i == HFSC_MAX_CLASSES) - return (EBUSY); - - if ((cl = hfsc_class_create(hif, &ap->service_curve, NULL, NULL, - parent, ap->qlimit, ap->flags, i)) == NULL) - return (ENOMEM); - - /* return a class handle to the user */ - ap->class_handle = i; - - return (0); -} - -static int -hfsccmd_delete_class(ap) - struct hfsc_delete_class *ap; -{ - struct hfsc_if *hif; - struct hfsc_class *cl; - - if ((hif = altq_lookup(ap->iface.hfsc_ifname, ALTQT_HFSC)) == NULL) - return (EBADF); - - if ((cl = clh_to_clp(hif, ap->class_handle)) == NULL) - return (EINVAL); - - return hfsc_class_destroy(cl); -} - -static int -hfsccmd_modify_class(ap) - struct hfsc_modify_class *ap; -{ - struct hfsc_if *hif; - struct hfsc_class *cl; - struct service_curve *rsc = NULL; - struct service_curve *fsc = NULL; - struct service_curve *usc = NULL; - - if ((hif = altq_lookup(ap->iface.hfsc_ifname, ALTQT_HFSC)) == NULL) - return (EBADF); - - if ((cl = clh_to_clp(hif, ap->class_handle)) == NULL) - return (EINVAL); - - if (ap->sctype & HFSC_REALTIMESC) - rsc = &ap->service_curve; - if (ap->sctype & HFSC_LINKSHARINGSC) - fsc = &ap->service_curve; - if (ap->sctype & HFSC_UPPERLIMITSC) - usc = &ap->service_curve; - - return hfsc_class_modify(cl, rsc, fsc, usc); -} - -static int -hfsccmd_add_filter(ap) - struct hfsc_add_filter *ap; -{ - struct hfsc_if *hif; - struct hfsc_class *cl; - - if ((hif = altq_lookup(ap->iface.hfsc_ifname, ALTQT_HFSC)) == NULL) - return (EBADF); - - if ((cl = clh_to_clp(hif, ap->class_handle)) == NULL) - return (EINVAL); - - if (is_a_parent_class(cl)) { -#ifdef ALTQ_DEBUG - printf("hfsccmd_add_filter: not a leaf class!\n"); -#endif - return (EINVAL); - } - - return acc_add_filter(&hif->hif_classifier, &ap->filter, - cl, &ap->filter_handle); -} - -static int -hfsccmd_delete_filter(ap) - struct hfsc_delete_filter *ap; -{ - struct hfsc_if *hif; - - if ((hif = altq_lookup(ap->iface.hfsc_ifname, ALTQT_HFSC)) == NULL) - return (EBADF); - - return acc_delete_filter(&hif->hif_classifier, - ap->filter_handle); -} - -static int -hfsccmd_class_stats(ap) - struct hfsc_class_stats *ap; -{ - struct hfsc_if *hif; - struct hfsc_class *cl; - struct hfsc_classstats stats, *usp; - int n, nclasses, error; - - if ((hif = altq_lookup(ap->iface.hfsc_ifname, ALTQT_HFSC)) == NULL) - return (EBADF); - - ap->cur_time = read_machclk(); - ap->machclk_freq = machclk_freq; - ap->hif_classes = hif->hif_classes; - ap->hif_packets = hif->hif_packets; - - /* skip the first N classes in the tree */ - nclasses = ap->nskip; - for (cl = hif->hif_rootclass, n = 0; cl != NULL && n < nclasses; - cl = hfsc_nextclass(cl), n++) - ; - if (n != nclasses) - return (EINVAL); - - /* then, read the next N classes in the tree */ - nclasses = ap->nclasses; - usp = ap->stats; - for (n = 0; cl != NULL && n < nclasses; cl = hfsc_nextclass(cl), n++) { - - get_class_stats(&stats, cl); - - if ((error = copyout((caddr_t)&stats, (caddr_t)usp++, - sizeof(stats))) != 0) - return (error); - } - - ap->nclasses = n; - - return (0); -} - -#ifdef KLD_MODULE - -static struct altqsw hfsc_sw = - {"hfsc", hfscopen, hfscclose, hfscioctl}; - -ALTQ_MODULE(altq_hfsc, ALTQT_HFSC, &hfsc_sw); -MODULE_DEPEND(altq_hfsc, altq_red, 1, 1, 1); -MODULE_DEPEND(altq_hfsc, altq_rio, 1, 1, 1); - -#endif /* KLD_MODULE */ -#endif /* ALTQ3_COMPAT */ #endif /* ALTQ_HFSC */ diff --git a/freebsd/sys/net/altq/altq_hfsc.h b/freebsd/sys/net/altq/altq_hfsc.h index 67ec0036..fa4aa811 100644 --- a/freebsd/sys/net/altq/altq_hfsc.h +++ b/freebsd/sys/net/altq/altq_hfsc.h @@ -168,74 +168,6 @@ struct hfsc_classstats_v1 { * header. */ -#ifdef ALTQ3_COMPAT -struct hfsc_interface { - char hfsc_ifname[IFNAMSIZ]; /* interface name (e.g., fxp0) */ -}; - -struct hfsc_attach { - struct hfsc_interface iface; - u_int bandwidth; /* link bandwidth in bits/sec */ -}; - -struct hfsc_add_class { - struct hfsc_interface iface; - u_int32_t parent_handle; - struct service_curve service_curve; - int qlimit; - int flags; - - u_int32_t class_handle; /* return value */ -}; - -struct hfsc_delete_class { - struct hfsc_interface iface; - u_int32_t class_handle; -}; - -struct hfsc_modify_class { - struct hfsc_interface iface; - u_int32_t class_handle; - struct service_curve service_curve; - int sctype; -}; - -struct hfsc_add_filter { - struct hfsc_interface iface; - u_int32_t class_handle; - struct flow_filter filter; - - u_long filter_handle; /* return value */ -}; - -struct hfsc_delete_filter { - struct hfsc_interface iface; - u_long filter_handle; -}; - -struct hfsc_class_stats { - struct hfsc_interface iface; - int nskip; /* skip # of classes */ - int nclasses; /* # of class stats (WR) */ - u_int64_t cur_time; /* current time */ - u_int32_t machclk_freq; /* machine clock frequency */ - u_int hif_classes; /* # of classes in the tree */ - u_int hif_packets; /* # of packets in the tree */ - struct hfsc_classstats *stats; /* pointer to stats array */ -}; - -#define HFSC_IF_ATTACH _IOW('Q', 1, struct hfsc_attach) -#define HFSC_IF_DETACH _IOW('Q', 2, struct hfsc_interface) -#define HFSC_ENABLE _IOW('Q', 3, struct hfsc_interface) -#define HFSC_DISABLE _IOW('Q', 4, struct hfsc_interface) -#define HFSC_CLEAR_HIERARCHY _IOW('Q', 5, struct hfsc_interface) -#define HFSC_ADD_CLASS _IOWR('Q', 7, struct hfsc_add_class) -#define HFSC_DEL_CLASS _IOW('Q', 8, struct hfsc_delete_class) -#define HFSC_MOD_CLASS _IOW('Q', 9, struct hfsc_modify_class) -#define HFSC_ADD_FILTER _IOWR('Q', 10, struct hfsc_add_filter) -#define HFSC_DEL_FILTER _IOW('Q', 11, struct hfsc_delete_filter) -#define HFSC_GETSTATS _IOWR('Q', 12, struct hfsc_class_stats) -#endif /* ALTQ3_COMPAT */ #ifdef _KERNEL /* diff --git a/freebsd/sys/net/altq/altq_priq.c b/freebsd/sys/net/altq/altq_priq.c index ce0830eb..5e77aef2 100644 --- a/freebsd/sys/net/altq/altq_priq.c +++ b/freebsd/sys/net/altq/altq_priq.c @@ -57,18 +57,11 @@ #include <netpfil/pf/pf_altq.h> #include <netpfil/pf/pf_mtag.h> #include <net/altq/altq.h> -#ifdef ALTQ3_COMPAT -#include <net/altq/altq_conf.h> -#endif #include <net/altq/altq_priq.h> /* * function prototypes */ -#ifdef ALTQ3_COMPAT -static struct priq_if *priq_attach(struct ifaltq *, u_int); -static int priq_detach(struct priq_if *); -#endif static int priq_clear_interface(struct priq_if *); static int priq_request(struct ifaltq *, int, void *); static void priq_purge(struct priq_if *); @@ -83,26 +76,10 @@ static struct mbuf *priq_getq(struct priq_class *); static struct mbuf *priq_pollq(struct priq_class *); static void priq_purgeq(struct priq_class *); -#ifdef ALTQ3_COMPAT -static int priqcmd_if_attach(struct priq_interface *); -static int priqcmd_if_detach(struct priq_interface *); -static int priqcmd_add_class(struct priq_add_class *); -static int priqcmd_delete_class(struct priq_delete_class *); -static int priqcmd_modify_class(struct priq_modify_class *); -static int priqcmd_add_filter(struct priq_add_filter *); -static int priqcmd_delete_filter(struct priq_delete_filter *); -static int priqcmd_class_stats(struct priq_class_stats *); -#endif /* ALTQ3_COMPAT */ static void get_class_stats(struct priq_classstats *, struct priq_class *); static struct priq_class *clh_to_clp(struct priq_if *, u_int32_t); -#ifdef ALTQ3_COMPAT -altqdev_decl(priq); - -/* pif_list keeps all priq_if's allocated. */ -static struct priq_if *pif_list = NULL; -#endif /* ALTQ3_COMPAT */ int priq_pfattach(struct pf_altq *a) @@ -491,10 +468,6 @@ priq_enqueue(struct ifaltq *ifq, struct mbuf *m, struct altq_pktattr *pktattr) cl = NULL; if ((t = pf_find_mtag(m)) != NULL) cl = clh_to_clp(pif, t->qid); -#ifdef ALTQ3_COMPAT - else if ((ifq->altq_flags & ALTQF_CLASSIFY) && pktattr != NULL) - cl = pktattr->pattr_class; -#endif if (cl == NULL) { cl = pif->pif_default; if (cl == NULL) { @@ -502,12 +475,7 @@ priq_enqueue(struct ifaltq *ifq, struct mbuf *m, struct altq_pktattr *pktattr) return (ENOBUFS); } } -#ifdef ALTQ3_COMPAT - if (pktattr != NULL) - cl->cl_pktattr = pktattr; /* save proto hdr used by ECN */ - else -#endif - cl->cl_pktattr = NULL; + cl->cl_pktattr = NULL; len = m_pktlen(m); if (priq_addq(cl, m) != 0) { /* drop occurred. mbuf was freed in priq_addq. */ @@ -676,397 +644,4 @@ clh_to_clp(struct priq_if *pif, u_int32_t chandle) } -#ifdef ALTQ3_COMPAT - -static struct priq_if * -priq_attach(ifq, bandwidth) - struct ifaltq *ifq; - u_int bandwidth; -{ - struct priq_if *pif; - - pif = malloc(sizeof(struct priq_if), - M_DEVBUF, M_WAITOK); - if (pif == NULL) - return (NULL); - bzero(pif, sizeof(struct priq_if)); - pif->pif_bandwidth = bandwidth; - pif->pif_maxpri = -1; - pif->pif_ifq = ifq; - - /* add this state to the priq list */ - pif->pif_next = pif_list; - pif_list = pif; - - return (pif); -} - -static int -priq_detach(pif) - struct priq_if *pif; -{ - (void)priq_clear_interface(pif); - - /* remove this interface from the pif list */ - if (pif_list == pif) - pif_list = pif->pif_next; - else { - struct priq_if *p; - - for (p = pif_list; p != NULL; p = p->pif_next) - if (p->pif_next == pif) { - p->pif_next = pif->pif_next; - break; - } - ASSERT(p != NULL); - } - - free(pif, M_DEVBUF); - return (0); -} - -/* - * priq device interface - */ -int -priqopen(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - /* everything will be done when the queueing scheme is attached. */ - return 0; -} - -int -priqclose(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - struct priq_if *pif; - int err, error = 0; - - while ((pif = pif_list) != NULL) { - /* destroy all */ - if (ALTQ_IS_ENABLED(pif->pif_ifq)) - altq_disable(pif->pif_ifq); - - err = altq_detach(pif->pif_ifq); - if (err == 0) - err = priq_detach(pif); - if (err != 0 && error == 0) - error = err; - } - - return error; -} - -int -priqioctl(dev, cmd, addr, flag, p) - dev_t dev; - ioctlcmd_t cmd; - caddr_t addr; - int flag; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - struct priq_if *pif; - struct priq_interface *ifacep; - int error = 0; - - /* check super-user privilege */ - switch (cmd) { - case PRIQ_GETSTATS: - break; - default: -#if (__FreeBSD_version > 700000) - if ((error = priv_check(p, PRIV_ALTQ_MANAGE)) != 0) - return (error); -#elsif (__FreeBSD_version > 400000) - if ((error = suser(p)) != 0) - return (error); -#else - if ((error = suser(p->p_ucred, &p->p_acflag)) != 0) - return (error); -#endif - break; - } - - switch (cmd) { - - case PRIQ_IF_ATTACH: - error = priqcmd_if_attach((struct priq_interface *)addr); - break; - - case PRIQ_IF_DETACH: - error = priqcmd_if_detach((struct priq_interface *)addr); - break; - - case PRIQ_ENABLE: - case PRIQ_DISABLE: - case PRIQ_CLEAR: - ifacep = (struct priq_interface *)addr; - if ((pif = altq_lookup(ifacep->ifname, - ALTQT_PRIQ)) == NULL) { - error = EBADF; - break; - } - - switch (cmd) { - case PRIQ_ENABLE: - if (pif->pif_default == NULL) { -#ifdef ALTQ_DEBUG - printf("priq: no default class\n"); -#endif - error = EINVAL; - break; - } - error = altq_enable(pif->pif_ifq); - break; - - case PRIQ_DISABLE: - error = altq_disable(pif->pif_ifq); - break; - - case PRIQ_CLEAR: - priq_clear_interface(pif); - break; - } - break; - - case PRIQ_ADD_CLASS: - error = priqcmd_add_class((struct priq_add_class *)addr); - break; - - case PRIQ_DEL_CLASS: - error = priqcmd_delete_class((struct priq_delete_class *)addr); - break; - - case PRIQ_MOD_CLASS: - error = priqcmd_modify_class((struct priq_modify_class *)addr); - break; - - case PRIQ_ADD_FILTER: - error = priqcmd_add_filter((struct priq_add_filter *)addr); - break; - - case PRIQ_DEL_FILTER: - error = priqcmd_delete_filter((struct priq_delete_filter *)addr); - break; - - case PRIQ_GETSTATS: - error = priqcmd_class_stats((struct priq_class_stats *)addr); - break; - - default: - error = EINVAL; - break; - } - return error; -} - -static int -priqcmd_if_attach(ap) - struct priq_interface *ap; -{ - struct priq_if *pif; - struct ifnet *ifp; - int error; - - if ((ifp = ifunit(ap->ifname)) == NULL) - return (ENXIO); - - if ((pif = priq_attach(&ifp->if_snd, ap->arg)) == NULL) - return (ENOMEM); - - /* - * set PRIQ to this ifnet structure. - */ - if ((error = altq_attach(&ifp->if_snd, ALTQT_PRIQ, pif, - priq_enqueue, priq_dequeue, priq_request, - &pif->pif_classifier, acc_classify)) != 0) - (void)priq_detach(pif); - - return (error); -} - -static int -priqcmd_if_detach(ap) - struct priq_interface *ap; -{ - struct priq_if *pif; - int error; - - if ((pif = altq_lookup(ap->ifname, ALTQT_PRIQ)) == NULL) - return (EBADF); - - if (ALTQ_IS_ENABLED(pif->pif_ifq)) - altq_disable(pif->pif_ifq); - - if ((error = altq_detach(pif->pif_ifq))) - return (error); - - return priq_detach(pif); -} - -static int -priqcmd_add_class(ap) - struct priq_add_class *ap; -{ - struct priq_if *pif; - struct priq_class *cl; - int qid; - - if ((pif = altq_lookup(ap->iface.ifname, ALTQT_PRIQ)) == NULL) - return (EBADF); - - if (ap->pri < 0 || ap->pri >= PRIQ_MAXPRI) - return (EINVAL); - if (pif->pif_classes[ap->pri] != NULL) - return (EBUSY); - - qid = ap->pri + 1; - if ((cl = priq_class_create(pif, ap->pri, - ap->qlimit, ap->flags, qid)) == NULL) - return (ENOMEM); - - /* return a class handle to the user */ - ap->class_handle = cl->cl_handle; - - return (0); -} - -static int -priqcmd_delete_class(ap) - struct priq_delete_class *ap; -{ - struct priq_if *pif; - struct priq_class *cl; - - if ((pif = altq_lookup(ap->iface.ifname, ALTQT_PRIQ)) == NULL) - return (EBADF); - - if ((cl = clh_to_clp(pif, ap->class_handle)) == NULL) - return (EINVAL); - - return priq_class_destroy(cl); -} - -static int -priqcmd_modify_class(ap) - struct priq_modify_class *ap; -{ - struct priq_if *pif; - struct priq_class *cl; - - if ((pif = altq_lookup(ap->iface.ifname, ALTQT_PRIQ)) == NULL) - return (EBADF); - - if (ap->pri < 0 || ap->pri >= PRIQ_MAXPRI) - return (EINVAL); - - if ((cl = clh_to_clp(pif, ap->class_handle)) == NULL) - return (EINVAL); - - /* - * if priority is changed, move the class to the new priority - */ - if (pif->pif_classes[ap->pri] != cl) { - if (pif->pif_classes[ap->pri] != NULL) - return (EEXIST); - pif->pif_classes[cl->cl_pri] = NULL; - pif->pif_classes[ap->pri] = cl; - cl->cl_pri = ap->pri; - } - - /* call priq_class_create to change class parameters */ - if ((cl = priq_class_create(pif, ap->pri, - ap->qlimit, ap->flags, ap->class_handle)) == NULL) - return (ENOMEM); - return 0; -} - -static int -priqcmd_add_filter(ap) - struct priq_add_filter *ap; -{ - struct priq_if *pif; - struct priq_class *cl; - - if ((pif = altq_lookup(ap->iface.ifname, ALTQT_PRIQ)) == NULL) - return (EBADF); - - if ((cl = clh_to_clp(pif, ap->class_handle)) == NULL) - return (EINVAL); - - return acc_add_filter(&pif->pif_classifier, &ap->filter, - cl, &ap->filter_handle); -} - -static int -priqcmd_delete_filter(ap) - struct priq_delete_filter *ap; -{ - struct priq_if *pif; - - if ((pif = altq_lookup(ap->iface.ifname, ALTQT_PRIQ)) == NULL) - return (EBADF); - - return acc_delete_filter(&pif->pif_classifier, - ap->filter_handle); -} - -static int -priqcmd_class_stats(ap) - struct priq_class_stats *ap; -{ - struct priq_if *pif; - struct priq_class *cl; - struct priq_classstats stats, *usp; - int pri, error; - - if ((pif = altq_lookup(ap->iface.ifname, ALTQT_PRIQ)) == NULL) - return (EBADF); - - ap->maxpri = pif->pif_maxpri; - - /* then, read the next N classes in the tree */ - usp = ap->stats; - for (pri = 0; pri <= pif->pif_maxpri; pri++) { - cl = pif->pif_classes[pri]; - if (cl != NULL) - get_class_stats(&stats, cl); - else - bzero(&stats, sizeof(stats)); - if ((error = copyout((caddr_t)&stats, (caddr_t)usp++, - sizeof(stats))) != 0) - return (error); - } - return (0); -} - -#ifdef KLD_MODULE - -static struct altqsw priq_sw = - {"priq", priqopen, priqclose, priqioctl}; - -ALTQ_MODULE(altq_priq, ALTQT_PRIQ, &priq_sw); -MODULE_DEPEND(altq_priq, altq_red, 1, 1, 1); -MODULE_DEPEND(altq_priq, altq_rio, 1, 1, 1); - -#endif /* KLD_MODULE */ - -#endif /* ALTQ3_COMPAT */ #endif /* ALTQ_PRIQ */ diff --git a/freebsd/sys/net/altq/altq_priq.h b/freebsd/sys/net/altq/altq_priq.h index 1a824d60..a5ee3289 100644 --- a/freebsd/sys/net/altq/altq_priq.h +++ b/freebsd/sys/net/altq/altq_priq.h @@ -42,21 +42,6 @@ extern "C" { #define PRIQ_MAXPRI 16 /* upper limit of the number of priorities */ -#ifdef ALTQ3_COMPAT -struct priq_interface { - char ifname[IFNAMSIZ]; /* interface name (e.g., fxp0) */ - u_long arg; /* request-specific argument */ -}; - -struct priq_add_class { - struct priq_interface iface; - int pri; /* priority (0 is the lowest) */ - int qlimit; /* queue size limit */ - int flags; /* misc flags (see below) */ - - u_int32_t class_handle; /* return value */ -}; -#endif /* ALTQ3_COMPAT */ /* priq class flags */ #define PRCF_RED 0x0001 /* use RED */ @@ -69,33 +54,6 @@ struct priq_add_class { /* special class handles */ #define PRIQ_NULLCLASS_HANDLE 0 -#ifdef ALTQ3_COMPAT -struct priq_delete_class { - struct priq_interface iface; - u_int32_t class_handle; -}; - -struct priq_modify_class { - struct priq_interface iface; - u_int32_t class_handle; - int pri; - int qlimit; - int flags; -}; - -struct priq_add_filter { - struct priq_interface iface; - u_int32_t class_handle; - struct flow_filter filter; - - u_long filter_handle; /* return value */ -}; - -struct priq_delete_filter { - struct priq_interface iface; - u_long filter_handle; -}; -#endif /* ALTQ3_COMPAT */ struct priq_classstats { u_int32_t class_handle; @@ -118,27 +76,6 @@ struct priq_classstats { * header. */ -#ifdef ALTQ3_COMPAT -struct priq_class_stats { - struct priq_interface iface; - int maxpri; /* in/out */ - - struct priq_classstats *stats; /* pointer to stats array */ -}; - -#define PRIQ_IF_ATTACH _IOW('Q', 1, struct priq_interface) -#define PRIQ_IF_DETACH _IOW('Q', 2, struct priq_interface) -#define PRIQ_ENABLE _IOW('Q', 3, struct priq_interface) -#define PRIQ_DISABLE _IOW('Q', 4, struct priq_interface) -#define PRIQ_CLEAR _IOW('Q', 5, struct priq_interface) -#define PRIQ_ADD_CLASS _IOWR('Q', 7, struct priq_add_class) -#define PRIQ_DEL_CLASS _IOW('Q', 8, struct priq_delete_class) -#define PRIQ_MOD_CLASS _IOW('Q', 9, struct priq_modify_class) -#define PRIQ_ADD_FILTER _IOWR('Q', 10, struct priq_add_filter) -#define PRIQ_DEL_FILTER _IOW('Q', 11, struct priq_delete_filter) -#define PRIQ_GETSTATS _IOWR('Q', 12, struct priq_class_stats) - -#endif /* ALTQ3_COMPAT */ #ifdef _KERNEL diff --git a/freebsd/sys/net/altq/altq_red.c b/freebsd/sys/net/altq/altq_red.c index 3ac0e0ec..f6ba0da8 100644 --- a/freebsd/sys/net/altq/altq_red.c +++ b/freebsd/sys/net/altq/altq_red.c @@ -98,12 +98,6 @@ #include <netpfil/pf/pf_mtag.h> #include <net/altq/altq.h> #include <net/altq/altq_red.h> -#ifdef ALTQ3_COMPAT -#include <net/altq/altq_conf.h> -#ifdef ALTQ_FLOWVALVE -#include <net/altq/altq_flowvalve.h> -#endif -#endif /* * ALTQ/RED (Random Early Detection) implementation using 32-bit @@ -170,56 +164,12 @@ * to switch to the random-drop policy, define "RED_RANDOM_DROP". */ -#ifdef ALTQ3_COMPAT -#ifdef ALTQ_FLOWVALVE -/* - * flow-valve is an extension to protect red from unresponsive flows - * and to promote end-to-end congestion control. - * flow-valve observes the average drop rates of the flows that have - * experienced packet drops in the recent past. - * when the average drop rate exceeds the threshold, the flow is - * blocked by the flow-valve. the trapped flow should back off - * exponentially to escape from the flow-valve. - */ -#ifdef RED_RANDOM_DROP -#error "random-drop can't be used with flow-valve!" -#endif -#endif /* ALTQ_FLOWVALVE */ - -/* red_list keeps all red_queue_t's allocated. */ -static red_queue_t *red_list = NULL; - -#endif /* ALTQ3_COMPAT */ /* default red parameter values */ static int default_th_min = TH_MIN; static int default_th_max = TH_MAX; static int default_inv_pmax = INV_P_MAX; -#ifdef ALTQ3_COMPAT -/* internal function prototypes */ -static int red_enqueue(struct ifaltq *, struct mbuf *, struct altq_pktattr *); -static struct mbuf *red_dequeue(struct ifaltq *, int); -static int red_request(struct ifaltq *, int, void *); -static void red_purgeq(red_queue_t *); -static int red_detach(red_queue_t *); -#ifdef ALTQ_FLOWVALVE -static __inline struct fve *flowlist_lookup(struct flowvalve *, - struct altq_pktattr *, struct timeval *); -static __inline struct fve *flowlist_reclaim(struct flowvalve *, - struct altq_pktattr *); -static __inline void flowlist_move_to_head(struct flowvalve *, struct fve *); -static __inline int fv_p2f(struct flowvalve *, int); -#if 0 /* XXX: make the compiler happy (fv_alloc unused) */ -static struct flowvalve *fv_alloc(struct red *); -#endif -static void fv_destroy(struct flowvalve *); -static int fv_checkflow(struct flowvalve *, struct altq_pktattr *, - struct fve **); -static void fv_dropbyred(struct flowvalve *fv, struct altq_pktattr *, - struct fve *); -#endif -#endif /* ALTQ3_COMPAT */ /* * red support routines @@ -317,12 +267,6 @@ red_alloc(int weight, int inv_pmax, int th_min, int th_max, int flags, void red_destroy(red_t *rp) { -#ifdef ALTQ3_COMPAT -#ifdef ALTQ_FLOWVALVE - if (rp->red_flowvalve != NULL) - fv_destroy(rp->red_flowvalve); -#endif -#endif /* ALTQ3_COMPAT */ wtab_destroy(rp->red_wtab); free(rp, M_DEVBUF); } @@ -344,17 +288,6 @@ red_addq(red_t *rp, class_queue_t *q, struct mbuf *m, { int avg, droptype; int n; -#ifdef ALTQ3_COMPAT -#ifdef ALTQ_FLOWVALVE - struct fve *fve = NULL; - - if (rp->red_flowvalve != NULL && rp->red_flowvalve->fv_flows > 0) - if (fv_checkflow(rp->red_flowvalve, pktattr, &fve)) { - m_freem(m); - return (-1); - } -#endif -#endif /* ALTQ3_COMPAT */ avg = rp->red_avg; @@ -460,12 +393,6 @@ red_addq(red_t *rp, class_queue_t *q, struct mbuf *m, PKTCNTR_ADD(&rp->red_stats.drop_cnt, m_pktlen(m)); #endif rp->red_count = 0; -#ifdef ALTQ3_COMPAT -#ifdef ALTQ_FLOWVALVE - if (rp->red_flowvalve != NULL) - fv_dropbyred(rp->red_flowvalve, pktattr, fve); -#endif -#endif /* ALTQ3_COMPAT */ m_freem(m); return (-1); } @@ -523,11 +450,6 @@ mark_ecn(struct mbuf *m, struct altq_pktattr *pktattr, int flags) at = pf_find_mtag(m); if (at != NULL) { hdr = at->hdr; -#ifdef ALTQ3_COMPAT - } else if (pktattr != NULL) { - af = pktattr->pattr_af; - hdr = pktattr->pattr_hdr; -#endif /* ALTQ3_COMPAT */ } else return (0); @@ -709,786 +631,5 @@ pow_w(struct wtab *w, int n) return (val); } -#ifdef ALTQ3_COMPAT -/* - * red device interface - */ -altqdev_decl(red); - -int -redopen(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - /* everything will be done when the queueing scheme is attached. */ - return 0; -} - -int -redclose(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - red_queue_t *rqp; - int err, error = 0; - - while ((rqp = red_list) != NULL) { - /* destroy all */ - err = red_detach(rqp); - if (err != 0 && error == 0) - error = err; - } - - return error; -} - -int -redioctl(dev, cmd, addr, flag, p) - dev_t dev; - ioctlcmd_t cmd; - caddr_t addr; - int flag; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - red_queue_t *rqp; - struct red_interface *ifacep; - struct ifnet *ifp; - int error = 0; - - /* check super-user privilege */ - switch (cmd) { - case RED_GETSTATS: - break; - default: -#if (__FreeBSD_version > 700000) - if ((error = priv_check(p, PRIV_ALTQ_MANAGE)) != 0) -#elsif (__FreeBSD_version > 400000) - if ((error = suser(p)) != 0) -#else - if ((error = suser(p->p_ucred, &p->p_acflag)) != 0) -#endif - return (error); - break; - } - - switch (cmd) { - - case RED_ENABLE: - ifacep = (struct red_interface *)addr; - if ((rqp = altq_lookup(ifacep->red_ifname, ALTQT_RED)) == NULL) { - error = EBADF; - break; - } - error = altq_enable(rqp->rq_ifq); - break; - - case RED_DISABLE: - ifacep = (struct red_interface *)addr; - if ((rqp = altq_lookup(ifacep->red_ifname, ALTQT_RED)) == NULL) { - error = EBADF; - break; - } - error = altq_disable(rqp->rq_ifq); - break; - - case RED_IF_ATTACH: - ifp = ifunit(((struct red_interface *)addr)->red_ifname); - if (ifp == NULL) { - error = ENXIO; - break; - } - - /* allocate and initialize red_queue_t */ - rqp = malloc(sizeof(red_queue_t), M_DEVBUF, M_WAITOK); - if (rqp == NULL) { - error = ENOMEM; - break; - } - bzero(rqp, sizeof(red_queue_t)); - - rqp->rq_q = malloc(sizeof(class_queue_t), - M_DEVBUF, M_WAITOK); - if (rqp->rq_q == NULL) { - free(rqp, M_DEVBUF); - error = ENOMEM; - break; - } - bzero(rqp->rq_q, sizeof(class_queue_t)); - - rqp->rq_red = red_alloc(0, 0, 0, 0, 0, 0); - if (rqp->rq_red == NULL) { - free(rqp->rq_q, M_DEVBUF); - free(rqp, M_DEVBUF); - error = ENOMEM; - break; - } - - rqp->rq_ifq = &ifp->if_snd; - qtail(rqp->rq_q) = NULL; - qlen(rqp->rq_q) = 0; - qlimit(rqp->rq_q) = RED_LIMIT; - qtype(rqp->rq_q) = Q_RED; - - /* - * set RED to this ifnet structure. - */ - error = altq_attach(rqp->rq_ifq, ALTQT_RED, rqp, - red_enqueue, red_dequeue, red_request, - NULL, NULL); - if (error) { - red_destroy(rqp->rq_red); - free(rqp->rq_q, M_DEVBUF); - free(rqp, M_DEVBUF); - break; - } - - /* add this state to the red list */ - rqp->rq_next = red_list; - red_list = rqp; - break; - - case RED_IF_DETACH: - ifacep = (struct red_interface *)addr; - if ((rqp = altq_lookup(ifacep->red_ifname, ALTQT_RED)) == NULL) { - error = EBADF; - break; - } - error = red_detach(rqp); - break; - - case RED_GETSTATS: - do { - struct red_stats *q_stats; - red_t *rp; - - q_stats = (struct red_stats *)addr; - if ((rqp = altq_lookup(q_stats->iface.red_ifname, - ALTQT_RED)) == NULL) { - error = EBADF; - break; - } - - q_stats->q_len = qlen(rqp->rq_q); - q_stats->q_limit = qlimit(rqp->rq_q); - - rp = rqp->rq_red; - q_stats->q_avg = rp->red_avg >> rp->red_wshift; - q_stats->xmit_cnt = rp->red_stats.xmit_cnt; - q_stats->drop_cnt = rp->red_stats.drop_cnt; - q_stats->drop_forced = rp->red_stats.drop_forced; - q_stats->drop_unforced = rp->red_stats.drop_unforced; - q_stats->marked_packets = rp->red_stats.marked_packets; - - q_stats->weight = rp->red_weight; - q_stats->inv_pmax = rp->red_inv_pmax; - q_stats->th_min = rp->red_thmin; - q_stats->th_max = rp->red_thmax; - -#ifdef ALTQ_FLOWVALVE - if (rp->red_flowvalve != NULL) { - struct flowvalve *fv = rp->red_flowvalve; - q_stats->fv_flows = fv->fv_flows; - q_stats->fv_pass = fv->fv_stats.pass; - q_stats->fv_predrop = fv->fv_stats.predrop; - q_stats->fv_alloc = fv->fv_stats.alloc; - q_stats->fv_escape = fv->fv_stats.escape; - } else { -#endif /* ALTQ_FLOWVALVE */ - q_stats->fv_flows = 0; - q_stats->fv_pass = 0; - q_stats->fv_predrop = 0; - q_stats->fv_alloc = 0; - q_stats->fv_escape = 0; -#ifdef ALTQ_FLOWVALVE - } -#endif /* ALTQ_FLOWVALVE */ - } while (/*CONSTCOND*/ 0); - break; - - case RED_CONFIG: - do { - struct red_conf *fc; - red_t *new; - int s, limit; - - fc = (struct red_conf *)addr; - if ((rqp = altq_lookup(fc->iface.red_ifname, - ALTQT_RED)) == NULL) { - error = EBADF; - break; - } - new = red_alloc(fc->red_weight, - fc->red_inv_pmax, - fc->red_thmin, - fc->red_thmax, - fc->red_flags, - fc->red_pkttime); - if (new == NULL) { - error = ENOMEM; - break; - } - - s = splnet(); - red_purgeq(rqp); - limit = fc->red_limit; - if (limit < fc->red_thmax) - limit = fc->red_thmax; - qlimit(rqp->rq_q) = limit; - fc->red_limit = limit; /* write back the new value */ - - red_destroy(rqp->rq_red); - rqp->rq_red = new; - - splx(s); - - /* write back new values */ - fc->red_limit = limit; - fc->red_inv_pmax = rqp->rq_red->red_inv_pmax; - fc->red_thmin = rqp->rq_red->red_thmin; - fc->red_thmax = rqp->rq_red->red_thmax; - - } while (/*CONSTCOND*/ 0); - break; - - case RED_SETDEFAULTS: - do { - struct redparams *rp; - - rp = (struct redparams *)addr; - - default_th_min = rp->th_min; - default_th_max = rp->th_max; - default_inv_pmax = rp->inv_pmax; - } while (/*CONSTCOND*/ 0); - break; - - default: - error = EINVAL; - break; - } - return error; -} - -static int -red_detach(rqp) - red_queue_t *rqp; -{ - red_queue_t *tmp; - int error = 0; - - if (ALTQ_IS_ENABLED(rqp->rq_ifq)) - altq_disable(rqp->rq_ifq); - - if ((error = altq_detach(rqp->rq_ifq))) - return (error); - - if (red_list == rqp) - red_list = rqp->rq_next; - else { - for (tmp = red_list; tmp != NULL; tmp = tmp->rq_next) - if (tmp->rq_next == rqp) { - tmp->rq_next = rqp->rq_next; - break; - } - if (tmp == NULL) - printf("red_detach: no state found in red_list!\n"); - } - - red_destroy(rqp->rq_red); - free(rqp->rq_q, M_DEVBUF); - free(rqp, M_DEVBUF); - return (error); -} - -/* - * enqueue routine: - * - * returns: 0 when successfully queued. - * ENOBUFS when drop occurs. - */ -static int -red_enqueue(ifq, m, pktattr) - struct ifaltq *ifq; - struct mbuf *m; - struct altq_pktattr *pktattr; -{ - red_queue_t *rqp = (red_queue_t *)ifq->altq_disc; - - IFQ_LOCK_ASSERT(ifq); - - if (red_addq(rqp->rq_red, rqp->rq_q, m, pktattr) < 0) - return ENOBUFS; - ifq->ifq_len++; - return 0; -} - -/* - * dequeue routine: - * must be called in splimp. - * - * returns: mbuf dequeued. - * NULL when no packet is available in the queue. - */ - -static struct mbuf * -red_dequeue(ifq, op) - struct ifaltq *ifq; - int op; -{ - red_queue_t *rqp = (red_queue_t *)ifq->altq_disc; - struct mbuf *m; - - IFQ_LOCK_ASSERT(ifq); - - if (op == ALTDQ_POLL) - return qhead(rqp->rq_q); - - /* op == ALTDQ_REMOVE */ - m = red_getq(rqp->rq_red, rqp->rq_q); - if (m != NULL) - ifq->ifq_len--; - return (m); -} - -static int -red_request(ifq, req, arg) - struct ifaltq *ifq; - int req; - void *arg; -{ - red_queue_t *rqp = (red_queue_t *)ifq->altq_disc; - - IFQ_LOCK_ASSERT(ifq); - - switch (req) { - case ALTRQ_PURGE: - red_purgeq(rqp); - break; - } - return (0); -} - -static void -red_purgeq(rqp) - red_queue_t *rqp; -{ - _flushq(rqp->rq_q); - if (ALTQ_IS_ENABLED(rqp->rq_ifq)) - rqp->rq_ifq->ifq_len = 0; -} - -#ifdef ALTQ_FLOWVALVE - -#define FV_PSHIFT 7 /* weight of average drop rate -- 1/128 */ -#define FV_PSCALE(x) ((x) << FV_PSHIFT) -#define FV_PUNSCALE(x) ((x) >> FV_PSHIFT) -#define FV_FSHIFT 5 /* weight of average fraction -- 1/32 */ -#define FV_FSCALE(x) ((x) << FV_FSHIFT) -#define FV_FUNSCALE(x) ((x) >> FV_FSHIFT) - -#define FV_TIMER (3 * hz) /* timer value for garbage collector */ -#define FV_FLOWLISTSIZE 64 /* how many flows in flowlist */ - -#define FV_N 10 /* update fve_f every FV_N packets */ - -#define FV_BACKOFFTHRESH 1 /* backoff threshold interval in second */ -#define FV_TTHRESH 3 /* time threshold to delete fve */ -#define FV_ALPHA 5 /* extra packet count */ - -#define FV_STATS - -#if (__FreeBSD_version > 300000) -#define FV_TIMESTAMP(tp) getmicrotime(tp) -#else -#define FV_TIMESTAMP(tp) { (*(tp)) = time; } -#endif - -/* - * Brtt table: 127 entry table to convert drop rate (p) to - * the corresponding bandwidth fraction (f) - * the following equation is implemented to use scaled values, - * fve_p and fve_f, in the fixed point format. - * - * Brtt(p) = 1 /(sqrt(4*p/3) + min(1,3*sqrt(p*6/8)) * p * (1+32 * p*p)) - * f = Brtt(p) / (max_th + alpha) - */ -#define BRTT_SIZE 128 -#define BRTT_SHIFT 12 -#define BRTT_MASK 0x0007f000 -#define BRTT_PMAX (1 << (FV_PSHIFT + FP_SHIFT)) - -const int brtt_tab[BRTT_SIZE] = { - 0, 1262010, 877019, 703694, 598706, 525854, 471107, 427728, - 392026, 361788, 335598, 312506, 291850, 273158, 256081, 240361, - 225800, 212247, 199585, 187788, 178388, 169544, 161207, 153333, - 145888, 138841, 132165, 125836, 119834, 114141, 108739, 103612, - 98747, 94129, 89746, 85585, 81637, 77889, 74333, 70957, - 67752, 64711, 61824, 59084, 56482, 54013, 51667, 49440, - 47325, 45315, 43406, 41591, 39866, 38227, 36667, 35184, - 33773, 32430, 31151, 29933, 28774, 27668, 26615, 25611, - 24653, 23740, 22868, 22035, 21240, 20481, 19755, 19062, - 18399, 17764, 17157, 16576, 16020, 15487, 14976, 14487, - 14017, 13567, 13136, 12721, 12323, 11941, 11574, 11222, - 10883, 10557, 10243, 9942, 9652, 9372, 9103, 8844, - 8594, 8354, 8122, 7898, 7682, 7474, 7273, 7079, - 6892, 6711, 6536, 6367, 6204, 6046, 5893, 5746, - 5603, 5464, 5330, 5201, 5075, 4954, 4836, 4722, - 4611, 4504, 4400, 4299, 4201, 4106, 4014, 3924 -}; - -static __inline struct fve * -flowlist_lookup(fv, pktattr, now) - struct flowvalve *fv; - struct altq_pktattr *pktattr; - struct timeval *now; -{ - struct fve *fve; - int flows; - struct ip *ip; -#ifdef INET6 - struct ip6_hdr *ip6; -#endif - struct timeval tthresh; - - if (pktattr == NULL) - return (NULL); - - tthresh.tv_sec = now->tv_sec - FV_TTHRESH; - flows = 0; - /* - * search the flow list - */ - switch (pktattr->pattr_af) { - case AF_INET: - ip = (struct ip *)pktattr->pattr_hdr; - TAILQ_FOREACH(fve, &fv->fv_flowlist, fve_lru){ - if (fve->fve_lastdrop.tv_sec == 0) - break; - if (fve->fve_lastdrop.tv_sec < tthresh.tv_sec) { - fve->fve_lastdrop.tv_sec = 0; - break; - } - if (fve->fve_flow.flow_af == AF_INET && - fve->fve_flow.flow_ip.ip_src.s_addr == - ip->ip_src.s_addr && - fve->fve_flow.flow_ip.ip_dst.s_addr == - ip->ip_dst.s_addr) - return (fve); - flows++; - } - break; -#ifdef INET6 - case AF_INET6: - ip6 = (struct ip6_hdr *)pktattr->pattr_hdr; - TAILQ_FOREACH(fve, &fv->fv_flowlist, fve_lru){ - if (fve->fve_lastdrop.tv_sec == 0) - break; - if (fve->fve_lastdrop.tv_sec < tthresh.tv_sec) { - fve->fve_lastdrop.tv_sec = 0; - break; - } - if (fve->fve_flow.flow_af == AF_INET6 && - IN6_ARE_ADDR_EQUAL(&fve->fve_flow.flow_ip6.ip6_src, - &ip6->ip6_src) && - IN6_ARE_ADDR_EQUAL(&fve->fve_flow.flow_ip6.ip6_dst, - &ip6->ip6_dst)) - return (fve); - flows++; - } - break; -#endif /* INET6 */ - - default: - /* unknown protocol. no drop. */ - return (NULL); - } - fv->fv_flows = flows; /* save the number of active fve's */ - return (NULL); -} - -static __inline struct fve * -flowlist_reclaim(fv, pktattr) - struct flowvalve *fv; - struct altq_pktattr *pktattr; -{ - struct fve *fve; - struct ip *ip; -#ifdef INET6 - struct ip6_hdr *ip6; -#endif - - /* - * get an entry from the tail of the LRU list. - */ - fve = TAILQ_LAST(&fv->fv_flowlist, fv_flowhead); - - switch (pktattr->pattr_af) { - case AF_INET: - ip = (struct ip *)pktattr->pattr_hdr; - fve->fve_flow.flow_af = AF_INET; - fve->fve_flow.flow_ip.ip_src = ip->ip_src; - fve->fve_flow.flow_ip.ip_dst = ip->ip_dst; - break; -#ifdef INET6 - case AF_INET6: - ip6 = (struct ip6_hdr *)pktattr->pattr_hdr; - fve->fve_flow.flow_af = AF_INET6; - fve->fve_flow.flow_ip6.ip6_src = ip6->ip6_src; - fve->fve_flow.flow_ip6.ip6_dst = ip6->ip6_dst; - break; -#endif - } - - fve->fve_state = Green; - fve->fve_p = 0.0; - fve->fve_f = 0.0; - fve->fve_ifseq = fv->fv_ifseq - 1; - fve->fve_count = 0; - - fv->fv_flows++; -#ifdef FV_STATS - fv->fv_stats.alloc++; -#endif - return (fve); -} - -static __inline void -flowlist_move_to_head(fv, fve) - struct flowvalve *fv; - struct fve *fve; -{ - if (TAILQ_FIRST(&fv->fv_flowlist) != fve) { - TAILQ_REMOVE(&fv->fv_flowlist, fve, fve_lru); - TAILQ_INSERT_HEAD(&fv->fv_flowlist, fve, fve_lru); - } -} - -#if 0 /* XXX: make the compiler happy (fv_alloc unused) */ -/* - * allocate flowvalve structure - */ -static struct flowvalve * -fv_alloc(rp) - struct red *rp; -{ - struct flowvalve *fv; - struct fve *fve; - int i, num; - - num = FV_FLOWLISTSIZE; - fv = malloc(sizeof(struct flowvalve), - M_DEVBUF, M_WAITOK); - if (fv == NULL) - return (NULL); - bzero(fv, sizeof(struct flowvalve)); - - fv->fv_fves = malloc(sizeof(struct fve) * num, - M_DEVBUF, M_WAITOK); - if (fv->fv_fves == NULL) { - free(fv, M_DEVBUF); - return (NULL); - } - bzero(fv->fv_fves, sizeof(struct fve) * num); - - fv->fv_flows = 0; - TAILQ_INIT(&fv->fv_flowlist); - for (i = 0; i < num; i++) { - fve = &fv->fv_fves[i]; - fve->fve_lastdrop.tv_sec = 0; - TAILQ_INSERT_TAIL(&fv->fv_flowlist, fve, fve_lru); - } - - /* initialize drop rate threshold in scaled fixed-point */ - fv->fv_pthresh = (FV_PSCALE(1) << FP_SHIFT) / rp->red_inv_pmax; - - /* initialize drop rate to fraction table */ - fv->fv_p2ftab = malloc(sizeof(int) * BRTT_SIZE, - M_DEVBUF, M_WAITOK); - if (fv->fv_p2ftab == NULL) { - free(fv->fv_fves, M_DEVBUF); - free(fv, M_DEVBUF); - return (NULL); - } - /* - * create the p2f table. - * (shift is used to keep the precision) - */ - for (i = 1; i < BRTT_SIZE; i++) { - int f; - - f = brtt_tab[i] << 8; - fv->fv_p2ftab[i] = (f / (rp->red_thmax + FV_ALPHA)) >> 8; - } - - return (fv); -} -#endif - -static void fv_destroy(fv) - struct flowvalve *fv; -{ - free(fv->fv_p2ftab, M_DEVBUF); - free(fv->fv_fves, M_DEVBUF); - free(fv, M_DEVBUF); -} - -static __inline int -fv_p2f(fv, p) - struct flowvalve *fv; - int p; -{ - int val, f; - - if (p >= BRTT_PMAX) - f = fv->fv_p2ftab[BRTT_SIZE-1]; - else if ((val = (p & BRTT_MASK))) - f = fv->fv_p2ftab[(val >> BRTT_SHIFT)]; - else - f = fv->fv_p2ftab[1]; - return (f); -} - -/* - * check if an arriving packet should be pre-dropped. - * called from red_addq() when a packet arrives. - * returns 1 when the packet should be pre-dropped. - * should be called in splimp. - */ -static int -fv_checkflow(fv, pktattr, fcache) - struct flowvalve *fv; - struct altq_pktattr *pktattr; - struct fve **fcache; -{ - struct fve *fve; - struct timeval now; - - fv->fv_ifseq++; - FV_TIMESTAMP(&now); - - if ((fve = flowlist_lookup(fv, pktattr, &now)) == NULL) - /* no matching entry in the flowlist */ - return (0); - - *fcache = fve; - - /* update fraction f for every FV_N packets */ - if (++fve->fve_count == FV_N) { - /* - * f = Wf * N / (fv_ifseq - fve_ifseq) + (1 - Wf) * f - */ - fve->fve_f = - (FV_N << FP_SHIFT) / (fv->fv_ifseq - fve->fve_ifseq) - + fve->fve_f - FV_FUNSCALE(fve->fve_f); - fve->fve_ifseq = fv->fv_ifseq; - fve->fve_count = 0; - } - - /* - * overpumping test - */ - if (fve->fve_state == Green && fve->fve_p > fv->fv_pthresh) { - int fthresh; - - /* calculate a threshold */ - fthresh = fv_p2f(fv, fve->fve_p); - if (fve->fve_f > fthresh) - fve->fve_state = Red; - } - - if (fve->fve_state == Red) { - /* - * backoff test - */ - if (now.tv_sec - fve->fve_lastdrop.tv_sec > FV_BACKOFFTHRESH) { - /* no drop for at least FV_BACKOFFTHRESH sec */ - fve->fve_p = 0; - fve->fve_state = Green; -#ifdef FV_STATS - fv->fv_stats.escape++; -#endif - } else { - /* block this flow */ - flowlist_move_to_head(fv, fve); - fve->fve_lastdrop = now; -#ifdef FV_STATS - fv->fv_stats.predrop++; -#endif - return (1); - } - } - - /* - * p = (1 - Wp) * p - */ - fve->fve_p -= FV_PUNSCALE(fve->fve_p); - if (fve->fve_p < 0) - fve->fve_p = 0; -#ifdef FV_STATS - fv->fv_stats.pass++; -#endif - return (0); -} - -/* - * called from red_addq when a packet is dropped by red. - * should be called in splimp. - */ -static void fv_dropbyred(fv, pktattr, fcache) - struct flowvalve *fv; - struct altq_pktattr *pktattr; - struct fve *fcache; -{ - struct fve *fve; - struct timeval now; - - if (pktattr == NULL) - return; - FV_TIMESTAMP(&now); - - if (fcache != NULL) - /* the fve of this packet is already cached */ - fve = fcache; - else if ((fve = flowlist_lookup(fv, pktattr, &now)) == NULL) - fve = flowlist_reclaim(fv, pktattr); - - flowlist_move_to_head(fv, fve); - - /* - * update p: the following line cancels the update - * in fv_checkflow() and calculate - * p = Wp + (1 - Wp) * p - */ - fve->fve_p = (1 << FP_SHIFT) + fve->fve_p; - - fve->fve_lastdrop = now; -} - -#endif /* ALTQ_FLOWVALVE */ - -#ifdef KLD_MODULE - -static struct altqsw red_sw = - {"red", redopen, redclose, redioctl}; - -ALTQ_MODULE(altq_red, ALTQT_RED, &red_sw); -MODULE_VERSION(altq_red, 1); - -#endif /* KLD_MODULE */ -#endif /* ALTQ3_COMPAT */ #endif /* ALTQ_RED */ diff --git a/freebsd/sys/net/altq/altq_red.h b/freebsd/sys/net/altq/altq_red.h index 8ae8d291..d4cd3d8b 100644 --- a/freebsd/sys/net/altq/altq_red.h +++ b/freebsd/sys/net/altq/altq_red.h @@ -32,48 +32,6 @@ #include <net/altq/altq_classq.h> -#ifdef ALTQ3_COMPAT -struct red_interface { - char red_ifname[IFNAMSIZ]; -}; - -struct red_stats { - struct red_interface iface; - int q_len; - int q_avg; - - struct pktcntr xmit_cnt; - struct pktcntr drop_cnt; - u_int drop_forced; - u_int drop_unforced; - u_int marked_packets; - - /* static red parameters */ - int q_limit; - int weight; - int inv_pmax; - int th_min; - int th_max; - - /* flowvalve related stuff */ - u_int fv_flows; - u_int fv_pass; - u_int fv_predrop; - u_int fv_alloc; - u_int fv_escape; -}; - -struct red_conf { - struct red_interface iface; - int red_weight; /* weight for EWMA */ - int red_inv_pmax; /* inverse of max drop probability */ - int red_thmin; /* red min threshold */ - int red_thmax; /* red max threshold */ - int red_limit; /* max queue length */ - int red_pkttime; /* average packet time in usec */ - int red_flags; /* see below */ -}; -#endif /* ALTQ3_COMPAT */ /* red flags */ #define REDF_ECN4 0x01 /* use packet marking for IPv4 packets */ @@ -100,24 +58,9 @@ struct redstats { u_int marked_packets; }; -#ifdef ALTQ3_COMPAT -/* - * IOCTLs for RED - */ -#define RED_IF_ATTACH _IOW('Q', 1, struct red_interface) -#define RED_IF_DETACH _IOW('Q', 2, struct red_interface) -#define RED_ENABLE _IOW('Q', 3, struct red_interface) -#define RED_DISABLE _IOW('Q', 4, struct red_interface) -#define RED_CONFIG _IOWR('Q', 6, struct red_conf) -#define RED_GETSTATS _IOWR('Q', 12, struct red_stats) -#define RED_SETDEFAULTS _IOW('Q', 30, struct redparams) -#endif /* ALTQ3_COMPAT */ #ifdef _KERNEL -#ifdef ALTQ3_COMPAT -struct flowvalve; -#endif /* weight table structure for idle time calibration */ struct wtab { @@ -153,9 +96,6 @@ typedef struct red { struct wtab *red_wtab; /* weight table */ struct timeval red_last; /* time when the queue becomes idle */ -#ifdef ALTQ3_COMPAT - struct flowvalve *red_flowvalve; /* flowvalve state */ -#endif struct { struct pktcntr xmit_cnt; @@ -166,16 +106,6 @@ typedef struct red { } red_stats; } red_t; -#ifdef ALTQ3_COMPAT -typedef struct red_queue { - struct red_queue *rq_next; /* next red_state in the list */ - struct ifaltq *rq_ifq; /* backpointer to ifaltq */ - - class_queue_t *rq_q; - - red_t *rq_red; -} red_queue_t; -#endif /* ALTQ3_COMPAT */ /* red drop types */ #define DTYPE_NODROP 0 /* no drop */ diff --git a/freebsd/sys/net/altq/altq_rio.c b/freebsd/sys/net/altq/altq_rio.c index af9b5835..f75660ad 100644 --- a/freebsd/sys/net/altq/altq_rio.c +++ b/freebsd/sys/net/altq/altq_rio.c @@ -94,9 +94,6 @@ #include <net/altq/altq_cdnr.h> #include <net/altq/altq_red.h> #include <net/altq/altq_rio.h> -#ifdef ALTQ3_COMPAT -#include <net/altq/altq_conf.h> -#endif /* * RIO: RED with IN/OUT bit @@ -170,10 +167,6 @@ } \ } -#ifdef ALTQ3_COMPAT -/* rio_list keeps all rio_queue_t's allocated. */ -static rio_queue_t *rio_list = NULL; -#endif /* default rio parameter values */ static struct redparams default_rio_params[RIO_NDROPPREC] = { /* th_min, th_max, inv_pmax */ @@ -184,18 +177,6 @@ static struct redparams default_rio_params[RIO_NDROPPREC] = { /* internal function prototypes */ static int dscp2index(u_int8_t); -#ifdef ALTQ3_COMPAT -static int rio_enqueue(struct ifaltq *, struct mbuf *, struct altq_pktattr *); -static struct mbuf *rio_dequeue(struct ifaltq *, int); -static int rio_request(struct ifaltq *, int, void *); -static int rio_detach(rio_queue_t *); - -/* - * rio device interface - */ -altqdev_decl(rio); - -#endif /* ALTQ3_COMPAT */ rio_t * rio_alloc(int weight, struct redparams *params, int flags, int pkttime) @@ -468,379 +449,5 @@ rio_getq(rio_t *rp, class_queue_t *q) return (m); } -#ifdef ALTQ3_COMPAT -int -rioopen(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - /* everything will be done when the queueing scheme is attached. */ - return 0; -} - -int -rioclose(dev, flag, fmt, p) - dev_t dev; - int flag, fmt; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - rio_queue_t *rqp; - int err, error = 0; - - while ((rqp = rio_list) != NULL) { - /* destroy all */ - err = rio_detach(rqp); - if (err != 0 && error == 0) - error = err; - } - - return error; -} - -int -rioioctl(dev, cmd, addr, flag, p) - dev_t dev; - ioctlcmd_t cmd; - caddr_t addr; - int flag; -#if (__FreeBSD_version > 500000) - struct thread *p; -#else - struct proc *p; -#endif -{ - rio_queue_t *rqp; - struct rio_interface *ifacep; - struct ifnet *ifp; - int error = 0; - - /* check super-user privilege */ - switch (cmd) { - case RIO_GETSTATS: - break; - default: -#if (__FreeBSD_version > 700000) - if ((error = priv_check(p, PRIV_ALTQ_MANAGE)) != 0) - return (error); -#elsif (__FreeBSD_version > 400000) - if ((error = suser(p)) != 0) - return (error); -#else - if ((error = suser(p->p_ucred, &p->p_acflag)) != 0) - return (error); -#endif - break; - } - - switch (cmd) { - - case RIO_ENABLE: - ifacep = (struct rio_interface *)addr; - if ((rqp = altq_lookup(ifacep->rio_ifname, ALTQT_RIO)) == NULL) { - error = EBADF; - break; - } - error = altq_enable(rqp->rq_ifq); - break; - - case RIO_DISABLE: - ifacep = (struct rio_interface *)addr; - if ((rqp = altq_lookup(ifacep->rio_ifname, ALTQT_RIO)) == NULL) { - error = EBADF; - break; - } - error = altq_disable(rqp->rq_ifq); - break; - - case RIO_IF_ATTACH: - ifp = ifunit(((struct rio_interface *)addr)->rio_ifname); - if (ifp == NULL) { - error = ENXIO; - break; - } - - /* allocate and initialize rio_queue_t */ - rqp = malloc(sizeof(rio_queue_t), M_DEVBUF, M_WAITOK); - if (rqp == NULL) { - error = ENOMEM; - break; - } - bzero(rqp, sizeof(rio_queue_t)); - - rqp->rq_q = malloc(sizeof(class_queue_t), - M_DEVBUF, M_WAITOK); - if (rqp->rq_q == NULL) { - free(rqp, M_DEVBUF); - error = ENOMEM; - break; - } - bzero(rqp->rq_q, sizeof(class_queue_t)); - - rqp->rq_rio = rio_alloc(0, NULL, 0, 0); - if (rqp->rq_rio == NULL) { - free(rqp->rq_q, M_DEVBUF); - free(rqp, M_DEVBUF); - error = ENOMEM; - break; - } - - rqp->rq_ifq = &ifp->if_snd; - qtail(rqp->rq_q) = NULL; - qlen(rqp->rq_q) = 0; - qlimit(rqp->rq_q) = RIO_LIMIT; - qtype(rqp->rq_q) = Q_RIO; - - /* - * set RIO to this ifnet structure. - */ - error = altq_attach(rqp->rq_ifq, ALTQT_RIO, rqp, - rio_enqueue, rio_dequeue, rio_request, - NULL, NULL); - if (error) { - rio_destroy(rqp->rq_rio); - free(rqp->rq_q, M_DEVBUF); - free(rqp, M_DEVBUF); - break; - } - - /* add this state to the rio list */ - rqp->rq_next = rio_list; - rio_list = rqp; - break; - - case RIO_IF_DETACH: - ifacep = (struct rio_interface *)addr; - if ((rqp = altq_lookup(ifacep->rio_ifname, ALTQT_RIO)) == NULL) { - error = EBADF; - break; - } - error = rio_detach(rqp); - break; - - case RIO_GETSTATS: - do { - struct rio_stats *q_stats; - rio_t *rp; - int i; - - q_stats = (struct rio_stats *)addr; - if ((rqp = altq_lookup(q_stats->iface.rio_ifname, - ALTQT_RIO)) == NULL) { - error = EBADF; - break; - } - - rp = rqp->rq_rio; - - q_stats->q_limit = qlimit(rqp->rq_q); - q_stats->weight = rp->rio_weight; - q_stats->flags = rp->rio_flags; - - for (i = 0; i < RIO_NDROPPREC; i++) { - q_stats->q_len[i] = rp->rio_precstate[i].qlen; - bcopy(&rp->q_stats[i], &q_stats->q_stats[i], - sizeof(struct redstats)); - q_stats->q_stats[i].q_avg = - rp->rio_precstate[i].avg >> rp->rio_wshift; - - q_stats->q_params[i].inv_pmax - = rp->rio_precstate[i].inv_pmax; - q_stats->q_params[i].th_min - = rp->rio_precstate[i].th_min; - q_stats->q_params[i].th_max - = rp->rio_precstate[i].th_max; - } - } while (/*CONSTCOND*/ 0); - break; - - case RIO_CONFIG: - do { - struct rio_conf *fc; - rio_t *new; - int s, limit, i; - - fc = (struct rio_conf *)addr; - if ((rqp = altq_lookup(fc->iface.rio_ifname, - ALTQT_RIO)) == NULL) { - error = EBADF; - break; - } - - new = rio_alloc(fc->rio_weight, &fc->q_params[0], - fc->rio_flags, fc->rio_pkttime); - if (new == NULL) { - error = ENOMEM; - break; - } - - s = splnet(); - _flushq(rqp->rq_q); - limit = fc->rio_limit; - if (limit < fc->q_params[RIO_NDROPPREC-1].th_max) - limit = fc->q_params[RIO_NDROPPREC-1].th_max; - qlimit(rqp->rq_q) = limit; - - rio_destroy(rqp->rq_rio); - rqp->rq_rio = new; - - splx(s); - - /* write back new values */ - fc->rio_limit = limit; - for (i = 0; i < RIO_NDROPPREC; i++) { - fc->q_params[i].inv_pmax = - rqp->rq_rio->rio_precstate[i].inv_pmax; - fc->q_params[i].th_min = - rqp->rq_rio->rio_precstate[i].th_min; - fc->q_params[i].th_max = - rqp->rq_rio->rio_precstate[i].th_max; - } - } while (/*CONSTCOND*/ 0); - break; - - case RIO_SETDEFAULTS: - do { - struct redparams *rp; - int i; - - rp = (struct redparams *)addr; - for (i = 0; i < RIO_NDROPPREC; i++) - default_rio_params[i] = rp[i]; - } while (/*CONSTCOND*/ 0); - break; - - default: - error = EINVAL; - break; - } - - return error; -} - -static int -rio_detach(rqp) - rio_queue_t *rqp; -{ - rio_queue_t *tmp; - int error = 0; - - if (ALTQ_IS_ENABLED(rqp->rq_ifq)) - altq_disable(rqp->rq_ifq); - - if ((error = altq_detach(rqp->rq_ifq))) - return (error); - - if (rio_list == rqp) - rio_list = rqp->rq_next; - else { - for (tmp = rio_list; tmp != NULL; tmp = tmp->rq_next) - if (tmp->rq_next == rqp) { - tmp->rq_next = rqp->rq_next; - break; - } - if (tmp == NULL) - printf("rio_detach: no state found in rio_list!\n"); - } - - rio_destroy(rqp->rq_rio); - free(rqp->rq_q, M_DEVBUF); - free(rqp, M_DEVBUF); - return (error); -} - -/* - * rio support routines - */ -static int -rio_request(ifq, req, arg) - struct ifaltq *ifq; - int req; - void *arg; -{ - rio_queue_t *rqp = (rio_queue_t *)ifq->altq_disc; - - IFQ_LOCK_ASSERT(ifq); - - switch (req) { - case ALTRQ_PURGE: - _flushq(rqp->rq_q); - if (ALTQ_IS_ENABLED(ifq)) - ifq->ifq_len = 0; - break; - } - return (0); -} - -/* - * enqueue routine: - * - * returns: 0 when successfully queued. - * ENOBUFS when drop occurs. - */ -static int -rio_enqueue(ifq, m, pktattr) - struct ifaltq *ifq; - struct mbuf *m; - struct altq_pktattr *pktattr; -{ - rio_queue_t *rqp = (rio_queue_t *)ifq->altq_disc; - int error = 0; - - IFQ_LOCK_ASSERT(ifq); - - if (rio_addq(rqp->rq_rio, rqp->rq_q, m, pktattr) == 0) - ifq->ifq_len++; - else - error = ENOBUFS; - return error; -} - -/* - * dequeue routine: - * must be called in splimp. - * - * returns: mbuf dequeued. - * NULL when no packet is available in the queue. - */ - -static struct mbuf * -rio_dequeue(ifq, op) - struct ifaltq *ifq; - int op; -{ - rio_queue_t *rqp = (rio_queue_t *)ifq->altq_disc; - struct mbuf *m = NULL; - - IFQ_LOCK_ASSERT(ifq); - - if (op == ALTDQ_POLL) - return qhead(rqp->rq_q); - - m = rio_getq(rqp->rq_rio, rqp->rq_q); - if (m != NULL) - ifq->ifq_len--; - return m; -} - -#ifdef KLD_MODULE - -static struct altqsw rio_sw = - {"rio", rioopen, rioclose, rioioctl}; - -ALTQ_MODULE(altq_rio, ALTQT_RIO, &rio_sw); -MODULE_VERSION(altq_rio, 1); -MODULE_DEPEND(altq_rio, altq_red, 1, 1, 1); - -#endif /* KLD_MODULE */ -#endif /* ALTQ3_COMPAT */ #endif /* ALTQ_RIO */ diff --git a/freebsd/sys/net/altq/altq_rio.h b/freebsd/sys/net/altq/altq_rio.h index ce9dc0e0..2ae71394 100644 --- a/freebsd/sys/net/altq/altq_rio.h +++ b/freebsd/sys/net/altq/altq_rio.h @@ -38,32 +38,6 @@ */ #define RIO_NDROPPREC 3 /* number of drop precedence values */ -#ifdef ALTQ3_COMPAT -struct rio_interface { - char rio_ifname[IFNAMSIZ]; -}; - -struct rio_stats { - struct rio_interface iface; - int q_len[RIO_NDROPPREC]; - struct redstats q_stats[RIO_NDROPPREC]; - - /* static red parameters */ - int q_limit; - int weight; - int flags; - struct redparams q_params[RIO_NDROPPREC]; -}; - -struct rio_conf { - struct rio_interface iface; - struct redparams q_params[RIO_NDROPPREC]; - int rio_weight; /* weight for EWMA */ - int rio_limit; /* max queue length */ - int rio_pkttime; /* average packet time in usec */ - int rio_flags; /* see below */ -}; -#endif /* ALTQ3_COMPAT */ /* rio flags */ #define RIOF_ECN4 0x01 /* use packet marking for IPv4 packets */ @@ -71,18 +45,6 @@ struct rio_conf { #define RIOF_ECN (RIOF_ECN4 | RIOF_ECN6) #define RIOF_CLEARDSCP 0x200 /* clear diffserv codepoint */ -#ifdef ALTQ3_COMPAT -/* - * IOCTLs for RIO - */ -#define RIO_IF_ATTACH _IOW('Q', 1, struct rio_interface) -#define RIO_IF_DETACH _IOW('Q', 2, struct rio_interface) -#define RIO_ENABLE _IOW('Q', 3, struct rio_interface) -#define RIO_DISABLE _IOW('Q', 4, struct rio_interface) -#define RIO_CONFIG _IOWR('Q', 6, struct rio_conf) -#define RIO_GETSTATS _IOWR('Q', 12, struct rio_stats) -#define RIO_SETDEFAULTS _IOW('Q', 30, struct redparams[RIO_NDROPPREC]) -#endif /* ALTQ3_COMPAT */ #ifdef _KERNEL @@ -122,16 +84,6 @@ typedef struct rio { struct redstats q_stats[RIO_NDROPPREC]; /* statistics */ } rio_t; -#ifdef ALTQ3_COMPAT -typedef struct rio_queue { - struct rio_queue *rq_next; /* next red_state in the list */ - struct ifaltq *rq_ifq; /* backpointer to ifaltq */ - - class_queue_t *rq_q; - - rio_t *rq_rio; -} rio_queue_t; -#endif /* ALTQ3_COMPAT */ extern rio_t *rio_alloc(int, struct redparams *, int, int); extern void rio_destroy(rio_t *); diff --git a/freebsd/sys/net/altq/altq_rmclass.c b/freebsd/sys/net/altq/altq_rmclass.c index 9ba3eb25..0f4a4899 100644 --- a/freebsd/sys/net/altq/altq_rmclass.c +++ b/freebsd/sys/net/altq/altq_rmclass.c @@ -51,17 +51,9 @@ #include <sys/systm.h> #include <sys/errno.h> #include <sys/time.h> -#ifdef ALTQ3_COMPAT -#include <sys/kernel.h> -#endif #include <net/if.h> #include <net/if_var.h> -#ifdef ALTQ3_COMPAT -#include <netinet/in.h> -#include <netinet/in_systm.h> -#include <netinet/ip.h> -#endif #include <net/altq/if_altq.h> #include <net/altq/altq.h> diff --git a/freebsd/sys/net/altq/altq_rmclass.h b/freebsd/sys/net/altq/altq_rmclass.h index 268ed63f..7e58ec49 100644 --- a/freebsd/sys/net/altq/altq_rmclass.h +++ b/freebsd/sys/net/altq/altq_rmclass.h @@ -233,13 +233,13 @@ struct rm_ifdat { }; /* flags for rmc_init and rmc_newclass */ -/* class flags */ +/* class flags; must be the same as class flags in altq_cbq.h */ #define RMCF_RED 0x0001 #define RMCF_ECN 0x0002 #define RMCF_RIO 0x0004 #define RMCF_FLOWVALVE 0x0008 /* use flowvalve (aka penalty-box) */ #define RMCF_CLEARDSCP 0x0010 /* clear diffserv codepoint */ -#define RMCF_CODEL 0x0020 +#define RMCF_CODEL 0x0040 /* flags for rmc_init */ #define RMCF_WRR 0x0100 diff --git a/freebsd/sys/net/altq/altq_subr.c b/freebsd/sys/net/altq/altq_subr.c index 6da36129..61aaec59 100644 --- a/freebsd/sys/net/altq/altq_subr.c +++ b/freebsd/sys/net/altq/altq_subr.c @@ -64,9 +64,6 @@ #include <netpfil/pf/pf.h> #include <netpfil/pf/pf_altq.h> #include <net/altq/altq.h> -#ifdef ALTQ3_COMPAT -#include <net/altq/altq_conf.h> -#endif /* machine dependent clock related includes */ #include <sys/bus.h> @@ -157,22 +154,6 @@ altq_attach(ifq, type, discipline, enqueue, dequeue, request, clfier, classify) return ENXIO; } -#ifdef ALTQ3_COMPAT - /* - * pfaltq can override the existing discipline, but altq3 cannot. - * check these if clfier is not NULL (which implies altq3). - */ - if (clfier != NULL) { - if (ALTQ_IS_ENABLED(ifq)) { - IFQ_UNLOCK(ifq); - return EBUSY; - } - if (ALTQ_IS_ATTACHED(ifq)) { - IFQ_UNLOCK(ifq); - return EEXIST; - } - } -#endif ifq->altq_type = type; ifq->altq_disc = discipline; ifq->altq_enqueue = enqueue; @@ -181,11 +162,6 @@ altq_attach(ifq, type, discipline, enqueue, dequeue, request, clfier, classify) ifq->altq_clfier = clfier; ifq->altq_classify = classify; ifq->altq_flags &= (ALTQF_CANTCHANGE|ALTQF_ENABLED); -#ifdef ALTQ3_COMPAT -#ifdef ALTQ_KLD - altq_module_incref(type); -#endif -#endif IFQ_UNLOCK(ifq); return 0; } @@ -208,11 +184,6 @@ altq_detach(ifq) IFQ_UNLOCK(ifq); return (0); } -#ifdef ALTQ3_COMPAT -#ifdef ALTQ_KLD - altq_module_declref(ifq->altq_type); -#endif -#endif ifq->altq_type = ALTQT_NONE; ifq->altq_disc = NULL; diff --git a/freebsd/sys/net/if.c b/freebsd/sys/net/if.c index d846482e..9d233444 100644 --- a/freebsd/sys/net/if.c +++ b/freebsd/sys/net/if.c @@ -604,8 +604,6 @@ if_free_internal(struct ifnet *ifp) #ifdef MAC mac_ifnet_destroy(ifp); #endif /* MAC */ - if (ifp->if_description != NULL) - free(ifp->if_description, M_IFDESCR); IF_AFDATA_DESTROY(ifp); IF_ADDR_LOCK_DESTROY(ifp); ifq_delete(&ifp->if_snd); @@ -613,6 +611,8 @@ if_free_internal(struct ifnet *ifp) for (int i = 0; i < IFCOUNTERS; i++) counter_u64_free(ifp->if_counters[i]); + free(ifp->if_description, M_IFDESCR); + free(ifp->if_hw_addr, M_IFADDR); free(ifp, M_IFNET); } @@ -1077,6 +1077,8 @@ if_detach_internal(struct ifnet *ifp, int vmove, struct if_clone **ifcp) CK_STAILQ_FOREACH(iter, &V_ifnet, if_link) if (iter == ifp) { CK_STAILQ_REMOVE(&V_ifnet, ifp, ifnet, if_link); + if (!vmove) + ifp->if_flags |= IFF_DYING; found = 1; break; } @@ -1191,14 +1193,8 @@ if_detach_internal(struct ifnet *ifp, int vmove, struct if_clone **ifcp) if_dead(ifp); /* - * Remove link ifaddr pointer and maybe decrement if_index. * Clean up all addresses. */ - free(ifp->if_hw_addr, M_IFADDR); - ifp->if_hw_addr = NULL; - ifp->if_addr = NULL; - - /* We can now free link ifaddr. */ IF_ADDR_WLOCK(ifp); if (!CK_STAILQ_EMPTY(&ifp->if_addrhead)) { ifa = CK_STAILQ_FIRST(&ifp->if_addrhead); diff --git a/freebsd/sys/net/if_ipsec.c b/freebsd/sys/net/if_ipsec.c index 7cc2c961..4b4d9ec5 100644 --- a/freebsd/sys/net/if_ipsec.c +++ b/freebsd/sys/net/if_ipsec.c @@ -974,7 +974,7 @@ static int ipsec_set_addresses(struct ifnet *ifp, struct sockaddr *src, struct sockaddr *dst) { - struct ipsec_softc *sc, *tsc; + struct ipsec_softc *sc; struct secasindex *saidx; sx_assert(&ipsec_ioctl_sx, SA_XLOCKED); @@ -989,18 +989,6 @@ ipsec_set_addresses(struct ifnet *ifp, struct sockaddr *src, return (0); /* Nothing has been changed. */ } - /* Check that given addresses aren't already configured */ - CK_LIST_FOREACH(tsc, ipsec_srchash(src), srchash) { - if (tsc == sc) - continue; - MPASS(tsc->family == src->sa_family); - saidx = ipsec_getsaidx(tsc, IPSEC_DIR_OUTBOUND, tsc->family); - if (key_sockaddrcmp(&saidx->src.sa, src, 0) == 0 && - key_sockaddrcmp(&saidx->dst.sa, dst, 0) == 0) { - /* We already have tunnel with such addresses */ - return (EADDRNOTAVAIL); - } - } /* If reqid is not set, generate new one. */ if (ipsec_init_reqid(sc) != 0) return (EEXIST); diff --git a/freebsd/sys/net/if_tap.c b/freebsd/sys/net/if_tap.c index a540a59a..dbf3e599 100644 --- a/freebsd/sys/net/if_tap.c +++ b/freebsd/sys/net/if_tap.c @@ -348,7 +348,7 @@ tapclone(void *arg, struct ucred *cred, char *name, int namelen, struct cdev **d return; if (!tapdclone || - (!tapuopen && priv_check_cred(cred, PRIV_NET_IFCREATE, 0) != 0)) + (!tapuopen && priv_check_cred(cred, PRIV_NET_IFCREATE) != 0)) return; unit = 0; diff --git a/freebsd/sys/net/if_tun.c b/freebsd/sys/net/if_tun.c index 328b1963..44441773 100644 --- a/freebsd/sys/net/if_tun.c +++ b/freebsd/sys/net/if_tun.c @@ -206,7 +206,7 @@ tunclone(void *arg, struct ucred *cred, char *name, int namelen, * If tun cloning is enabled, only the superuser can create an * interface. */ - if (!tundclone || priv_check_cred(cred, PRIV_NET_IFCREATE, 0) != 0) + if (!tundclone || priv_check_cred(cred, PRIV_NET_IFCREATE) != 0) return; if (strcmp(name, tunname) == 0) { diff --git a/freebsd/sys/net/if_vlan.c b/freebsd/sys/net/if_vlan.c index 6f07b4b4..893bb2cf 100644 --- a/freebsd/sys/net/if_vlan.c +++ b/freebsd/sys/net/if_vlan.c @@ -316,9 +316,6 @@ VNET_DEFINE_STATIC(struct if_clone *, vlan_cloner); #define V_vlan_cloner VNET(vlan_cloner) #endif -#ifndef VLAN_ARRAY -#define HASH(n, m) ((((n) >> 8) ^ ((n) >> 4) ^ (n)) & (m)) - static void vlan_mc_free(struct epoch_context *ctx) { @@ -326,6 +323,9 @@ vlan_mc_free(struct epoch_context *ctx) free(mc, M_VLAN); } +#ifndef VLAN_ARRAY +#define HASH(n, m) ((((n) >> 8) ^ ((n) >> 4) ^ (n)) & (m)) + static void vlan_inithash(struct ifvlantrunk *trunk) { diff --git a/freebsd/sys/net/route.h b/freebsd/sys/net/route.h index 15ec1b3e..c4333838 100644 --- a/freebsd/sys/net/route.h +++ b/freebsd/sys/net/route.h @@ -251,6 +251,7 @@ struct rt_msghdr { u_char rtm_version; /* future binary compatibility */ u_char rtm_type; /* message type */ u_short rtm_index; /* index for associated ifp */ + u_short _rtm_spare1; int rtm_flags; /* flags, incl. kern & message, e.g. DONE */ int rtm_addrs; /* bitmask identifying sockaddrs in msg */ pid_t rtm_pid; /* identify sender */ diff --git a/freebsd/sys/net/rtsock.c b/freebsd/sys/net/rtsock.c index 84afd627..e1b87095 100644 --- a/freebsd/sys/net/rtsock.c +++ b/freebsd/sys/net/rtsock.c @@ -85,6 +85,7 @@ struct if_msghdr32 { int32_t ifm_addrs; int32_t ifm_flags; uint16_t ifm_index; + uint16_t _ifm_spare1; struct if_data ifm_data; }; @@ -98,6 +99,7 @@ struct if_msghdrl32 { uint16_t _ifm_spare1; uint16_t ifm_len; uint16_t ifm_data_off; + uint32_t _ifm_spare2; struct if_data ifm_data; }; @@ -1231,8 +1233,11 @@ rtsock_msg_buffer(int type, struct rt_addrinfo *rtinfo, struct walkarg *w, int * dlen = ALIGN(len) - len; if (buflen < dlen) cp = NULL; - else + else { + bzero(cp, dlen); + cp += dlen; buflen -= dlen; + } } len = ALIGN(len); @@ -1566,6 +1571,8 @@ sysctl_dumpentry(struct radix_node *rn, void *vw) struct rt_addrinfo info; struct sockaddr_storage ss; + IFNET_RLOCK_NOSLEEP_ASSERT(); + if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg)) return 0; if ((rt->rt_flags & RTF_HOST) == 0 @@ -1578,7 +1585,7 @@ sysctl_dumpentry(struct radix_node *rn, void *vw) info.rti_info[RTAX_NETMASK] = rtsock_fix_netmask(rt_key(rt), rt_mask(rt), &ss); info.rti_info[RTAX_GENMASK] = 0; - if (rt->rt_ifp) { + if (rt->rt_ifp && !(rt->rt_ifp->if_flags & IFF_DYING)) { info.rti_info[RTAX_IFP] = rt->rt_ifp->if_addr->ifa_addr; info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr; if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) @@ -1589,6 +1596,8 @@ sysctl_dumpentry(struct radix_node *rn, void *vw) if (w->w_req && w->w_tmem) { struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem; + bzero(&rtm->rtm_index, + sizeof(*rtm) - offsetof(struct rt_msghdr, rtm_index)); if (rt->rt_flags & RTF_GWFLAG_COMPAT) rtm->rtm_flags = RTF_GATEWAY | (rt->rt_flags & ~RTF_GWFLAG_COMPAT); @@ -1596,7 +1605,6 @@ sysctl_dumpentry(struct radix_node *rn, void *vw) rtm->rtm_flags = rt->rt_flags; rt_getmetrics(rt, &rtm->rtm_rmx); rtm->rtm_index = rt->rt_ifp->if_index; - rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0; rtm->rtm_addrs = info.rti_addrs; error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size); return (error); @@ -1624,6 +1632,7 @@ sysctl_iflist_ifml(struct ifnet *ifp, const struct if_data *src_ifd, ifm32->_ifm_spare1 = 0; ifm32->ifm_len = sizeof(*ifm32); ifm32->ifm_data_off = offsetof(struct if_msghdrl32, ifm_data); + ifm32->_ifm_spare2 = 0; ifd = &ifm32->ifm_data; } else #endif @@ -1634,6 +1643,7 @@ sysctl_iflist_ifml(struct ifnet *ifp, const struct if_data *src_ifd, ifm->_ifm_spare1 = 0; ifm->ifm_len = sizeof(*ifm); ifm->ifm_data_off = offsetof(struct if_msghdrl, ifm_data); + ifm->_ifm_spare2 = 0; ifd = &ifm->ifm_data; } @@ -1659,6 +1669,7 @@ sysctl_iflist_ifm(struct ifnet *ifp, const struct if_data *src_ifd, ifm32->ifm_addrs = info->rti_addrs; ifm32->ifm_flags = ifp->if_flags | ifp->if_drv_flags; ifm32->ifm_index = ifp->if_index; + ifm32->_ifm_spare1 = 0; ifd = &ifm32->ifm_data; } else #endif @@ -1666,6 +1677,7 @@ sysctl_iflist_ifm(struct ifnet *ifp, const struct if_data *src_ifd, ifm->ifm_addrs = info->rti_addrs; ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags; ifm->ifm_index = ifp->if_index; + ifm->_ifm_spare1 = 0; ifd = &ifm->ifm_data; } @@ -1734,6 +1746,7 @@ sysctl_iflist_ifam(struct ifaddr *ifa, struct rt_addrinfo *info, ifam->ifam_addrs = info->rti_addrs; ifam->ifam_flags = ifa->ifa_flags; ifam->ifam_index = ifa->ifa_ifp->if_index; + ifam->_ifam_spare1 = 0; ifam->ifam_metric = ifa->ifa_ifp->if_metric; return (SYSCTL_OUT(w->w_req, w->w_tmem, len)); @@ -1845,6 +1858,7 @@ sysctl_ifmalist(int af, struct walkarg *w) ifmam->ifmam_index = ifma->ifma_ifp->if_index; ifmam->ifmam_flags = 0; ifmam->ifmam_addrs = info.rti_addrs; + ifmam->_ifmam_spare1 = 0; error = SYSCTL_OUT(w->w_req, w->w_tmem, len); if (error != 0) break; @@ -1942,8 +1956,10 @@ sysctl_rtsock(SYSCTL_HANDLER_ARGS) rnh = rt_tables_get_rnh(fib, i); if (rnh != NULL) { RIB_RLOCK(rnh); + IFNET_RLOCK_NOSLEEP(); error = rnh->rnh_walktree(&rnh->head, sysctl_dumpentry, &w); + IFNET_RUNLOCK_NOSLEEP(); RIB_RUNLOCK(rnh); } else if (af != 0) error = EAFNOSUPPORT; diff --git a/freebsd/sys/net80211/ieee80211_node.c b/freebsd/sys/net80211/ieee80211_node.c index 55e51299..ec726880 100644 --- a/freebsd/sys/net80211/ieee80211_node.c +++ b/freebsd/sys/net80211/ieee80211_node.c @@ -1419,8 +1419,6 @@ ieee80211_alloc_node(struct ieee80211_node_table *nt, IEEE80211_NOTE(vap, IEEE80211_MSG_INACT, ni, "%s: inact_reload %u", __func__, ni->ni_inact_reload); - ieee80211_ratectl_node_init(ni); - return ni; } diff --git a/freebsd/sys/netinet/cc/cc.h b/freebsd/sys/netinet/cc/cc.h index 1389b27a..b1a69ca1 100644 --- a/freebsd/sys/netinet/cc/cc.h +++ b/freebsd/sys/netinet/cc/cc.h @@ -102,8 +102,6 @@ struct cc_var { #define CCF_ACKNOW 0x0008 /* Will this ack be sent now? */ #define CCF_IPHDR_CE 0x0010 /* Does this packet set CE bit? */ #define CCF_TCPHDR_CWR 0x0020 /* Does this packet set CWR bit? */ -#define CCF_MAX_CWND 0x0040 /* Have we reached maximum cwnd? */ -#define CCF_CHG_MAX_CWND 0x0080 /* Cubic max_cwnd changed, for K */ /* ACK types passed to the ack_received() hook. */ #define CC_ACK 0x0001 /* Regular in sequence ACK. */ @@ -184,4 +182,6 @@ extern struct rwlock cc_list_lock; #define CC_LIST_WUNLOCK() rw_wunlock(&cc_list_lock) #define CC_LIST_LOCK_ASSERT() rw_assert(&cc_list_lock, RA_LOCKED) +#define CC_ALGOOPT_LIMIT 2048 + #endif /* _NETINET_CC_CC_H_ */ diff --git a/freebsd/sys/netinet/cc/cc_newreno.c b/freebsd/sys/netinet/cc/cc_newreno.c index 2450f08e..b1307c92 100644 --- a/freebsd/sys/netinet/cc/cc_newreno.c +++ b/freebsd/sys/netinet/cc/cc_newreno.c @@ -81,8 +81,6 @@ __FBSDID("$FreeBSD$"); static MALLOC_DEFINE(M_NEWRENO, "newreno data", "newreno beta values"); -#define CAST_PTR_INT(X) (*((int*)(X))) - static void newreno_cb_destroy(struct cc_var *ccv); static void newreno_ack_received(struct cc_var *ccv, uint16_t type); static void newreno_after_idle(struct cc_var *ccv); @@ -366,15 +364,21 @@ newreno_ctl_output(struct cc_var *ccv, struct sockopt *sopt, void *buf) static int newreno_beta_handler(SYSCTL_HANDLER_ARGS) { + int error; + uint32_t new; - if (req->newptr != NULL ) { + new = *(uint32_t *)arg1; + error = sysctl_handle_int(oidp, &new, 0, req); + if (error == 0 && req->newptr != NULL ) { if (arg1 == &VNET_NAME(newreno_beta_ecn) && !V_cc_do_abe) - return (EACCES); - if (CAST_PTR_INT(req->newptr) <= 0 || CAST_PTR_INT(req->newptr) > 100) - return (EINVAL); + error = EACCES; + else if (new == 0 || new > 100) + error = EINVAL; + else + *(uint32_t *)arg1 = new; } - return (sysctl_handle_int(oidp, arg1, arg2, req)); + return (error); } SYSCTL_DECL(_net_inet_tcp_cc_newreno); diff --git a/freebsd/sys/netinet/if_ether.c b/freebsd/sys/netinet/if_ether.c index 6ee6b71c..e9efd89e 100644 --- a/freebsd/sys/netinet/if_ether.c +++ b/freebsd/sys/netinet/if_ether.c @@ -435,10 +435,10 @@ arprequest(struct ifnet *ifp, const struct in_addr *sip, /* * Resolve an IP address into an ethernet address - heavy version. * Used internally by arpresolve(). - * We have already checked than we can't use existing lle without - * modification so we have to acquire LLE_EXCLUSIVE lle lock. + * We have already checked that we can't use an existing lle without + * modification so we have to acquire an LLE_EXCLUSIVE lle lock. * - * On success, desten and flags are filled in and the function returns 0; + * On success, desten and pflags are filled in and the function returns 0; * If the packet must be held pending resolution, we return EWOULDBLOCK * On other errors, we return the corresponding error code. * Note that m_freem() handles NULL. @@ -575,21 +575,6 @@ arpresolve_full(struct ifnet *ifp, int is_gw, int flags, struct mbuf *m, } /* - * Resolve an IP address into an ethernet address. - */ -int -arpresolve_addr(struct ifnet *ifp, int flags, const struct sockaddr *dst, - char *desten, uint32_t *pflags, struct llentry **plle) -{ - int error; - - flags |= LLE_ADDRONLY; - error = arpresolve_full(ifp, 0, flags, NULL, dst, desten, pflags, plle); - return (error); -} - - -/* * Lookups link header based on an IP address. * On input: * ifp is the interface we use diff --git a/freebsd/sys/netinet/if_ether.h b/freebsd/sys/netinet/if_ether.h index 028e45a7..2f56e7a9 100644 --- a/freebsd/sys/netinet/if_ether.h +++ b/freebsd/sys/netinet/if_ether.h @@ -117,9 +117,6 @@ extern u_char ether_ipmulticast_max[ETHER_ADDR_LEN]; struct ifaddr; struct llentry; -int arpresolve_addr(struct ifnet *ifp, int flags, - const struct sockaddr *dst, char *desten, uint32_t *pflags, - struct llentry **plle); int arpresolve(struct ifnet *ifp, int is_gw, struct mbuf *m, const struct sockaddr *dst, u_char *desten, uint32_t *pflags, struct llentry **plle); diff --git a/freebsd/sys/netinet/in_pcb.c b/freebsd/sys/netinet/in_pcb.c index 6bf43464..c00593e5 100644 --- a/freebsd/sys/netinet/in_pcb.c +++ b/freebsd/sys/netinet/in_pcb.c @@ -345,8 +345,7 @@ in_pcbinslbgrouphash(struct inpcb *inp) } #endif - idx = INP_PCBLBGROUP_PORTHASH(inp->inp_lport, - pcbinfo->ipi_lbgrouphashmask); + idx = INP_PCBPORTHASH(inp->inp_lport, pcbinfo->ipi_lbgrouphashmask); hdr = &pcbinfo->ipi_lbgrouphashbase[idx]; CK_LIST_FOREACH(grp, hdr, il_list) { if (grp->il_vflag == inp->inp_vflag && @@ -403,9 +402,7 @@ in_pcbremlbgrouphash(struct inpcb *inp) INP_HASH_WLOCK_ASSERT(pcbinfo); hdr = &pcbinfo->ipi_lbgrouphashbase[ - INP_PCBLBGROUP_PORTHASH(inp->inp_lport, - pcbinfo->ipi_lbgrouphashmask)]; - + INP_PCBPORTHASH(inp->inp_lport, pcbinfo->ipi_lbgrouphashmask)]; CK_LIST_FOREACH(grp, hdr, il_list) { for (i = 0; i < grp->il_inpcnt; ++i) { if (grp->il_inp[i] != inp) @@ -445,6 +442,8 @@ in_pcbinfo_init(struct inpcbinfo *pcbinfo, const char *name, char *inpcbzone_name, uma_init inpcbzone_init, u_int hashfields) { + porthash_nelements = imin(porthash_nelements, IPPORT_MAX + 1); + INP_INFO_LOCK_INIT(pcbinfo, name); INP_HASH_LOCK_INIT(pcbinfo, "pcbinfohash"); /* XXXRW: argument? */ INP_LIST_LOCK_INIT(pcbinfo, "pcbinfolist"); @@ -458,7 +457,7 @@ in_pcbinfo_init(struct inpcbinfo *pcbinfo, const char *name, &pcbinfo->ipi_hashmask); pcbinfo->ipi_porthashbase = hashinit(porthash_nelements, M_PCB, &pcbinfo->ipi_porthashmask); - pcbinfo->ipi_lbgrouphashbase = hashinit(hash_nelements, M_PCB, + pcbinfo->ipi_lbgrouphashbase = hashinit(porthash_nelements, M_PCB, &pcbinfo->ipi_lbgrouphashmask); #ifdef PCBGROUP in_pcbgroup_init(pcbinfo, hashfields, hash_nelements); @@ -629,7 +628,7 @@ in_pcb_lport(struct inpcb *inp, struct in_addr *laddrp, u_short *lportp, last = V_ipport_hilastauto; lastport = &pcbinfo->ipi_lasthi; } else if (inp->inp_flags & INP_LOWPORT) { - error = priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, 0); + error = priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT); if (error) return (error); first = V_ipport_lowfirstauto; /* 1023 */ @@ -875,12 +874,10 @@ in_pcbbind_setup(struct inpcb *inp, struct sockaddr *nam, in_addr_t *laddrp, /* GROSS */ if (ntohs(lport) <= V_ipport_reservedhigh && ntohs(lport) >= V_ipport_reservedlow && - priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, - 0)) + priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT)) return (EACCES); if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)) && - priv_check_cred(inp->inp_cred, - PRIV_NETINET_REUSEPORT, 0) != 0) { + priv_check_cred(inp->inp_cred, PRIV_NETINET_REUSEPORT) != 0) { t = in_pcblookup_local(pcbinfo, sin->sin_addr, lport, INPLOOKUP_WILDCARD, cred); /* @@ -1962,8 +1959,8 @@ in_pcblookup_lbgroup(const struct inpcbinfo *pcbinfo, INP_HASH_LOCK_ASSERT(pcbinfo); - hdr = &pcbinfo->ipi_lbgrouphashbase[INP_PCBLBGROUP_PORTHASH(lport, - pcbinfo->ipi_lbgrouphashmask)]; + hdr = &pcbinfo->ipi_lbgrouphashbase[ + INP_PCBPORTHASH(lport, pcbinfo->ipi_lbgrouphashmask)]; /* * Order of socket selection: @@ -2895,11 +2892,10 @@ void in_pcbtoxinpcb(const struct inpcb *inp, struct xinpcb *xi) { + bzero(xi, sizeof(*xi)); xi->xi_len = sizeof(struct xinpcb); if (inp->inp_socket) sotoxsocket(inp->inp_socket, &xi->xi_socket); - else - bzero(&xi->xi_socket, sizeof(struct xsocket)); bcopy(&inp->inp_inc, &xi->inp_inc, sizeof(struct in_conninfo)); xi->inp_gencnt = inp->inp_gencnt; xi->inp_ppcb = (uintptr_t)inp->inp_ppcb; diff --git a/freebsd/sys/netinet/in_pcb.h b/freebsd/sys/netinet/in_pcb.h index 6d2c86d5..ecbd7a22 100644 --- a/freebsd/sys/netinet/in_pcb.h +++ b/freebsd/sys/netinet/in_pcb.h @@ -688,8 +688,6 @@ int inp_so_options(const struct inpcb *inp); (((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) & (mask)) #define INP_PCBPORTHASH(lport, mask) \ (ntohs((lport)) & (mask)) -#define INP_PCBLBGROUP_PORTHASH(lport, mask) \ - (ntohs((lport)) & (mask)) #define INP_PCBLBGROUP_PKTHASH(faddr, lport, fport) \ ((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) #define INP6_PCBHASHKEY(faddr) ((faddr)->s6_addr32[3]) diff --git a/freebsd/sys/netinet/ip_divert.c b/freebsd/sys/netinet/ip_divert.c index fbf74ca1..6a99645e 100644 --- a/freebsd/sys/netinet/ip_divert.c +++ b/freebsd/sys/netinet/ip_divert.c @@ -666,6 +666,7 @@ div_pcblist(SYSCTL_HANDLER_ARGS) if (error != 0) return (error); + bzero(&xig, sizeof(xig)); xig.xig_len = sizeof xig; xig.xig_count = n; xig.xig_gen = gencnt; diff --git a/freebsd/sys/netinet/ip_fw.h b/freebsd/sys/netinet/ip_fw.h index cfcdaa29..41351215 100644 --- a/freebsd/sys/netinet/ip_fw.h +++ b/freebsd/sys/netinet/ip_fw.h @@ -551,11 +551,12 @@ typedef struct _ipfw_insn_nat { } ipfw_insn_nat; /* Apply ipv6 mask on ipv6 addr */ -#define APPLY_MASK(addr,mask) \ +#define APPLY_MASK(addr,mask) do { \ (addr)->__u6_addr.__u6_addr32[0] &= (mask)->__u6_addr.__u6_addr32[0]; \ (addr)->__u6_addr.__u6_addr32[1] &= (mask)->__u6_addr.__u6_addr32[1]; \ (addr)->__u6_addr.__u6_addr32[2] &= (mask)->__u6_addr.__u6_addr32[2]; \ - (addr)->__u6_addr.__u6_addr32[3] &= (mask)->__u6_addr.__u6_addr32[3]; + (addr)->__u6_addr.__u6_addr32[3] &= (mask)->__u6_addr.__u6_addr32[3]; \ +} while (0) /* Structure for ipv6 */ typedef struct _ipfw_insn_ip6 { @@ -707,6 +708,7 @@ struct _ipfw_dyn_rule { u_int32_t state; /* state of this rule (typically a * combination of TCP flags) */ +#define IPFW_DYN_ORPHANED 0x40000 /* state's parent rule was deleted */ u_int32_t ack_fwd; /* most recent ACKs in forward */ u_int32_t ack_rev; /* and reverse directions (used */ /* to generate keepalives) */ @@ -937,9 +939,10 @@ typedef struct _ipfw_range_tlv { #define IPFW_RCFLAG_RANGE 0x01 /* rule range is set */ #define IPFW_RCFLAG_ALL 0x02 /* match ALL rules */ #define IPFW_RCFLAG_SET 0x04 /* match rules in given set */ +#define IPFW_RCFLAG_DYNAMIC 0x08 /* match only dynamic states */ /* User-settable flags */ #define IPFW_RCFLAG_USER (IPFW_RCFLAG_RANGE | IPFW_RCFLAG_ALL | \ - IPFW_RCFLAG_SET) + IPFW_RCFLAG_SET | IPFW_RCFLAG_DYNAMIC) /* Internally used flags */ #define IPFW_RCFLAG_DEFAULT 0x0100 /* Do not skip defaul rule */ diff --git a/freebsd/sys/netinet/ip_reass.c b/freebsd/sys/netinet/ip_reass.c index 8bc5b53b..70a6edae 100644 --- a/freebsd/sys/netinet/ip_reass.c +++ b/freebsd/sys/netinet/ip_reass.c @@ -213,19 +213,21 @@ ip_reass(struct mbuf *m) * convert offset of this to bytes. */ ip->ip_len = htons(ntohs(ip->ip_len) - hlen); - if (ip->ip_off & htons(IP_MF)) { - /* - * Make sure that fragments have a data length - * that's a non-zero multiple of 8 bytes. - */ - if (ip->ip_len == htons(0) || (ntohs(ip->ip_len) & 0x7) != 0) { - IPSTAT_INC(ips_toosmall); /* XXX */ - IPSTAT_INC(ips_fragdropped); - m_freem(m); - return (NULL); - } + /* + * Make sure that fragments have a data length + * that's a non-zero multiple of 8 bytes, unless + * this is the last fragment. + */ + if (ip->ip_len == htons(0) || + ((ip->ip_off & htons(IP_MF)) && (ntohs(ip->ip_len) & 0x7) != 0)) { + IPSTAT_INC(ips_toosmall); /* XXX */ + IPSTAT_INC(ips_fragdropped); + m_freem(m); + return (NULL); + } + if (ip->ip_off & htons(IP_MF)) m->m_flags |= M_IP_FRAG; - } else + else m->m_flags &= ~M_IP_FRAG; ip->ip_off = htons(ntohs(ip->ip_off) << 3); @@ -303,9 +305,28 @@ ip_reass(struct mbuf *m) fp->ipq_src = ip->ip_src; fp->ipq_dst = ip->ip_dst; fp->ipq_frags = m; + if (m->m_flags & M_IP_FRAG) + fp->ipq_maxoff = -1; + else + fp->ipq_maxoff = ntohs(ip->ip_off) + ntohs(ip->ip_len); m->m_nextpkt = NULL; goto done; } else { + /* + * If we already saw the last fragment, make sure + * this fragment's offset looks sane. Otherwise, if + * this is the last fragment, record its endpoint. + */ + if (fp->ipq_maxoff > 0) { + i = ntohs(ip->ip_off) + ntohs(ip->ip_len); + if (((m->m_flags & M_IP_FRAG) && i >= fp->ipq_maxoff) || + ((m->m_flags & M_IP_FRAG) == 0 && + i != fp->ipq_maxoff)) { + fp = NULL; + goto dropfrag; + } + } else if ((m->m_flags & M_IP_FRAG) == 0) + fp->ipq_maxoff = ntohs(ip->ip_off) + ntohs(ip->ip_len); fp->ipq_nfrags++; atomic_add_int(&nfrags, 1); #ifdef MAC diff --git a/freebsd/sys/netinet/ip_var.h b/freebsd/sys/netinet/ip_var.h index f874628a..86615a15 100644 --- a/freebsd/sys/netinet/ip_var.h +++ b/freebsd/sys/netinet/ip_var.h @@ -61,6 +61,7 @@ struct ipq { u_char ipq_ttl; /* time for reass q to live */ u_char ipq_p; /* protocol of this fragment */ u_short ipq_id; /* sequence id for reassembly */ + int ipq_maxoff; /* total length of packet */ struct mbuf *ipq_frags; /* to ip headers of fragments */ struct in_addr ipq_src,ipq_dst; u_char ipq_nfrags; /* # frags in this packet */ diff --git a/freebsd/sys/netinet/raw_ip.c b/freebsd/sys/netinet/raw_ip.c index a97eadae..1cac8d12 100644 --- a/freebsd/sys/netinet/raw_ip.c +++ b/freebsd/sys/netinet/raw_ip.c @@ -1062,6 +1062,7 @@ rip_pcblist(SYSCTL_HANDLER_ARGS) n = V_ripcbinfo.ipi_count; INP_INFO_WUNLOCK(&V_ripcbinfo); + bzero(&xig, sizeof(xig)); xig.xig_len = sizeof xig; xig.xig_count = n; xig.xig_gen = gencnt; diff --git a/freebsd/sys/netinet/sctp_sysctl.c b/freebsd/sys/netinet/sctp_sysctl.c index a4343cbe..1ba7e261 100644 --- a/freebsd/sys/netinet/sctp_sysctl.c +++ b/freebsd/sys/netinet/sctp_sysctl.c @@ -397,6 +397,9 @@ sctp_sysctl_handle_assoclist(SYSCTL_HANDLER_ARGS) SCTP_LTRACE_ERR_RET(NULL, NULL, NULL, SCTP_FROM_SCTP_SYSCTL, EPERM); return (EPERM); } + memset(&xinpcb, 0, sizeof(xinpcb)); + memset(&xstcb, 0, sizeof(xstcb)); + memset(&xraddr, 0, sizeof(xraddr)); LIST_FOREACH(inp, &SCTP_BASE_INFO(listhead), sctp_list) { SCTP_INP_RLOCK(inp); if (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) { diff --git a/freebsd/sys/netinet/tcp_output.c b/freebsd/sys/netinet/tcp_output.c index 60276348..b641f04c 100644 --- a/freebsd/sys/netinet/tcp_output.c +++ b/freebsd/sys/netinet/tcp_output.c @@ -1174,14 +1174,18 @@ send: /* * Calculate receive window. Don't shrink window, * but avoid silly window syndrome. + * If a RST segment is sent, advertise a window of zero. */ - if (recwin < (so->so_rcv.sb_hiwat / 4) && - recwin < tp->t_maxseg) + if (flags & TH_RST) { recwin = 0; - if (SEQ_GT(tp->rcv_adv, tp->rcv_nxt) && - recwin < (tp->rcv_adv - tp->rcv_nxt)) - recwin = (tp->rcv_adv - tp->rcv_nxt); - + } else { + if (recwin < (so->so_rcv.sb_hiwat / 4) && + recwin < tp->t_maxseg) + recwin = 0; + if (SEQ_GT(tp->rcv_adv, tp->rcv_nxt) && + recwin < (tp->rcv_adv - tp->rcv_nxt)) + recwin = (tp->rcv_adv - tp->rcv_nxt); + } /* * According to RFC1323 the window field in a SYN (i.e., a <SYN> * or <SYN,ACK>) segment itself is never scaled. The <SYN,ACK> diff --git a/freebsd/sys/netinet/tcp_subr.c b/freebsd/sys/netinet/tcp_subr.c index 4852ffaf..52d2ca2a 100644 --- a/freebsd/sys/netinet/tcp_subr.c +++ b/freebsd/sys/netinet/tcp_subr.c @@ -562,6 +562,7 @@ sysctl_net_inet_list_func_info(SYSCTL_HANDLER_ARGS) cnt++; #endif if (req->oldptr != NULL) { + bzero(&tfi, sizeof(tfi)); tfi.tfi_refcnt = f->tf_fb->tfb_refcnt; tfi.tfi_id = f->tf_fb->tfb_id; (void)strncpy(tfi.tfi_alias, f->tf_name, @@ -2160,6 +2161,7 @@ tcp_pcblist(SYSCTL_HANDLER_ARGS) if (error != 0) return (error); + bzero(&xig, sizeof(xig)); xig.xig_len = sizeof xig; xig.xig_count = n + m; xig.xig_gen = gencnt; @@ -3221,8 +3223,8 @@ tcp_inptoxtp(const struct inpcb *inp, struct xtcpcb *xt) struct tcpcb *tp = intotcpcb(inp); sbintime_t now; + bzero(xt, sizeof(*xt)); if (inp->inp_flags & INP_TIMEWAIT) { - bzero(xt, sizeof(struct xtcpcb)); xt->t_state = TCPS_TIME_WAIT; } else { xt->t_state = tp->t_state; @@ -3250,7 +3252,6 @@ tcp_inptoxtp(const struct inpcb *inp, struct xtcpcb *xt) bcopy(tp->t_fb->tfb_tcp_block_name, xt->xt_stack, TCP_FUNCTION_NAME_LEN_MAX); - bzero(xt->xt_logid, TCP_LOG_ID_LEN); #ifdef TCP_BLACKBOX (void)tcp_log_get_id(tp, xt->xt_logid); #endif diff --git a/freebsd/sys/netinet/tcp_usrreq.c b/freebsd/sys/netinet/tcp_usrreq.c index 617f60d0..27ab745b 100644 --- a/freebsd/sys/netinet/tcp_usrreq.c +++ b/freebsd/sys/netinet/tcp_usrreq.c @@ -1771,6 +1771,8 @@ tcp_default_ctloutput(struct socket *so, struct sockopt *sopt, struct inpcb *inp switch (sopt->sopt_name) { case TCP_CCALGOOPT: INP_WUNLOCK(inp); + if (sopt->sopt_valsize > CC_ALGOOPT_LIMIT) + return (EINVAL); pbuf = malloc(sopt->sopt_valsize, M_TEMP, M_WAITOK | M_ZERO); error = sooptcopyin(sopt, pbuf, sopt->sopt_valsize, sopt->sopt_valsize); diff --git a/freebsd/sys/netinet/udp_usrreq.c b/freebsd/sys/netinet/udp_usrreq.c index ca4fecc0..33b89c21 100644 --- a/freebsd/sys/netinet/udp_usrreq.c +++ b/freebsd/sys/netinet/udp_usrreq.c @@ -893,6 +893,7 @@ udp_pcblist(SYSCTL_HANDLER_ARGS) if (error != 0) return (error); + bzero(&xig, sizeof(xig)); xig.xig_len = sizeof xig; xig.xig_count = n; xig.xig_gen = gencnt; diff --git a/freebsd/sys/netinet6/in6_pcb.c b/freebsd/sys/netinet6/in6_pcb.c index f5b22db7..a6beba43 100644 --- a/freebsd/sys/netinet6/in6_pcb.c +++ b/freebsd/sys/netinet6/in6_pcb.c @@ -218,12 +218,10 @@ in6_pcbbind(struct inpcb *inp, struct sockaddr *nam, /* GROSS */ if (ntohs(lport) <= V_ipport_reservedhigh && ntohs(lport) >= V_ipport_reservedlow && - priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT, - 0)) + priv_check_cred(cred, PRIV_NETINET_RESERVEDPORT)) return (EACCES); if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr) && - priv_check_cred(inp->inp_cred, - PRIV_NETINET_REUSEPORT, 0) != 0) { + priv_check_cred(inp->inp_cred, PRIV_NETINET_REUSEPORT) != 0) { t = in6_pcblookup_local(pcbinfo, &sin6->sin6_addr, lport, INPLOOKUP_WILDCARD, cred); @@ -894,8 +892,8 @@ in6_pcblookup_lbgroup(const struct inpcbinfo *pcbinfo, INP_HASH_LOCK_ASSERT(pcbinfo); - hdr = &pcbinfo->ipi_lbgrouphashbase[INP_PCBLBGROUP_PORTHASH( - lport, pcbinfo->ipi_lbgrouphashmask)]; + hdr = &pcbinfo->ipi_lbgrouphashbase[ + INP_PCBPORTHASH(lport, pcbinfo->ipi_lbgrouphashmask)]; /* * Order of socket selection: diff --git a/freebsd/sys/netinet6/ip6_mroute.c b/freebsd/sys/netinet6/ip6_mroute.c index c1f66028..9dee53b0 100644 --- a/freebsd/sys/netinet6/ip6_mroute.c +++ b/freebsd/sys/netinet6/ip6_mroute.c @@ -205,7 +205,8 @@ sysctl_mif6table(SYSCTL_HANDLER_ARGS) struct mif6_sctl *out; int error; - out = malloc(sizeof(struct mif6_sctl) * MAXMIFS, M_TEMP, M_WAITOK); + out = malloc(sizeof(struct mif6_sctl) * MAXMIFS, M_TEMP, + M_WAITOK | M_ZERO); for (int i = 0; i < MAXMIFS; i++) { out[i].m6_flags = mif6table[i].m6_flags; out[i].m6_rate_limit = mif6table[i].m6_rate_limit; diff --git a/freebsd/sys/netinet6/ip6_output.c b/freebsd/sys/netinet6/ip6_output.c index d3e530a6..0851bef8 100644 --- a/freebsd/sys/netinet6/ip6_output.c +++ b/freebsd/sys/netinet6/ip6_output.c @@ -2783,8 +2783,7 @@ ip6_setpktopt(int optname, u_char *buf, int len, struct ip6_pktopts *opt, case IPV6_2292NEXTHOP: case IPV6_NEXTHOP: if (cred != NULL) { - error = priv_check_cred(cred, - PRIV_NETINET_SETHDROPTS, 0); + error = priv_check_cred(cred, PRIV_NETINET_SETHDROPTS); if (error) return (error); } @@ -2842,8 +2841,7 @@ ip6_setpktopt(int optname, u_char *buf, int len, struct ip6_pktopts *opt, * overhead. */ if (cred != NULL) { - error = priv_check_cred(cred, - PRIV_NETINET_SETHDROPTS, 0); + error = priv_check_cred(cred, PRIV_NETINET_SETHDROPTS); if (error) return (error); } @@ -2879,8 +2877,7 @@ ip6_setpktopt(int optname, u_char *buf, int len, struct ip6_pktopts *opt, int destlen; if (cred != NULL) { /* XXX: see the comment for IPV6_HOPOPTS */ - error = priv_check_cred(cred, - PRIV_NETINET_SETHDROPTS, 0); + error = priv_check_cred(cred, PRIV_NETINET_SETHDROPTS); if (error) return (error); } diff --git a/freebsd/sys/netipsec/ipsec_pcb.c b/freebsd/sys/netipsec/ipsec_pcb.c index a8992d56..3b98ab65 100644 --- a/freebsd/sys/netipsec/ipsec_pcb.c +++ b/freebsd/sys/netipsec/ipsec_pcb.c @@ -298,7 +298,7 @@ ipsec_set_pcbpolicy(struct inpcb *inp, struct ucred *cred, case IPSEC_POLICY_IPSEC: case IPSEC_POLICY_BYPASS: if (cred != NULL && - priv_check_cred(cred, PRIV_NETINET_IPSEC, 0) != 0) + priv_check_cred(cred, PRIV_NETINET_IPSEC) != 0) return (EACCES); /* Allocate new SP entry. */ newsp = key_msg2sp(xpl, len, &error); diff --git a/freebsd/sys/netpfil/ipfw/ip_fw_private.h b/freebsd/sys/netpfil/ipfw/ip_fw_private.h index c389e01a..7e966d0a 100644 --- a/freebsd/sys/netpfil/ipfw/ip_fw_private.h +++ b/freebsd/sys/netpfil/ipfw/ip_fw_private.h @@ -146,6 +146,9 @@ enum { /* * Function definitions. */ +int ipfw_chk(struct ip_fw_args *args); +struct mbuf *ipfw_send_pkt(struct mbuf *, struct ipfw_flow_id *, + u_int32_t, u_int32_t, int); /* attach (arg = 1) or detach (arg = 0) hooks */ int ipfw_attach_hooks(int); @@ -156,6 +159,7 @@ void ipfw_nat_destroy(void); /* In ip_fw_log.c */ struct ip; struct ip_fw_chain; + void ipfw_bpf_init(int); void ipfw_bpf_uninit(int); void ipfw_bpf_mtap2(void *, u_int, struct mbuf *); @@ -168,6 +172,7 @@ VNET_DECLARE(int, verbose_limit); #define V_verbose_limit VNET(verbose_limit) /* In ip_fw_dynamic.c */ +struct sockopt_data; enum { /* result for matching dynamic rules */ MATCH_REVERSE = 0, @@ -177,19 +182,6 @@ enum { /* result for matching dynamic rules */ }; /* - * The lock for dynamic rules is only used once outside the file, - * and only to release the result of lookup_dyn_rule(). - * Eventually we may implement it with a callback on the function. - */ -struct ip_fw_chain; -struct sockopt_data; -int ipfw_is_dyn_rule(struct ip_fw *rule); -void ipfw_expire_dyn_states(struct ip_fw_chain *, ipfw_range_tlv *); - -struct tcphdr; -struct mbuf *ipfw_send_pkt(struct mbuf *, struct ipfw_flow_id *, - u_int32_t, u_int32_t, int); -/* * Macro to determine that we need to do or redo dynamic state lookup. * direction == MATCH_UNKNOWN means that this is first lookup, then we need * to do lookup. @@ -219,13 +211,17 @@ struct ip_fw *ipfw_dyn_lookup_state(const struct ip_fw_args *args, const void *ulp, int pktlen, const ipfw_insn *cmd, struct ipfw_dyn_info *info); +int ipfw_is_dyn_rule(struct ip_fw *rule); +void ipfw_expire_dyn_states(struct ip_fw_chain *, ipfw_range_tlv *); void ipfw_get_dynamic(struct ip_fw_chain *chain, char **bp, const char *ep); int ipfw_dump_states(struct ip_fw_chain *chain, struct sockopt_data *sd); void ipfw_dyn_init(struct ip_fw_chain *); /* per-vnet initialization */ void ipfw_dyn_uninit(int); /* per-vnet deinitialization */ int ipfw_dyn_len(void); -uint32_t ipfw_dyn_get_count(void); +uint32_t ipfw_dyn_get_count(uint32_t *, int *); +void ipfw_dyn_reset_eaction(struct ip_fw_chain *ch, uint16_t eaction_id, + uint16_t default_id, uint16_t instance_id); /* common variables */ VNET_DECLARE(int, fw_one_pass); @@ -280,7 +276,9 @@ struct ip_fw { uint32_t id; /* rule id */ uint32_t cached_id; /* used by jump_fast */ uint32_t cached_pos; /* used by jump_fast */ + uint32_t refcnt; /* number of references */ + struct ip_fw *next; /* linked list of deleted rules */ ipfw_insn cmd[1]; /* storage for commands */ }; @@ -650,7 +648,6 @@ void ipfw_init_skipto_cache(struct ip_fw_chain *chain); void ipfw_destroy_skipto_cache(struct ip_fw_chain *chain); int ipfw_find_rule(struct ip_fw_chain *chain, uint32_t key, uint32_t id); int ipfw_ctl3(struct sockopt *sopt); -int ipfw_chk(struct ip_fw_args *args); int ipfw_add_protected_rule(struct ip_fw_chain *chain, struct ip_fw *rule, int locked); void ipfw_reap_add(struct ip_fw_chain *chain, struct ip_fw **head, @@ -659,7 +656,9 @@ void ipfw_reap_rules(struct ip_fw *head); void ipfw_init_counters(void); void ipfw_destroy_counters(void); struct ip_fw *ipfw_alloc_rule(struct ip_fw_chain *chain, size_t rulesize); +void ipfw_free_rule(struct ip_fw *rule); int ipfw_match_range(struct ip_fw *rule, ipfw_range_tlv *rt); +int ipfw_mark_object_kidx(uint32_t *bmask, uint16_t etlv, uint16_t kidx); typedef int (sopt_handler_f)(struct ip_fw_chain *ch, ip_fw3_opheader *op3, struct sockopt_data *sd); @@ -758,6 +757,10 @@ uint16_t ipfw_add_eaction(struct ip_fw_chain *ch, ipfw_eaction_t handler, int ipfw_del_eaction(struct ip_fw_chain *ch, uint16_t eaction_id); int ipfw_run_eaction(struct ip_fw_chain *ch, struct ip_fw_args *args, ipfw_insn *cmd, int *done); +int ipfw_reset_eaction(struct ip_fw_chain *ch, struct ip_fw *rule, + uint16_t eaction_id, uint16_t default_id, uint16_t instance_id); +int ipfw_reset_eaction_instance(struct ip_fw_chain *ch, uint16_t eaction_id, + uint16_t instance_id); /* In ip_fw_table.c */ struct table_info; diff --git a/freebsd/sys/netpfil/pf/if_pfsync.c b/freebsd/sys/netpfil/pf/if_pfsync.c index 9612ac99..026d19a3 100644 --- a/freebsd/sys/netpfil/pf/if_pfsync.c +++ b/freebsd/sys/netpfil/pf/if_pfsync.c @@ -79,6 +79,7 @@ __FBSDID("$FreeBSD$"); #include <sys/mutex.h> #include <sys/priv.h> #include <sys/protosw.h> +#include <sys/smp.h> #include <sys/socket.h> #include <sys/sockio.h> #include <sys/sysctl.h> @@ -108,6 +109,8 @@ __FBSDID("$FreeBSD$"); sizeof(struct pfsync_header) + \ sizeof(struct pfsync_subheader) ) +struct pfsync_bucket; + struct pfsync_pkt { struct ip *ip; struct in_addr src; @@ -166,7 +169,7 @@ static struct pfsync_q pfsync_qs[] = { }; static void pfsync_q_ins(struct pf_state *, int, bool); -static void pfsync_q_del(struct pf_state *, bool); +static void pfsync_q_del(struct pf_state *, bool, struct pfsync_bucket *); static void pfsync_update_state(struct pf_state *); @@ -185,6 +188,28 @@ struct pfsync_deferral { struct mbuf *pd_m; }; +struct pfsync_sofct; + +struct pfsync_bucket +{ + int b_id; + struct pfsync_softc *b_sc; + struct mtx b_mtx; + struct callout b_tmo; + int b_flags; +#define PFSYNCF_BUCKET_PUSH 0x00000001 + + size_t b_len; + TAILQ_HEAD(, pf_state) b_qs[PFSYNC_S_COUNT]; + TAILQ_HEAD(, pfsync_upd_req_item) b_upd_req_list; + TAILQ_HEAD(, pfsync_deferral) b_deferrals; + u_int b_deferred; + void *b_plus; + size_t b_pluslen; + + struct ifaltq b_snd; +}; + struct pfsync_softc { /* Configuration */ struct ifnet *sc_ifp; @@ -194,20 +219,12 @@ struct pfsync_softc { uint32_t sc_flags; #define PFSYNCF_OK 0x00000001 #define PFSYNCF_DEFER 0x00000002 -#define PFSYNCF_PUSH 0x00000004 uint8_t sc_maxupdates; struct ip sc_template; - struct callout sc_tmo; struct mtx sc_mtx; /* Queued data */ - size_t sc_len; - TAILQ_HEAD(, pf_state) sc_qs[PFSYNC_S_COUNT]; - TAILQ_HEAD(, pfsync_upd_req_item) sc_upd_req_list; - TAILQ_HEAD(, pfsync_deferral) sc_deferrals; - u_int sc_deferred; - void *sc_plus; - size_t sc_pluslen; + struct pfsync_bucket *sc_buckets; /* Bulk update info */ struct mtx sc_bulk_mtx; @@ -225,6 +242,10 @@ struct pfsync_softc { #define PFSYNC_UNLOCK(sc) mtx_unlock(&(sc)->sc_mtx) #define PFSYNC_LOCK_ASSERT(sc) mtx_assert(&(sc)->sc_mtx, MA_OWNED) +#define PFSYNC_BUCKET_LOCK(b) mtx_lock(&(b)->b_mtx) +#define PFSYNC_BUCKET_UNLOCK(b) mtx_unlock(&(b)->b_mtx) +#define PFSYNC_BUCKET_LOCK_ASSERT(b) mtx_assert(&(b)->b_mtx, MA_OWNED) + #define PFSYNC_BLOCK(sc) mtx_lock(&(sc)->sc_bulk_mtx) #define PFSYNC_BUNLOCK(sc) mtx_unlock(&(sc)->sc_bulk_mtx) #define PFSYNC_BLOCK_ASSERT(sc) mtx_assert(&(sc)->sc_bulk_mtx, MA_OWNED) @@ -241,7 +262,8 @@ VNET_DEFINE_STATIC(int, pfsync_carp_adj) = CARP_MAXSKEW; #define V_pfsync_carp_adj VNET(pfsync_carp_adj) static void pfsync_timeout(void *); -static void pfsync_push(struct pfsync_softc *); +static void pfsync_push(struct pfsync_bucket *); +static void pfsync_push_all(struct pfsync_softc *); static void pfsyncintr(void *); static int pfsync_multicast_setup(struct pfsync_softc *, struct ifnet *, void *); @@ -251,12 +273,16 @@ static void pfsync_pointers_uninit(void); static int pfsync_init(void); static void pfsync_uninit(void); +static unsigned long pfsync_buckets; + SYSCTL_NODE(_net, OID_AUTO, pfsync, CTLFLAG_RW, 0, "PFSYNC"); SYSCTL_STRUCT(_net_pfsync, OID_AUTO, stats, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(pfsyncstats), pfsyncstats, "PFSYNC statistics (struct pfsyncstats, net/if_pfsync.h)"); SYSCTL_INT(_net_pfsync, OID_AUTO, carp_demotion_factor, CTLFLAG_RW, &VNET_NAME(pfsync_carp_adj), 0, "pfsync's CARP demotion factor adjustment"); +SYSCTL_ULONG(_net_pfsync, OID_AUTO, pfsync_buckets, CTLFLAG_RDTUN, + &pfsync_buckets, 0, "Number of pfsync hash buckets"); static int pfsync_clone_create(struct if_clone *, int, caddr_t); static void pfsync_clone_destroy(struct ifnet *); @@ -272,10 +298,10 @@ static void pfsync_undefer_state(struct pf_state *, int); static void pfsync_defer_tmo(void *); static void pfsync_request_update(u_int32_t, u_int64_t); -static void pfsync_update_state_req(struct pf_state *); +static bool pfsync_update_state_req(struct pf_state *); static void pfsync_drop(struct pfsync_softc *); -static void pfsync_sendout(int); +static void pfsync_sendout(int, int); static void pfsync_send_plus(void *, size_t); static void pfsync_bulk_start(void); @@ -287,6 +313,9 @@ static void pfsync_detach_ifnet(struct ifnet *); #ifdef IPSEC static void pfsync_update_net_tdb(struct pfsync_tdb *); #endif +static struct pfsync_bucket *pfsync_get_bucket(struct pfsync_softc *, + struct pf_state *); + #define PFSYNC_MAX_BULKTRIES 12 @@ -298,21 +327,16 @@ pfsync_clone_create(struct if_clone *ifc, int unit, caddr_t param) { struct pfsync_softc *sc; struct ifnet *ifp; - int q; + struct pfsync_bucket *b; + int c, q; if (unit != 0) return (EINVAL); - sc = malloc(sizeof(struct pfsync_softc), M_PFSYNC, M_WAITOK | M_ZERO); - sc->sc_flags |= PFSYNCF_OK; - - for (q = 0; q < PFSYNC_S_COUNT; q++) - TAILQ_INIT(&sc->sc_qs[q]); - - TAILQ_INIT(&sc->sc_upd_req_list); - TAILQ_INIT(&sc->sc_deferrals); + if (! pfsync_buckets) + pfsync_buckets = mp_ncpus * 2; - sc->sc_len = PFSYNC_MINPKT; + sc = malloc(sizeof(struct pfsync_softc), M_PFSYNC, M_WAITOK | M_ZERO); sc->sc_maxupdates = 128; ifp = sc->sc_ifp = if_alloc(IFT_PFSYNC); @@ -325,12 +349,10 @@ pfsync_clone_create(struct if_clone *ifc, int unit, caddr_t param) ifp->if_ioctl = pfsyncioctl; ifp->if_output = pfsyncoutput; ifp->if_type = IFT_PFSYNC; - ifp->if_snd.ifq_maxlen = ifqmaxlen; ifp->if_hdrlen = sizeof(struct pfsync_header); ifp->if_mtu = ETHERMTU; mtx_init(&sc->sc_mtx, pfsyncname, NULL, MTX_DEF); mtx_init(&sc->sc_bulk_mtx, "pfsync bulk", NULL, MTX_DEF); - callout_init(&sc->sc_tmo, 1); callout_init_mtx(&sc->sc_bulk_tmo, &sc->sc_bulk_mtx, 0); callout_init_mtx(&sc->sc_bulkfail_tmo, &sc->sc_bulk_mtx, 0); @@ -338,6 +360,27 @@ pfsync_clone_create(struct if_clone *ifc, int unit, caddr_t param) bpfattach(ifp, DLT_PFSYNC, PFSYNC_HDRLEN); + sc->sc_buckets = mallocarray(pfsync_buckets, sizeof(*sc->sc_buckets), + M_PFSYNC, M_ZERO | M_WAITOK); + for (c = 0; c < pfsync_buckets; c++) { + b = &sc->sc_buckets[c]; + mtx_init(&b->b_mtx, pfsyncname, NULL, MTX_DEF); + + b->b_id = c; + b->b_sc = sc; + b->b_len = PFSYNC_MINPKT; + + for (q = 0; q < PFSYNC_S_COUNT; q++) + TAILQ_INIT(&b->b_qs[q]); + + TAILQ_INIT(&b->b_upd_req_list); + TAILQ_INIT(&b->b_deferrals); + + callout_init(&b->b_tmo, 1); + + b->b_snd.ifq_maxlen = ifqmaxlen; + } + V_pfsyncif = sc; return (0); @@ -347,29 +390,36 @@ static void pfsync_clone_destroy(struct ifnet *ifp) { struct pfsync_softc *sc = ifp->if_softc; + struct pfsync_bucket *b; + int c; - /* - * At this stage, everything should have already been - * cleared by pfsync_uninit(), and we have only to - * drain callouts. - */ - while (sc->sc_deferred > 0) { - struct pfsync_deferral *pd = TAILQ_FIRST(&sc->sc_deferrals); - - TAILQ_REMOVE(&sc->sc_deferrals, pd, pd_entry); - sc->sc_deferred--; - if (callout_stop(&pd->pd_tmo) > 0) { - pf_release_state(pd->pd_st); - m_freem(pd->pd_m); - free(pd, M_PFSYNC); - } else { - pd->pd_refs++; - callout_drain(&pd->pd_tmo); - free(pd, M_PFSYNC); + for (c = 0; c < pfsync_buckets; c++) { + b = &sc->sc_buckets[c]; + /* + * At this stage, everything should have already been + * cleared by pfsync_uninit(), and we have only to + * drain callouts. + */ + while (b->b_deferred > 0) { + struct pfsync_deferral *pd = + TAILQ_FIRST(&b->b_deferrals); + + TAILQ_REMOVE(&b->b_deferrals, pd, pd_entry); + b->b_deferred--; + if (callout_stop(&pd->pd_tmo) > 0) { + pf_release_state(pd->pd_st); + m_freem(pd->pd_m); + free(pd, M_PFSYNC); + } else { + pd->pd_refs++; + callout_drain(&pd->pd_tmo); + free(pd, M_PFSYNC); + } } + + callout_drain(&b->b_tmo); } - callout_drain(&sc->sc_tmo); callout_drain(&sc->sc_bulkfail_tmo); callout_drain(&sc->sc_bulk_tmo); @@ -385,6 +435,8 @@ pfsync_clone_destroy(struct ifnet *ifp) pfsync_multicast_cleanup(sc); mtx_destroy(&sc->sc_mtx); mtx_destroy(&sc->sc_bulk_mtx); + + free(sc->sc_buckets, M_PFSYNC); free(sc, M_PFSYNC); V_pfsyncif = NULL; @@ -548,7 +600,7 @@ pfsync_state_import(struct pfsync_state *sp, u_int8_t flags) st->state_flags &= ~PFSTATE_NOSYNC; if (st->state_flags & PFSTATE_ACK) { pfsync_q_ins(st, PFSYNC_S_IACK, true); - pfsync_push(sc); + pfsync_push_all(sc); } } st->state_flags &= ~PFSTATE_ACK; @@ -787,9 +839,7 @@ pfsync_in_iack(struct pfsync_pkt *pkt, struct mbuf *m, int offset, int count) continue; if (st->state_flags & PFSTATE_ACK) { - PFSYNC_LOCK(V_pfsyncif); pfsync_undefer_state(st, 0); - PFSYNC_UNLOCK(V_pfsyncif); } PF_STATE_UNLOCK(st); } @@ -878,9 +928,7 @@ pfsync_in_upd(struct pfsync_pkt *pkt, struct mbuf *m, int offset, int count) } if (st->state_flags & PFSTATE_ACK) { - PFSYNC_LOCK(sc); pfsync_undefer_state(st, 1); - PFSYNC_UNLOCK(sc); } if (st->key[PF_SK_WIRE]->proto == IPPROTO_TCP) @@ -914,9 +962,7 @@ pfsync_in_upd(struct pfsync_pkt *pkt, struct mbuf *m, int offset, int count) pfsync_update_state(st); PF_STATE_UNLOCK(st); - PFSYNC_LOCK(sc); - pfsync_push(sc); - PFSYNC_UNLOCK(sc); + pfsync_push_all(sc); continue; } PF_STATE_UNLOCK(st); @@ -962,16 +1008,14 @@ pfsync_in_upd_c(struct pfsync_pkt *pkt, struct mbuf *m, int offset, int count) st = pf_find_state_byid(up->id, up->creatorid); if (st == NULL) { /* We don't have this state. Ask for it. */ - PFSYNC_LOCK(sc); + PFSYNC_BUCKET_LOCK(&sc->sc_buckets[0]); pfsync_request_update(up->creatorid, up->id); - PFSYNC_UNLOCK(sc); + PFSYNC_BUCKET_UNLOCK(&sc->sc_buckets[0]); continue; } if (st->state_flags & PFSTATE_ACK) { - PFSYNC_LOCK(sc); pfsync_undefer_state(st, 1); - PFSYNC_UNLOCK(sc); } if (st->key[PF_SK_WIRE]->proto == IPPROTO_TCP) @@ -1005,9 +1049,7 @@ pfsync_in_upd_c(struct pfsync_pkt *pkt, struct mbuf *m, int offset, int count) pfsync_update_state(st); PF_STATE_UNLOCK(st); - PFSYNC_LOCK(sc); - pfsync_push(sc); - PFSYNC_UNLOCK(sc); + pfsync_push_all(sc); continue; } PF_STATE_UNLOCK(st); @@ -1285,6 +1327,7 @@ pfsyncioctl(struct ifnet *ifp, u_long cmd, caddr_t data) struct ifreq *ifr = (struct ifreq *)data; struct pfsyncreq pfsyncr; int error; + int c; switch (cmd) { case SIOCSIFFLAGS: @@ -1305,10 +1348,12 @@ pfsyncioctl(struct ifnet *ifp, u_long cmd, caddr_t data) ifr->ifr_mtu > sc->sc_sync_if->if_mtu) return (EINVAL); if (ifr->ifr_mtu < ifp->if_mtu) { - PFSYNC_LOCK(sc); - if (sc->sc_len > PFSYNC_MINPKT) - pfsync_sendout(1); - PFSYNC_UNLOCK(sc); + for (c = 0; c < pfsync_buckets; c++) { + PFSYNC_BUCKET_LOCK(&sc->sc_buckets[c]); + if (sc->sc_buckets[c].b_len > PFSYNC_MINPKT) + pfsync_sendout(1, c); + PFSYNC_BUCKET_UNLOCK(&sc->sc_buckets[c]); + } } ifp->if_mtu = ifr->ifr_mtu; break; @@ -1381,12 +1426,16 @@ pfsyncioctl(struct ifnet *ifp, u_long cmd, caddr_t data) break; } - if (sc->sc_len > PFSYNC_MINPKT && - (sifp->if_mtu < sc->sc_ifp->if_mtu || - (sc->sc_sync_if != NULL && - sifp->if_mtu < sc->sc_sync_if->if_mtu) || - sifp->if_mtu < MCLBYTES - sizeof(struct ip))) - pfsync_sendout(1); + for (c = 0; c < pfsync_buckets; c++) { + PFSYNC_BUCKET_LOCK(&sc->sc_buckets[c]); + if (sc->sc_buckets[c].b_len > PFSYNC_MINPKT && + (sifp->if_mtu < sc->sc_ifp->if_mtu || + (sc->sc_sync_if != NULL && + sifp->if_mtu < sc->sc_sync_if->if_mtu) || + sifp->if_mtu < MCLBYTES - sizeof(struct ip))) + pfsync_sendout(1, c); + PFSYNC_BUCKET_UNLOCK(&sc->sc_buckets[c]); + } if (imo->imo_membership) pfsync_multicast_cleanup(sc); @@ -1423,8 +1472,10 @@ pfsyncioctl(struct ifnet *ifp, u_long cmd, caddr_t data) sc->sc_flags &= ~PFSYNCF_OK; if (V_pf_status.debug >= PF_DEBUG_MISC) printf("pfsync: requesting bulk update\n"); - pfsync_request_update(0, 0); PFSYNC_UNLOCK(sc); + PFSYNC_BUCKET_LOCK(&sc->sc_buckets[0]); + pfsync_request_update(0, 0); + PFSYNC_BUCKET_UNLOCK(&sc->sc_buckets[0]); PFSYNC_BLOCK(sc); sc->sc_ureq_sent = time_uptime; callout_reset(&sc->sc_bulkfail_tmo, 5 * hz, pfsync_bulk_fail, @@ -1485,33 +1536,37 @@ pfsync_drop(struct pfsync_softc *sc) { struct pf_state *st, *next; struct pfsync_upd_req_item *ur; - int q; + struct pfsync_bucket *b; + int c, q; - for (q = 0; q < PFSYNC_S_COUNT; q++) { - if (TAILQ_EMPTY(&sc->sc_qs[q])) - continue; + for (c = 0; c < pfsync_buckets; c++) { + b = &sc->sc_buckets[c]; + for (q = 0; q < PFSYNC_S_COUNT; q++) { + if (TAILQ_EMPTY(&b->b_qs[q])) + continue; - TAILQ_FOREACH_SAFE(st, &sc->sc_qs[q], sync_list, next) { - KASSERT(st->sync_state == q, - ("%s: st->sync_state == q", - __func__)); - st->sync_state = PFSYNC_S_NONE; - pf_release_state(st); + TAILQ_FOREACH_SAFE(st, &b->b_qs[q], sync_list, next) { + KASSERT(st->sync_state == q, + ("%s: st->sync_state == q", + __func__)); + st->sync_state = PFSYNC_S_NONE; + pf_release_state(st); + } + TAILQ_INIT(&b->b_qs[q]); } - TAILQ_INIT(&sc->sc_qs[q]); - } - while ((ur = TAILQ_FIRST(&sc->sc_upd_req_list)) != NULL) { - TAILQ_REMOVE(&sc->sc_upd_req_list, ur, ur_entry); - free(ur, M_PFSYNC); - } + while ((ur = TAILQ_FIRST(&b->b_upd_req_list)) != NULL) { + TAILQ_REMOVE(&b->b_upd_req_list, ur, ur_entry); + free(ur, M_PFSYNC); + } - sc->sc_plus = NULL; - sc->sc_len = PFSYNC_MINPKT; + b->b_len = PFSYNC_MINPKT; + b->b_plus = NULL; + } } static void -pfsync_sendout(int schedswi) +pfsync_sendout(int schedswi, int c) { struct pfsync_softc *sc = V_pfsyncif; struct ifnet *ifp = sc->sc_ifp; @@ -1521,27 +1576,28 @@ pfsync_sendout(int schedswi) struct pfsync_subheader *subh; struct pf_state *st, *st_next; struct pfsync_upd_req_item *ur; + struct pfsync_bucket *b = &sc->sc_buckets[c]; int offset; int q, count = 0; KASSERT(sc != NULL, ("%s: null sc", __func__)); - KASSERT(sc->sc_len > PFSYNC_MINPKT, - ("%s: sc_len %zu", __func__, sc->sc_len)); - PFSYNC_LOCK_ASSERT(sc); + KASSERT(b->b_len > PFSYNC_MINPKT, + ("%s: sc_len %zu", __func__, b->b_len)); + PFSYNC_BUCKET_LOCK_ASSERT(b); if (ifp->if_bpf == NULL && sc->sc_sync_if == NULL) { pfsync_drop(sc); return; } - m = m_get2(max_linkhdr + sc->sc_len, M_NOWAIT, MT_DATA, M_PKTHDR); + m = m_get2(max_linkhdr + b->b_len, M_NOWAIT, MT_DATA, M_PKTHDR); if (m == NULL) { if_inc_counter(sc->sc_ifp, IFCOUNTER_OERRORS, 1); V_pfsyncstats.pfsyncs_onomem++; return; } m->m_data += max_linkhdr; - m->m_len = m->m_pkthdr.len = sc->sc_len; + m->m_len = m->m_pkthdr.len = b->b_len; /* build the ip header */ ip = (struct ip *)m->m_data; @@ -1557,19 +1613,19 @@ pfsync_sendout(int schedswi) offset += sizeof(*ph); ph->version = PFSYNC_VERSION; - ph->len = htons(sc->sc_len - sizeof(*ip)); + ph->len = htons(b->b_len - sizeof(*ip)); bcopy(V_pf_status.pf_chksum, ph->pfcksum, PF_MD5_DIGEST_LENGTH); /* walk the queues */ for (q = 0; q < PFSYNC_S_COUNT; q++) { - if (TAILQ_EMPTY(&sc->sc_qs[q])) + if (TAILQ_EMPTY(&b->b_qs[q])) continue; subh = (struct pfsync_subheader *)(m->m_data + offset); offset += sizeof(*subh); count = 0; - TAILQ_FOREACH_SAFE(st, &sc->sc_qs[q], sync_list, st_next) { + TAILQ_FOREACH_SAFE(st, &b->b_qs[q], sync_list, st_next) { KASSERT(st->sync_state == q, ("%s: st->sync_state == q", __func__)); @@ -1583,7 +1639,7 @@ pfsync_sendout(int schedswi) pf_release_state(st); count++; } - TAILQ_INIT(&sc->sc_qs[q]); + TAILQ_INIT(&b->b_qs[q]); bzero(subh, sizeof(*subh)); subh->action = pfsync_qs[q].action; @@ -1591,13 +1647,13 @@ pfsync_sendout(int schedswi) V_pfsyncstats.pfsyncs_oacts[pfsync_qs[q].action] += count; } - if (!TAILQ_EMPTY(&sc->sc_upd_req_list)) { + if (!TAILQ_EMPTY(&b->b_upd_req_list)) { subh = (struct pfsync_subheader *)(m->m_data + offset); offset += sizeof(*subh); count = 0; - while ((ur = TAILQ_FIRST(&sc->sc_upd_req_list)) != NULL) { - TAILQ_REMOVE(&sc->sc_upd_req_list, ur, ur_entry); + while ((ur = TAILQ_FIRST(&b->b_upd_req_list)) != NULL) { + TAILQ_REMOVE(&b->b_upd_req_list, ur, ur_entry); bcopy(&ur->ur_msg, m->m_data + offset, sizeof(ur->ur_msg)); @@ -1613,11 +1669,11 @@ pfsync_sendout(int schedswi) } /* has someone built a custom region for us to add? */ - if (sc->sc_plus != NULL) { - bcopy(sc->sc_plus, m->m_data + offset, sc->sc_pluslen); - offset += sc->sc_pluslen; + if (b->b_plus != NULL) { + bcopy(b->b_plus, m->m_data + offset, b->b_pluslen); + offset += b->b_pluslen; - sc->sc_plus = NULL; + b->b_plus = NULL; } subh = (struct pfsync_subheader *)(m->m_data + offset); @@ -1631,24 +1687,24 @@ pfsync_sendout(int schedswi) /* we're done, let's put it on the wire */ if (ifp->if_bpf) { m->m_data += sizeof(*ip); - m->m_len = m->m_pkthdr.len = sc->sc_len - sizeof(*ip); + m->m_len = m->m_pkthdr.len = b->b_len - sizeof(*ip); BPF_MTAP(ifp, m); m->m_data -= sizeof(*ip); - m->m_len = m->m_pkthdr.len = sc->sc_len; + m->m_len = m->m_pkthdr.len = b->b_len; } if (sc->sc_sync_if == NULL) { - sc->sc_len = PFSYNC_MINPKT; + b->b_len = PFSYNC_MINPKT; m_freem(m); return; } if_inc_counter(sc->sc_ifp, IFCOUNTER_OPACKETS, 1); if_inc_counter(sc->sc_ifp, IFCOUNTER_OBYTES, m->m_pkthdr.len); - sc->sc_len = PFSYNC_MINPKT; + b->b_len = PFSYNC_MINPKT; - if (!_IF_QFULL(&sc->sc_ifp->if_snd)) - _IF_ENQUEUE(&sc->sc_ifp->if_snd, m); + if (!_IF_QFULL(&b->b_snd)) + _IF_ENQUEUE(&b->b_snd, m); else { m_freem(m); if_inc_counter(sc->sc_ifp, IFCOUNTER_OQDROPS, 1); @@ -1661,6 +1717,7 @@ static void pfsync_insert_state(struct pf_state *st) { struct pfsync_softc *sc = V_pfsyncif; + struct pfsync_bucket *b = pfsync_get_bucket(sc, st); if (st->state_flags & PFSTATE_NOSYNC) return; @@ -1674,12 +1731,12 @@ pfsync_insert_state(struct pf_state *st) KASSERT(st->sync_state == PFSYNC_S_NONE, ("%s: st->sync_state %u", __func__, st->sync_state)); - PFSYNC_LOCK(sc); - if (sc->sc_len == PFSYNC_MINPKT) - callout_reset(&sc->sc_tmo, 1 * hz, pfsync_timeout, V_pfsyncif); + PFSYNC_BUCKET_LOCK(b); + if (b->b_len == PFSYNC_MINPKT) + callout_reset(&b->b_tmo, 1 * hz, pfsync_timeout, b); pfsync_q_ins(st, PFSYNC_S_INS, true); - PFSYNC_UNLOCK(sc); + PFSYNC_BUCKET_UNLOCK(b); st->sync_updates = 0; } @@ -1689,6 +1746,7 @@ pfsync_defer(struct pf_state *st, struct mbuf *m) { struct pfsync_softc *sc = V_pfsyncif; struct pfsync_deferral *pd; + struct pfsync_bucket *b = pfsync_get_bucket(sc, st); if (m->m_flags & (M_BCAST|M_MCAST)) return (0); @@ -1701,13 +1759,13 @@ pfsync_defer(struct pf_state *st, struct mbuf *m) return (0); } - if (sc->sc_deferred >= 128) - pfsync_undefer(TAILQ_FIRST(&sc->sc_deferrals), 0); + if (b->b_deferred >= 128) + pfsync_undefer(TAILQ_FIRST(&b->b_deferrals), 0); pd = malloc(sizeof(*pd), M_PFSYNC, M_NOWAIT); if (pd == NULL) return (0); - sc->sc_deferred++; + b->b_deferred++; m->m_flags |= M_SKIP_FIREWALL; st->state_flags |= PFSTATE_ACK; @@ -1718,11 +1776,11 @@ pfsync_defer(struct pf_state *st, struct mbuf *m) pf_ref_state(st); pd->pd_m = m; - TAILQ_INSERT_TAIL(&sc->sc_deferrals, pd, pd_entry); - callout_init_mtx(&pd->pd_tmo, &sc->sc_mtx, CALLOUT_RETURNUNLOCKED); + TAILQ_INSERT_TAIL(&b->b_deferrals, pd, pd_entry); + callout_init_mtx(&pd->pd_tmo, &b->b_mtx, CALLOUT_RETURNUNLOCKED); callout_reset(&pd->pd_tmo, 10, pfsync_defer_tmo, pd); - pfsync_push(sc); + pfsync_push(b); return (1); } @@ -1733,11 +1791,12 @@ pfsync_undefer(struct pfsync_deferral *pd, int drop) struct pfsync_softc *sc = pd->pd_sc; struct mbuf *m = pd->pd_m; struct pf_state *st = pd->pd_st; + struct pfsync_bucket *b = pfsync_get_bucket(sc, st); - PFSYNC_LOCK_ASSERT(sc); + PFSYNC_BUCKET_LOCK_ASSERT(b); - TAILQ_REMOVE(&sc->sc_deferrals, pd, pd_entry); - sc->sc_deferred--; + TAILQ_REMOVE(&b->b_deferrals, pd, pd_entry); + b->b_deferred--; pd->pd_st->state_flags &= ~PFSTATE_ACK; /* XXX: locking! */ free(pd, M_PFSYNC); pf_release_state(st); @@ -1745,8 +1804,8 @@ pfsync_undefer(struct pfsync_deferral *pd, int drop) if (drop) m_freem(m); else { - _IF_ENQUEUE(&sc->sc_ifp->if_snd, m); - pfsync_push(sc); + _IF_ENQUEUE(&b->b_snd, m); + pfsync_push(b); } } @@ -1757,13 +1816,14 @@ pfsync_defer_tmo(void *arg) struct pfsync_softc *sc = pd->pd_sc; struct mbuf *m = pd->pd_m; struct pf_state *st = pd->pd_st; + struct pfsync_bucket *b = pfsync_get_bucket(sc, st); - PFSYNC_LOCK_ASSERT(sc); + PFSYNC_BUCKET_LOCK_ASSERT(b); CURVNET_SET(m->m_pkthdr.rcvif->if_vnet); - TAILQ_REMOVE(&sc->sc_deferrals, pd, pd_entry); - sc->sc_deferred--; + TAILQ_REMOVE(&b->b_deferrals, pd, pd_entry); + b->b_deferred--; pd->pd_st->state_flags &= ~PFSTATE_ACK; /* XXX: locking! */ if (pd->pd_refs == 0) free(pd, M_PFSYNC); @@ -1781,40 +1841,52 @@ pfsync_undefer_state(struct pf_state *st, int drop) { struct pfsync_softc *sc = V_pfsyncif; struct pfsync_deferral *pd; + struct pfsync_bucket *b = pfsync_get_bucket(sc, st); - PFSYNC_LOCK_ASSERT(sc); + PFSYNC_BUCKET_LOCK(b); - TAILQ_FOREACH(pd, &sc->sc_deferrals, pd_entry) { + TAILQ_FOREACH(pd, &b->b_deferrals, pd_entry) { if (pd->pd_st == st) { if (callout_stop(&pd->pd_tmo) > 0) pfsync_undefer(pd, drop); + + PFSYNC_BUCKET_UNLOCK(b); return; } } + PFSYNC_BUCKET_UNLOCK(b); panic("%s: unable to find deferred state", __func__); } +static struct pfsync_bucket* +pfsync_get_bucket(struct pfsync_softc *sc, struct pf_state *st) +{ + int c = PF_IDHASH(st) % pfsync_buckets; + return &sc->sc_buckets[c]; +} + static void pfsync_update_state(struct pf_state *st) { struct pfsync_softc *sc = V_pfsyncif; bool sync = false, ref = true; + struct pfsync_bucket *b = pfsync_get_bucket(sc, st); PF_STATE_LOCK_ASSERT(st); - PFSYNC_LOCK(sc); + PFSYNC_BUCKET_LOCK(b); if (st->state_flags & PFSTATE_ACK) pfsync_undefer_state(st, 0); if (st->state_flags & PFSTATE_NOSYNC) { if (st->sync_state != PFSYNC_S_NONE) - pfsync_q_del(st, true); - PFSYNC_UNLOCK(sc); + pfsync_q_del(st, true, b); + PFSYNC_BUCKET_UNLOCK(b); return; } - if (sc->sc_len == PFSYNC_MINPKT) - callout_reset(&sc->sc_tmo, 1 * hz, pfsync_timeout, V_pfsyncif); + if (b->b_len == PFSYNC_MINPKT) + callout_reset(&b->b_tmo, 1 * hz, pfsync_timeout, b); switch (st->sync_state) { case PFSYNC_S_UPD_C: @@ -1830,7 +1902,7 @@ pfsync_update_state(struct pf_state *st) break; case PFSYNC_S_IACK: - pfsync_q_del(st, false); + pfsync_q_del(st, false, b); ref = false; /* FALLTHROUGH */ @@ -1844,26 +1916,27 @@ pfsync_update_state(struct pf_state *st) } if (sync || (time_uptime - st->pfsync_time) < 2) - pfsync_push(sc); + pfsync_push(b); - PFSYNC_UNLOCK(sc); + PFSYNC_BUCKET_UNLOCK(b); } static void pfsync_request_update(u_int32_t creatorid, u_int64_t id) { struct pfsync_softc *sc = V_pfsyncif; + struct pfsync_bucket *b = &sc->sc_buckets[0]; struct pfsync_upd_req_item *item; size_t nlen = sizeof(struct pfsync_upd_req); - PFSYNC_LOCK_ASSERT(sc); + PFSYNC_BUCKET_LOCK_ASSERT(b); /* * This code does a bit to prevent multiple update requests for the * same state being generated. It searches current subheader queue, * but it doesn't lookup into queue of already packed datagrams. */ - TAILQ_FOREACH(item, &sc->sc_upd_req_list, ur_entry) + TAILQ_FOREACH(item, &b->b_upd_req_list, ur_entry) if (item->ur_msg.id == id && item->ur_msg.creatorid == creatorid) return; @@ -1875,46 +1948,47 @@ pfsync_request_update(u_int32_t creatorid, u_int64_t id) item->ur_msg.id = id; item->ur_msg.creatorid = creatorid; - if (TAILQ_EMPTY(&sc->sc_upd_req_list)) + if (TAILQ_EMPTY(&b->b_upd_req_list)) nlen += sizeof(struct pfsync_subheader); - if (sc->sc_len + nlen > sc->sc_ifp->if_mtu) { - pfsync_sendout(1); + if (b->b_len + nlen > sc->sc_ifp->if_mtu) { + pfsync_sendout(1, 0); nlen = sizeof(struct pfsync_subheader) + sizeof(struct pfsync_upd_req); } - TAILQ_INSERT_TAIL(&sc->sc_upd_req_list, item, ur_entry); - sc->sc_len += nlen; + TAILQ_INSERT_TAIL(&b->b_upd_req_list, item, ur_entry); + b->b_len += nlen; } -static void +static bool pfsync_update_state_req(struct pf_state *st) { struct pfsync_softc *sc = V_pfsyncif; - bool ref = true; + bool ref = true, full = false; + struct pfsync_bucket *b = pfsync_get_bucket(sc, st); PF_STATE_LOCK_ASSERT(st); - PFSYNC_LOCK(sc); + PFSYNC_BUCKET_LOCK(b); if (st->state_flags & PFSTATE_NOSYNC) { if (st->sync_state != PFSYNC_S_NONE) - pfsync_q_del(st, true); - PFSYNC_UNLOCK(sc); - return; + pfsync_q_del(st, true, b); + PFSYNC_BUCKET_UNLOCK(b); + return (full); } switch (st->sync_state) { case PFSYNC_S_UPD_C: case PFSYNC_S_IACK: - pfsync_q_del(st, false); + pfsync_q_del(st, false, b); ref = false; /* FALLTHROUGH */ case PFSYNC_S_NONE: pfsync_q_ins(st, PFSYNC_S_UPD, ref); - pfsync_push(sc); + pfsync_push(b); break; case PFSYNC_S_INS: @@ -1927,38 +2001,44 @@ pfsync_update_state_req(struct pf_state *st) panic("%s: unexpected sync state %d", __func__, st->sync_state); } - PFSYNC_UNLOCK(sc); + if ((sc->sc_ifp->if_mtu - b->b_len) < sizeof(struct pfsync_state)) + full = true; + + PFSYNC_BUCKET_UNLOCK(b); + + return (full); } static void pfsync_delete_state(struct pf_state *st) { struct pfsync_softc *sc = V_pfsyncif; + struct pfsync_bucket *b = pfsync_get_bucket(sc, st); bool ref = true; - PFSYNC_LOCK(sc); + PFSYNC_BUCKET_LOCK(b); if (st->state_flags & PFSTATE_ACK) pfsync_undefer_state(st, 1); if (st->state_flags & PFSTATE_NOSYNC) { if (st->sync_state != PFSYNC_S_NONE) - pfsync_q_del(st, true); - PFSYNC_UNLOCK(sc); + pfsync_q_del(st, true, b); + PFSYNC_BUCKET_UNLOCK(b); return; } - if (sc->sc_len == PFSYNC_MINPKT) - callout_reset(&sc->sc_tmo, 1 * hz, pfsync_timeout, V_pfsyncif); + if (b->b_len == PFSYNC_MINPKT) + callout_reset(&b->b_tmo, 1 * hz, pfsync_timeout, b); switch (st->sync_state) { case PFSYNC_S_INS: /* We never got to tell the world so just forget about it. */ - pfsync_q_del(st, true); + pfsync_q_del(st, true, b); break; case PFSYNC_S_UPD_C: case PFSYNC_S_UPD: case PFSYNC_S_IACK: - pfsync_q_del(st, false); + pfsync_q_del(st, false, b); ref = false; /* FALLTHROUGH */ @@ -1970,13 +2050,12 @@ pfsync_delete_state(struct pf_state *st) panic("%s: unexpected sync state %d", __func__, st->sync_state); } - PFSYNC_UNLOCK(sc); + PFSYNC_BUCKET_UNLOCK(b); } static void pfsync_clear_states(u_int32_t creatorid, const char *ifname) { - struct pfsync_softc *sc = V_pfsyncif; struct { struct pfsync_subheader subh; struct pfsync_clr clr; @@ -1991,9 +2070,7 @@ pfsync_clear_states(u_int32_t creatorid, const char *ifname) strlcpy(r.clr.ifname, ifname, sizeof(r.clr.ifname)); r.clr.creatorid = creatorid; - PFSYNC_LOCK(sc); pfsync_send_plus(&r, sizeof(r)); - PFSYNC_UNLOCK(sc); } static void @@ -2001,48 +2078,48 @@ pfsync_q_ins(struct pf_state *st, int q, bool ref) { struct pfsync_softc *sc = V_pfsyncif; size_t nlen = pfsync_qs[q].len; + struct pfsync_bucket *b = pfsync_get_bucket(sc, st); - PFSYNC_LOCK_ASSERT(sc); + PFSYNC_BUCKET_LOCK_ASSERT(b); KASSERT(st->sync_state == PFSYNC_S_NONE, ("%s: st->sync_state %u", __func__, st->sync_state)); - KASSERT(sc->sc_len >= PFSYNC_MINPKT, ("pfsync pkt len is too low %zu", - sc->sc_len)); + KASSERT(b->b_len >= PFSYNC_MINPKT, ("pfsync pkt len is too low %zu", + b->b_len)); - if (TAILQ_EMPTY(&sc->sc_qs[q])) + if (TAILQ_EMPTY(&b->b_qs[q])) nlen += sizeof(struct pfsync_subheader); - if (sc->sc_len + nlen > sc->sc_ifp->if_mtu) { - pfsync_sendout(1); + if (b->b_len + nlen > sc->sc_ifp->if_mtu) { + pfsync_sendout(1, b->b_id); nlen = sizeof(struct pfsync_subheader) + pfsync_qs[q].len; } - sc->sc_len += nlen; - TAILQ_INSERT_TAIL(&sc->sc_qs[q], st, sync_list); + b->b_len += nlen; + TAILQ_INSERT_TAIL(&b->b_qs[q], st, sync_list); st->sync_state = q; if (ref) pf_ref_state(st); } static void -pfsync_q_del(struct pf_state *st, bool unref) +pfsync_q_del(struct pf_state *st, bool unref, struct pfsync_bucket *b) { - struct pfsync_softc *sc = V_pfsyncif; int q = st->sync_state; - PFSYNC_LOCK_ASSERT(sc); + PFSYNC_BUCKET_LOCK_ASSERT(b); KASSERT(st->sync_state != PFSYNC_S_NONE, ("%s: st->sync_state != PFSYNC_S_NONE", __func__)); - sc->sc_len -= pfsync_qs[q].len; - TAILQ_REMOVE(&sc->sc_qs[q], st, sync_list); + b->b_len -= pfsync_qs[q].len; + TAILQ_REMOVE(&b->b_qs[q], st, sync_list); st->sync_state = PFSYNC_S_NONE; if (unref) pf_release_state(st); - if (TAILQ_EMPTY(&sc->sc_qs[q])) - sc->sc_len -= sizeof(struct pfsync_subheader); + if (TAILQ_EMPTY(&b->b_qs[q])) + b->b_len -= sizeof(struct pfsync_subheader); } static void @@ -2096,23 +2173,19 @@ pfsync_bulk_update(void *arg) } for (; s; s = LIST_NEXT(s, entry)) { - - if (sent > 1 && (sc->sc_ifp->if_mtu - sc->sc_len) < - sizeof(struct pfsync_state)) { - /* We've filled a packet. */ - sc->sc_bulk_hashid = i; - sc->sc_bulk_stateid = s->id; - sc->sc_bulk_creatorid = s->creatorid; - PF_HASHROW_UNLOCK(ih); - callout_reset(&sc->sc_bulk_tmo, 1, - pfsync_bulk_update, sc); - goto full; - } - if (s->sync_state == PFSYNC_S_NONE && s->timeout < PFTM_MAX && s->pfsync_time <= sc->sc_ureq_received) { - pfsync_update_state_req(s); + if (pfsync_update_state_req(s)) { + /* We've filled a packet. */ + sc->sc_bulk_hashid = i; + sc->sc_bulk_stateid = s->id; + sc->sc_bulk_creatorid = s->creatorid; + PF_HASHROW_UNLOCK(ih); + callout_reset(&sc->sc_bulk_tmo, 1, + pfsync_bulk_update, sc); + goto full; + } sent++; } } @@ -2121,7 +2194,6 @@ pfsync_bulk_update(void *arg) /* We're done. */ pfsync_bulk_status(PFSYNC_BUS_END); - full: CURVNET_RESTORE(); } @@ -2146,15 +2218,14 @@ pfsync_bulk_status(u_int8_t status) r.bus.endtime = htonl(time_uptime - sc->sc_ureq_received); r.bus.status = status; - PFSYNC_LOCK(sc); pfsync_send_plus(&r, sizeof(r)); - PFSYNC_UNLOCK(sc); } static void pfsync_bulk_fail(void *arg) { struct pfsync_softc *sc = arg; + struct pfsync_bucket *b = &sc->sc_buckets[0]; CURVNET_SET(sc->sc_ifp->if_vnet); @@ -2164,9 +2235,9 @@ pfsync_bulk_fail(void *arg) /* Try again */ callout_reset(&sc->sc_bulkfail_tmo, 5 * hz, pfsync_bulk_fail, V_pfsyncif); - PFSYNC_LOCK(sc); + PFSYNC_BUCKET_LOCK(b); pfsync_request_update(0, 0); - PFSYNC_UNLOCK(sc); + PFSYNC_BUCKET_UNLOCK(b); } else { /* Pretend like the transfer was ok. */ sc->sc_ureq_sent = 0; @@ -2188,73 +2259,96 @@ static void pfsync_send_plus(void *plus, size_t pluslen) { struct pfsync_softc *sc = V_pfsyncif; + struct pfsync_bucket *b = &sc->sc_buckets[0]; - PFSYNC_LOCK_ASSERT(sc); + PFSYNC_BUCKET_LOCK(b); - if (sc->sc_len + pluslen > sc->sc_ifp->if_mtu) - pfsync_sendout(1); + if (b->b_len + pluslen > sc->sc_ifp->if_mtu) + pfsync_sendout(1, b->b_id); - sc->sc_plus = plus; - sc->sc_len += (sc->sc_pluslen = pluslen); + b->b_plus = plus; + b->b_len += (b->b_pluslen = pluslen); - pfsync_sendout(1); + pfsync_sendout(1, b->b_id); + PFSYNC_BUCKET_UNLOCK(b); } static void pfsync_timeout(void *arg) { - struct pfsync_softc *sc = arg; + struct pfsync_bucket *b = arg; - CURVNET_SET(sc->sc_ifp->if_vnet); - PFSYNC_LOCK(sc); - pfsync_push(sc); - PFSYNC_UNLOCK(sc); + CURVNET_SET(b->b_sc->sc_ifp->if_vnet); + PFSYNC_BUCKET_LOCK(b); + pfsync_push(b); + PFSYNC_BUCKET_UNLOCK(b); CURVNET_RESTORE(); } static void -pfsync_push(struct pfsync_softc *sc) +pfsync_push(struct pfsync_bucket *b) { - PFSYNC_LOCK_ASSERT(sc); + PFSYNC_BUCKET_LOCK_ASSERT(b); - sc->sc_flags |= PFSYNCF_PUSH; + b->b_flags |= PFSYNCF_BUCKET_PUSH; swi_sched(V_pfsync_swi_cookie, 0); } static void +pfsync_push_all(struct pfsync_softc *sc) +{ + int c; + struct pfsync_bucket *b; + + for (c = 0; c < pfsync_buckets; c++) { + b = &sc->sc_buckets[c]; + + PFSYNC_BUCKET_LOCK(b); + pfsync_push(b); + PFSYNC_BUCKET_UNLOCK(b); + } +} + +static void pfsyncintr(void *arg) { struct pfsync_softc *sc = arg; + struct pfsync_bucket *b; struct mbuf *m, *n; + int c; CURVNET_SET(sc->sc_ifp->if_vnet); - PFSYNC_LOCK(sc); - if ((sc->sc_flags & PFSYNCF_PUSH) && sc->sc_len > PFSYNC_MINPKT) { - pfsync_sendout(0); - sc->sc_flags &= ~PFSYNCF_PUSH; - } - _IF_DEQUEUE_ALL(&sc->sc_ifp->if_snd, m); - PFSYNC_UNLOCK(sc); + for (c = 0; c < pfsync_buckets; c++) { + b = &sc->sc_buckets[c]; - for (; m != NULL; m = n) { + PFSYNC_BUCKET_LOCK(b); + if ((b->b_flags & PFSYNCF_BUCKET_PUSH) && b->b_len > PFSYNC_MINPKT) { + pfsync_sendout(0, b->b_id); + b->b_flags &= ~PFSYNCF_BUCKET_PUSH; + } + _IF_DEQUEUE_ALL(&b->b_snd, m); + PFSYNC_BUCKET_UNLOCK(b); - n = m->m_nextpkt; - m->m_nextpkt = NULL; + for (; m != NULL; m = n) { - /* - * We distinguish between a deferral packet and our - * own pfsync packet based on M_SKIP_FIREWALL - * flag. This is XXX. - */ - if (m->m_flags & M_SKIP_FIREWALL) - ip_output(m, NULL, NULL, 0, NULL, NULL); - else if (ip_output(m, NULL, NULL, IP_RAWOUTPUT, &sc->sc_imo, - NULL) == 0) - V_pfsyncstats.pfsyncs_opackets++; - else - V_pfsyncstats.pfsyncs_oerrors++; + n = m->m_nextpkt; + m->m_nextpkt = NULL; + + /* + * We distinguish between a deferral packet and our + * own pfsync packet based on M_SKIP_FIREWALL + * flag. This is XXX. + */ + if (m->m_flags & M_SKIP_FIREWALL) + ip_output(m, NULL, NULL, 0, NULL, NULL); + else if (ip_output(m, NULL, NULL, IP_RAWOUTPUT, &sc->sc_imo, + NULL) == 0) + V_pfsyncstats.pfsyncs_opackets++; + else + V_pfsyncstats.pfsyncs_oerrors++; + } } CURVNET_RESTORE(); } diff --git a/freebsd/sys/netpfil/pf/pf.c b/freebsd/sys/netpfil/pf/pf.c index e115061a..9b4653e2 100644 --- a/freebsd/sys/netpfil/pf/pf.c +++ b/freebsd/sys/netpfil/pf/pf.c @@ -1569,9 +1569,11 @@ pf_state_expires(const struct pf_state *state) states = V_pf_status.states; } if (end && states > start && start < end) { - if (states < end) - return (state->expire + timeout * (end - states) / - (end - start)); + if (states < end) { + timeout = (u_int64_t)timeout * (end - states) / + (end - start); + return (state->expire + timeout); + } else return (time_uptime); } @@ -5523,6 +5525,8 @@ pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, dst.sin_len = sizeof(dst); dst.sin_addr = ip->ip_dst; + bzero(&naddr, sizeof(naddr)); + if (TAILQ_EMPTY(&r->rpool.list)) { DPFPRINTF(PF_DEBUG_URGENT, ("%s: TAILQ_EMPTY(&r->rpool.list)\n", __func__)); @@ -5682,6 +5686,8 @@ pf_route6(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, dst.sin6_len = sizeof(dst); dst.sin6_addr = ip6->ip6_dst; + bzero(&naddr, sizeof(naddr)); + if (TAILQ_EMPTY(&r->rpool.list)) { DPFPRINTF(PF_DEBUG_URGENT, ("%s: TAILQ_EMPTY(&r->rpool.list)\n", __func__)); diff --git a/freebsd/sys/netpfil/pf/pf_if.c b/freebsd/sys/netpfil/pf/pf_if.c index ed69acad..4314bbce 100644 --- a/freebsd/sys/netpfil/pf/pf_if.c +++ b/freebsd/sys/netpfil/pf/pf_if.c @@ -855,7 +855,8 @@ pfi_detach_ifnet_event(void *arg __unused, struct ifnet *ifp) V_pfi_update++; pfi_kif_update(kif); - if_rele(kif->pfik_ifp); + if (kif->pfik_ifp) + if_rele(kif->pfik_ifp); kif->pfik_ifp = NULL; ifp->if_pf_kif = NULL; diff --git a/freebsd/sys/netpfil/pf/pf_lb.c b/freebsd/sys/netpfil/pf/pf_lb.c index 1fd02a0b..4e2bbbee 100644 --- a/freebsd/sys/netpfil/pf/pf_lb.c +++ b/freebsd/sys/netpfil/pf/pf_lb.c @@ -295,6 +295,10 @@ pf_get_sport(sa_family_t af, u_int8_t proto, struct pf_rule *r, switch (r->rpool.opts & PF_POOL_TYPEMASK) { case PF_POOL_RANDOM: case PF_POOL_ROUNDROBIN: + /* + * pick a different source address since we're out + * of free port choices for the current one. + */ if (pf_map_addr(af, r, saddr, naddr, &init_addr, sn)) return (1); break; @@ -326,6 +330,12 @@ pf_map_addr(sa_family_t af, struct pf_rule *r, struct pf_addr *saddr, src node was created just a moment ago in pf_create_state and it needs to be filled in with routing decision calculated here. */ if (*sn != NULL && !PF_AZERO(&(*sn)->raddr, af)) { + /* If the supplied address is the same as the current one we've + * been asked before, so tell the caller that there's no other + * address to be had. */ + if (PF_AEQ(naddr, &(*sn)->raddr, af)) + return (1); + PF_ACPY(naddr, &(*sn)->raddr, af); if (V_pf_status.debug >= PF_DEBUG_MISC) { printf("pf_map_addr: src tracking maps "); diff --git a/freebsd/sys/opencrypto/criov.c b/freebsd/sys/opencrypto/criov.c index 7fc7d392..a1893691 100644 --- a/freebsd/sys/opencrypto/criov.c +++ b/freebsd/sys/opencrypto/criov.c @@ -40,6 +40,8 @@ __FBSDID("$FreeBSD$"); #include <sys/kernel.h> #include <sys/mbuf.h> #include <sys/uio.h> +#include <sys/limits.h> +#include <sys/lock.h> #include <opencrypto/cryptodev.h> @@ -241,3 +243,55 @@ crypto_mbuftoiov(struct mbuf *mbuf, struct iovec **iovptr, int *cnt, *cnt = i; return 0; } + +static inline void * +m_contiguous_subsegment(struct mbuf *m, size_t skip, size_t len) +{ + int rel_off; + + MPASS(skip <= INT_MAX); + + m = m_getptr(m, (int)skip, &rel_off); + if (m == NULL) + return (NULL); + + MPASS(rel_off >= 0); + skip = rel_off; + if (skip + len > m->m_len) + return (NULL); + + return (mtod(m, char*) + skip); +} + +static inline void * +cuio_contiguous_segment(struct uio *uio, size_t skip, size_t len) +{ + int rel_off, idx; + + MPASS(skip <= INT_MAX); + idx = cuio_getptr(uio, (int)skip, &rel_off); + if (idx < 0) + return (NULL); + + MPASS(rel_off >= 0); + skip = rel_off; + if (skip + len > uio->uio_iov[idx].iov_len) + return (NULL); + return ((char *)uio->uio_iov[idx].iov_base + skip); +} + +void * +crypto_contiguous_subsegment(int crp_flags, void *crpbuf, + size_t skip, size_t len) +{ + if ((crp_flags & CRYPTO_F_IMBUF) != 0) + return (m_contiguous_subsegment(crpbuf, skip, len)); + else if ((crp_flags & CRYPTO_F_IOV) != 0) + return (cuio_contiguous_segment(crpbuf, skip, len)); + else { + MPASS((crp_flags & (CRYPTO_F_IMBUF | CRYPTO_F_IOV)) != + (CRYPTO_F_IMBUF | CRYPTO_F_IOV)); + return ((char*)crpbuf + skip); + } +} + diff --git a/freebsd/sys/opencrypto/cryptodev.h b/freebsd/sys/opencrypto/cryptodev.h index b3f81563..6431e6d8 100644 --- a/freebsd/sys/opencrypto/cryptodev.h +++ b/freebsd/sys/opencrypto/cryptodev.h @@ -564,5 +564,7 @@ extern void crypto_copydata(int flags, caddr_t buf, int off, int size, extern int crypto_apply(int flags, caddr_t buf, int off, int len, int (*f)(void *, void *, u_int), void *arg); +extern void *crypto_contiguous_subsegment(int, void *, size_t, size_t); + #endif /* _KERNEL */ #endif /* _CRYPTO_CRYPTO_H_ */ diff --git a/freebsd/sys/opencrypto/cryptosoft.c b/freebsd/sys/opencrypto/cryptosoft.c index 81c5bde8..43455b48 100644 --- a/freebsd/sys/opencrypto/cryptosoft.c +++ b/freebsd/sys/opencrypto/cryptosoft.c @@ -1093,6 +1093,9 @@ swcr_freesession(device_t dev, crypto_session_t cses) case CRYPTO_SHA2_256: case CRYPTO_SHA2_384: case CRYPTO_SHA2_512: + case CRYPTO_AES_128_NIST_GMAC: + case CRYPTO_AES_192_NIST_GMAC: + case CRYPTO_AES_256_NIST_GMAC: axf = swd->sw_axf; if (swd->sw_ictx) { diff --git a/freebsd/sys/security/audit/audit.h b/freebsd/sys/security/audit/audit.h index 2e47b8c4..2aeb5329 100644 --- a/freebsd/sys/security/audit/audit.h +++ b/freebsd/sys/security/audit/audit.h @@ -71,7 +71,7 @@ extern u_int audit_dtrace_enabled; extern int audit_trail_enabled; extern int audit_trail_suspended; -extern int audit_syscalls_enabled; +extern bool audit_syscalls_enabled; void audit_syscall_enter(unsigned short code, struct thread *td); void audit_syscall_exit(int error, struct thread *td); @@ -150,7 +150,7 @@ void audit_thread_free(struct thread *td); * Define macros to wrap the audit_arg_* calls by checking the global * audit_syscalls_enabled flag before performing the actual call. */ -#define AUDITING_TD(td) ((td)->td_pflags & TDP_AUDITREC) +#define AUDITING_TD(td) (__predict_false((td)->td_pflags & TDP_AUDITREC)) #define AUDIT_ARG_ADDR(addr) do { \ if (AUDITING_TD(curthread)) \ @@ -389,7 +389,7 @@ void audit_thread_free(struct thread *td); * auditing is disabled, so we don't just check audit_syscalls_enabled here. */ #define AUDIT_SYSCALL_EXIT(error, td) do { \ - if (td->td_pflags & TDP_AUDITREC) \ + if (AUDITING_TD(td)) \ audit_syscall_exit(error, td); \ } while (0) @@ -397,7 +397,7 @@ void audit_thread_free(struct thread *td); * A Macro to wrap the audit_sysclose() function. */ #define AUDIT_SYSCLOSE(td, fd) do { \ - if (td->td_pflags & TDP_AUDITREC) \ + if (AUDITING_TD(td)) \ audit_sysclose(td, fd); \ } while (0) diff --git a/freebsd/sys/sys/bus.h b/freebsd/sys/sys/bus.h index 74a48f81..8e1ba763 100644 --- a/freebsd/sys/sys/bus.h +++ b/freebsd/sys/sys/bus.h @@ -485,6 +485,10 @@ int bus_generic_suspend(device_t dev); int bus_generic_suspend_child(device_t dev, device_t child); int bus_generic_teardown_intr(device_t dev, device_t child, struct resource *irq, void *cookie); +int bus_generic_suspend_intr(device_t dev, device_t child, + struct resource *irq); +int bus_generic_resume_intr(device_t dev, device_t child, + struct resource *irq); int bus_generic_unmap_resource(device_t dev, device_t child, int type, struct resource *r, struct resource_map *map); @@ -535,6 +539,8 @@ int bus_setup_intr(device_t dev, struct resource *r, int flags, driver_filter_t filter, driver_intr_t handler, void *arg, void **cookiep); int bus_teardown_intr(device_t dev, struct resource *r, void *cookie); +int bus_suspend_intr(device_t dev, struct resource *r); +int bus_resume_intr(device_t dev, struct resource *r); int bus_bind_intr(device_t dev, struct resource *r, int cpu); int bus_describe_intr(device_t dev, struct resource *irq, void *cookie, const char *fmt, ...) __printflike(4, 5); @@ -612,6 +618,7 @@ void device_set_desc(device_t dev, const char* desc); void device_set_desc_copy(device_t dev, const char* desc); int device_set_devclass(device_t dev, const char *classname); int device_set_devclass_fixed(device_t dev, const char *classname); +bool device_is_devclass_fixed(device_t dev); int device_set_driver(device_t dev, driver_t *driver); void device_set_flags(device_t dev, u_int32_t flags); void device_set_softc(device_t dev, void *softc); diff --git a/freebsd/sys/sys/filedesc.h b/freebsd/sys/sys/filedesc.h index a3224f42..857d1fc9 100644 --- a/freebsd/sys/sys/filedesc.h +++ b/freebsd/sys/sys/filedesc.h @@ -173,7 +173,13 @@ enum { struct thread; -void filecaps_init(struct filecaps *fcaps); +static __inline void +filecaps_init(struct filecaps *fcaps) +{ + + bzero(fcaps, sizeof(*fcaps)); + fcaps->fc_nioctls = -1; +} bool filecaps_copy(const struct filecaps *src, struct filecaps *dst, bool locked); void filecaps_move(struct filecaps *src, struct filecaps *dst); @@ -289,7 +295,7 @@ fget_locked(struct filedesc *fdp, int fd) FILEDESC_LOCK_ASSERT(fdp); - if (fd < 0 || fd > fdp->fd_lastfile) + if (__predict_false((u_int)fd >= fdp->fd_nfiles)) return (NULL); return (fdp->fd_ofiles[fd].fde_file); @@ -302,11 +308,11 @@ fdeget_locked(struct filedesc *fdp, int fd) FILEDESC_LOCK_ASSERT(fdp); - if (fd < 0 || fd > fdp->fd_lastfile) + if (__predict_false((u_int)fd >= fdp->fd_nfiles)) return (NULL); fde = &fdp->fd_ofiles[fd]; - if (fde->fde_file == NULL) + if (__predict_false(fde->fde_file == NULL)) return (NULL); return (fde); diff --git a/freebsd/sys/sys/interrupt.h b/freebsd/sys/sys/interrupt.h index 105bb968..5c634054 100644 --- a/freebsd/sys/sys/interrupt.h +++ b/freebsd/sys/sys/interrupt.h @@ -62,6 +62,8 @@ struct intr_handler { #define IH_EXCLUSIVE 0x00000002 /* Exclusive interrupt. */ #define IH_ENTROPY 0x00000004 /* Device is a good entropy source. */ #define IH_DEAD 0x00000008 /* Handler should be removed. */ +#define IH_SUSP 0x00000010 /* Device is powered down. */ +#define IH_CHANGED 0x40000000 /* Handler state is changed. */ #define IH_MPSAFE 0x80000000 /* Handler does not need Giant. */ /* @@ -184,6 +186,8 @@ int intr_event_describe_handler(struct intr_event *ie, void *cookie, int intr_event_destroy(struct intr_event *ie); int intr_event_handle(struct intr_event *ie, struct trapframe *frame); int intr_event_remove_handler(void *cookie); +int intr_event_suspend_handler(void *cookie); +int intr_event_resume_handler(void *cookie); int intr_getaffinity(int irq, int mode, void *mask); void *intr_handler_source(void *cookie); int intr_setaffinity(int irq, int mode, void *mask); diff --git a/freebsd/sys/sys/jail.h b/freebsd/sys/sys/jail.h index c9f24bdd..ccd9f57f 100644 --- a/freebsd/sys/sys/jail.h +++ b/freebsd/sys/sys/jail.h @@ -229,9 +229,16 @@ struct prison_racct { #define PR_ALLOW_SOCKET_AF 0x00000040 #define PR_ALLOW_MLOCK 0x00000080 #define PR_ALLOW_READ_MSGBUF 0x00000100 +#define PR_ALLOW_UNPRIV_DEBUG 0x00000200 #define PR_ALLOW_RESERVED_PORTS 0x00008000 #define PR_ALLOW_KMEM_ACCESS 0x00010000 /* reserved, not used yet */ -#define PR_ALLOW_ALL_STATIC 0x000181ff +#define PR_ALLOW_ALL_STATIC 0x000183ff + +/* + * PR_ALLOW_DIFFERENCES determines which flags are able to be + * different between the parent and child jail upon creation. + */ +#define PR_ALLOW_DIFFERENCES (PR_ALLOW_UNPRIV_DEBUG) /* * OSD methods diff --git a/freebsd/sys/sys/mount.h b/freebsd/sys/sys/mount.h index ac1c215b..2a5d4cff 100644 --- a/freebsd/sys/sys/mount.h +++ b/freebsd/sys/sys/mount.h @@ -932,11 +932,15 @@ void syncer_resume(void); struct stat; __BEGIN_DECLS +int fhlink(struct fhandle *, const char *); +int fhlinkat(struct fhandle *, int, const char *); int fhopen(const struct fhandle *, int); +int fhreadlink(struct fhandle *, char *, size_t); int fhstat(const struct fhandle *, struct stat *); int fhstatfs(const struct fhandle *, struct statfs *); int fstatfs(int, struct statfs *); int getfh(const char *, fhandle_t *); +int getfhat(int, char *, struct fhandle *, int); int getfsstat(struct statfs *, long, int); int getmntinfo(struct statfs **, int); int lgetfh(const char *, fhandle_t *); diff --git a/freebsd/sys/sys/mutex.h b/freebsd/sys/sys/mutex.h index e9c91f80..8d2f01ba 100644 --- a/freebsd/sys/sys/mutex.h +++ b/freebsd/sys/sys/mutex.h @@ -531,7 +531,7 @@ do { \ int _giantcnt = 0; \ WITNESS_SAVE_DECL(Giant); \ \ - if (mtx_owned(&Giant)) { \ + if (__predict_false(mtx_owned(&Giant))) { \ WITNESS_SAVE(&Giant.lock_object, Giant); \ for (_giantcnt = 0; mtx_owned(&Giant) && \ !SCHEDULER_STOPPED(); _giantcnt++) \ @@ -544,7 +544,7 @@ do { \ #define PARTIAL_PICKUP_GIANT() \ mtx_assert(&Giant, MA_NOTOWNED); \ - if (_giantcnt > 0) { \ + if (__predict_false(_giantcnt > 0)) { \ while (_giantcnt--) \ mtx_lock(&Giant); \ WITNESS_RESTORE(&Giant.lock_object, Giant); \ diff --git a/freebsd/sys/sys/priv.h b/freebsd/sys/sys/priv.h index 55453f5a..3027cbf5 100644 --- a/freebsd/sys/sys/priv.h +++ b/freebsd/sys/sys/priv.h @@ -533,10 +533,10 @@ struct thread; struct ucred; #ifndef __rtems__ int priv_check(struct thread *td, int priv); -int priv_check_cred(struct ucred *cred, int priv, int flags); +int priv_check_cred(struct ucred *cred, int priv); #else /* __rtems__ */ #define priv_check(td, priv) 0 -#define priv_check_cred(cred, priv, flags) 0 +#define priv_check_cred(cred, priv) 0 #endif /* __rtems__ */ #endif diff --git a/freebsd/sys/sys/proc.h b/freebsd/sys/sys/proc.h index 9868f9cf..6c352059 100644 --- a/freebsd/sys/sys/proc.h +++ b/freebsd/sys/sys/proc.h @@ -643,10 +643,10 @@ struct proc { struct ksiginfo *p_ksi; /* Locked by parent proc lock */ sigqueue_t p_sigqueue; /* (c) Sigs not delivered to a td. */ #define p_siglist p_sigqueue.sq_signals + pid_t p_oppid; /* (c + e) Real parent pid. */ /* The following fields are all zeroed upon creation in fork. */ -#define p_startzero p_oppid - pid_t p_oppid; /* (c + e) Save ppid in ptrace. XXX */ +#define p_startzero p_vmspace struct vmspace *p_vmspace; /* (b) Address space. */ u_int p_swtick; /* (c) Tick when swapped in or out. */ u_int p_cowgen; /* (c) Generation of COW pointers. */ @@ -691,6 +691,7 @@ struct proc { u_int p_magic; /* (b) Magic number. */ int p_osrel; /* (x) osreldate for the binary (from ELF note, if any) */ + uint32_t p_fctl0; /* (x) ABI feature control, ELF note */ char p_comm[MAXCOMLEN + 1]; /* (x) Process name. */ struct sysentvec *p_sysent; /* (b) Syscall dispatch info. */ struct pargs *p_args; /* (c) Process arguments. */ @@ -997,8 +998,11 @@ extern pid_t pid_max; #define THREAD_CAN_SLEEP() ((curthread)->td_no_sleeping == 0) #define PIDHASH(pid) (&pidhashtbl[(pid) & pidhash]) +#define PIDHASHLOCK(pid) (&pidhashtbl_lock[((pid) & pidhashlock)]) extern LIST_HEAD(pidhashhead, proc) *pidhashtbl; +extern struct sx *pidhashtbl_lock; extern u_long pidhash; +extern u_long pidhashlock; #define TIDHASH(tid) (&tidhashtbl[(tid) & tidhash]) extern LIST_HEAD(tidhashhead, thread) *tidhashtbl; extern u_long tidhash; @@ -1010,8 +1014,10 @@ extern u_long pgrphash; extern struct sx allproc_lock; extern int allproc_gen; +extern struct sx zombproc_lock; extern struct sx proctree_lock; extern struct mtx ppeers_lock; +extern struct mtx procid_lock; extern struct proc proc0; /* Process slot for swapper. */ extern struct thread0_storage thread0_st; /* Primary thread in proc0. */ #define thread0 (thread0_st.t0st_thread) @@ -1033,7 +1039,6 @@ extern struct uma_zone *proc_zone; struct proc *pfind(pid_t); /* Find process by id. */ struct proc *pfind_any(pid_t); /* Find (zombie) process by id. */ -struct proc *pfind_locked(pid_t pid); struct pgrp *pgfind(pid_t); /* Find process group by id. */ struct proc *zpfind(pid_t); /* Find zombie process by id. */ @@ -1081,6 +1086,7 @@ int enterthispgrp(struct proc *p, struct pgrp *pgrp); void faultin(struct proc *p); void fixjobc(struct proc *p, struct pgrp *pgrp, int entering); int fork1(struct thread *, struct fork_req *); +void fork_rfppwait(struct thread *); void fork_exit(void (*)(void *, struct trapframe *), void *, struct trapframe *); void fork_return(struct thread *, struct trapframe *); @@ -1111,11 +1117,12 @@ int proc_getargv(struct thread *td, struct proc *p, struct sbuf *sb); int proc_getauxv(struct thread *td, struct proc *p, struct sbuf *sb); int proc_getenvv(struct thread *td, struct proc *p, struct sbuf *sb); void procinit(void); +int proc_iterate(int (*cb)(struct proc *, void *), void *cbarg); void proc_linkup0(struct proc *p, struct thread *td); void proc_linkup(struct proc *p, struct thread *td); struct proc *proc_realparent(struct proc *child); void proc_reap(struct thread *td, struct proc *p, int *status, int options); -void proc_reparent(struct proc *child, struct proc *newparent); +void proc_reparent(struct proc *child, struct proc *newparent, bool set_oppid); void proc_set_traced(struct proc *p, bool stop); void proc_wkilled(struct proc *p); struct pstats *pstats_alloc(void); @@ -1234,6 +1241,15 @@ td_softdep_cleanup(struct thread *td) } #endif /* __rtems__ */ +#define PROC_ID_PID 0 +#define PROC_ID_GROUP 1 +#define PROC_ID_SESSION 2 +#define PROC_ID_REAP 3 + +void proc_id_set(int type, pid_t id); +void proc_id_set_cond(int type, pid_t id); +void proc_id_clear(int type, pid_t id); + #endif /* _KERNEL */ #endif /* !_SYS_PROC_H_ */ diff --git a/freebsd/sys/sys/racct.h b/freebsd/sys/sys/racct.h index 84de705f..17af3276 100644 --- a/freebsd/sys/sys/racct.h +++ b/freebsd/sys/sys/racct.h @@ -91,7 +91,7 @@ struct ucred; #define RACCT_DECAYING 0x20 extern int racct_types[]; -extern int racct_enable; +extern bool racct_enable; #define ASSERT_RACCT_ENABLED() KASSERT(racct_enable, \ ("%s called with !racct_enable", __func__)) @@ -164,12 +164,14 @@ extern struct mtx racct_lock; #define RACCT_UNLOCK() mtx_unlock(&racct_lock) #define RACCT_LOCK_ASSERT() mtx_assert(&racct_lock, MA_OWNED) +#define RACCT_ENABLED() __predict_false(racct_enable) + #define RACCT_PROC_LOCK(p) do { \ - if (__predict_false(racct_enable)) \ + if (RACCT_ENABLED()) \ PROC_LOCK(p); \ } while (0) #define RACCT_PROC_UNLOCK(p) do { \ - if (__predict_false(racct_enable)) \ + if (RACCT_ENABLED()) \ PROC_UNLOCK(p); \ } while (0) @@ -178,6 +180,7 @@ void racct_add_cred(struct ucred *cred, int resource, uint64_t amount); void racct_add_force(struct proc *p, int resource, uint64_t amount); void racct_add_buf(struct proc *p, const struct buf *bufp, int is_write); int racct_set(struct proc *p, int resource, uint64_t amount); +int racct_set_unlocked(struct proc *p, int resource, uint64_t amount); void racct_set_force(struct proc *p, int resource, uint64_t amount); void racct_sub(struct proc *p, int resource, uint64_t amount); void racct_sub_cred(struct ucred *cred, int resource, uint64_t amount); @@ -194,6 +197,7 @@ void racct_proc_exit(struct proc *p); void racct_proc_ucred_changed(struct proc *p, struct ucred *oldcred, struct ucred *newcred); void racct_move(struct racct *dest, struct racct *src); +void racct_proc_throttled(struct proc *p); void racct_proc_throttle(struct proc *p, int timeout); #else diff --git a/freebsd/sys/sys/refcount.h b/freebsd/sys/sys/refcount.h index 41713147..0cc4eb41 100644 --- a/freebsd/sys/sys/refcount.h +++ b/freebsd/sys/sys/refcount.h @@ -79,8 +79,6 @@ refcount_release(volatile u_int *count) /* * This functions returns non-zero if the refcount was * incremented. Else zero is returned. - * - * A temporary hack until refcount_* APIs are sorted out. */ static __inline __result_use_check int refcount_acquire_if_not_zero(volatile u_int *count) diff --git a/freebsd/sys/sys/resourcevar.h b/freebsd/sys/sys/resourcevar.h index 7c7e8458..c7576d2d 100644 --- a/freebsd/sys/sys/resourcevar.h +++ b/freebsd/sys/sys/resourcevar.h @@ -147,6 +147,19 @@ struct plimit *lim_alloc(void); void lim_copy(struct plimit *dst, struct plimit *src); rlim_t lim_cur(struct thread *td, int which); +#define lim_cur(td, which) ({ \ + rlim_t _rlim; \ + struct thread *_td = (td); \ + int _which = (which); \ + if (__builtin_constant_p(which) && which != RLIMIT_DATA && \ + which != RLIMIT_STACK && which != RLIMIT_VMEM) { \ + _rlim = td->td_limit->pl_rlimit[which].rlim_cur; \ + } else { \ + _rlim = lim_cur(_td, _which); \ + } \ + _rlim; \ +}) + rlim_t lim_cur_proc(struct proc *p, int which); void lim_fork(struct proc *p1, struct proc *p2); void lim_free(struct plimit *limp); diff --git a/freebsd/sys/sys/rman.h b/freebsd/sys/sys/rman.h index 9c2f4653..e955cd58 100644 --- a/freebsd/sys/sys/rman.h +++ b/freebsd/sys/sys/rman.h @@ -131,6 +131,7 @@ bus_space_tag_t rman_get_bustag(struct resource *); rman_res_t rman_get_end(struct resource *); device_t rman_get_device(struct resource *); u_int rman_get_flags(struct resource *); +void *rman_get_irq_cookie(struct resource *); void rman_get_mapping(struct resource *, struct resource_map *); int rman_get_rid(struct resource *); rman_res_t rman_get_size(struct resource *); @@ -155,6 +156,7 @@ void rman_set_bushandle(struct resource *_r, bus_space_handle_t _h); void rman_set_bustag(struct resource *_r, bus_space_tag_t _t); void rman_set_device(struct resource *_r, device_t _dev); void rman_set_end(struct resource *_r, rman_res_t _end); +void rman_set_irq_cookie(struct resource *_r, void *_c); void rman_set_mapping(struct resource *, struct resource_map *); void rman_set_rid(struct resource *_r, int _rid); void rman_set_start(struct resource *_r, rman_res_t _start); diff --git a/freebsd/sys/sys/sdt.h b/freebsd/sys/sys/sdt.h index 424a0e3a..fb5c3d67 100644 --- a/freebsd/sys/sys/sdt.h +++ b/freebsd/sys/sys/sdt.h @@ -90,6 +90,7 @@ extern volatile bool sdt_probes_enabled; #define SDT_PROVIDER_DECLARE(prov) #define SDT_PROBE_DEFINE(prov, mod, func, name) #define SDT_PROBE_DECLARE(prov, mod, func, name) +#define SDT_PROBES_ENABLED() 0 #define SDT_PROBE(prov, mod, func, name, arg0, arg1, arg2, arg3, arg4) #define SDT_PROBE_ARGTYPE(prov, mod, func, name, num, type, xtype) @@ -164,8 +165,10 @@ SET_DECLARE(sdt_argtypes_set, struct sdt_argtype); #define SDT_PROBE_DECLARE(prov, mod, func, name) \ extern struct sdt_probe sdt_##prov##_##mod##_##func##_##name[1] +#define SDT_PROBES_ENABLED() __predict_false(sdt_probes_enabled) + #define SDT_PROBE(prov, mod, func, name, arg0, arg1, arg2, arg3, arg4) do { \ - if (__predict_false(sdt_probes_enabled)) { \ + if (SDT_PROBES_ENABLED()) { \ if (__predict_false(sdt_##prov##_##mod##_##func##_##name->id)) \ (*sdt_probe_func)(sdt_##prov##_##mod##_##func##_##name->id, \ (uintptr_t) arg0, (uintptr_t) arg1, (uintptr_t) arg2, \ diff --git a/freebsd/sys/sys/sockbuf.h b/freebsd/sys/sys/sockbuf.h index fc287023..3b716283 100644 --- a/freebsd/sys/sys/sockbuf.h +++ b/freebsd/sys/sys/sockbuf.h @@ -165,8 +165,6 @@ void sbrelease_locked(struct sockbuf *sb, struct socket *so); int sbsetopt(struct socket *so, int cmd, u_long cc); int sbreserve_locked(struct sockbuf *sb, u_long cc, struct socket *so, struct thread *td); -struct mbuf * - sbsndptr(struct sockbuf *sb, u_int off, u_int len, u_int *moff); void sbsndptr_adv(struct sockbuf *sb, struct mbuf *mb, u_int len); struct mbuf * sbsndptr_noadv(struct sockbuf *sb, u_int off, u_int *moff); diff --git a/freebsd/sys/sys/sx.h b/freebsd/sys/sys/sx.h index 10cfb10a..9e0da83a 100644 --- a/freebsd/sys/sys/sx.h +++ b/freebsd/sys/sys/sx.h @@ -313,7 +313,6 @@ int sx_xlocked(struct sx *sx); #define SX_NOPROFILE 0x02 #define SX_NOWITNESS 0x04 #define SX_QUIET 0x08 -#define SX_NOADAPTIVE 0x10 #define SX_RECURSE 0x20 #define SX_NEW 0x40 diff --git a/freebsd/sys/sys/sysctl.h b/freebsd/sys/sys/sysctl.h index e270a4b2..cf5a5581 100644 --- a/freebsd/sys/sys/sysctl.h +++ b/freebsd/sys/sys/sysctl.h @@ -165,11 +165,7 @@ struct sysctl_req { size_t oldlen; size_t oldidx; int (*oldfunc)(struct sysctl_req *, const void *, size_t); -#ifndef __rtems__ - void *newptr; -#else /* __rtems__ */ - const void *newptr; -#endif /* __rtems__ */ + const void *newptr; size_t newlen; size_t newidx; int (*newfunc)(struct sysctl_req *, void *, size_t); @@ -1135,11 +1131,7 @@ int kernel_sysctlbyname(struct thread *td, char *name, void *old, #endif /* __rtems__ */ int flags); int userland_sysctl(struct thread *td, int *name, u_int namelen, void *old, -#ifndef __rtems__ - size_t *oldlenp, int inkernel, void *new, size_t newlen, -#else /* __rtems__ */ - size_t *oldlenp, int inkernel, const void *newp, size_t newlen, -#endif /* __rtems__ */ + size_t *oldlenp, int inkernel, const void *new, size_t newlen, size_t *retval, int flags); int sysctl_find_oid(int *name, u_int namelen, struct sysctl_oid **noid, int *nindx, struct sysctl_req *req); diff --git a/freebsd/sys/sys/sysproto.h b/freebsd/sys/sys/sysproto.h index a39f5cd3..faa5a62f 100644 --- a/freebsd/sys/sys/sysproto.h +++ b/freebsd/sys/sys/sysproto.h @@ -254,7 +254,7 @@ struct umask_args { struct chroot_args { char path_l_[PADL_(const char *)]; const char * path; char path_r_[PADR_(const char *)]; }; -struct getpagesize_args { +struct ogetpagesize_args { register_t dummy; }; struct msync_args { @@ -1817,6 +1817,26 @@ struct getrandom_args { char buflen_l_[PADL_(size_t)]; size_t buflen; char buflen_r_[PADR_(size_t)]; char flags_l_[PADL_(unsigned int)]; unsigned int flags; char flags_r_[PADR_(unsigned int)]; }; +struct getfhat_args { + char fd_l_[PADL_(int)]; int fd; char fd_r_[PADR_(int)]; + char path_l_[PADL_(char *)]; char * path; char path_r_[PADR_(char *)]; + char fhp_l_[PADL_(struct fhandle *)]; struct fhandle * fhp; char fhp_r_[PADR_(struct fhandle *)]; + char flags_l_[PADL_(int)]; int flags; char flags_r_[PADR_(int)]; +}; +struct fhlink_args { + char fhp_l_[PADL_(struct fhandle *)]; struct fhandle * fhp; char fhp_r_[PADR_(struct fhandle *)]; + char to_l_[PADL_(const char *)]; const char * to; char to_r_[PADR_(const char *)]; +}; +struct fhlinkat_args { + char fhp_l_[PADL_(struct fhandle *)]; struct fhandle * fhp; char fhp_r_[PADR_(struct fhandle *)]; + char tofd_l_[PADL_(int)]; int tofd; char tofd_r_[PADR_(int)]; + char to_l_[PADL_(const char *)]; const char * to; char to_r_[PADR_(const char *)]; +}; +struct fhreadlink_args { + char fhp_l_[PADL_(struct fhandle *)]; struct fhandle * fhp; char fhp_r_[PADR_(struct fhandle *)]; + char buf_l_[PADL_(char *)]; char * buf; char buf_r_[PADR_(char *)]; + char bufsize_l_[PADL_(size_t)]; size_t bufsize; char bufsize_r_[PADR_(size_t)]; +}; int nosys(struct thread *, struct nosys_args *); void sys_sys_exit(struct thread *, struct sys_exit_args *); int sys_fork(struct thread *, struct fork_args *); @@ -2197,6 +2217,10 @@ int sys_kevent(struct thread *, struct kevent_args *); int sys_cpuset_getdomain(struct thread *, struct cpuset_getdomain_args *); int sys_cpuset_setdomain(struct thread *, struct cpuset_setdomain_args *); int sys_getrandom(struct thread *, struct getrandom_args *); +int sys_getfhat(struct thread *, struct getfhat_args *); +int sys_fhlink(struct thread *, struct fhlink_args *); +int sys_fhlinkat(struct thread *, struct fhlinkat_args *); +int sys_fhreadlink(struct thread *, struct fhreadlink_args *); #ifdef COMPAT_43 @@ -2230,7 +2254,7 @@ struct ofstat_args { char fd_l_[PADL_(int)]; int fd; char fd_r_[PADR_(int)]; char sb_l_[PADL_(struct ostat *)]; struct ostat * sb; char sb_r_[PADR_(struct ostat *)]; }; -struct getkerninfo_args { +struct ogetkerninfo_args { char op_l_[PADL_(int)]; int op; char op_r_[PADR_(int)]; char where_l_[PADL_(char *)]; char * where; char where_r_[PADR_(char *)]; char size_l_[PADL_(size_t *)]; size_t * size; char size_r_[PADR_(size_t *)]; @@ -2244,14 +2268,19 @@ struct ommap_args { char fd_l_[PADL_(int)]; int fd; char fd_r_[PADR_(int)]; char pos_l_[PADL_(long)]; long pos; char pos_r_[PADR_(long)]; }; -struct gethostname_args { +struct ogethostname_args { char hostname_l_[PADL_(char *)]; char * hostname; char hostname_r_[PADR_(char *)]; char len_l_[PADL_(u_int)]; u_int len; char len_r_[PADR_(u_int)]; }; -struct sethostname_args { +struct osethostname_args { char hostname_l_[PADL_(char *)]; char * hostname; char hostname_r_[PADR_(char *)]; char len_l_[PADL_(u_int)]; u_int len; char len_r_[PADR_(u_int)]; }; +struct oaccept_args { + char s_l_[PADL_(int)]; int s; char s_r_[PADR_(int)]; + char name_l_[PADL_(struct sockaddr *)]; struct sockaddr * name; char name_r_[PADR_(struct sockaddr *)]; + char anamelen_l_[PADL_(int *)]; int * anamelen; char anamelen_r_[PADR_(int *)]; +}; struct osend_args { char s_l_[PADL_(int)]; int s; char s_r_[PADR_(int)]; char buf_l_[PADL_(const void *)]; const void * buf; char buf_r_[PADR_(const void *)]; @@ -2337,13 +2366,13 @@ int osigaction(struct thread *, struct osigaction_args *); int osigprocmask(struct thread *, struct osigprocmask_args *); int osigpending(struct thread *, struct osigpending_args *); int ofstat(struct thread *, struct ofstat_args *); -int ogetkerninfo(struct thread *, struct getkerninfo_args *); -int ogetpagesize(struct thread *, struct getpagesize_args *); +int ogetkerninfo(struct thread *, struct ogetkerninfo_args *); +int ogetpagesize(struct thread *, struct ogetpagesize_args *); int ommap(struct thread *, struct ommap_args *); int owait(struct thread *, struct owait_args *); -int ogethostname(struct thread *, struct gethostname_args *); -int osethostname(struct thread *, struct sethostname_args *); -int oaccept(struct thread *, struct accept_args *); +int ogethostname(struct thread *, struct ogethostname_args *); +int osethostname(struct thread *, struct osethostname_args *); +int oaccept(struct thread *, struct oaccept_args *); int osend(struct thread *, struct osend_args *); int orecv(struct thread *, struct orecv_args *); int osigreturn(struct thread *, struct osigreturn_args *); @@ -3094,6 +3123,10 @@ int freebsd11_mknodat(struct thread *, struct freebsd11_mknodat_args *); #define SYS_AUE_cpuset_getdomain AUE_NULL #define SYS_AUE_cpuset_setdomain AUE_NULL #define SYS_AUE_getrandom AUE_NULL +#define SYS_AUE_getfhat AUE_NULL +#define SYS_AUE_fhlink AUE_NULL +#define SYS_AUE_fhlinkat AUE_NULL +#define SYS_AUE_fhreadlink AUE_NULL #endif /* __rtems__ */ #undef PAD_ diff --git a/freebsd/sys/sys/systm.h b/freebsd/sys/sys/systm.h index d74a1e4b..1aa57670 100644 --- a/freebsd/sys/sys/systm.h +++ b/freebsd/sys/sys/systm.h @@ -654,6 +654,32 @@ int alloc_unr_specific(struct unrhdr *uh, u_int item); int alloc_unrl(struct unrhdr *uh); void free_unr(struct unrhdr *uh, u_int item); +#ifndef __LP64__ +#define UNR64_LOCKED +#endif + +struct unrhdr64 { + uint64_t counter; +}; + +static __inline void +new_unrhdr64(struct unrhdr64 *unr64, uint64_t low) +{ + + unr64->counter = low; +} + +#ifdef UNR64_LOCKED +uint64_t alloc_unr64(struct unrhdr64 *); +#else +static __inline uint64_t +alloc_unr64(struct unrhdr64 *unr64) +{ + + return (atomic_fetchadd_64(&unr64->counter, 1)); +} +#endif + void intr_prof_stack_use(struct thread *td, struct trapframe *frame); void counted_warning(unsigned *counter, const char *msg); diff --git a/freebsd/sys/sys/taskqueue.h b/freebsd/sys/sys/taskqueue.h index 7ba9e268..4af1e0a3 100644 --- a/freebsd/sys/sys/taskqueue.h +++ b/freebsd/sys/sys/taskqueue.h @@ -93,6 +93,7 @@ void taskqueue_drain(struct taskqueue *queue, struct task *task); void taskqueue_drain_timeout(struct taskqueue *queue, struct timeout_task *timeout_task); void taskqueue_drain_all(struct taskqueue *queue); +void taskqueue_quiesce(struct taskqueue *queue); void taskqueue_free(struct taskqueue *queue); void taskqueue_run(struct taskqueue *queue); void taskqueue_block(struct taskqueue *queue); diff --git a/freebsd/sys/sys/user.h b/freebsd/sys/sys/user.h index 1218deec..80716460 100644 --- a/freebsd/sys/sys/user.h +++ b/freebsd/sys/sys/user.h @@ -264,6 +264,7 @@ struct user { #define KF_TYPE_SEM 9 #define KF_TYPE_PTS 10 #define KF_TYPE_PROCDESC 11 +#define KF_TYPE_DEV 12 #define KF_TYPE_UNKNOWN 255 #define KF_VTYPE_VNON 0 diff --git a/freebsd/sys/vm/uma_core.c b/freebsd/sys/vm/uma_core.c index db624654..ade3328a 100644 --- a/freebsd/sys/vm/uma_core.c +++ b/freebsd/sys/vm/uma_core.c @@ -1537,7 +1537,7 @@ keg_small_init(uma_keg_t keg) if (keg->uk_flags & UMA_ZONE_OFFPAGE) shsize = 0; else - shsize = sizeof(struct uma_slab); + shsize = SIZEOF_UMA_SLAB; if (rsize <= slabsize - shsize) keg->uk_ipers = (slabsize - shsize) / rsize; @@ -1607,7 +1607,6 @@ keg_small_init(uma_keg_t keg) static void keg_large_init(uma_keg_t keg) { - u_int shsize; KASSERT(keg != NULL, ("Keg is null in keg_large_init")); KASSERT((keg->uk_flags & UMA_ZFLAG_CACHEONLY) == 0, @@ -1620,23 +1619,17 @@ keg_large_init(uma_keg_t keg) keg->uk_rsize = keg->uk_size; /* Check whether we have enough space to not do OFFPAGE. */ - if ((keg->uk_flags & UMA_ZONE_OFFPAGE) == 0) { - shsize = sizeof(struct uma_slab); - if (shsize & UMA_ALIGN_PTR) - shsize = (shsize & ~UMA_ALIGN_PTR) + - (UMA_ALIGN_PTR + 1); - - if (PAGE_SIZE * keg->uk_ppera - keg->uk_rsize < shsize) { - /* - * We can't do OFFPAGE if we're internal, in which case - * we need an extra page per allocation to contain the - * slab header. - */ - if ((keg->uk_flags & UMA_ZFLAG_INTERNAL) == 0) - keg->uk_flags |= UMA_ZONE_OFFPAGE; - else - keg->uk_ppera++; - } + if ((keg->uk_flags & UMA_ZONE_OFFPAGE) == 0 && + PAGE_SIZE * keg->uk_ppera - keg->uk_rsize < SIZEOF_UMA_SLAB) { + /* + * We can't do OFFPAGE if we're internal, in which case + * we need an extra page per allocation to contain the + * slab header. + */ + if ((keg->uk_flags & UMA_ZFLAG_INTERNAL) == 0) + keg->uk_flags |= UMA_ZONE_OFFPAGE; + else + keg->uk_ppera++; } if ((keg->uk_flags & UMA_ZONE_OFFPAGE) && @@ -1783,20 +1776,11 @@ keg_ctor(void *mem, int size, void *udata, int flags) /* * If we're putting the slab header in the actual page we need to - * figure out where in each page it goes. This calculates a right - * justified offset into the memory on an ALIGN_PTR boundary. + * figure out where in each page it goes. See SIZEOF_UMA_SLAB + * macro definition. */ if (!(keg->uk_flags & UMA_ZONE_OFFPAGE)) { - u_int totsize; - - /* Size of the slab struct and free list */ - totsize = sizeof(struct uma_slab); - - if (totsize & UMA_ALIGN_PTR) - totsize = (totsize & ~UMA_ALIGN_PTR) + - (UMA_ALIGN_PTR + 1); - keg->uk_pgoff = (PAGE_SIZE * keg->uk_ppera) - totsize; - + keg->uk_pgoff = (PAGE_SIZE * keg->uk_ppera) - SIZEOF_UMA_SLAB; /* * The only way the following is possible is if with our * UMA_ALIGN_PTR adjustments we are now bigger than @@ -1804,13 +1788,10 @@ keg_ctor(void *mem, int size, void *udata, int flags) * mathematically possible for all cases, so we make * sure here anyway. */ - totsize = keg->uk_pgoff + sizeof(struct uma_slab); - if (totsize > PAGE_SIZE * keg->uk_ppera) { - printf("zone %s ipers %d rsize %d size %d\n", - zone->uz_name, keg->uk_ipers, keg->uk_rsize, - keg->uk_size); - panic("UMA slab won't fit."); - } + KASSERT(keg->uk_pgoff + sizeof(struct uma_slab) <= + PAGE_SIZE * keg->uk_ppera, + ("zone %s ipers %d rsize %d size %d slab won't fit", + zone->uz_name, keg->uk_ipers, keg->uk_rsize, keg->uk_size)); } if (keg->uk_flags & UMA_ZONE_HASH) @@ -2105,10 +2086,17 @@ uma_startup_count(int vm_zones) #endif /* Memory for the rest of startup zones, UMA and VM, ... */ - if (zsize > UMA_SLAB_SPACE) - pages += (zones + vm_zones) * - howmany(roundup2(zsize, UMA_BOOT_ALIGN), UMA_SLAB_SIZE); - else if (roundup2(zsize, UMA_BOOT_ALIGN) > UMA_SLAB_SPACE) + if (zsize > UMA_SLAB_SPACE) { + /* See keg_large_init(). */ + u_int ppera; + + ppera = howmany(roundup2(zsize, UMA_BOOT_ALIGN), PAGE_SIZE); + if (PAGE_SIZE * ppera - roundup2(zsize, UMA_BOOT_ALIGN) < + SIZEOF_UMA_SLAB) + ppera++; + pages += (zones + vm_zones) * ppera; + } else if (roundup2(zsize, UMA_BOOT_ALIGN) > UMA_SLAB_SPACE) + /* See keg_small_init() special case for uk_ppera = 1. */ pages += zones; else pages += howmany(zones, diff --git a/freebsd/sys/vm/uma_int.h b/freebsd/sys/vm/uma_int.h index 0fbc0512..0429bac6 100644 --- a/freebsd/sys/vm/uma_int.h +++ b/freebsd/sys/vm/uma_int.h @@ -139,9 +139,17 @@ #define UMA_MAX_WASTE 10 /* - * Size of memory in a not offpage slab available for actual items. + * Actual size of uma_slab when it is placed at an end of a page + * with pointer sized alignment requirement. */ -#define UMA_SLAB_SPACE (UMA_SLAB_SIZE - sizeof(struct uma_slab)) +#define SIZEOF_UMA_SLAB ((sizeof(struct uma_slab) & UMA_ALIGN_PTR) ? \ + (sizeof(struct uma_slab) & ~UMA_ALIGN_PTR) + \ + (UMA_ALIGN_PTR + 1) : sizeof(struct uma_slab)) + +/* + * Size of memory in a not offpage single page slab available for actual items. + */ +#define UMA_SLAB_SPACE (PAGE_SIZE - SIZEOF_UMA_SLAB) /* * I doubt there will be many cases where this is exceeded. This is the initial |