Index: head/share/man/man4/Makefile =================================================================== --- head/share/man/man4/Makefile (revision 277053) +++ head/share/man/man4/Makefile (revision 277054) @@ -1,870 +1,871 @@ # @(#)Makefile 8.1 (Berkeley) 6/18/93 # $FreeBSD$ .include MAN= aac.4 \ aacraid.4 \ acpi.4 \ ${_acpi_asus.4} \ ${_acpi_asus_wmi.4} \ ${_acpi_dock.4} \ ${_acpi_fujitsu.4} \ ${_acpi_hp.4} \ ${_acpi_ibm.4} \ ${_acpi_panasonic.4} \ ${_acpi_rapidstart.4} \ ${_acpi_sony.4} \ acpi_thermal.4 \ ${_acpi_toshiba.4} \ acpi_video.4 \ ${_acpi_wmi.4} \ ada.4 \ adv.4 \ adw.4 \ ae.4 \ ${_aesni.4} \ age.4 \ agp.4 \ aha.4 \ ahb.4 \ ahc.4 \ ahci.4 \ ahd.4 \ ${_aibs.4} \ aio.4 \ alc.4 \ ale.4 \ alpm.4 \ altera_atse.4 \ altera_avgen.4 \ altera_jtag_uart.4 \ altera_sdcard.4 \ altq.4 \ amdpm.4 \ ${_amdsbwd.4} \ ${_amdsmb.4} \ ${_amdtemp.4} \ ${_bxe.4} \ amr.4 \ an.4 \ ${_aout.4} \ ${_apic.4} \ arcmsr.4 \ ${_asmc.4} \ ata.4 \ ath.4 \ ath_ahb.4 \ ath_hal.4 \ ath_pci.4 \ atkbd.4 \ atkbdc.4 \ atp.4 \ ${_atf_test_case.4} \ ${_atrtc.4} \ ${_attimer.4} \ audit.4 \ auditpipe.4 \ aue.4 \ axe.4 \ axge.4 \ bce.4 \ bfe.4 \ bge.4 \ ${_bhyve.4} \ bktr.4 \ blackhole.4 \ bpf.4 \ bridge.4 \ bt.4 \ bwi.4 \ bwn.4 \ capsicum.4 \ cardbus.4 \ carp.4 \ cas.4 \ cc_cdg.4 \ cc_chd.4 \ cc_cubic.4 \ + cc_dctcp.4 \ cc_hd.4 \ cc_htcp.4 \ cc_newreno.4 \ cc_vegas.4 \ ccd.4 \ cd.4 \ cdce.4 \ ch.4 \ ciss.4 \ cm.4 \ cmx.4 \ ${_coretemp.4} \ ${_cpuctl.4} \ cpufreq.4 \ crypto.4 \ ctl.4 \ cue.4 \ cxgb.4 \ cxgbe.4 \ cy.4 \ da.4 \ dc.4 \ dcons.4 \ dcons_crom.4 \ ddb.4 \ de.4 \ devctl.4 \ digi.4 \ disc.4 \ divert.4 \ ${_dpms.4} \ dpt.4 \ dummynet.4 \ ed.4 \ edsc.4 \ ehci.4 \ em.4 \ en.4 \ enc.4 \ epair.4 \ esp.4 \ est.4 \ et.4 \ etherswitch.4 \ eventtimers.4 \ exca.4 \ fatm.4 \ fd.4 \ fdc.4 \ fdt.4 \ fdtbus.4 \ ffclock.4 \ filemon.4 \ firewire.4 \ fpa.4 \ full.4 \ fwe.4 \ fwip.4 \ fwohci.4 \ fxp.4 \ gbde.4 \ gdb.4 \ gem.4 \ geom.4 \ geom_fox.4 \ geom_linux_lvm.4 \ geom_map.4 \ geom_uncompress.4 \ geom_uzip.4 \ gif.4 \ gpio.4 \ gpioiic.4 \ gpioled.4 \ gre.4 \ h_ertt.4 \ hatm.4 \ hifn.4 \ hme.4 \ hpet.4 \ ${_hpt27xx.4} \ ${_hptiop.4} \ ${_hptmv.4} \ ${_hptnr.4} \ ${_hptrr.4} \ hv_ata_pci_disengage.4 \ hv_kvp.4 \ hv_netvsc.4 \ hv_storvsc.4 \ hv_utils.4 \ hv_vmbus.4 \ hwpmc.4 \ ichsmb.4 \ ${_ichwd.4} \ icmp.4 \ icmp6.4 \ ida.4 \ ifmib.4 \ igb.4 \ igmp.4 \ iic.4 \ iicbb.4 \ iicbus.4 \ iicsmb.4 \ iir.4 \ inet.4 \ inet6.4 \ intpm.4 \ intro.4 \ ${_io.4} \ ip.4 \ ip6.4 \ ipfirewall.4 \ ipheth.4 \ ${_ipmi.4} \ ips.4 \ ipsec.4 \ ipw.4 \ ipwfw.4 \ isci.4 \ iscsi.4 \ iscsi_initiator.4 \ ismt.4 \ isp.4 \ ispfw.4 \ iwi.4 \ iwifw.4 \ iwn.4 \ iwnfw.4 \ ixgb.4 \ ixgbe.4 \ jme.4 \ joy.4 \ kbdmux.4 \ keyboard.4 \ kld.4 \ ksyms.4 \ ktr.4 \ kue.4 \ lagg.4 \ le.4 \ led.4 \ lge.4 \ ${_linux.4} \ lm75.4 \ lmc.4 \ lo.4 \ lp.4 \ lpbb.4 \ lpt.4 \ mac.4 \ mac_biba.4 \ mac_bsdextended.4 \ mac_ifoff.4 \ mac_lomac.4 \ mac_mls.4 \ mac_none.4 \ mac_partition.4 \ mac_portacl.4 \ mac_seeotheruids.4 \ mac_stub.4 \ mac_test.4 \ malo.4 \ mcd.4 \ md.4 \ me.4 \ mem.4 \ meteor.4 \ mfi.4 \ miibus.4 \ mk48txx.4 \ mld.4 \ mlx.4 \ mly.4 \ mmc.4 \ mmcsd.4 \ mn.4 \ mod_cc.4 \ mos.4 \ mouse.4 \ mpr.4 \ mps.4 \ mpt.4 \ mrsas.4 \ msk.4 \ mtio.4 \ multicast.4 \ mvs.4 \ mwl.4 \ mwlfw.4 \ mxge.4 \ my.4 \ nand.4 \ nandsim.4 \ natm.4 \ natmip.4 \ ncr.4 \ ncv.4 \ ${_ndis.4} \ net80211.4 \ netfpga10g_nf10bmac.4 \ netgraph.4 \ netintro.4 \ netmap.4 \ ${_nfe.4} \ ${_nfsmb.4} \ ng_async.4 \ ng_atm.4 \ ngatmbase.4 \ ng_atmllc.4 \ ng_bluetooth.4 \ ng_bpf.4 \ ng_bridge.4 \ ng_bt3c.4 \ ng_btsocket.4 \ ng_car.4 \ ng_ccatm.4 \ ng_cisco.4 \ ng_deflate.4 \ ng_device.4 \ nge.4 \ ng_echo.4 \ ng_eiface.4 \ ng_etf.4 \ ng_ether.4 \ ng_ether_echo.4 \ ng_frame_relay.4 \ ng_gif.4 \ ng_gif_demux.4 \ ng_h4.4 \ ng_hci.4 \ ng_hole.4 \ ng_hub.4 \ ng_iface.4 \ ng_ipfw.4 \ ng_ip_input.4 \ ng_ksocket.4 \ ng_l2cap.4 \ ng_l2tp.4 \ ng_lmi.4 \ ng_mppc.4 \ ng_nat.4 \ ng_netflow.4 \ ng_one2many.4 \ ng_patch.4 \ ng_ppp.4 \ ng_pppoe.4 \ ng_pptpgre.4 \ ng_pred1.4 \ ng_rfc1490.4 \ ng_socket.4 \ ng_source.4 \ ng_split.4 \ ng_sppp.4 \ ng_sscfu.4 \ ng_sscop.4 \ ng_tag.4 \ ng_tcpmss.4 \ ng_tee.4 \ ng_tty.4 \ ng_ubt.4 \ ng_UI.4 \ ng_uni.4 \ ng_vjc.4 \ ng_vlan.4 \ nmdm.4 \ nsp.4 \ ${_ntb.4} \ null.4 \ ${_nvd.4} \ ${_nvme.4} \ ${_nvram.4} \ ${_nvram2env.4} \ ${_nxge.4} \ oce.4 \ ohci.4 \ orm.4 \ ${_padlock.4} \ pass.4 \ patm.4 \ pccard.4 \ pccbb.4 \ pcf.4 \ pci.4 \ pcib.4 \ pcic.4 \ pcm.4 \ pcn.4 \ ${_pf.4} \ ${_pflog.4} \ ${_pfsync.4} \ pim.4 \ polling.4 \ ppbus.4 \ ppc.4 \ ppi.4 \ procdesc.4 \ proto.4 \ psm.4 \ pst.4 \ pt.4 \ pts.4 \ pty.4 \ puc.4 \ ${_qlxge.4} \ ${_qlxgb.4} \ ${_qlxgbe.4} \ ral.4 \ random.4 \ rc.4 \ re.4 \ rgephy.4 \ rights.4 \ rl.4 \ rndtest.4 \ route.4 \ rp.4 \ rsu.4 \ rsufw.4 \ rue.4 \ rum.4 \ run.4 \ runfw.4 \ sa.4 \ safe.4 \ sbp.4 \ sbp_targ.4 \ scc.4 \ scd.4 \ sched_4bsd.4 \ sched_ule.4 \ screen.4 \ scsi.4 \ sctp.4 \ sdhci.4 \ sem.4 \ send.4 \ ses.4 \ sf.4 \ ${_sfxge.4} \ sge.4 \ si.4 \ siba.4 \ siftr.4 \ siis.4 \ simplebus.4 \ sio.4 \ sis.4 \ sk.4 \ smb.4 \ smbus.4 \ smp.4 \ smsc.4 \ sn.4 \ snd_ad1816.4 \ snd_als4000.4 \ snd_atiixp.4 \ snd_cmi.4 \ snd_cs4281.4 \ snd_csa.4 \ snd_ds1.4 \ snd_emu10k1.4 \ snd_emu10kx.4 \ snd_envy24.4 \ snd_envy24ht.4 \ snd_es137x.4 \ snd_ess.4 \ snd_fm801.4 \ snd_gusc.4 \ snd_hda.4 \ snd_hdspe.4 \ snd_ich.4 \ snd_maestro3.4 \ snd_maestro.4 \ snd_mss.4 \ snd_neomagic.4 \ snd_sbc.4 \ snd_solo.4 \ snd_spicds.4 \ snd_t4dwave.4 \ snd_uaudio.4 \ snd_via8233.4 \ snd_via82c686.4 \ snd_vibes.4 \ snp.4 \ spic.4 \ ${_spkr.4} \ splash.4 \ sppp.4 \ ste.4 \ stf.4 \ stg.4 \ stge.4 \ sym.4 \ syncache.4 \ syncer.4 \ syscons.4 \ sysmouse.4 \ tap.4 \ targ.4 \ tcp.4 \ tdfx.4 \ terasic_mtl.4 \ termios.4 \ textdump.4 \ ti.4 \ timecounters.4 \ tl.4 \ ${_tpm.4} \ trm.4 \ tty.4 \ tun.4 \ twa.4 \ twe.4 \ tws.4 \ tx.4 \ txp.4 \ u3g.4 \ uark.4 \ uart.4 \ uath.4 \ ubsa.4 \ ubsec.4 \ ubser.4 \ ubtbcmfw.4 \ uchcom.4 \ ucom.4 \ ucycom.4 \ udav.4 \ udbp.4 \ udp.4 \ udplite.4 \ uep.4 \ ufm.4 \ ufoma.4 \ uftdi.4 \ ugen.4 \ uhci.4 \ uhid.4 \ uhso.4 \ uipaq.4 \ ukbd.4 \ uled.4 \ ulpt.4 \ umass.4 \ umcs.4 \ umct.4 \ umodem.4 \ umoscom.4 \ ums.4 \ unix.4 \ upgt.4 \ uplcom.4 \ ural.4 \ urio.4 \ urndis.4 \ ${_urtw.4} \ urtwn.4 \ urtwnfw.4 \ usb.4 \ usb_template.4 \ usb_quirk.4 \ uslcom.4 \ usfs.4 \ utopia.4 \ uvisor.4 \ uvscom.4 \ vale.4 \ vga.4 \ vge.4 \ viapm.4 \ ${_viawd.4} \ ${_virtio.4} \ ${_virtio_balloon.4} \ ${_virtio_blk.4} \ ${_virtio_console.4} \ ${_virtio_random.4} \ ${_virtio_scsi.4} \ vkbd.4 \ vlan.4 \ vxlan.4 \ ${_vmx.4} \ vpo.4 \ vr.4 \ vt.4 \ vte.4 \ ${_vtnet.4} \ ${_vxge.4} \ watchdog.4 \ wb.4 \ ${_wbwd.4} \ wi.4 \ witness.4 \ wlan.4 \ wlan_acl.4 \ wlan_amrr.4 \ wlan_ccmp.4 \ wlan_tkip.4 \ wlan_wep.4 \ wlan_xauth.4 \ ${_wpi.4} \ wsp.4 \ xe.4 \ ${_xen.4} \ xhci.4 \ xl.4 \ ${_xnb.4} \ xpt.4 \ zero.4 \ zyd.4 MLINKS= ae.4 if_ae.4 MLINKS+=age.4 if_age.4 MLINKS+=agp.4 agpgart.4 MLINKS+=alc.4 if_alc.4 MLINKS+=ale.4 if_ale.4 MLINKS+=altera_atse.4 atse.4 MLINKS+=altera_sdcard.4 altera_sdcardc.4 MLINKS+=altq.4 ALTQ.4 MLINKS+=ath.4 if_ath.4 MLINKS+=ath_pci.4 if_ath_pci.4 MLINKS+=an.4 if_an.4 MLINKS+=aue.4 if_aue.4 MLINKS+=axe.4 if_axe.4 MLINKS+=bce.4 if_bce.4 MLINKS+=bfe.4 if_bfe.4 MLINKS+=bge.4 if_bge.4 MLINKS+=bktr.4 brooktree.4 MLINKS+=bridge.4 if_bridge.4 MLINKS+=bwi.4 if_bwi.4 MLINKS+=bwn.4 if_bwn.4 MLINKS+=${_bxe.4} ${_if_bxe.4} MLINKS+=cas.4 if_cas.4 MLINKS+=cdce.4 if_cdce.4 MLINKS+=crypto.4 cryptodev.4 MLINKS+=cue.4 if_cue.4 MLINKS+=cxgb.4 if_cxgb.4 MLINKS+=cxgbe.4 if_cxgbe.4 MLINKS+=dc.4 if_dc.4 MLINKS+=de.4 if_de.4 MLINKS+=disc.4 if_disc.4 MLINKS+=ed.4 if_ed.4 MLINKS+=edsc.4 if_edsc.4 MLINKS+=em.4 if_em.4 MLINKS+=en.4 if_en.4 MLINKS+=enc.4 if_enc.4 MLINKS+=epair.4 if_epair.4 MLINKS+=et.4 if_et.4 MLINKS+=fatm.4 if_fatm.4 MLINKS+=fd.4 stderr.4 \ fd.4 stdin.4 \ fd.4 stdout.4 MLINKS+=fdt.4 FDT.4 MLINKS+=firewire.4 ieee1394.4 MLINKS+=fpa.4 fea.4 MLINKS+=fwe.4 if_fwe.4 MLINKS+=fwip.4 if_fwip.4 MLINKS+=fxp.4 if_fxp.4 MLINKS+=gem.4 if_gem.4 MLINKS+=geom.4 GEOM.4 MLINKS+=gif.4 if_gif.4 MLINKS+=gpio.4 gpiobus.4 MLINKS+=gre.4 if_gre.4 MLINKS+=hatm.4 if_hatm.4 MLINKS+=hme.4 if_hme.4 MLINKS+=hpet.4 acpi_hpet.4 MLINKS+=${_hptrr.4} ${_rr232x.4} MLINKS+=${_attimer.4} ${_i8254.4} MLINKS+=igb.4 if_igb.4 MLINKS+=ip.4 rawip.4 MLINKS+=ipfirewall.4 ipaccounting.4 \ ipfirewall.4 ipacct.4 \ ipfirewall.4 ipfw.4 MLINKS+=ipheth.4 if_ipheth.4 MLINKS+=ipw.4 if_ipw.4 MLINKS+=iwi.4 if_iwi.4 MLINKS+=iwn.4 if_iwn.4 MLINKS+=ixgb.4 if_ixgb.4 MLINKS+=ixgbe.4 ix.4 MLINKS+=ixgbe.4 if_ix.4 MLINKS+=ixgbe.4 if_ixgbe.4 MLINKS+=jme.4 if_jme.4 MLINKS+=kue.4 if_kue.4 MLINKS+=lagg.4 trunk.4 MLINKS+=lagg.4 if_lagg.4 MLINKS+=le.4 if_le.4 MLINKS+=lge.4 if_lge.4 MLINKS+=lmc.4 if_lmc.4 MLINKS+=lo.4 loop.4 MLINKS+=lp.4 plip.4 MLINKS+=malo.4 if_malo.4 MLINKS+=md.4 vn.4 MLINKS+=mem.4 kmem.4 MLINKS+=mn.4 if_mn.4 MLINKS+=mos.4 if_mos.4 MLINKS+=msk.4 if_msk.4 MLINKS+=mwl.4 if_mwl.4 MLINKS+=mxge.4 if_mxge.4 MLINKS+=my.4 if_my.4 MLINKS+=${_ndis.4} ${_if_ndis.4} MLINKS+=netfpga10g_nf10bmac.4 if_nf10bmac.4 MLINKS+=netintro.4 net.4 \ netintro.4 networking.4 MLINKS+=${_nfe.4} ${_if_nfe.4} MLINKS+=nge.4 if_nge.4 MLINKS+=${_ntb.4} ${_if_ntb.4} \ ${_ntb.4} ${_ntb_hw.4} MLINKS+=${_nxge.4} ${_if_nxge.4} MLINKS+=patm.4 if_patm.4 MLINKS+=pccbb.4 cbb.4 MLINKS+=pcm.4 snd.4 \ pcm.4 sound.4 MLINKS+=pcn.4 if_pcn.4 MLINKS+=ral.4 if_ral.4 MLINKS+=re.4 if_re.4 MLINKS+=rl.4 if_rl.4 MLINKS+=rue.4 if_rue.4 MLINKS+=rum.4 if_rum.4 MLINKS+=run.4 if_run.4 MLINKS+=scsi.4 CAM.4 \ scsi.4 cam.4 \ scsi.4 scbus.4 \ scsi.4 SCSI.4 MLINKS+=sf.4 if_sf.4 MLINKS+=sge.4 if_sge.4 MLINKS+=sis.4 if_sis.4 MLINKS+=sk.4 if_sk.4 MLINKS+=smp.4 SMP.4 MLINKS+=smsc.4 if_smsc.4 MLINKS+=sn.4 if_sn.4 MLINKS+=snd_envy24.4 snd_ak452x.4 MLINKS+=snd_sbc.4 snd_sb16.4 \ snd_sbc.4 snd_sb8.4 MLINKS+=${_spkr.4} ${_speaker.4} MLINKS+=splash.4 screensaver.4 MLINKS+=ste.4 if_ste.4 MLINKS+=stf.4 if_stf.4 MLINKS+=stge.4 if_stge.4 MLINKS+=syncache.4 syncookies.4 MLINKS+=syscons.4 sc.4 MLINKS+=tap.4 if_tap.4 MLINKS+=tdfx.4 tdfx_linux.4 MLINKS+=ti.4 if_ti.4 MLINKS+=tl.4 if_tl.4 MLINKS+=tun.4 if_tun.4 MLINKS+=tx.4 if_tx.4 MLINKS+=txp.4 if_txp.4 MLINKS+=u3g.4 u3gstub.4 MLINKS+=uath.4 if_uath.4 MLINKS+=udav.4 if_udav.4 MLINKS+=upgt.4 if_upgt.4 MLINKS+=ural.4 if_ural.4 MLINKS+=urndis.4 if_urndis.4 MLINKS+=${_urtw.4} ${_if_urtw.4} MLINKS+=vge.4 if_vge.4 MLINKS+=vlan.4 if_vlan.4 MLINKS+=vxlan.4 if_vxlan.4 MLINKS+=${_vmx.4} ${_if_vmx.4} MLINKS+=vpo.4 imm.4 MLINKS+=vr.4 if_vr.4 MLINKS+=vte.4 if_vte.4 MLINKS+=${_vtnet.4} ${_if_vtnet.4} MLINKS+=${_vxge.4} ${_if_vxge.4} MLINKS+=watchdog.4 SW_WATCHDOG.4 MLINKS+=wb.4 if_wb.4 MLINKS+=wi.4 if_wi.4 MLINKS+=${_wpi.4} ${_if_wpi.4} MLINKS+=xe.4 if_xe.4 MLINKS+=xl.4 if_xl.4 MLINKS+=zyd.4 if_zyd.4 .if ${MACHINE_CPUARCH} == "amd64" || ${MACHINE_CPUARCH} == "i386" _acpi_asus.4= acpi_asus.4 _acpi_asus_wmi.4= acpi_asus_wmi.4 _acpi_dock.4= acpi_dock.4 _acpi_fujitsu.4=acpi_fujitsu.4 _acpi_hp.4= acpi_hp.4 _acpi_ibm.4= acpi_ibm.4 _acpi_panasonic.4=acpi_panasonic.4 _acpi_rapidstart.4=acpi_rapidstart.4 _acpi_sony.4= acpi_sony.4 _acpi_toshiba.4=acpi_toshiba.4 _acpi_wmi.4= acpi_wmi.4 _aesni.4= aesni.4 _aout.4= aout.4 _apic.4= apic.4 _atrtc.4= atrtc.4 _attimer.4= attimer.4 _aibs.4= aibs.4 _amdsbwd.4= amdsbwd.4 _amdsmb.4= amdsmb.4 _amdtemp.4= amdtemp.4 _asmc.4= asmc.4 _bxe.4= bxe.4 _coretemp.4= coretemp.4 _cpuctl.4= cpuctl.4 _dpms.4= dpms.4 _hpt27xx.4= hpt27xx.4 _hptiop.4= hptiop.4 _hptmv.4= hptmv.4 _hptnr.4= hptnr.4 _hptrr.4= hptrr.4 _i8254.4= i8254.4 _ichwd.4= ichwd.4 _if_bxe.4= if_bxe.4 _if_ndis.4= if_ndis.4 _if_nfe.4= if_nfe.4 _if_nxge.4= if_nxge.4 _if_urtw.4= if_urtw.4 _if_vmx.4= if_vmx.4 _if_vtnet.4= if_vtnet.4 _if_vxge.4= if_vxge.4 _if_wpi.4= if_wpi.4 _ipmi.4= ipmi.4 _io.4= io.4 _linux.4= linux.4 _ndis.4= ndis.4 _nfe.4= nfe.4 _nfsmb.4= nfsmb.4 _nvd.4= nvd.4 _nvme.4= nvme.4 _nvram.4= nvram.4 _nxge.4= nxge.4 _virtio.4= virtio.4 _virtio_balloon.4=virtio_balloon.4 _virtio_blk.4= virtio_blk.4 _virtio_console.4=virtio_console.4 _virtio_random.4= virtio_random.4 _virtio_scsi.4= virtio_scsi.4 _vmx.4= vmx.4 _vtnet.4= vtnet.4 _vxge.4= vxge.4 _padlock.4= padlock.4 _rr232x.4= rr232x.4 _speaker.4= speaker.4 _spkr.4= spkr.4 _tpm.4= tpm.4 _urtw.4= urtw.4 _viawd.4= viawd.4 _wbwd.4= wbwd.4 _wpi.4= wpi.4 _xen.4= xen.4 _xnb.4= xnb.4 .endif .if ${MACHINE_CPUARCH} == "amd64" _bhyve.4= bhyve.4 _if_ntb.4= if_ntb.4 _ntb.4= ntb.4 _ntb_hw.4= ntb_hw.4 _qlxge.4= qlxge.4 _qlxgb.4= qlxgb.4 _qlxgbe.4= qlxgbe.4 _sfxge.4= sfxge.4 MLINKS+=qlxge.4 if_qlxge.4 MLINKS+=qlxgb.4 if_qlxgb.4 MLINKS+=qlxgbe.4 if_qlxgbe.4 MLINKS+=sfxge.4 if_sfxge.4 .endif .if ${MACHINE_CPUARCH} == "mips" _nvram2env.4= nvram2env.4 .endif .if exists(${.CURDIR}/man4.${MACHINE_CPUARCH}) SUBDIR= man4.${MACHINE_CPUARCH} .endif .if ${MK_TESTS} != "no" ATF= ${.CURDIR}/../../../contrib/atf .PATH: ${ATF}/doc _atf_test_case.4= atf-test-case.4 .endif .if ${MK_PF} != "no" _pf.4= pf.4 _pflog.4= pflog.4 _pfsync.4= pfsync.4 .endif .include Index: head/share/man/man4/cc_dctcp.4 =================================================================== --- head/share/man/man4/cc_dctcp.4 (nonexistent) +++ head/share/man/man4/cc_dctcp.4 (revision 277054) @@ -0,0 +1,127 @@ +.\" +.\" Copyright (c) 2014 Midori Kato +.\" Copyright (c) 2014 The FreeBSD Foundation +.\" All rights reserved. +.\" +.\" Portions of this documentation were written at Keio University, Japan. +.\" +.\" 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 THE AUTHOR 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 THE AUTHOR 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. +.\" +.\" $FreeBSD$ +.\" +.Dd January 12, 2015 +.Dt CC_DCTCP 4 +.Os +.Sh NAME +.Nm cc_dctcp +.Nd DCTCP Congestion Control Algorithm +.Sh DESCRIPTION +The DCTCP (data center TCP) congestion control algorithm aims to maximise +throughput and minimise latency in data center networks by utilising the +proportion of Explicit Congestion Notification (ECN) marks received from capable +hardware as a congestion signal. +.Pp +DCTCP uses fraction of ECN marked packets to update congestion window. The +window reduction ratio is always <= 1/2. Only when all of the packets are +marked, congestion window is halved. +.Pp +In order to keep the accuracy of the ECN marked fraction, a DCTCP receiver +mirrors back incoming (or missing) CE marks by setting (or clearing) ECE marks. +This feedback methodology is also adopted when the receiver uses delayed ACK. +.Pp +FreeBSD DCTCP implementation includes two minor modifications for the one-sided +deployment. Considering the situation that DCTCP is used as sender and classic +ECN is used as receiver, DCTCP sets the CWR flag as the reaction to the ECE +flag. In addition, when classic ECN is used as sender and DCTCP is used as +receiver, DCTCP avoids to mirror back ACKs only when the CWR flag is +set in the incoming packet. +.Pp +The other specifications are based on the paper and Internet Draft referenced +in the +.Sx SEE ALSO +section below. +.Sh MIB Variables +The algorithm exposes the following tunable variables in the +.Va net.inet.tcp.cc.dctcp +branch of the +.Xr sysctl 3 +MIB: +.Bl -tag -width ".Va alpha" +.It Va alpha +An initial estimator of the congestion on the link. +Default is 0. +.It Va dctcp_shift_g +An estimation gain in the alpha calculation. +Default is 16. +.It Va slowstart +A trigger to halve congestion window after slow start. +Default does nothing to halve window. +.Sh SEE ALSO +.Xr cc_chd 4 , +.Xr cc_cubic 4 , +.Xr cc_hd 4 , +.Xr cc_htcp 4 , +.Xr cc_newreno 4 , +.Xr cc_vegas 4 , +.Xr mod_cc 4 , +.Xr tcp 4 , +.Xr mod_cc 9 +.Rs +.%A "Mohammad Alizadeh" +.%A "Albert Greenberg" +.%A "David A. Maltz" +.%A "Jitendra Padhye" +.%A "Parveen Patel" +.%A "Balaji Prabhakar" +.%A "Sudipta Sengupta" +.%A "Murari Sridharan" +.%T "Data Center TCP (DCTCP)" +.%U "http://research.microsoft.com/pubs/121386/dctcp-public.pdf" +.%J "ACM SIGCOMM 2010" +.%D "July 2010" +.%P "63-74" +.Re +.Rs +.%A "Stephen Bensley" +.%A "Lars Eggert" +.%A "Dave Thaler" +.%T "Microsoft's Datacenter TCP (DCTCP): TCP Congestion Control for Datacenters" +.%U "http://tools.ietf.org/html/draft-bensley-tcpm-dctcp-01" +.Re +.Sh HISTORY +The +.Nm +congestion control module first appeared in +.Fx 11.0 . +.Pp +The module was first released in 2014 by Midori Kato studying at Keio +University, Japan. +.Sh AUTHORS +.An -nosplit +The +.Nm +congestion control module and this manual page were written by +.An Midori Kato katoon@sfc.wide.ad.jp +and +.An Lars Eggert lars@netapp.com +with help and modifications from +.An Hiren Panchasara hiren@FreeBSD.org Property changes on: head/share/man/man4/cc_dctcp.4 ___________________________________________________________________ Added: svn:eol-style ## -0,0 +1 ## +native \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Added: svn:mime-type ## -0,0 +1 ## +text/plain \ No newline at end of property Index: head/share/man/man4/mod_cc.4 =================================================================== --- head/share/man/man4/mod_cc.4 (revision 277053) +++ head/share/man/man4/mod_cc.4 (revision 277054) @@ -1,118 +1,119 @@ .\" .\" Copyright (c) 2010-2011 The FreeBSD Foundation .\" All rights reserved. .\" .\" This documentation was written at the Centre for Advanced Internet .\" Architectures, Swinburne University of Technology, Melbourne, Australia by .\" David Hayes and Lawrence Stewart under sponsorship from the FreeBSD .\" Foundation. .\" .\" 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 THE AUTHOR 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 THE AUTHOR 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. .\" .\" $FreeBSD$ .\" -.Dd December 26, 2014 +.Dd January 12, 2015 .Dt MOD_CC 4 .Os .Sh NAME .Nm mod_cc .Nd Modular congestion control .Sh DESCRIPTION The modular congestion control framework allows the TCP implementation to dynamically change the congestion control algorithm used by new and existing connections. Algorithms are identified by a unique .Xr ascii 7 name. Algorithm modules can be compiled into the kernel or loaded as kernel modules using the .Xr kld 4 facility. .Pp The default algorithm is NewReno, and all connections use the default unless explicitly overridden using the TCP_CONGESTION socket option (see .Xr tcp 4 for details). The default can be changed using a .Xr sysctl 3 MIB variable detailed in the .Sx MIB Variables section below. .Sh MIB Variables The framework exposes the following variables in the .Va net.inet.tcp.cc branch of the .Xr sysctl 3 MIB: .Bl -tag -width ".Va available" .It Va available Read-only list of currently available congestion control algorithms by name. .It Va algorithm Returns the current default congestion control algorithm when read, and changes the default when set. When attempting to change the default algorithm, this variable should be set to one of the names listed by the .Va net.inet.tcp.cc.available MIB variable. .El .Sh SEE ALSO .Xr cc_cdg 4 , .Xr cc_chd 4 , .Xr cc_cubic 4 , +.Xr cc_dctcp 4 , .Xr cc_hd 4 , .Xr cc_htcp 4 , .Xr cc_newreno 4 , .Xr cc_vegas 4 , .Xr tcp 4 , .Xr mod_cc 9 .Sh ACKNOWLEDGEMENTS Development and testing of this software were made possible in part by grants from the FreeBSD Foundation and Cisco University Research Program Fund at Community Foundation Silicon Valley. .Sh HISTORY The .Nm modular congestion control framework first appeared in .Fx 9.0 . .Pp The framework was first released in 2007 by James Healy and Lawrence Stewart whilst working on the NewTCP research project at Swinburne University of Technology's Centre for Advanced Internet Architectures, Melbourne, Australia, which was made possible in part by a grant from the Cisco University Research Program Fund at Community Foundation Silicon Valley. More details are available at: .Pp http://caia.swin.edu.au/urp/newtcp/ .Sh AUTHORS .An -nosplit The .Nm facility was written by .An Lawrence Stewart Aq Mt lstewart@FreeBSD.org , .An James Healy Aq Mt jimmy@deefa.com and .An David Hayes Aq Mt david.hayes@ieee.org . .Pp This manual page was written by .An David Hayes Aq Mt david.hayes@ieee.org and .An Lawrence Stewart Aq Mt lstewart@FreeBSD.org . Index: head/sys/modules/cc/Makefile =================================================================== --- head/sys/modules/cc/Makefile (revision 277053) +++ head/sys/modules/cc/Makefile (revision 277054) @@ -1,10 +1,11 @@ # $FreeBSD$ SUBDIR= cc_cdg \ cc_chd \ cc_cubic \ + cc_dctcp \ cc_hd \ cc_htcp \ cc_vegas .include Index: head/sys/modules/cc/cc_dctcp/Makefile =================================================================== --- head/sys/modules/cc/cc_dctcp/Makefile (nonexistent) +++ head/sys/modules/cc/cc_dctcp/Makefile (revision 277054) @@ -0,0 +1,7 @@ +# $FreeBSD$ + +.PATH: ${.CURDIR}/../../../netinet/cc +KMOD= cc_dctcp +SRCS= cc_dctcp.c + +.include Property changes on: head/sys/modules/cc/cc_dctcp/Makefile ___________________________________________________________________ Added: svn:eol-style ## -0,0 +1 ## +native \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Added: svn:mime-type ## -0,0 +1 ## +text/plain \ No newline at end of property Index: head/sys/netinet/cc/cc_dctcp.c =================================================================== --- head/sys/netinet/cc/cc_dctcp.c (nonexistent) +++ head/sys/netinet/cc/cc_dctcp.c (revision 277054) @@ -0,0 +1,474 @@ +/*- + * Copyright (c) 2007-2008 + * Swinburne University of Technology, Melbourne, Australia + * Copyright (c) 2009-2010 Lawrence Stewart + * Copyright (c) 2014 Midori Kato + * Copyright (c) 2014 The FreeBSD Foundation + * 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 THE AUTHOR 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 THE AUTHOR 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. + */ + +/* + * An implementation of the DCTCP algorithm for FreeBSD, based on + * "Data Center TCP (DCTCP)" by M. Alizadeh, A. Greenberg, D. A. Maltz, + * J. Padhye, P. Patel, B. Prabhakar, S. Sengupta, and M. Sridharan., + * in ACM Conference on SIGCOMM 2010, New York, USA, + * Originally released as the contribution of Microsoft Research project. + */ + +#include +__FBSDID("$FreeBSD$"); + +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include +#include +#include +#include +#include + +#include + +#define CAST_PTR_INT(X) (*((int*)(X))) + +#define MAX_ALPHA_VALUE 1024 +static VNET_DEFINE(uint32_t, dctcp_alpha) = 0; +#define V_dctcp_alpha VNET(dctcp_alpha) +static VNET_DEFINE(uint32_t, dctcp_shift_g) = 4; +#define V_dctcp_shift_g VNET(dctcp_shift_g) +static VNET_DEFINE(uint32_t, dctcp_slowstart) = 0; +#define V_dctcp_slowstart VNET(dctcp_slowstart) + +struct dctcp { + int bytes_ecn; /* # of marked bytes during a RTT */ + int bytes_total; /* # of acked bytes during a RTT */ + int alpha; /* the fraction of marked bytes */ + int ce_prev; /* CE state of the last segment */ + int save_sndnxt; /* end sequence number of the current window */ + int ece_curr; /* ECE flag in this segment */ + int ece_prev; /* ECE flag in the last segment */ + uint32_t num_cong_events; /* # of congestion events */ +}; + +static MALLOC_DEFINE(M_dctcp, "dctcp data", + "Per connection data required for the dctcp algorithm"); + +static void dctcp_ack_received(struct cc_var *ccv, uint16_t type); +static void dctcp_after_idle(struct cc_var *ccv); +static void dctcp_cb_destroy(struct cc_var *ccv); +static int dctcp_cb_init(struct cc_var *ccv); +static void dctcp_cong_signal(struct cc_var *ccv, uint32_t type); +static void dctcp_conn_init(struct cc_var *ccv); +static void dctcp_post_recovery(struct cc_var *ccv); +static void dctcp_ecnpkt_handler(struct cc_var *ccv); +static void dctcp_update_alpha(struct cc_var *ccv); + +struct cc_algo dctcp_cc_algo = { + .name = "dctcp", + .ack_received = dctcp_ack_received, + .cb_destroy = dctcp_cb_destroy, + .cb_init = dctcp_cb_init, + .cong_signal = dctcp_cong_signal, + .conn_init = dctcp_conn_init, + .post_recovery = dctcp_post_recovery, + .ecnpkt_handler = dctcp_ecnpkt_handler, + .after_idle = dctcp_after_idle, +}; + +static void +dctcp_ack_received(struct cc_var *ccv, uint16_t type) +{ + struct dctcp *dctcp_data; + int bytes_acked = 0; + + dctcp_data = ccv->cc_data; + + if (CCV(ccv, t_flags) & TF_ECN_PERMIT) { + /* + * DCTCP doesn't treat receipt of ECN marked packet as a + * congestion event. Thus, DCTCP always executes the ACK + * processing out of congestion recovery. + */ + if (IN_CONGRECOVERY(CCV(ccv, t_flags))) { + EXIT_CONGRECOVERY(CCV(ccv, t_flags)); + newreno_cc_algo.ack_received(ccv, type); + ENTER_CONGRECOVERY(CCV(ccv, t_flags)); + } else + newreno_cc_algo.ack_received(ccv, type); + + if (type == CC_DUPACK) + bytes_acked = CCV(ccv, t_maxseg); + + if (type == CC_ACK) + bytes_acked = ccv->bytes_this_ack; + + /* Update total bytes. */ + dctcp_data->bytes_total += bytes_acked; + + /* Update total marked bytes. */ + if (dctcp_data->ece_curr) { + if (!dctcp_data->ece_prev + && bytes_acked > CCV(ccv, t_maxseg)) { + dctcp_data->bytes_ecn += + (bytes_acked - CCV(ccv, t_maxseg)); + } else + dctcp_data->bytes_ecn += bytes_acked; + dctcp_data->ece_prev = 1; + } else { + if (dctcp_data->ece_prev + && bytes_acked > CCV(ccv, t_maxseg)) + dctcp_data->bytes_ecn += CCV(ccv, t_maxseg); + dctcp_data->ece_prev = 0; + } + dctcp_data->ece_curr = 0; + + /* + * Update the fraction of marked bytes at the end of + * current window size. + */ + if ((IN_FASTRECOVERY(CCV(ccv, t_flags)) && + SEQ_GEQ(ccv->curack, CCV(ccv, snd_recover))) || + (!IN_FASTRECOVERY(CCV(ccv, t_flags)) && + SEQ_GT(ccv->curack, dctcp_data->save_sndnxt))) + dctcp_update_alpha(ccv); + } else + newreno_cc_algo.ack_received(ccv, type); +} + +static void +dctcp_after_idle(struct cc_var *ccv) +{ + struct dctcp *dctcp_data; + + dctcp_data = ccv->cc_data; + + /* Initialize internal parameters after idle time */ + dctcp_data->bytes_ecn = 0; + dctcp_data->bytes_total = 0; + dctcp_data->save_sndnxt = CCV(ccv, snd_nxt); + dctcp_data->alpha = V_dctcp_alpha; + dctcp_data->ece_curr = 0; + dctcp_data->ece_prev = 0; + dctcp_data->num_cong_events = 0; + + dctcp_cc_algo.after_idle = newreno_cc_algo.after_idle; +} + +static void +dctcp_cb_destroy(struct cc_var *ccv) +{ + if (ccv->cc_data != NULL) + free(ccv->cc_data, M_dctcp); +} + +static int +dctcp_cb_init(struct cc_var *ccv) +{ + struct dctcp *dctcp_data; + + dctcp_data = malloc(sizeof(struct dctcp), M_dctcp, M_NOWAIT|M_ZERO); + + if (dctcp_data == NULL) + return (ENOMEM); + + /* Initialize some key variables with sensible defaults. */ + dctcp_data->bytes_ecn = 0; + dctcp_data->bytes_total = 0; + /* + * When alpha is set to 0 in the beggining, DCTCP sender transfers as + * much data as possible until the value converges which may expand the + * queueing delay at the switch. When alpha is set to 1, queueing delay + * is kept small. + * Throughput-sensitive applications should have alpha = 0 + * Latency-sensitive applications should have alpha = 1 + * + * Note: DCTCP draft suggests initial alpha to be 1 but we've decided to + * keep it 0 as default. + */ + dctcp_data->alpha = V_dctcp_alpha; + dctcp_data->save_sndnxt = 0; + dctcp_data->ce_prev = 0; + dctcp_data->ece_curr = 0; + dctcp_data->ece_prev = 0; + dctcp_data->num_cong_events = 0; + + ccv->cc_data = dctcp_data; + return (0); +} + +/* + * Perform any necessary tasks before we enter congestion recovery. + */ +static void +dctcp_cong_signal(struct cc_var *ccv, uint32_t type) +{ + struct dctcp *dctcp_data; + u_int win, mss; + + dctcp_data = ccv->cc_data; + win = CCV(ccv, snd_cwnd); + mss = CCV(ccv, t_maxseg); + + switch (type) { + case CC_NDUPACK: + if (!IN_FASTRECOVERY(CCV(ccv, t_flags))) { + if (!IN_CONGRECOVERY(CCV(ccv, t_flags))) { + CCV(ccv, snd_ssthresh) = mss * + max(win / 2 / mss, 2); + dctcp_data->num_cong_events++; + } else { + /* cwnd has already updated as congestion + * recovery. Reverse cwnd value using + * snd_cwnd_prev and recalculate snd_ssthresh + */ + win = CCV(ccv, snd_cwnd_prev); + CCV(ccv, snd_ssthresh) = + max(win / 2 / mss, 2) * mss; + } + ENTER_RECOVERY(CCV(ccv, t_flags)); + } + break; + case CC_ECN: + /* + * Save current snd_cwnd when the host encounters both + * congestion recovery and fast recovery. + */ + CCV(ccv, snd_cwnd_prev) = win; + if (!IN_CONGRECOVERY(CCV(ccv, t_flags))) { + if (V_dctcp_slowstart && + dctcp_data->num_cong_events++ == 0) { + CCV(ccv, snd_ssthresh) = + mss * max(win / 2 / mss, 2); + dctcp_data->alpha = MAX_ALPHA_VALUE; + dctcp_data->bytes_ecn = 0; + dctcp_data->bytes_total = 0; + dctcp_data->save_sndnxt = CCV(ccv, snd_nxt); + } else + CCV(ccv, snd_ssthresh) = max((win - ((win * + dctcp_data->alpha) >> 11)) / mss, 2) * mss; + CCV(ccv, snd_cwnd) = CCV(ccv, snd_ssthresh); + ENTER_CONGRECOVERY(CCV(ccv, t_flags)); + } + dctcp_data->ece_curr = 1; + break; + case CC_RTO: + if (CCV(ccv, t_flags) & TF_ECN_PERMIT) { + CCV(ccv, t_flags) |= TF_ECN_SND_CWR; + dctcp_update_alpha(ccv); + dctcp_data->save_sndnxt += CCV(ccv, t_maxseg); + dctcp_data->num_cong_events++; + } + break; + } +} + +static void +dctcp_conn_init(struct cc_var *ccv) +{ + struct dctcp *dctcp_data; + + dctcp_data = ccv->cc_data; + + if (CCV(ccv, t_flags) & TF_ECN_PERMIT) + dctcp_data->save_sndnxt = CCV(ccv, snd_nxt); +} + +/* + * Perform any necessary tasks before we exit congestion recovery. + */ +static void +dctcp_post_recovery(struct cc_var *ccv) +{ + dctcp_cc_algo.post_recovery = newreno_cc_algo.post_recovery; + + if (CCV(ccv, t_flags) & TF_ECN_PERMIT) + dctcp_update_alpha(ccv); +} + +/* + * Execute an additional ECN processing using ECN field in IP header and the CWR + * bit in TCP header. + * + * delay_ack == 0 - Delayed ACK disabled + * delay_ack == 1 - Delayed ACK enabled + */ + +static void +dctcp_ecnpkt_handler(struct cc_var *ccv) +{ + struct dctcp *dctcp_data; + uint32_t ccflag; + int delay_ack; + + dctcp_data = ccv->cc_data; + ccflag = ccv->flags; + delay_ack = 1; + + /* + * DCTCP responses an ACK immediately when the CE state + * in between this segment and the last segment is not same. + */ + if (ccflag & CCF_IPHDR_CE) { + if (!dctcp_data->ce_prev && (ccflag & CCF_DELACK)) + delay_ack = 0; + dctcp_data->ce_prev = 1; + CCV(ccv, t_flags) |= TF_ECN_SND_ECE; + } else { + if (dctcp_data->ce_prev && (ccflag & CCF_DELACK)) + delay_ack = 0; + dctcp_data->ce_prev = 0; + CCV(ccv, t_flags) &= ~TF_ECN_SND_ECE; + } + + /* DCTCP sets delayed ack when this segment sets the CWR flag. */ + if ((ccflag & CCF_DELACK) && (ccflag & CCF_TCPHDR_CWR)) + delay_ack = 1; + + if (delay_ack == 0) + ccv->flags |= CCF_ACKNOW; + else + ccv->flags &= ~CCF_ACKNOW; +} + +/* + * Update the fraction of marked bytes represented as 'alpha'. + * Also initialize several internal parameters at the end of this function. + */ +static void +dctcp_update_alpha(struct cc_var *ccv) +{ + struct dctcp *dctcp_data; + int alpha_prev; + + dctcp_data = ccv->cc_data; + alpha_prev = dctcp_data->alpha; + dctcp_data->bytes_total = max(dctcp_data->bytes_total, 1); + + /* + * Update alpha: alpha = (1 - g) * alpha + g * F. + * Here: + * g is weight factor + * recommaded to be set to 1/16 + * small g = slow convergence between competitive DCTCP flows + * large g = impacts low utilization of bandwidth at switches + * F is fraction of marked segments in last RTT + * updated every RTT + * Alpha must be round to 0 - MAX_ALPHA_VALUE. + */ + dctcp_data->alpha = min(alpha_prev - (alpha_prev >> V_dctcp_shift_g) + + (dctcp_data->bytes_ecn << (10 - V_dctcp_shift_g)) / + dctcp_data->bytes_total, MAX_ALPHA_VALUE); + + /* Initialize internal parameters for next alpha calculation */ + dctcp_data->bytes_ecn = 0; + dctcp_data->bytes_total = 0; + dctcp_data->save_sndnxt = CCV(ccv, snd_nxt); +} + +static int +dctcp_alpha_handler(SYSCTL_HANDLER_ARGS) +{ + uint32_t new; + int error; + + new = V_dctcp_alpha; + error = sysctl_handle_int(oidp, &new, 0, req); + if (error == 0 && req->newptr != NULL) { + if (CAST_PTR_INT(req->newptr) > 1) + error = EINVAL; + else { + if (new > MAX_ALPHA_VALUE) + V_dctcp_alpha = MAX_ALPHA_VALUE; + else + V_dctcp_alpha = new; + } + } + + return (error); +} + +static int +dctcp_shift_g_handler(SYSCTL_HANDLER_ARGS) +{ + uint32_t new; + int error; + + new = V_dctcp_shift_g; + error = sysctl_handle_int(oidp, &new, 0, req); + if (error == 0 && req->newptr != NULL) { + if (CAST_PTR_INT(req->newptr) > 1) + error = EINVAL; + else + V_dctcp_shift_g = new; + } + + return (error); +} + +static int +dctcp_slowstart_handler(SYSCTL_HANDLER_ARGS) +{ + uint32_t new; + int error; + + new = V_dctcp_slowstart; + error = sysctl_handle_int(oidp, &new, 0, req); + if (error == 0 && req->newptr != NULL) { + if (CAST_PTR_INT(req->newptr) > 1) + error = EINVAL; + else + V_dctcp_slowstart = new; + } + + return (error); +} + +SYSCTL_DECL(_net_inet_tcp_cc_dctcp); +SYSCTL_NODE(_net_inet_tcp_cc, OID_AUTO, dctcp, CTLFLAG_RW, NULL, + "dctcp congestion control related settings"); + +SYSCTL_PROC(_net_inet_tcp_cc_dctcp, OID_AUTO, alpha, + CTLFLAG_VNET|CTLTYPE_UINT|CTLFLAG_RW, &VNET_NAME(dctcp_alpha), 0, + &dctcp_alpha_handler, + "IU", "dctcp alpha parameter"); + +SYSCTL_PROC(_net_inet_tcp_cc_dctcp, OID_AUTO, shift_g, + CTLFLAG_VNET|CTLTYPE_UINT|CTLFLAG_RW, &VNET_NAME(dctcp_shift_g), 4, + &dctcp_shift_g_handler, + "IU", "dctcp shift parameter"); + +SYSCTL_PROC(_net_inet_tcp_cc_dctcp, OID_AUTO, slowstart, + CTLFLAG_VNET|CTLTYPE_UINT|CTLFLAG_RW, &VNET_NAME(dctcp_slowstart), 0, + &dctcp_slowstart_handler, + "IU", "half CWND reduction after the first slow start"); + +DECLARE_CC_MODULE(dctcp, &dctcp_cc_algo); Property changes on: head/sys/netinet/cc/cc_dctcp.c ___________________________________________________________________ Added: svn:eol-style ## -0,0 +1 ## +native \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Added: svn:mime-type ## -0,0 +1 ## +text/plain \ No newline at end of property Index: head/sys/netinet/cc.h =================================================================== --- head/sys/netinet/cc.h (revision 277053) +++ head/sys/netinet/cc.h (revision 277054) @@ -1,167 +1,174 @@ /*- * Copyright (c) 2007-2008 * Swinburne University of Technology, Melbourne, Australia. * Copyright (c) 2009-2010 Lawrence Stewart * Copyright (c) 2010 The FreeBSD Foundation * All rights reserved. * * This software was developed at the Centre for Advanced Internet * Architectures, Swinburne University of Technology, by Lawrence Stewart and * James Healy, made possible in part by a grant from the Cisco University * Research Program Fund at Community Foundation Silicon Valley. * * Portions of this software were developed at the Centre for Advanced * Internet Architectures, Swinburne University of Technology, Melbourne, * Australia by David Hayes under sponsorship from the FreeBSD Foundation. * * 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 THE AUTHOR 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 THE AUTHOR 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. * * $FreeBSD$ */ /* * This software was first released in 2007 by James Healy and Lawrence Stewart * whilst working on the NewTCP research project at Swinburne University of * Technology's Centre for Advanced Internet Architectures, Melbourne, * Australia, which was made possible in part by a grant from the Cisco * University Research Program Fund at Community Foundation Silicon Valley. * More details are available at: * http://caia.swin.edu.au/urp/newtcp/ */ #ifndef _NETINET_CC_H_ #define _NETINET_CC_H_ /* XXX: TCP_CA_NAME_MAX define lives in tcp.h for compat reasons. */ #include /* Global CC vars. */ extern STAILQ_HEAD(cc_head, cc_algo) cc_list; extern const int tcprexmtthresh; extern struct cc_algo newreno_cc_algo; /* Per-netstack bits. */ VNET_DECLARE(struct cc_algo *, default_cc_ptr); #define V_default_cc_ptr VNET(default_cc_ptr) /* Define the new net.inet.tcp.cc sysctl tree. */ SYSCTL_DECL(_net_inet_tcp_cc); /* CC housekeeping functions. */ int cc_register_algo(struct cc_algo *add_cc); int cc_deregister_algo(struct cc_algo *remove_cc); /* * Wrapper around transport structs that contain same-named congestion * control variables. Allows algos to be shared amongst multiple CC aware * transprots. */ struct cc_var { void *cc_data; /* Per-connection private CC algorithm data. */ int bytes_this_ack; /* # bytes acked by the current ACK. */ tcp_seq curack; /* Most recent ACK. */ uint32_t flags; /* Flags for cc_var (see below) */ int type; /* Indicates which ptr is valid in ccvc. */ union ccv_container { struct tcpcb *tcp; struct sctp_nets *sctp; } ccvc; }; /* cc_var flags. */ #define CCF_ABC_SENTAWND 0x0001 /* ABC counted cwnd worth of bytes? */ #define CCF_CWND_LIMITED 0x0002 /* Are we currently cwnd limited? */ +#define CCF_DELACK 0x0004 /* Is this ack delayed? */ +#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? */ /* ACK types passed to the ack_received() hook. */ #define CC_ACK 0x0001 /* Regular in sequence ACK. */ #define CC_DUPACK 0x0002 /* Duplicate ACK. */ #define CC_PARTIALACK 0x0004 /* Not yet. */ #define CC_SACK 0x0008 /* Not yet. */ /* * Congestion signal types passed to the cong_signal() hook. The highest order 8 * bits (0x01000000 - 0x80000000) are reserved for CC algos to declare their own * congestion signal types. */ #define CC_ECN 0x00000001 /* ECN marked packet received. */ #define CC_RTO 0x00000002 /* RTO fired. */ #define CC_RTO_ERR 0x00000004 /* RTO fired in error. */ #define CC_NDUPACK 0x00000008 /* Threshold of dupack's reached. */ #define CC_SIGPRIVMASK 0xFF000000 /* Mask to check if sig is private. */ /* * Structure to hold data and function pointers that together represent a * congestion control algorithm. */ struct cc_algo { char name[TCP_CA_NAME_MAX]; /* Init global module state on kldload. */ int (*mod_init)(void); /* Cleanup global module state on kldunload. */ int (*mod_destroy)(void); /* Init CC state for a new control block. */ int (*cb_init)(struct cc_var *ccv); /* Cleanup CC state for a terminating control block. */ void (*cb_destroy)(struct cc_var *ccv); /* Init variables for a newly established connection. */ void (*conn_init)(struct cc_var *ccv); /* Called on receipt of an ack. */ void (*ack_received)(struct cc_var *ccv, uint16_t type); /* Called on detection of a congestion signal. */ void (*cong_signal)(struct cc_var *ccv, uint32_t type); /* Called after exiting congestion recovery. */ void (*post_recovery)(struct cc_var *ccv); /* Called when data transfer resumes after an idle period. */ void (*after_idle)(struct cc_var *ccv); + + /* Called for an additional ECN processing apart from RFC3168. */ + void (*ecnpkt_handler)(struct cc_var *ccv); STAILQ_ENTRY (cc_algo) entries; }; /* Macro to obtain the CC algo's struct ptr. */ #define CC_ALGO(tp) ((tp)->cc_algo) /* Macro to obtain the CC algo's data ptr. */ #define CC_DATA(tp) ((tp)->ccv->cc_data) /* Macro to obtain the system default CC algo's struct ptr. */ #define CC_DEFAULT() V_default_cc_ptr extern struct rwlock cc_list_lock; #define CC_LIST_LOCK_INIT() rw_init(&cc_list_lock, "cc_list") #define CC_LIST_LOCK_DESTROY() rw_destroy(&cc_list_lock) #define CC_LIST_RLOCK() rw_rlock(&cc_list_lock) #define CC_LIST_RUNLOCK() rw_runlock(&cc_list_lock) #define CC_LIST_WLOCK() rw_wlock(&cc_list_lock) #define CC_LIST_WUNLOCK() rw_wunlock(&cc_list_lock) #define CC_LIST_LOCK_ASSERT() rw_assert(&cc_list_lock, RA_LOCKED) #endif /* _NETINET_CC_H_ */ Index: head/sys/netinet/tcp_input.c =================================================================== --- head/sys/netinet/tcp_input.c (revision 277053) +++ head/sys/netinet/tcp_input.c (revision 277054) @@ -1,3676 +1,3715 @@ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995 * The Regents of the University of California. All rights reserved. * Copyright (c) 2007-2008,2010 * Swinburne University of Technology, Melbourne, Australia. * Copyright (c) 2009-2010 Lawrence Stewart * Copyright (c) 2010 The FreeBSD Foundation * Copyright (c) 2010-2011 Juniper Networks, Inc. * All rights reserved. * * Portions of this software were developed at the Centre for Advanced Internet * Architectures, Swinburne University of Technology, by Lawrence Stewart, * James Healy and David Hayes, made possible in part by a grant from the Cisco * University Research Program Fund at Community Foundation Silicon Valley. * * Portions of this software were developed at the Centre for Advanced * Internet Architectures, Swinburne University of Technology, Melbourne, * Australia by David Hayes under sponsorship from the FreeBSD Foundation. * * Portions of this software were developed by Robert N. M. Watson under * contract to Juniper Networks, Inc. * * 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. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 THE REGENTS 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. * * @(#)tcp_input.c 8.12 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_ipfw.h" /* for ipfw_fwd */ #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_tcpdebug.h" #include #include #include #include #include #include /* for proc0 declaration */ #include #include #include #include #include #include #include #include #include /* before tcp_seq.h, for tcp_random18() */ #include #include #include #include #include #define TCPSTATES /* for logging */ #include #include #include #include #include #include #include #include /* required for icmp_var.h */ #include /* for ICMP_BANDLIM */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef TCPDEBUG #include #endif /* TCPDEBUG */ #ifdef TCP_OFFLOAD #include #endif #ifdef IPSEC #include #include #endif /*IPSEC*/ #include #include const int tcprexmtthresh = 3; int tcp_log_in_vain = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, log_in_vain, CTLFLAG_RW, &tcp_log_in_vain, 0, "Log all incoming TCP segments to closed ports"); VNET_DEFINE(int, blackhole) = 0; #define V_blackhole VNET(blackhole) SYSCTL_INT(_net_inet_tcp, OID_AUTO, blackhole, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(blackhole), 0, "Do not send RST on segments to closed ports"); VNET_DEFINE(int, tcp_delack_enabled) = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, delayed_ack, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_delack_enabled), 0, "Delay ACK to try and piggyback it onto a data packet"); VNET_DEFINE(int, drop_synfin) = 0; #define V_drop_synfin VNET(drop_synfin) SYSCTL_INT(_net_inet_tcp, OID_AUTO, drop_synfin, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(drop_synfin), 0, "Drop TCP packets with SYN+FIN set"); VNET_DEFINE(int, tcp_do_rfc3042) = 1; #define V_tcp_do_rfc3042 VNET(tcp_do_rfc3042) SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3042, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_rfc3042), 0, "Enable RFC 3042 (Limited Transmit)"); VNET_DEFINE(int, tcp_do_rfc3390) = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3390, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_rfc3390), 0, "Enable RFC 3390 (Increasing TCP's Initial Congestion Window)"); SYSCTL_NODE(_net_inet_tcp, OID_AUTO, experimental, CTLFLAG_RW, 0, "Experimental TCP extensions"); VNET_DEFINE(int, tcp_do_initcwnd10) = 1; SYSCTL_INT(_net_inet_tcp_experimental, OID_AUTO, initcwnd10, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_initcwnd10), 0, "Enable RFC 6928 (Increasing initial CWND to 10)"); VNET_DEFINE(int, tcp_do_rfc3465) = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3465, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_rfc3465), 0, "Enable RFC 3465 (Appropriate Byte Counting)"); VNET_DEFINE(int, tcp_abc_l_var) = 2; SYSCTL_INT(_net_inet_tcp, OID_AUTO, abc_l_var, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_abc_l_var), 2, "Cap the max cwnd increment during slow-start to this number of segments"); static SYSCTL_NODE(_net_inet_tcp, OID_AUTO, ecn, CTLFLAG_RW, 0, "TCP ECN"); VNET_DEFINE(int, tcp_do_ecn) = 0; SYSCTL_INT(_net_inet_tcp_ecn, OID_AUTO, enable, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_ecn), 0, "TCP ECN support"); VNET_DEFINE(int, tcp_ecn_maxretries) = 1; SYSCTL_INT(_net_inet_tcp_ecn, OID_AUTO, maxretries, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_ecn_maxretries), 0, "Max retries before giving up on ECN"); VNET_DEFINE(int, tcp_insecure_syn) = 0; #define V_tcp_insecure_syn VNET(tcp_insecure_syn) SYSCTL_INT(_net_inet_tcp, OID_AUTO, insecure_syn, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_insecure_syn), 0, "Follow RFC793 instead of RFC5961 criteria for accepting SYN packets"); VNET_DEFINE(int, tcp_insecure_rst) = 0; #define V_tcp_insecure_rst VNET(tcp_insecure_rst) SYSCTL_INT(_net_inet_tcp, OID_AUTO, insecure_rst, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_insecure_rst), 0, "Follow RFC793 instead of RFC5961 criteria for accepting RST packets"); VNET_DEFINE(int, tcp_recvspace) = 1024*64; #define V_tcp_recvspace VNET(tcp_recvspace) SYSCTL_INT(_net_inet_tcp, TCPCTL_RECVSPACE, recvspace, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_recvspace), 0, "Initial receive socket buffer size"); VNET_DEFINE(int, tcp_do_autorcvbuf) = 1; #define V_tcp_do_autorcvbuf VNET(tcp_do_autorcvbuf) SYSCTL_INT(_net_inet_tcp, OID_AUTO, recvbuf_auto, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_autorcvbuf), 0, "Enable automatic receive buffer sizing"); VNET_DEFINE(int, tcp_autorcvbuf_inc) = 16*1024; #define V_tcp_autorcvbuf_inc VNET(tcp_autorcvbuf_inc) SYSCTL_INT(_net_inet_tcp, OID_AUTO, recvbuf_inc, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_autorcvbuf_inc), 0, "Incrementor step size of automatic receive buffer"); VNET_DEFINE(int, tcp_autorcvbuf_max) = 2*1024*1024; #define V_tcp_autorcvbuf_max VNET(tcp_autorcvbuf_max) SYSCTL_INT(_net_inet_tcp, OID_AUTO, recvbuf_max, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_autorcvbuf_max), 0, "Max size of automatic receive buffer"); VNET_DEFINE(struct inpcbhead, tcb); #define tcb6 tcb /* for KAME src sync over BSD*'s */ VNET_DEFINE(struct inpcbinfo, tcbinfo); static void tcp_dooptions(struct tcpopt *, u_char *, int, int); static void tcp_do_segment(struct mbuf *, struct tcphdr *, struct socket *, struct tcpcb *, int, int, uint8_t, int); static void tcp_dropwithreset(struct mbuf *, struct tcphdr *, struct tcpcb *, int, int); static void tcp_pulloutofband(struct socket *, struct tcphdr *, struct mbuf *, int); static void tcp_xmit_timer(struct tcpcb *, int); static void tcp_newreno_partial_ack(struct tcpcb *, struct tcphdr *); static void inline cc_ack_received(struct tcpcb *tp, struct tcphdr *th, uint16_t type); static void inline cc_conn_init(struct tcpcb *tp); static void inline cc_post_recovery(struct tcpcb *tp, struct tcphdr *th); static void inline hhook_run_tcp_est_in(struct tcpcb *tp, struct tcphdr *th, struct tcpopt *to); /* * TCP statistics are stored in an "array" of counter(9)s. */ VNET_PCPUSTAT_DEFINE(struct tcpstat, tcpstat); VNET_PCPUSTAT_SYSINIT(tcpstat); SYSCTL_VNET_PCPUSTAT(_net_inet_tcp, TCPCTL_STATS, stats, struct tcpstat, tcpstat, "TCP statistics (struct tcpstat, netinet/tcp_var.h)"); #ifdef VIMAGE VNET_PCPUSTAT_SYSUNINIT(tcpstat); #endif /* VIMAGE */ /* * Kernel module interface for updating tcpstat. The argument is an index * into tcpstat treated as an array. */ void kmod_tcpstat_inc(int statnum) { counter_u64_add(VNET(tcpstat)[statnum], 1); } /* * Wrapper for the TCP established input helper hook. */ static void inline hhook_run_tcp_est_in(struct tcpcb *tp, struct tcphdr *th, struct tcpopt *to) { struct tcp_hhook_data hhook_data; if (V_tcp_hhh[HHOOK_TCP_EST_IN]->hhh_nhooks > 0) { hhook_data.tp = tp; hhook_data.th = th; hhook_data.to = to; hhook_run_hooks(V_tcp_hhh[HHOOK_TCP_EST_IN], &hhook_data, tp->osd); } } /* * CC wrapper hook functions */ static void inline cc_ack_received(struct tcpcb *tp, struct tcphdr *th, uint16_t type) { INP_WLOCK_ASSERT(tp->t_inpcb); tp->ccv->bytes_this_ack = BYTES_THIS_ACK(tp, th); if (tp->snd_cwnd <= tp->snd_wnd) tp->ccv->flags |= CCF_CWND_LIMITED; else tp->ccv->flags &= ~CCF_CWND_LIMITED; if (type == CC_ACK) { if (tp->snd_cwnd > tp->snd_ssthresh) { tp->t_bytes_acked += min(tp->ccv->bytes_this_ack, V_tcp_abc_l_var * tp->t_maxseg); if (tp->t_bytes_acked >= tp->snd_cwnd) { tp->t_bytes_acked -= tp->snd_cwnd; tp->ccv->flags |= CCF_ABC_SENTAWND; } } else { tp->ccv->flags &= ~CCF_ABC_SENTAWND; tp->t_bytes_acked = 0; } } if (CC_ALGO(tp)->ack_received != NULL) { /* XXXLAS: Find a way to live without this */ tp->ccv->curack = th->th_ack; CC_ALGO(tp)->ack_received(tp->ccv, type); } } static void inline cc_conn_init(struct tcpcb *tp) { struct hc_metrics_lite metrics; struct inpcb *inp = tp->t_inpcb; int rtt; INP_WLOCK_ASSERT(tp->t_inpcb); tcp_hc_get(&inp->inp_inc, &metrics); if (tp->t_srtt == 0 && (rtt = metrics.rmx_rtt)) { tp->t_srtt = rtt; tp->t_rttbest = tp->t_srtt + TCP_RTT_SCALE; TCPSTAT_INC(tcps_usedrtt); if (metrics.rmx_rttvar) { tp->t_rttvar = metrics.rmx_rttvar; TCPSTAT_INC(tcps_usedrttvar); } else { /* default variation is +- 1 rtt */ tp->t_rttvar = tp->t_srtt * TCP_RTTVAR_SCALE / TCP_RTT_SCALE; } TCPT_RANGESET(tp->t_rxtcur, ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1, tp->t_rttmin, TCPTV_REXMTMAX); } if (metrics.rmx_ssthresh) { /* * There's some sort of gateway or interface * buffer limit on the path. Use this to set * the slow start threshhold, but set the * threshold to no less than 2*mss. */ tp->snd_ssthresh = max(2 * tp->t_maxseg, metrics.rmx_ssthresh); TCPSTAT_INC(tcps_usedssthresh); } /* * Set the initial slow-start flight size. * * RFC5681 Section 3.1 specifies the default conservative values. * RFC3390 specifies slightly more aggressive values. * RFC6928 increases it to ten segments. * * If a SYN or SYN/ACK was lost and retransmitted, we have to * reduce the initial CWND to one segment as congestion is likely * requiring us to be cautious. */ if (tp->snd_cwnd == 1) tp->snd_cwnd = tp->t_maxseg; /* SYN(-ACK) lost */ else if (V_tcp_do_initcwnd10) tp->snd_cwnd = min(10 * tp->t_maxseg, max(2 * tp->t_maxseg, 14600)); else if (V_tcp_do_rfc3390) tp->snd_cwnd = min(4 * tp->t_maxseg, max(2 * tp->t_maxseg, 4380)); else { /* Per RFC5681 Section 3.1 */ if (tp->t_maxseg > 2190) tp->snd_cwnd = 2 * tp->t_maxseg; else if (tp->t_maxseg > 1095) tp->snd_cwnd = 3 * tp->t_maxseg; else tp->snd_cwnd = 4 * tp->t_maxseg; } if (CC_ALGO(tp)->conn_init != NULL) CC_ALGO(tp)->conn_init(tp->ccv); } void inline cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type) { INP_WLOCK_ASSERT(tp->t_inpcb); switch(type) { case CC_NDUPACK: if (!IN_FASTRECOVERY(tp->t_flags)) { tp->snd_recover = tp->snd_max; if (tp->t_flags & TF_ECN_PERMIT) tp->t_flags |= TF_ECN_SND_CWR; } break; case CC_ECN: if (!IN_CONGRECOVERY(tp->t_flags)) { TCPSTAT_INC(tcps_ecn_rcwnd); tp->snd_recover = tp->snd_max; if (tp->t_flags & TF_ECN_PERMIT) tp->t_flags |= TF_ECN_SND_CWR; } break; case CC_RTO: tp->t_dupacks = 0; tp->t_bytes_acked = 0; EXIT_RECOVERY(tp->t_flags); tp->snd_ssthresh = max(2, min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg) * tp->t_maxseg; tp->snd_cwnd = tp->t_maxseg; break; case CC_RTO_ERR: TCPSTAT_INC(tcps_sndrexmitbad); /* RTO was unnecessary, so reset everything. */ tp->snd_cwnd = tp->snd_cwnd_prev; tp->snd_ssthresh = tp->snd_ssthresh_prev; tp->snd_recover = tp->snd_recover_prev; if (tp->t_flags & TF_WASFRECOVERY) ENTER_FASTRECOVERY(tp->t_flags); if (tp->t_flags & TF_WASCRECOVERY) ENTER_CONGRECOVERY(tp->t_flags); tp->snd_nxt = tp->snd_max; tp->t_flags &= ~TF_PREVVALID; tp->t_badrxtwin = 0; break; } if (CC_ALGO(tp)->cong_signal != NULL) { if (th != NULL) tp->ccv->curack = th->th_ack; CC_ALGO(tp)->cong_signal(tp->ccv, type); } } static void inline cc_post_recovery(struct tcpcb *tp, struct tcphdr *th) { INP_WLOCK_ASSERT(tp->t_inpcb); /* XXXLAS: KASSERT that we're in recovery? */ if (CC_ALGO(tp)->post_recovery != NULL) { tp->ccv->curack = th->th_ack; CC_ALGO(tp)->post_recovery(tp->ccv); } /* XXXLAS: EXIT_RECOVERY ? */ tp->t_bytes_acked = 0; } #ifdef TCP_SIGNATURE static inline int tcp_signature_verify_input(struct mbuf *m, int off0, int tlen, int optlen, struct tcpopt *to, struct tcphdr *th, u_int tcpbflag) { int ret; tcp_fields_to_net(th); ret = tcp_signature_verify(m, off0, tlen, optlen, to, th, tcpbflag); tcp_fields_to_host(th); return (ret); } #endif /* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */ #ifdef INET6 #define ND6_HINT(tp) \ do { \ if ((tp) && (tp)->t_inpcb && \ ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \ nd6_nud_hint(NULL, NULL, 0); \ } while (0) #else #define ND6_HINT(tp) #endif /* * Indicate whether this ack should be delayed. We can delay the ack if * following conditions are met: * - There is no delayed ack timer in progress. * - Our last ack wasn't a 0-sized window. We never want to delay * the ack that opens up a 0-sized window. * - LRO wasn't used for this segment. We make sure by checking that the * segment size is not larger than the MSS. * - Delayed acks are enabled or this is a half-synchronized T/TCP * connection. */ #define DELAY_ACK(tp, tlen) \ ((!tcp_timer_active(tp, TT_DELACK) && \ (tp->t_flags & TF_RXWIN0SENT) == 0) && \ (tlen <= tp->t_maxopd) && \ (V_tcp_delack_enabled || (tp->t_flags & TF_NEEDSYN))) +static void inline +cc_ecnpkt_handler(struct tcpcb *tp, struct tcphdr *th, uint8_t iptos) +{ + INP_WLOCK_ASSERT(tp->t_inpcb); + + if (CC_ALGO(tp)->ecnpkt_handler != NULL) { + switch (iptos & IPTOS_ECN_MASK) { + case IPTOS_ECN_CE: + tp->ccv->flags |= CCF_IPHDR_CE; + break; + case IPTOS_ECN_ECT0: + tp->ccv->flags &= ~CCF_IPHDR_CE; + break; + case IPTOS_ECN_ECT1: + tp->ccv->flags &= ~CCF_IPHDR_CE; + break; + } + + if (th->th_flags & TH_CWR) + tp->ccv->flags |= CCF_TCPHDR_CWR; + else + tp->ccv->flags &= ~CCF_TCPHDR_CWR; + + if (tp->t_flags & TF_DELACK) + tp->ccv->flags |= CCF_DELACK; + else + tp->ccv->flags &= ~CCF_DELACK; + + CC_ALGO(tp)->ecnpkt_handler(tp->ccv); + + if (tp->ccv->flags & CCF_ACKNOW) + tcp_timer_activate(tp, TT_DELACK, tcp_delacktime); + } +} + /* * TCP input handling is split into multiple parts: * tcp6_input is a thin wrapper around tcp_input for the extended * ip6_protox[] call format in ip6_input * tcp_input handles primary segment validation, inpcb lookup and * SYN processing on listen sockets * tcp_do_segment processes the ACK and text of the segment for * establishing, established and closing connections */ #ifdef INET6 int tcp6_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m = *mp; struct in6_ifaddr *ia6; struct ip6_hdr *ip6; IP6_EXTHDR_CHECK(m, *offp, sizeof(struct tcphdr), IPPROTO_DONE); /* * draft-itojun-ipv6-tcp-to-anycast * better place to put this in? */ ip6 = mtod(m, struct ip6_hdr *); ia6 = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */); if (ia6 && (ia6->ia6_flags & IN6_IFF_ANYCAST)) { struct ip6_hdr *ip6; ifa_free(&ia6->ia_ifa); ip6 = mtod(m, struct ip6_hdr *); icmp6_error(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_ADDR, (caddr_t)&ip6->ip6_dst - (caddr_t)ip6); return (IPPROTO_DONE); } if (ia6) ifa_free(&ia6->ia_ifa); return (tcp_input(mp, offp, proto)); } #endif /* INET6 */ int tcp_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m = *mp; struct tcphdr *th = NULL; struct ip *ip = NULL; struct inpcb *inp = NULL; struct tcpcb *tp = NULL; struct socket *so = NULL; u_char *optp = NULL; int off0; int optlen = 0; #ifdef INET int len; #endif int tlen = 0, off; int drop_hdrlen; int thflags; int rstreason = 0; /* For badport_bandlim accounting purposes */ #ifdef TCP_SIGNATURE uint8_t sig_checked = 0; #endif uint8_t iptos = 0; struct m_tag *fwd_tag = NULL; #ifdef INET6 struct ip6_hdr *ip6 = NULL; int isipv6; #else const void *ip6 = NULL; #endif /* INET6 */ struct tcpopt to; /* options in this segment */ char *s = NULL; /* address and port logging */ int ti_locked; #define TI_UNLOCKED 1 #define TI_WLOCKED 2 #ifdef TCPDEBUG /* * The size of tcp_saveipgen must be the size of the max ip header, * now IPv6. */ u_char tcp_saveipgen[IP6_HDR_LEN]; struct tcphdr tcp_savetcp; short ostate = 0; #endif #ifdef INET6 isipv6 = (mtod(m, struct ip *)->ip_v == 6) ? 1 : 0; #endif off0 = *offp; m = *mp; *mp = NULL; to.to_flags = 0; TCPSTAT_INC(tcps_rcvtotal); #ifdef INET6 if (isipv6) { /* IP6_EXTHDR_CHECK() is already done at tcp6_input(). */ if (m->m_len < (sizeof(*ip6) + sizeof(*th))) { m = m_pullup(m, sizeof(*ip6) + sizeof(*th)); if (m == NULL) { TCPSTAT_INC(tcps_rcvshort); return (IPPROTO_DONE); } } ip6 = mtod(m, struct ip6_hdr *); th = (struct tcphdr *)((caddr_t)ip6 + off0); tlen = sizeof(*ip6) + ntohs(ip6->ip6_plen) - off0; if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID_IPV6) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) th->th_sum = m->m_pkthdr.csum_data; else th->th_sum = in6_cksum_pseudo(ip6, tlen, IPPROTO_TCP, m->m_pkthdr.csum_data); th->th_sum ^= 0xffff; } else th->th_sum = in6_cksum(m, IPPROTO_TCP, off0, tlen); if (th->th_sum) { TCPSTAT_INC(tcps_rcvbadsum); goto drop; } /* * Be proactive about unspecified IPv6 address in source. * As we use all-zero to indicate unbounded/unconnected pcb, * unspecified IPv6 address can be used to confuse us. * * Note that packets with unspecified IPv6 destination is * already dropped in ip6_input. */ if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { /* XXX stat */ goto drop; } } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { /* * Get IP and TCP header together in first mbuf. * Note: IP leaves IP header in first mbuf. */ if (off0 > sizeof (struct ip)) { ip_stripoptions(m); off0 = sizeof(struct ip); } if (m->m_len < sizeof (struct tcpiphdr)) { if ((m = m_pullup(m, sizeof (struct tcpiphdr))) == NULL) { TCPSTAT_INC(tcps_rcvshort); return (IPPROTO_DONE); } } ip = mtod(m, struct ip *); th = (struct tcphdr *)((caddr_t)ip + off0); tlen = ntohs(ip->ip_len) - off0; if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) th->th_sum = m->m_pkthdr.csum_data; else th->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htonl(m->m_pkthdr.csum_data + tlen + IPPROTO_TCP)); th->th_sum ^= 0xffff; } else { struct ipovly *ipov = (struct ipovly *)ip; /* * Checksum extended TCP header and data. */ len = off0 + tlen; bzero(ipov->ih_x1, sizeof(ipov->ih_x1)); ipov->ih_len = htons(tlen); th->th_sum = in_cksum(m, len); /* Reset length for SDT probes. */ ip->ip_len = htons(tlen + off0); } if (th->th_sum) { TCPSTAT_INC(tcps_rcvbadsum); goto drop; } /* Re-initialization for later version check */ ip->ip_v = IPVERSION; } #endif /* INET */ #ifdef INET6 if (isipv6) iptos = (ntohl(ip6->ip6_flow) >> 20) & 0xff; #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET iptos = ip->ip_tos; #endif /* * Check that TCP offset makes sense, * pull out TCP options and adjust length. XXX */ off = th->th_off << 2; if (off < sizeof (struct tcphdr) || off > tlen) { TCPSTAT_INC(tcps_rcvbadoff); goto drop; } tlen -= off; /* tlen is used instead of ti->ti_len */ if (off > sizeof (struct tcphdr)) { #ifdef INET6 if (isipv6) { IP6_EXTHDR_CHECK(m, off0, off, IPPROTO_DONE); ip6 = mtod(m, struct ip6_hdr *); th = (struct tcphdr *)((caddr_t)ip6 + off0); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { if (m->m_len < sizeof(struct ip) + off) { if ((m = m_pullup(m, sizeof (struct ip) + off)) == NULL) { TCPSTAT_INC(tcps_rcvshort); return (IPPROTO_DONE); } ip = mtod(m, struct ip *); th = (struct tcphdr *)((caddr_t)ip + off0); } } #endif optlen = off - sizeof (struct tcphdr); optp = (u_char *)(th + 1); } thflags = th->th_flags; /* * Convert TCP protocol specific fields to host format. */ tcp_fields_to_host(th); /* * Delay dropping TCP, IP headers, IPv6 ext headers, and TCP options. */ drop_hdrlen = off0 + off; /* * Locate pcb for segment; if we're likely to add or remove a * connection then first acquire pcbinfo lock. There are three cases * where we might discover later we need a write lock despite the * flags: ACKs moving a connection out of the syncache, ACKs for a * connection in TIMEWAIT and SYNs not targeting a listening socket. */ if ((thflags & (TH_FIN | TH_RST)) != 0) { INP_INFO_WLOCK(&V_tcbinfo); ti_locked = TI_WLOCKED; } else ti_locked = TI_UNLOCKED; /* * Grab info from PACKET_TAG_IPFORWARD tag prepended to the chain. */ if ( #ifdef INET6 (isipv6 && (m->m_flags & M_IP6_NEXTHOP)) #ifdef INET || (!isipv6 && (m->m_flags & M_IP_NEXTHOP)) #endif #endif #if defined(INET) && !defined(INET6) (m->m_flags & M_IP_NEXTHOP) #endif ) fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL); findpcb: #ifdef INVARIANTS if (ti_locked == TI_WLOCKED) { INP_INFO_WLOCK_ASSERT(&V_tcbinfo); } else { INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); } #endif #ifdef INET6 if (isipv6 && fwd_tag != NULL) { struct sockaddr_in6 *next_hop6; next_hop6 = (struct sockaddr_in6 *)(fwd_tag + 1); /* * Transparently forwarded. Pretend to be the destination. * Already got one like this? */ inp = in6_pcblookup_mbuf(&V_tcbinfo, &ip6->ip6_src, th->th_sport, &ip6->ip6_dst, th->th_dport, INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif, m); if (!inp) { /* * It's new. Try to find the ambushing socket. * Because we've rewritten the destination address, * any hardware-generated hash is ignored. */ inp = in6_pcblookup(&V_tcbinfo, &ip6->ip6_src, th->th_sport, &next_hop6->sin6_addr, next_hop6->sin6_port ? ntohs(next_hop6->sin6_port) : th->th_dport, INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif); } } else if (isipv6) { inp = in6_pcblookup_mbuf(&V_tcbinfo, &ip6->ip6_src, th->th_sport, &ip6->ip6_dst, th->th_dport, INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif, m); } #endif /* INET6 */ #if defined(INET6) && defined(INET) else #endif #ifdef INET if (fwd_tag != NULL) { struct sockaddr_in *next_hop; next_hop = (struct sockaddr_in *)(fwd_tag+1); /* * Transparently forwarded. Pretend to be the destination. * already got one like this? */ inp = in_pcblookup_mbuf(&V_tcbinfo, ip->ip_src, th->th_sport, ip->ip_dst, th->th_dport, INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif, m); if (!inp) { /* * It's new. Try to find the ambushing socket. * Because we've rewritten the destination address, * any hardware-generated hash is ignored. */ inp = in_pcblookup(&V_tcbinfo, ip->ip_src, th->th_sport, next_hop->sin_addr, next_hop->sin_port ? ntohs(next_hop->sin_port) : th->th_dport, INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif); } } else inp = in_pcblookup_mbuf(&V_tcbinfo, ip->ip_src, th->th_sport, ip->ip_dst, th->th_dport, INPLOOKUP_WILDCARD | INPLOOKUP_WLOCKPCB, m->m_pkthdr.rcvif, m); #endif /* INET */ /* * If the INPCB does not exist then all data in the incoming * segment is discarded and an appropriate RST is sent back. * XXX MRT Send RST using which routing table? */ if (inp == NULL) { /* * Log communication attempts to ports that are not * in use. */ if ((tcp_log_in_vain == 1 && (thflags & TH_SYN)) || tcp_log_in_vain == 2) { if ((s = tcp_log_vain(NULL, th, (void *)ip, ip6))) log(LOG_INFO, "%s; %s: Connection attempt " "to closed port\n", s, __func__); } /* * When blackholing do not respond with a RST but * completely ignore the segment and drop it. */ if ((V_blackhole == 1 && (thflags & TH_SYN)) || V_blackhole == 2) goto dropunlock; rstreason = BANDLIM_RST_CLOSEDPORT; goto dropwithreset; } INP_WLOCK_ASSERT(inp); if ((inp->inp_flowtype == M_HASHTYPE_NONE) && (M_HASHTYPE_GET(m) != M_HASHTYPE_NONE) && ((inp->inp_socket == NULL) || (inp->inp_socket->so_options & SO_ACCEPTCONN) == 0)) { inp->inp_flowid = m->m_pkthdr.flowid; inp->inp_flowtype = M_HASHTYPE_GET(m); } #ifdef IPSEC #ifdef INET6 if (isipv6 && ipsec6_in_reject(m, inp)) { goto dropunlock; } else #endif /* INET6 */ if (ipsec4_in_reject(m, inp) != 0) { goto dropunlock; } #endif /* IPSEC */ /* * Check the minimum TTL for socket. */ if (inp->inp_ip_minttl != 0) { #ifdef INET6 if (isipv6 && inp->inp_ip_minttl > ip6->ip6_hlim) goto dropunlock; else #endif if (inp->inp_ip_minttl > ip->ip_ttl) goto dropunlock; } /* * A previous connection in TIMEWAIT state is supposed to catch stray * or duplicate segments arriving late. If this segment was a * legitimate new connection attempt, the old INPCB gets removed and * we can try again to find a listening socket. * * At this point, due to earlier optimism, we may hold only an inpcb * lock, and not the inpcbinfo write lock. If so, we need to try to * acquire it, or if that fails, acquire a reference on the inpcb, * drop all locks, acquire a global write lock, and then re-acquire * the inpcb lock. We may at that point discover that another thread * has tried to free the inpcb, in which case we need to loop back * and try to find a new inpcb to deliver to. * * XXXRW: It may be time to rethink timewait locking. */ relocked: if (inp->inp_flags & INP_TIMEWAIT) { if (ti_locked == TI_UNLOCKED) { if (INP_INFO_TRY_WLOCK(&V_tcbinfo) == 0) { in_pcbref(inp); INP_WUNLOCK(inp); INP_INFO_WLOCK(&V_tcbinfo); ti_locked = TI_WLOCKED; INP_WLOCK(inp); if (in_pcbrele_wlocked(inp)) { inp = NULL; goto findpcb; } } else ti_locked = TI_WLOCKED; } INP_INFO_WLOCK_ASSERT(&V_tcbinfo); if (thflags & TH_SYN) tcp_dooptions(&to, optp, optlen, TO_SYN); /* * NB: tcp_twcheck unlocks the INP and frees the mbuf. */ if (tcp_twcheck(inp, &to, th, m, tlen)) goto findpcb; INP_INFO_WUNLOCK(&V_tcbinfo); return (IPPROTO_DONE); } /* * The TCPCB may no longer exist if the connection is winding * down or it is in the CLOSED state. Either way we drop the * segment and send an appropriate response. */ tp = intotcpcb(inp); if (tp == NULL || tp->t_state == TCPS_CLOSED) { rstreason = BANDLIM_RST_CLOSEDPORT; goto dropwithreset; } #ifdef TCP_OFFLOAD if (tp->t_flags & TF_TOE) { tcp_offload_input(tp, m); m = NULL; /* consumed by the TOE driver */ goto dropunlock; } #endif /* * We've identified a valid inpcb, but it could be that we need an * inpcbinfo write lock but don't hold it. In this case, attempt to * acquire using the same strategy as the TIMEWAIT case above. If we * relock, we have to jump back to 'relocked' as the connection might * now be in TIMEWAIT. */ #ifdef INVARIANTS if ((thflags & (TH_FIN | TH_RST)) != 0) INP_INFO_WLOCK_ASSERT(&V_tcbinfo); #endif if (!((tp->t_state == TCPS_ESTABLISHED && (thflags & TH_SYN) == 0) || (tp->t_state == TCPS_LISTEN && (thflags & TH_SYN)))) { if (ti_locked == TI_UNLOCKED) { if (INP_INFO_TRY_WLOCK(&V_tcbinfo) == 0) { in_pcbref(inp); INP_WUNLOCK(inp); INP_INFO_WLOCK(&V_tcbinfo); ti_locked = TI_WLOCKED; INP_WLOCK(inp); if (in_pcbrele_wlocked(inp)) { inp = NULL; goto findpcb; } goto relocked; } else ti_locked = TI_WLOCKED; } INP_INFO_WLOCK_ASSERT(&V_tcbinfo); } #ifdef MAC INP_WLOCK_ASSERT(inp); if (mac_inpcb_check_deliver(inp, m)) goto dropunlock; #endif so = inp->inp_socket; KASSERT(so != NULL, ("%s: so == NULL", __func__)); #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) { ostate = tp->t_state; #ifdef INET6 if (isipv6) { bcopy((char *)ip6, (char *)tcp_saveipgen, sizeof(*ip6)); } else #endif bcopy((char *)ip, (char *)tcp_saveipgen, sizeof(*ip)); tcp_savetcp = *th; } #endif /* TCPDEBUG */ /* * When the socket is accepting connections (the INPCB is in LISTEN * state) we look into the SYN cache if this is a new connection * attempt or the completion of a previous one. */ if (so->so_options & SO_ACCEPTCONN) { struct in_conninfo inc; KASSERT(tp->t_state == TCPS_LISTEN, ("%s: so accepting but " "tp not listening", __func__)); bzero(&inc, sizeof(inc)); #ifdef INET6 if (isipv6) { inc.inc_flags |= INC_ISIPV6; inc.inc6_faddr = ip6->ip6_src; inc.inc6_laddr = ip6->ip6_dst; } else #endif { inc.inc_faddr = ip->ip_src; inc.inc_laddr = ip->ip_dst; } inc.inc_fport = th->th_sport; inc.inc_lport = th->th_dport; inc.inc_fibnum = so->so_fibnum; /* * Check for an existing connection attempt in syncache if * the flag is only ACK. A successful lookup creates a new * socket appended to the listen queue in SYN_RECEIVED state. */ if ((thflags & (TH_RST|TH_ACK|TH_SYN)) == TH_ACK) { INP_INFO_WLOCK_ASSERT(&V_tcbinfo); /* * Parse the TCP options here because * syncookies need access to the reflected * timestamp. */ tcp_dooptions(&to, optp, optlen, 0); /* * NB: syncache_expand() doesn't unlock * inp and tcpinfo locks. */ if (!syncache_expand(&inc, &to, th, &so, m)) { /* * No syncache entry or ACK was not * for our SYN/ACK. Send a RST. * NB: syncache did its own logging * of the failure cause. */ rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } if (so == NULL) { /* * We completed the 3-way handshake * but could not allocate a socket * either due to memory shortage, * listen queue length limits or * global socket limits. Send RST * or wait and have the remote end * retransmit the ACK for another * try. */ if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Socket allocation failed due to " "limits or memory shortage, %s\n", s, __func__, V_tcp_sc_rst_sock_fail ? "sending RST" : "try again"); if (V_tcp_sc_rst_sock_fail) { rstreason = BANDLIM_UNLIMITED; goto dropwithreset; } else goto dropunlock; } /* * Socket is created in state SYN_RECEIVED. * Unlock the listen socket, lock the newly * created socket and update the tp variable. */ INP_WUNLOCK(inp); /* listen socket */ inp = sotoinpcb(so); INP_WLOCK(inp); /* new connection */ tp = intotcpcb(inp); KASSERT(tp->t_state == TCPS_SYN_RECEIVED, ("%s: ", __func__)); #ifdef TCP_SIGNATURE if (sig_checked == 0) { tcp_dooptions(&to, optp, optlen, (thflags & TH_SYN) ? TO_SYN : 0); if (!tcp_signature_verify_input(m, off0, tlen, optlen, &to, th, tp->t_flags)) { /* * In SYN_SENT state if it receives an * RST, it is allowed for further * processing. */ if ((thflags & TH_RST) == 0 || (tp->t_state == TCPS_SYN_SENT) == 0) goto dropunlock; } sig_checked = 1; } #endif /* * Process the segment and the data it * contains. tcp_do_segment() consumes * the mbuf chain and unlocks the inpcb. */ tcp_do_segment(m, th, so, tp, drop_hdrlen, tlen, iptos, ti_locked); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); return (IPPROTO_DONE); } /* * Segment flag validation for new connection attempts: * * Our (SYN|ACK) response was rejected. * Check with syncache and remove entry to prevent * retransmits. * * NB: syncache_chkrst does its own logging of failure * causes. */ if (thflags & TH_RST) { syncache_chkrst(&inc, th); goto dropunlock; } /* * We can't do anything without SYN. */ if ((thflags & TH_SYN) == 0) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "SYN is missing, segment ignored\n", s, __func__); TCPSTAT_INC(tcps_badsyn); goto dropunlock; } /* * (SYN|ACK) is bogus on a listen socket. */ if (thflags & TH_ACK) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "SYN|ACK invalid, segment rejected\n", s, __func__); syncache_badack(&inc); /* XXX: Not needed! */ TCPSTAT_INC(tcps_badsyn); rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } /* * If the drop_synfin option is enabled, drop all * segments with both the SYN and FIN bits set. * This prevents e.g. nmap from identifying the * TCP/IP stack. * XXX: Poor reasoning. nmap has other methods * and is constantly refining its stack detection * strategies. * XXX: This is a violation of the TCP specification * and was used by RFC1644. */ if ((thflags & TH_FIN) && V_drop_synfin) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "SYN|FIN segment ignored (based on " "sysctl setting)\n", s, __func__); TCPSTAT_INC(tcps_badsyn); goto dropunlock; } /* * Segment's flags are (SYN) or (SYN|FIN). * * TH_PUSH, TH_URG, TH_ECE, TH_CWR are ignored * as they do not affect the state of the TCP FSM. * The data pointed to by TH_URG and th_urp is ignored. */ KASSERT((thflags & (TH_RST|TH_ACK)) == 0, ("%s: Listen socket: TH_RST or TH_ACK set", __func__)); KASSERT(thflags & (TH_SYN), ("%s: Listen socket: TH_SYN not set", __func__)); #ifdef INET6 /* * If deprecated address is forbidden, * we do not accept SYN to deprecated interface * address to prevent any new inbound connection from * getting established. * When we do not accept SYN, we send a TCP RST, * with deprecated source address (instead of dropping * it). We compromise it as it is much better for peer * to send a RST, and RST will be the final packet * for the exchange. * * If we do not forbid deprecated addresses, we accept * the SYN packet. RFC2462 does not suggest dropping * SYN in this case. * If we decipher RFC2462 5.5.4, it says like this: * 1. use of deprecated addr with existing * communication is okay - "SHOULD continue to be * used" * 2. use of it with new communication: * (2a) "SHOULD NOT be used if alternate address * with sufficient scope is available" * (2b) nothing mentioned otherwise. * Here we fall into (2b) case as we have no choice in * our source address selection - we must obey the peer. * * The wording in RFC2462 is confusing, and there are * multiple description text for deprecated address * handling - worse, they are not exactly the same. * I believe 5.5.4 is the best one, so we follow 5.5.4. */ if (isipv6 && !V_ip6_use_deprecated) { struct in6_ifaddr *ia6; ia6 = in6ifa_ifwithaddr(&ip6->ip6_dst, 0 /* XXX */); if (ia6 != NULL && (ia6->ia6_flags & IN6_IFF_DEPRECATED)) { ifa_free(&ia6->ia_ifa); if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt to deprecated " "IPv6 address rejected\n", s, __func__); rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } if (ia6) ifa_free(&ia6->ia_ifa); } #endif /* INET6 */ /* * Basic sanity checks on incoming SYN requests: * Don't respond if the destination is a link layer * broadcast according to RFC1122 4.2.3.10, p. 104. * If it is from this socket it must be forged. * Don't respond if the source or destination is a * global or subnet broad- or multicast address. * Note that it is quite possible to receive unicast * link-layer packets with a broadcast IP address. Use * in_broadcast() to find them. */ if (m->m_flags & (M_BCAST|M_MCAST)) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt from broad- or multicast " "link layer address ignored\n", s, __func__); goto dropunlock; } #ifdef INET6 if (isipv6) { if (th->th_dport == th->th_sport && IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &ip6->ip6_src)) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt to/from self " "ignored\n", s, __func__); goto dropunlock; } if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || IN6_IS_ADDR_MULTICAST(&ip6->ip6_src)) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt from/to multicast " "address ignored\n", s, __func__); goto dropunlock; } } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { if (th->th_dport == th->th_sport && ip->ip_dst.s_addr == ip->ip_src.s_addr) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt from/to self " "ignored\n", s, __func__); goto dropunlock; } if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) || IN_MULTICAST(ntohl(ip->ip_src.s_addr)) || ip->ip_src.s_addr == htonl(INADDR_BROADCAST) || in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) { if ((s = tcp_log_addrs(&inc, th, NULL, NULL))) log(LOG_DEBUG, "%s; %s: Listen socket: " "Connection attempt from/to broad- " "or multicast address ignored\n", s, __func__); goto dropunlock; } } #endif /* * SYN appears to be valid. Create compressed TCP state * for syncache. */ #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif tcp_dooptions(&to, optp, optlen, TO_SYN); syncache_add(&inc, &to, th, inp, &so, m, NULL, NULL); /* * Entry added to syncache and mbuf consumed. * Only the listen socket is unlocked by syncache_add(). */ if (ti_locked == TI_WLOCKED) { INP_INFO_WUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; } INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); return (IPPROTO_DONE); } else if (tp->t_state == TCPS_LISTEN) { /* * When a listen socket is torn down the SO_ACCEPTCONN * flag is removed first while connections are drained * from the accept queue in a unlock/lock cycle of the * ACCEPT_LOCK, opening a race condition allowing a SYN * attempt go through unhandled. */ goto dropunlock; } #ifdef TCP_SIGNATURE if (sig_checked == 0) { tcp_dooptions(&to, optp, optlen, (thflags & TH_SYN) ? TO_SYN : 0); if (!tcp_signature_verify_input(m, off0, tlen, optlen, &to, th, tp->t_flags)) { /* * In SYN_SENT state if it receives an RST, it is * allowed for further processing. */ if ((thflags & TH_RST) == 0 || (tp->t_state == TCPS_SYN_SENT) == 0) goto dropunlock; } sig_checked = 1; } #endif TCP_PROBE5(receive, NULL, tp, mtod(m, const char *), tp, th); /* * Segment belongs to a connection in SYN_SENT, ESTABLISHED or later * state. tcp_do_segment() always consumes the mbuf chain, unlocks * the inpcb, and unlocks pcbinfo. */ tcp_do_segment(m, th, so, tp, drop_hdrlen, tlen, iptos, ti_locked); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); return (IPPROTO_DONE); dropwithreset: TCP_PROBE5(receive, NULL, tp, mtod(m, const char *), tp, th); if (ti_locked == TI_WLOCKED) { INP_INFO_WUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; } #ifdef INVARIANTS else { KASSERT(ti_locked == TI_UNLOCKED, ("%s: dropwithreset " "ti_locked: %d", __func__, ti_locked)); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); } #endif if (inp != NULL) { tcp_dropwithreset(m, th, tp, tlen, rstreason); INP_WUNLOCK(inp); } else tcp_dropwithreset(m, th, NULL, tlen, rstreason); m = NULL; /* mbuf chain got consumed. */ goto drop; dropunlock: if (m != NULL) TCP_PROBE5(receive, NULL, tp, mtod(m, const char *), tp, th); if (ti_locked == TI_WLOCKED) { INP_INFO_WUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; } #ifdef INVARIANTS else { KASSERT(ti_locked == TI_UNLOCKED, ("%s: dropunlock " "ti_locked: %d", __func__, ti_locked)); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); } #endif if (inp != NULL) INP_WUNLOCK(inp); drop: INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); if (s != NULL) free(s, M_TCPLOG); if (m != NULL) m_freem(m); return (IPPROTO_DONE); } static void tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so, struct tcpcb *tp, int drop_hdrlen, int tlen, uint8_t iptos, int ti_locked) { int thflags, acked, ourfinisacked, needoutput = 0; int rstreason, todrop, win; u_long tiwin; char *s; struct in_conninfo *inc; struct mbuf *mfree; struct tcpopt to; #ifdef TCPDEBUG /* * The size of tcp_saveipgen must be the size of the max ip header, * now IPv6. */ u_char tcp_saveipgen[IP6_HDR_LEN]; struct tcphdr tcp_savetcp; short ostate = 0; #endif thflags = th->th_flags; inc = &tp->t_inpcb->inp_inc; tp->sackhint.last_sack_ack = 0; /* * If this is either a state-changing packet or current state isn't * established, we require a write lock on tcbinfo. Otherwise, we * allow the tcbinfo to be in either alocked or unlocked, as the * caller may have unnecessarily acquired a write lock due to a race. */ if ((thflags & (TH_SYN | TH_FIN | TH_RST)) != 0 || tp->t_state != TCPS_ESTABLISHED) { KASSERT(ti_locked == TI_WLOCKED, ("%s ti_locked %d for " "SYN/FIN/RST/!EST", __func__, ti_locked)); INP_INFO_WLOCK_ASSERT(&V_tcbinfo); } else { #ifdef INVARIANTS if (ti_locked == TI_WLOCKED) INP_INFO_WLOCK_ASSERT(&V_tcbinfo); else { KASSERT(ti_locked == TI_UNLOCKED, ("%s: EST " "ti_locked: %d", __func__, ti_locked)); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); } #endif } INP_WLOCK_ASSERT(tp->t_inpcb); KASSERT(tp->t_state > TCPS_LISTEN, ("%s: TCPS_LISTEN", __func__)); KASSERT(tp->t_state != TCPS_TIME_WAIT, ("%s: TCPS_TIME_WAIT", __func__)); /* * Segment received on connection. * Reset idle time and keep-alive timer. * XXX: This should be done after segment * validation to ignore broken/spoofed segs. */ tp->t_rcvtime = ticks; if (TCPS_HAVEESTABLISHED(tp->t_state)) tcp_timer_activate(tp, TT_KEEP, TP_KEEPIDLE(tp)); /* * Unscale the window into a 32-bit value. * For the SYN_SENT state the scale is zero. */ tiwin = th->th_win << tp->snd_scale; /* * TCP ECN processing. */ if (tp->t_flags & TF_ECN_PERMIT) { if (thflags & TH_CWR) tp->t_flags &= ~TF_ECN_SND_ECE; switch (iptos & IPTOS_ECN_MASK) { case IPTOS_ECN_CE: tp->t_flags |= TF_ECN_SND_ECE; TCPSTAT_INC(tcps_ecn_ce); break; case IPTOS_ECN_ECT0: TCPSTAT_INC(tcps_ecn_ect0); break; case IPTOS_ECN_ECT1: TCPSTAT_INC(tcps_ecn_ect1); break; } + + /* Process a packet differently from RFC3168. */ + cc_ecnpkt_handler(tp, th, iptos); + /* Congestion experienced. */ if (thflags & TH_ECE) { cc_cong_signal(tp, th, CC_ECN); } } /* * Parse options on any incoming segment. */ tcp_dooptions(&to, (u_char *)(th + 1), (th->th_off << 2) - sizeof(struct tcphdr), (thflags & TH_SYN) ? TO_SYN : 0); /* * If echoed timestamp is later than the current time, * fall back to non RFC1323 RTT calculation. Normalize * timestamp if syncookies were used when this connection * was established. */ if ((to.to_flags & TOF_TS) && (to.to_tsecr != 0)) { to.to_tsecr -= tp->ts_offset; if (TSTMP_GT(to.to_tsecr, tcp_ts_getticks())) to.to_tsecr = 0; } /* * If timestamps were negotiated during SYN/ACK they should * appear on every segment during this session and vice versa. */ if ((tp->t_flags & TF_RCVD_TSTMP) && !(to.to_flags & TOF_TS)) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Timestamp missing, " "no action\n", s, __func__); free(s, M_TCPLOG); } } if (!(tp->t_flags & TF_RCVD_TSTMP) && (to.to_flags & TOF_TS)) { if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Timestamp not expected, " "no action\n", s, __func__); free(s, M_TCPLOG); } } /* * Process options only when we get SYN/ACK back. The SYN case * for incoming connections is handled in tcp_syncache. * According to RFC1323 the window field in a SYN (i.e., a * or ) segment itself is never scaled. * XXX this is traditional behavior, may need to be cleaned up. */ if (tp->t_state == TCPS_SYN_SENT && (thflags & TH_SYN)) { if ((to.to_flags & TOF_SCALE) && (tp->t_flags & TF_REQ_SCALE)) { tp->t_flags |= TF_RCVD_SCALE; tp->snd_scale = to.to_wscale; } /* * Initial send window. It will be updated with * the next incoming segment to the scaled value. */ tp->snd_wnd = th->th_win; if (to.to_flags & TOF_TS) { tp->t_flags |= TF_RCVD_TSTMP; tp->ts_recent = to.to_tsval; tp->ts_recent_age = tcp_ts_getticks(); } if (to.to_flags & TOF_MSS) tcp_mss(tp, to.to_mss); if ((tp->t_flags & TF_SACK_PERMIT) && (to.to_flags & TOF_SACKPERM) == 0) tp->t_flags &= ~TF_SACK_PERMIT; } /* * Header prediction: check for the two common cases * of a uni-directional data xfer. If the packet has * no control flags, is in-sequence, the window didn't * change and we're not retransmitting, it's a * candidate. If the length is zero and the ack moved * forward, we're the sender side of the xfer. Just * free the data acked & wake any higher level process * that was blocked waiting for space. If the length * is non-zero and the ack didn't move, we're the * receiver side. If we're getting packets in-order * (the reassembly queue is empty), add the data to * the socket buffer and note that we need a delayed ack. * Make sure that the hidden state-flags are also off. * Since we check for TCPS_ESTABLISHED first, it can only * be TH_NEEDSYN. */ if (tp->t_state == TCPS_ESTABLISHED && th->th_seq == tp->rcv_nxt && (thflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) == TH_ACK && tp->snd_nxt == tp->snd_max && tiwin && tiwin == tp->snd_wnd && ((tp->t_flags & (TF_NEEDSYN|TF_NEEDFIN)) == 0) && tp->t_segq == NULL && ((to.to_flags & TOF_TS) == 0 || TSTMP_GEQ(to.to_tsval, tp->ts_recent)) ) { /* * If last ACK falls within this segment's sequence numbers, * record the timestamp. * NOTE that the test is modified according to the latest * proposal of the tcplw@cray.com list (Braden 1993/04/26). */ if ((to.to_flags & TOF_TS) != 0 && SEQ_LEQ(th->th_seq, tp->last_ack_sent)) { tp->ts_recent_age = tcp_ts_getticks(); tp->ts_recent = to.to_tsval; } if (tlen == 0) { if (SEQ_GT(th->th_ack, tp->snd_una) && SEQ_LEQ(th->th_ack, tp->snd_max) && !IN_RECOVERY(tp->t_flags) && (to.to_flags & TOF_SACK) == 0 && TAILQ_EMPTY(&tp->snd_holes)) { /* * This is a pure ack for outstanding data. */ if (ti_locked == TI_WLOCKED) INP_INFO_WUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; TCPSTAT_INC(tcps_predack); /* * "bad retransmit" recovery. */ if (tp->t_rxtshift == 1 && tp->t_flags & TF_PREVVALID && (int)(ticks - tp->t_badrxtwin) < 0) { cc_cong_signal(tp, th, CC_RTO_ERR); } /* * Recalculate the transmit timer / rtt. * * Some boxes send broken timestamp replies * during the SYN+ACK phase, ignore * timestamps of 0 or we could calculate a * huge RTT and blow up the retransmit timer. */ if ((to.to_flags & TOF_TS) != 0 && to.to_tsecr) { u_int t; t = tcp_ts_getticks() - to.to_tsecr; if (!tp->t_rttlow || tp->t_rttlow > t) tp->t_rttlow = t; tcp_xmit_timer(tp, TCP_TS_TO_TICKS(t) + 1); } else if (tp->t_rtttime && SEQ_GT(th->th_ack, tp->t_rtseq)) { if (!tp->t_rttlow || tp->t_rttlow > ticks - tp->t_rtttime) tp->t_rttlow = ticks - tp->t_rtttime; tcp_xmit_timer(tp, ticks - tp->t_rtttime); } acked = BYTES_THIS_ACK(tp, th); /* Run HHOOK_TCP_ESTABLISHED_IN helper hooks. */ hhook_run_tcp_est_in(tp, th, &to); TCPSTAT_INC(tcps_rcvackpack); TCPSTAT_ADD(tcps_rcvackbyte, acked); sbdrop(&so->so_snd, acked); if (SEQ_GT(tp->snd_una, tp->snd_recover) && SEQ_LEQ(th->th_ack, tp->snd_recover)) tp->snd_recover = th->th_ack - 1; /* * Let the congestion control algorithm update * congestion control related information. This * typically means increasing the congestion * window. */ cc_ack_received(tp, th, CC_ACK); tp->snd_una = th->th_ack; /* * Pull snd_wl2 up to prevent seq wrap relative * to th_ack. */ tp->snd_wl2 = th->th_ack; tp->t_dupacks = 0; m_freem(m); ND6_HINT(tp); /* Some progress has been made. */ /* * If all outstanding data are acked, stop * retransmit timer, otherwise restart timer * using current (possibly backed-off) value. * If process is waiting for space, * wakeup/selwakeup/signal. If data * are ready to send, let tcp_output * decide between more output or persist. */ #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif if (tp->snd_una == tp->snd_max) tcp_timer_activate(tp, TT_REXMT, 0); else if (!tcp_timer_active(tp, TT_PERSIST)) tcp_timer_activate(tp, TT_REXMT, tp->t_rxtcur); sowwakeup(so); if (sbavail(&so->so_snd)) (void) tcp_output(tp); goto check_delack; } } else if (th->th_ack == tp->snd_una && tlen <= sbspace(&so->so_rcv)) { int newsize = 0; /* automatic sockbuf scaling */ /* * This is a pure, in-sequence data packet with * nothing on the reassembly queue and we have enough * buffer space to take it. */ if (ti_locked == TI_WLOCKED) INP_INFO_WUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; /* Clean receiver SACK report if present */ if ((tp->t_flags & TF_SACK_PERMIT) && tp->rcv_numsacks) tcp_clean_sackreport(tp); TCPSTAT_INC(tcps_preddat); tp->rcv_nxt += tlen; /* * Pull snd_wl1 up to prevent seq wrap relative to * th_seq. */ tp->snd_wl1 = th->th_seq; /* * Pull rcv_up up to prevent seq wrap relative to * rcv_nxt. */ tp->rcv_up = tp->rcv_nxt; TCPSTAT_INC(tcps_rcvpack); TCPSTAT_ADD(tcps_rcvbyte, tlen); ND6_HINT(tp); /* Some progress has been made */ #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif /* * Automatic sizing of receive socket buffer. Often the send * buffer size is not optimally adjusted to the actual network * conditions at hand (delay bandwidth product). Setting the * buffer size too small limits throughput on links with high * bandwidth and high delay (eg. trans-continental/oceanic links). * * On the receive side the socket buffer memory is only rarely * used to any significant extent. This allows us to be much * more aggressive in scaling the receive socket buffer. For * the case that the buffer space is actually used to a large * extent and we run out of kernel memory we can simply drop * the new segments; TCP on the sender will just retransmit it * later. Setting the buffer size too big may only consume too * much kernel memory if the application doesn't read() from * the socket or packet loss or reordering makes use of the * reassembly queue. * * The criteria to step up the receive buffer one notch are: * 1. Application has not set receive buffer size with * SO_RCVBUF. Setting SO_RCVBUF clears SB_AUTOSIZE. * 2. the number of bytes received during the time it takes * one timestamp to be reflected back to us (the RTT); * 3. received bytes per RTT is within seven eighth of the * current socket buffer size; * 4. receive buffer size has not hit maximal automatic size; * * This algorithm does one step per RTT at most and only if * we receive a bulk stream w/o packet losses or reorderings. * Shrinking the buffer during idle times is not necessary as * it doesn't consume any memory when idle. * * TODO: Only step up if the application is actually serving * the buffer to better manage the socket buffer resources. */ if (V_tcp_do_autorcvbuf && to.to_tsecr && (so->so_rcv.sb_flags & SB_AUTOSIZE)) { if (TSTMP_GT(to.to_tsecr, tp->rfbuf_ts) && to.to_tsecr - tp->rfbuf_ts < hz) { if (tp->rfbuf_cnt > (so->so_rcv.sb_hiwat / 8 * 7) && so->so_rcv.sb_hiwat < V_tcp_autorcvbuf_max) { newsize = min(so->so_rcv.sb_hiwat + V_tcp_autorcvbuf_inc, V_tcp_autorcvbuf_max); } /* Start over with next RTT. */ tp->rfbuf_ts = 0; tp->rfbuf_cnt = 0; } else tp->rfbuf_cnt += tlen; /* add up */ } /* Add data to socket buffer. */ SOCKBUF_LOCK(&so->so_rcv); if (so->so_rcv.sb_state & SBS_CANTRCVMORE) { m_freem(m); } else { /* * Set new socket buffer size. * Give up when limit is reached. */ if (newsize) if (!sbreserve_locked(&so->so_rcv, newsize, so, NULL)) so->so_rcv.sb_flags &= ~SB_AUTOSIZE; m_adj(m, drop_hdrlen); /* delayed header drop */ sbappendstream_locked(&so->so_rcv, m, 0); } /* NB: sorwakeup_locked() does an implicit unlock. */ sorwakeup_locked(so); if (DELAY_ACK(tp, tlen)) { tp->t_flags |= TF_DELACK; } else { tp->t_flags |= TF_ACKNOW; tcp_output(tp); } goto check_delack; } } /* * Calculate amount of space in receive window, * and then do TCP input processing. * Receive window is amount of space in rcv queue, * but not less than advertised window. */ win = sbspace(&so->so_rcv); if (win < 0) win = 0; tp->rcv_wnd = imax(win, (int)(tp->rcv_adv - tp->rcv_nxt)); /* Reset receive buffer auto scaling when not in bulk receive mode. */ tp->rfbuf_ts = 0; tp->rfbuf_cnt = 0; switch (tp->t_state) { /* * If the state is SYN_RECEIVED: * if seg contains an ACK, but not for our SYN/ACK, send a RST. */ case TCPS_SYN_RECEIVED: if ((thflags & TH_ACK) && (SEQ_LEQ(th->th_ack, tp->snd_una) || SEQ_GT(th->th_ack, tp->snd_max))) { rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } break; /* * If the state is SYN_SENT: * if seg contains an ACK, but not for our SYN, drop the input. * if seg contains a RST, then drop the connection. * if seg does not contain SYN, then drop it. * Otherwise this is an acceptable SYN segment * initialize tp->rcv_nxt and tp->irs * if seg contains ack then advance tp->snd_una * if seg contains an ECE and ECN support is enabled, the stream * is ECN capable. * if SYN has been acked change to ESTABLISHED else SYN_RCVD state * arrange for segment to be acked (eventually) * continue processing rest of data/controls, beginning with URG */ case TCPS_SYN_SENT: if ((thflags & TH_ACK) && (SEQ_LEQ(th->th_ack, tp->iss) || SEQ_GT(th->th_ack, tp->snd_max))) { rstreason = BANDLIM_UNLIMITED; goto dropwithreset; } if ((thflags & (TH_ACK|TH_RST)) == (TH_ACK|TH_RST)) { TCP_PROBE5(connect__refused, NULL, tp, mtod(m, const char *), tp, th); tp = tcp_drop(tp, ECONNREFUSED); } if (thflags & TH_RST) goto drop; if (!(thflags & TH_SYN)) goto drop; tp->irs = th->th_seq; tcp_rcvseqinit(tp); if (thflags & TH_ACK) { TCPSTAT_INC(tcps_connects); soisconnected(so); #ifdef MAC mac_socketpeer_set_from_mbuf(m, so); #endif /* Do window scaling on this connection? */ if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == (TF_RCVD_SCALE|TF_REQ_SCALE)) { tp->rcv_scale = tp->request_r_scale; } tp->rcv_adv += imin(tp->rcv_wnd, TCP_MAXWIN << tp->rcv_scale); tp->snd_una++; /* SYN is acked */ /* * If there's data, delay ACK; if there's also a FIN * ACKNOW will be turned on later. */ if (DELAY_ACK(tp, tlen) && tlen != 0) tcp_timer_activate(tp, TT_DELACK, tcp_delacktime); else tp->t_flags |= TF_ACKNOW; if ((thflags & TH_ECE) && V_tcp_do_ecn) { tp->t_flags |= TF_ECN_PERMIT; TCPSTAT_INC(tcps_ecn_shs); } /* * Received in SYN_SENT[*] state. * Transitions: * SYN_SENT --> ESTABLISHED * SYN_SENT* --> FIN_WAIT_1 */ tp->t_starttime = ticks; if (tp->t_flags & TF_NEEDFIN) { tcp_state_change(tp, TCPS_FIN_WAIT_1); tp->t_flags &= ~TF_NEEDFIN; thflags &= ~TH_SYN; } else { tcp_state_change(tp, TCPS_ESTABLISHED); TCP_PROBE5(connect__established, NULL, tp, mtod(m, const char *), tp, th); cc_conn_init(tp); tcp_timer_activate(tp, TT_KEEP, TP_KEEPIDLE(tp)); } } else { /* * Received initial SYN in SYN-SENT[*] state => * simultaneous open. * If it succeeds, connection is * half-synchronized. * Otherwise, do 3-way handshake: * SYN-SENT -> SYN-RECEIVED * SYN-SENT* -> SYN-RECEIVED* */ tp->t_flags |= (TF_ACKNOW | TF_NEEDSYN); tcp_timer_activate(tp, TT_REXMT, 0); tcp_state_change(tp, TCPS_SYN_RECEIVED); } KASSERT(ti_locked == TI_WLOCKED, ("%s: trimthenstep6: " "ti_locked %d", __func__, ti_locked)); INP_INFO_WLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(tp->t_inpcb); /* * Advance th->th_seq to correspond to first data byte. * If data, trim to stay within window, * dropping FIN if necessary. */ th->th_seq++; if (tlen > tp->rcv_wnd) { todrop = tlen - tp->rcv_wnd; m_adj(m, -todrop); tlen = tp->rcv_wnd; thflags &= ~TH_FIN; TCPSTAT_INC(tcps_rcvpackafterwin); TCPSTAT_ADD(tcps_rcvbyteafterwin, todrop); } tp->snd_wl1 = th->th_seq - 1; tp->rcv_up = th->th_seq; /* * Client side of transaction: already sent SYN and data. * If the remote host used T/TCP to validate the SYN, * our data will be ACK'd; if so, enter normal data segment * processing in the middle of step 5, ack processing. * Otherwise, goto step 6. */ if (thflags & TH_ACK) goto process_ACK; goto step6; /* * If the state is LAST_ACK or CLOSING or TIME_WAIT: * do normal processing. * * NB: Leftover from RFC1644 T/TCP. Cases to be reused later. */ case TCPS_LAST_ACK: case TCPS_CLOSING: break; /* continue normal processing */ } /* * States other than LISTEN or SYN_SENT. * First check the RST flag and sequence number since reset segments * are exempt from the timestamp and connection count tests. This * fixes a bug introduced by the Stevens, vol. 2, p. 960 bugfix * below which allowed reset segments in half the sequence space * to fall though and be processed (which gives forged reset * segments with a random sequence number a 50 percent chance of * killing a connection). * Then check timestamp, if present. * Then check the connection count, if present. * Then check that at least some bytes of segment are within * receive window. If segment begins before rcv_nxt, * drop leading data (and SYN); if nothing left, just ack. */ if (thflags & TH_RST) { /* * RFC5961 Section 3.2 * * - RST drops connection only if SEG.SEQ == RCV.NXT. * - If RST is in window, we send challenge ACK. * * Note: to take into account delayed ACKs, we should * test against last_ack_sent instead of rcv_nxt. * Note 2: we handle special case of closed window, not * covered by the RFC. */ if ((SEQ_GEQ(th->th_seq, tp->last_ack_sent) && SEQ_LT(th->th_seq, tp->last_ack_sent + tp->rcv_wnd)) || (tp->rcv_wnd == 0 && tp->last_ack_sent == th->th_seq)) { INP_INFO_WLOCK_ASSERT(&V_tcbinfo); KASSERT(ti_locked == TI_WLOCKED, ("%s: TH_RST ti_locked %d, th %p tp %p", __func__, ti_locked, th, tp)); KASSERT(tp->t_state != TCPS_SYN_SENT, ("%s: TH_RST for TCPS_SYN_SENT th %p tp %p", __func__, th, tp)); if (V_tcp_insecure_rst || tp->last_ack_sent == th->th_seq) { TCPSTAT_INC(tcps_drops); /* Drop the connection. */ switch (tp->t_state) { case TCPS_SYN_RECEIVED: so->so_error = ECONNREFUSED; goto close; case TCPS_ESTABLISHED: case TCPS_FIN_WAIT_1: case TCPS_FIN_WAIT_2: case TCPS_CLOSE_WAIT: so->so_error = ECONNRESET; close: tcp_state_change(tp, TCPS_CLOSED); /* FALLTHROUGH */ default: tp = tcp_close(tp); } } else { TCPSTAT_INC(tcps_badrst); /* Send challenge ACK. */ tcp_respond(tp, mtod(m, void *), th, m, tp->rcv_nxt, tp->snd_nxt, TH_ACK); tp->last_ack_sent = tp->rcv_nxt; m = NULL; } } goto drop; } /* * RFC5961 Section 4.2 * Send challenge ACK for any SYN in synchronized state. */ if ((thflags & TH_SYN) && tp->t_state != TCPS_SYN_SENT) { KASSERT(ti_locked == TI_WLOCKED, ("tcp_do_segment: TH_SYN ti_locked %d", ti_locked)); INP_INFO_WLOCK_ASSERT(&V_tcbinfo); TCPSTAT_INC(tcps_badsyn); if (V_tcp_insecure_syn && SEQ_GEQ(th->th_seq, tp->last_ack_sent) && SEQ_LT(th->th_seq, tp->last_ack_sent + tp->rcv_wnd)) { tp = tcp_drop(tp, ECONNRESET); rstreason = BANDLIM_UNLIMITED; } else { /* Send challenge ACK. */ tcp_respond(tp, mtod(m, void *), th, m, tp->rcv_nxt, tp->snd_nxt, TH_ACK); tp->last_ack_sent = tp->rcv_nxt; m = NULL; } goto drop; } /* * RFC 1323 PAWS: If we have a timestamp reply on this segment * and it's less than ts_recent, drop it. */ if ((to.to_flags & TOF_TS) != 0 && tp->ts_recent && TSTMP_LT(to.to_tsval, tp->ts_recent)) { /* Check to see if ts_recent is over 24 days old. */ if (tcp_ts_getticks() - tp->ts_recent_age > TCP_PAWS_IDLE) { /* * Invalidate ts_recent. If this segment updates * ts_recent, the age will be reset later and ts_recent * will get a valid value. If it does not, setting * ts_recent to zero will at least satisfy the * requirement that zero be placed in the timestamp * echo reply when ts_recent isn't valid. The * age isn't reset until we get a valid ts_recent * because we don't want out-of-order segments to be * dropped when ts_recent is old. */ tp->ts_recent = 0; } else { TCPSTAT_INC(tcps_rcvduppack); TCPSTAT_ADD(tcps_rcvdupbyte, tlen); TCPSTAT_INC(tcps_pawsdrop); if (tlen) goto dropafterack; goto drop; } } /* * In the SYN-RECEIVED state, validate that the packet belongs to * this connection before trimming the data to fit the receive * window. Check the sequence number versus IRS since we know * the sequence numbers haven't wrapped. This is a partial fix * for the "LAND" DoS attack. */ if (tp->t_state == TCPS_SYN_RECEIVED && SEQ_LT(th->th_seq, tp->irs)) { rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } todrop = tp->rcv_nxt - th->th_seq; if (todrop > 0) { if (thflags & TH_SYN) { thflags &= ~TH_SYN; th->th_seq++; if (th->th_urp > 1) th->th_urp--; else thflags &= ~TH_URG; todrop--; } /* * Following if statement from Stevens, vol. 2, p. 960. */ if (todrop > tlen || (todrop == tlen && (thflags & TH_FIN) == 0)) { /* * Any valid FIN must be to the left of the window. * At this point the FIN must be a duplicate or out * of sequence; drop it. */ thflags &= ~TH_FIN; /* * Send an ACK to resynchronize and drop any data. * But keep on processing for RST or ACK. */ tp->t_flags |= TF_ACKNOW; todrop = tlen; TCPSTAT_INC(tcps_rcvduppack); TCPSTAT_ADD(tcps_rcvdupbyte, todrop); } else { TCPSTAT_INC(tcps_rcvpartduppack); TCPSTAT_ADD(tcps_rcvpartdupbyte, todrop); } drop_hdrlen += todrop; /* drop from the top afterwards */ th->th_seq += todrop; tlen -= todrop; if (th->th_urp > todrop) th->th_urp -= todrop; else { thflags &= ~TH_URG; th->th_urp = 0; } } /* * If new data are received on a connection after the * user processes are gone, then RST the other end. */ if ((so->so_state & SS_NOFDREF) && tp->t_state > TCPS_CLOSE_WAIT && tlen) { KASSERT(ti_locked == TI_WLOCKED, ("%s: SS_NOFDEREF && " "CLOSE_WAIT && tlen ti_locked %d", __func__, ti_locked)); INP_INFO_WLOCK_ASSERT(&V_tcbinfo); if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: %s: Received %d bytes of data " "after socket was closed, " "sending RST and removing tcpcb\n", s, __func__, tcpstates[tp->t_state], tlen); free(s, M_TCPLOG); } tp = tcp_close(tp); TCPSTAT_INC(tcps_rcvafterclose); rstreason = BANDLIM_UNLIMITED; goto dropwithreset; } /* * If segment ends after window, drop trailing data * (and PUSH and FIN); if nothing left, just ACK. */ todrop = (th->th_seq + tlen) - (tp->rcv_nxt + tp->rcv_wnd); if (todrop > 0) { TCPSTAT_INC(tcps_rcvpackafterwin); if (todrop >= tlen) { TCPSTAT_ADD(tcps_rcvbyteafterwin, tlen); /* * If window is closed can only take segments at * window edge, and have to drop data and PUSH from * incoming segments. Continue processing, but * remember to ack. Otherwise, drop segment * and ack. */ if (tp->rcv_wnd == 0 && th->th_seq == tp->rcv_nxt) { tp->t_flags |= TF_ACKNOW; TCPSTAT_INC(tcps_rcvwinprobe); } else goto dropafterack; } else TCPSTAT_ADD(tcps_rcvbyteafterwin, todrop); m_adj(m, -todrop); tlen -= todrop; thflags &= ~(TH_PUSH|TH_FIN); } /* * If last ACK falls within this segment's sequence numbers, * record its timestamp. * NOTE: * 1) That the test incorporates suggestions from the latest * proposal of the tcplw@cray.com list (Braden 1993/04/26). * 2) That updating only on newer timestamps interferes with * our earlier PAWS tests, so this check should be solely * predicated on the sequence space of this segment. * 3) That we modify the segment boundary check to be * Last.ACK.Sent <= SEG.SEQ + SEG.Len * instead of RFC1323's * Last.ACK.Sent < SEG.SEQ + SEG.Len, * This modified check allows us to overcome RFC1323's * limitations as described in Stevens TCP/IP Illustrated * Vol. 2 p.869. In such cases, we can still calculate the * RTT correctly when RCV.NXT == Last.ACK.Sent. */ if ((to.to_flags & TOF_TS) != 0 && SEQ_LEQ(th->th_seq, tp->last_ack_sent) && SEQ_LEQ(tp->last_ack_sent, th->th_seq + tlen + ((thflags & (TH_SYN|TH_FIN)) != 0))) { tp->ts_recent_age = tcp_ts_getticks(); tp->ts_recent = to.to_tsval; } /* * If the ACK bit is off: if in SYN-RECEIVED state or SENDSYN * flag is on (half-synchronized state), then queue data for * later processing; else drop segment and return. */ if ((thflags & TH_ACK) == 0) { if (tp->t_state == TCPS_SYN_RECEIVED || (tp->t_flags & TF_NEEDSYN)) goto step6; else if (tp->t_flags & TF_ACKNOW) goto dropafterack; else goto drop; } /* * Ack processing. */ switch (tp->t_state) { /* * In SYN_RECEIVED state, the ack ACKs our SYN, so enter * ESTABLISHED state and continue processing. * The ACK was checked above. */ case TCPS_SYN_RECEIVED: TCPSTAT_INC(tcps_connects); soisconnected(so); /* Do window scaling? */ if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == (TF_RCVD_SCALE|TF_REQ_SCALE)) { tp->rcv_scale = tp->request_r_scale; tp->snd_wnd = tiwin; } /* * Make transitions: * SYN-RECEIVED -> ESTABLISHED * SYN-RECEIVED* -> FIN-WAIT-1 */ tp->t_starttime = ticks; if (tp->t_flags & TF_NEEDFIN) { tcp_state_change(tp, TCPS_FIN_WAIT_1); tp->t_flags &= ~TF_NEEDFIN; } else { tcp_state_change(tp, TCPS_ESTABLISHED); TCP_PROBE5(accept__established, NULL, tp, mtod(m, const char *), tp, th); cc_conn_init(tp); tcp_timer_activate(tp, TT_KEEP, TP_KEEPIDLE(tp)); } /* * If segment contains data or ACK, will call tcp_reass() * later; if not, do so now to pass queued data to user. */ if (tlen == 0 && (thflags & TH_FIN) == 0) (void) tcp_reass(tp, (struct tcphdr *)0, 0, (struct mbuf *)0); tp->snd_wl1 = th->th_seq - 1; /* FALLTHROUGH */ /* * In ESTABLISHED state: drop duplicate ACKs; ACK out of range * ACKs. If the ack is in the range * tp->snd_una < th->th_ack <= tp->snd_max * then advance tp->snd_una to th->th_ack and drop * data from the retransmission queue. If this ACK reflects * more up to date window information we update our window information. */ case TCPS_ESTABLISHED: case TCPS_FIN_WAIT_1: case TCPS_FIN_WAIT_2: case TCPS_CLOSE_WAIT: case TCPS_CLOSING: case TCPS_LAST_ACK: if (SEQ_GT(th->th_ack, tp->snd_max)) { TCPSTAT_INC(tcps_rcvacktoomuch); goto dropafterack; } if ((tp->t_flags & TF_SACK_PERMIT) && ((to.to_flags & TOF_SACK) || !TAILQ_EMPTY(&tp->snd_holes))) tcp_sack_doack(tp, &to, th->th_ack); /* Run HHOOK_TCP_ESTABLISHED_IN helper hooks. */ hhook_run_tcp_est_in(tp, th, &to); if (SEQ_LEQ(th->th_ack, tp->snd_una)) { if (tlen == 0 && tiwin == tp->snd_wnd) { /* * If this is the first time we've seen a * FIN from the remote, this is not a * duplicate and it needs to be processed * normally. This happens during a * simultaneous close. */ if ((thflags & TH_FIN) && (TCPS_HAVERCVDFIN(tp->t_state) == 0)) { tp->t_dupacks = 0; break; } TCPSTAT_INC(tcps_rcvdupack); /* * If we have outstanding data (other than * a window probe), this is a completely * duplicate ack (ie, window info didn't * change and FIN isn't set), * the ack is the biggest we've * seen and we've seen exactly our rexmt * threshhold of them, assume a packet * has been dropped and retransmit it. * Kludge snd_nxt & the congestion * window so we send only this one * packet. * * We know we're losing at the current * window size so do congestion avoidance * (set ssthresh to half the current window * and pull our congestion window back to * the new ssthresh). * * Dup acks mean that packets have left the * network (they're now cached at the receiver) * so bump cwnd by the amount in the receiver * to keep a constant cwnd packets in the * network. * * When using TCP ECN, notify the peer that * we reduced the cwnd. */ if (!tcp_timer_active(tp, TT_REXMT) || th->th_ack != tp->snd_una) tp->t_dupacks = 0; else if (++tp->t_dupacks > tcprexmtthresh || IN_FASTRECOVERY(tp->t_flags)) { cc_ack_received(tp, th, CC_DUPACK); if ((tp->t_flags & TF_SACK_PERMIT) && IN_FASTRECOVERY(tp->t_flags)) { int awnd; /* * Compute the amount of data in flight first. * We can inject new data into the pipe iff * we have less than 1/2 the original window's * worth of data in flight. */ awnd = (tp->snd_nxt - tp->snd_fack) + tp->sackhint.sack_bytes_rexmit; if (awnd < tp->snd_ssthresh) { tp->snd_cwnd += tp->t_maxseg; if (tp->snd_cwnd > tp->snd_ssthresh) tp->snd_cwnd = tp->snd_ssthresh; } } else tp->snd_cwnd += tp->t_maxseg; (void) tcp_output(tp); goto drop; } else if (tp->t_dupacks == tcprexmtthresh) { tcp_seq onxt = tp->snd_nxt; /* * If we're doing sack, check to * see if we're already in sack * recovery. If we're not doing sack, * check to see if we're in newreno * recovery. */ if (tp->t_flags & TF_SACK_PERMIT) { if (IN_FASTRECOVERY(tp->t_flags)) { tp->t_dupacks = 0; break; } } else { if (SEQ_LEQ(th->th_ack, tp->snd_recover)) { tp->t_dupacks = 0; break; } } /* Congestion signal before ack. */ cc_cong_signal(tp, th, CC_NDUPACK); cc_ack_received(tp, th, CC_DUPACK); tcp_timer_activate(tp, TT_REXMT, 0); tp->t_rtttime = 0; if (tp->t_flags & TF_SACK_PERMIT) { TCPSTAT_INC( tcps_sack_recovery_episode); tp->sack_newdata = tp->snd_nxt; tp->snd_cwnd = tp->t_maxseg; (void) tcp_output(tp); goto drop; } tp->snd_nxt = th->th_ack; tp->snd_cwnd = tp->t_maxseg; (void) tcp_output(tp); KASSERT(tp->snd_limited <= 2, ("%s: tp->snd_limited too big", __func__)); tp->snd_cwnd = tp->snd_ssthresh + tp->t_maxseg * (tp->t_dupacks - tp->snd_limited); if (SEQ_GT(onxt, tp->snd_nxt)) tp->snd_nxt = onxt; goto drop; } else if (V_tcp_do_rfc3042) { cc_ack_received(tp, th, CC_DUPACK); u_long oldcwnd = tp->snd_cwnd; tcp_seq oldsndmax = tp->snd_max; u_int sent; int avail; KASSERT(tp->t_dupacks == 1 || tp->t_dupacks == 2, ("%s: dupacks not 1 or 2", __func__)); if (tp->t_dupacks == 1) tp->snd_limited = 0; tp->snd_cwnd = (tp->snd_nxt - tp->snd_una) + (tp->t_dupacks - tp->snd_limited) * tp->t_maxseg; /* * Only call tcp_output when there * is new data available to be sent. * Otherwise we would send pure ACKs. */ SOCKBUF_LOCK(&so->so_snd); avail = sbavail(&so->so_snd) - (tp->snd_nxt - tp->snd_una); SOCKBUF_UNLOCK(&so->so_snd); if (avail > 0) (void) tcp_output(tp); sent = tp->snd_max - oldsndmax; if (sent > tp->t_maxseg) { KASSERT((tp->t_dupacks == 2 && tp->snd_limited == 0) || (sent == tp->t_maxseg + 1 && tp->t_flags & TF_SENTFIN), ("%s: sent too much", __func__)); tp->snd_limited = 2; } else if (sent > 0) ++tp->snd_limited; tp->snd_cwnd = oldcwnd; goto drop; } } else tp->t_dupacks = 0; break; } KASSERT(SEQ_GT(th->th_ack, tp->snd_una), ("%s: th_ack <= snd_una", __func__)); /* * If the congestion window was inflated to account * for the other side's cached packets, retract it. */ if (IN_FASTRECOVERY(tp->t_flags)) { if (SEQ_LT(th->th_ack, tp->snd_recover)) { if (tp->t_flags & TF_SACK_PERMIT) tcp_sack_partialack(tp, th); else tcp_newreno_partial_ack(tp, th); } else cc_post_recovery(tp, th); } tp->t_dupacks = 0; /* * If we reach this point, ACK is not a duplicate, * i.e., it ACKs something we sent. */ if (tp->t_flags & TF_NEEDSYN) { /* * T/TCP: Connection was half-synchronized, and our * SYN has been ACK'd (so connection is now fully * synchronized). Go to non-starred state, * increment snd_una for ACK of SYN, and check if * we can do window scaling. */ tp->t_flags &= ~TF_NEEDSYN; tp->snd_una++; /* Do window scaling? */ if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == (TF_RCVD_SCALE|TF_REQ_SCALE)) { tp->rcv_scale = tp->request_r_scale; /* Send window already scaled. */ } } process_ACK: INP_WLOCK_ASSERT(tp->t_inpcb); acked = BYTES_THIS_ACK(tp, th); TCPSTAT_INC(tcps_rcvackpack); TCPSTAT_ADD(tcps_rcvackbyte, acked); /* * If we just performed our first retransmit, and the ACK * arrives within our recovery window, then it was a mistake * to do the retransmit in the first place. Recover our * original cwnd and ssthresh, and proceed to transmit where * we left off. */ if (tp->t_rxtshift == 1 && tp->t_flags & TF_PREVVALID && (int)(ticks - tp->t_badrxtwin) < 0) cc_cong_signal(tp, th, CC_RTO_ERR); /* * If we have a timestamp reply, update smoothed * round trip time. If no timestamp is present but * transmit timer is running and timed sequence * number was acked, update smoothed round trip time. * Since we now have an rtt measurement, cancel the * timer backoff (cf., Phil Karn's retransmit alg.). * Recompute the initial retransmit timer. * * Some boxes send broken timestamp replies * during the SYN+ACK phase, ignore * timestamps of 0 or we could calculate a * huge RTT and blow up the retransmit timer. */ if ((to.to_flags & TOF_TS) != 0 && to.to_tsecr) { u_int t; t = tcp_ts_getticks() - to.to_tsecr; if (!tp->t_rttlow || tp->t_rttlow > t) tp->t_rttlow = t; tcp_xmit_timer(tp, TCP_TS_TO_TICKS(t) + 1); } else if (tp->t_rtttime && SEQ_GT(th->th_ack, tp->t_rtseq)) { if (!tp->t_rttlow || tp->t_rttlow > ticks - tp->t_rtttime) tp->t_rttlow = ticks - tp->t_rtttime; tcp_xmit_timer(tp, ticks - tp->t_rtttime); } /* * If all outstanding data is acked, stop retransmit * timer and remember to restart (more output or persist). * If there is more data to be acked, restart retransmit * timer, using current (possibly backed-off) value. */ if (th->th_ack == tp->snd_max) { tcp_timer_activate(tp, TT_REXMT, 0); needoutput = 1; } else if (!tcp_timer_active(tp, TT_PERSIST)) tcp_timer_activate(tp, TT_REXMT, tp->t_rxtcur); /* * If no data (only SYN) was ACK'd, * skip rest of ACK processing. */ if (acked == 0) goto step6; /* * Let the congestion control algorithm update congestion * control related information. This typically means increasing * the congestion window. */ cc_ack_received(tp, th, CC_ACK); SOCKBUF_LOCK(&so->so_snd); if (acked > sbavail(&so->so_snd)) { tp->snd_wnd -= sbavail(&so->so_snd); mfree = sbcut_locked(&so->so_snd, (int)sbavail(&so->so_snd)); ourfinisacked = 1; } else { mfree = sbcut_locked(&so->so_snd, acked); tp->snd_wnd -= acked; ourfinisacked = 0; } /* NB: sowwakeup_locked() does an implicit unlock. */ sowwakeup_locked(so); m_freem(mfree); /* Detect una wraparound. */ if (!IN_RECOVERY(tp->t_flags) && SEQ_GT(tp->snd_una, tp->snd_recover) && SEQ_LEQ(th->th_ack, tp->snd_recover)) tp->snd_recover = th->th_ack - 1; /* XXXLAS: Can this be moved up into cc_post_recovery? */ if (IN_RECOVERY(tp->t_flags) && SEQ_GEQ(th->th_ack, tp->snd_recover)) { EXIT_RECOVERY(tp->t_flags); } tp->snd_una = th->th_ack; if (tp->t_flags & TF_SACK_PERMIT) { if (SEQ_GT(tp->snd_una, tp->snd_recover)) tp->snd_recover = tp->snd_una; } if (SEQ_LT(tp->snd_nxt, tp->snd_una)) tp->snd_nxt = tp->snd_una; switch (tp->t_state) { /* * In FIN_WAIT_1 STATE in addition to the processing * for the ESTABLISHED state if our FIN is now acknowledged * then enter FIN_WAIT_2. */ case TCPS_FIN_WAIT_1: if (ourfinisacked) { /* * If we can't receive any more * data, then closing user can proceed. * Starting the timer is contrary to the * specification, but if we don't get a FIN * we'll hang forever. * * XXXjl: * we should release the tp also, and use a * compressed state. */ if (so->so_rcv.sb_state & SBS_CANTRCVMORE) { soisdisconnected(so); tcp_timer_activate(tp, TT_2MSL, (tcp_fast_finwait2_recycle ? tcp_finwait2_timeout : TP_MAXIDLE(tp))); } tcp_state_change(tp, TCPS_FIN_WAIT_2); } break; /* * In CLOSING STATE in addition to the processing for * the ESTABLISHED state if the ACK acknowledges our FIN * then enter the TIME-WAIT state, otherwise ignore * the segment. */ case TCPS_CLOSING: if (ourfinisacked) { INP_INFO_WLOCK_ASSERT(&V_tcbinfo); tcp_twstart(tp); INP_INFO_WUNLOCK(&V_tcbinfo); m_freem(m); return; } break; /* * In LAST_ACK, we may still be waiting for data to drain * and/or to be acked, as well as for the ack of our FIN. * If our FIN is now acknowledged, delete the TCB, * enter the closed state and return. */ case TCPS_LAST_ACK: if (ourfinisacked) { INP_INFO_WLOCK_ASSERT(&V_tcbinfo); tp = tcp_close(tp); goto drop; } break; } } step6: INP_WLOCK_ASSERT(tp->t_inpcb); /* * Update window information. * Don't look at window if no ACK: TAC's send garbage on first SYN. */ if ((thflags & TH_ACK) && (SEQ_LT(tp->snd_wl1, th->th_seq) || (tp->snd_wl1 == th->th_seq && (SEQ_LT(tp->snd_wl2, th->th_ack) || (tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd))))) { /* keep track of pure window updates */ if (tlen == 0 && tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd) TCPSTAT_INC(tcps_rcvwinupd); tp->snd_wnd = tiwin; tp->snd_wl1 = th->th_seq; tp->snd_wl2 = th->th_ack; if (tp->snd_wnd > tp->max_sndwnd) tp->max_sndwnd = tp->snd_wnd; needoutput = 1; } /* * Process segments with URG. */ if ((thflags & TH_URG) && th->th_urp && TCPS_HAVERCVDFIN(tp->t_state) == 0) { /* * This is a kludge, but if we receive and accept * random urgent pointers, we'll crash in * soreceive. It's hard to imagine someone * actually wanting to send this much urgent data. */ SOCKBUF_LOCK(&so->so_rcv); if (th->th_urp + sbavail(&so->so_rcv) > sb_max) { th->th_urp = 0; /* XXX */ thflags &= ~TH_URG; /* XXX */ SOCKBUF_UNLOCK(&so->so_rcv); /* XXX */ goto dodata; /* XXX */ } /* * If this segment advances the known urgent pointer, * then mark the data stream. This should not happen * in CLOSE_WAIT, CLOSING, LAST_ACK or TIME_WAIT STATES since * a FIN has been received from the remote side. * In these states we ignore the URG. * * According to RFC961 (Assigned Protocols), * the urgent pointer points to the last octet * of urgent data. We continue, however, * to consider it to indicate the first octet * of data past the urgent section as the original * spec states (in one of two places). */ if (SEQ_GT(th->th_seq+th->th_urp, tp->rcv_up)) { tp->rcv_up = th->th_seq + th->th_urp; so->so_oobmark = sbavail(&so->so_rcv) + (tp->rcv_up - tp->rcv_nxt) - 1; if (so->so_oobmark == 0) so->so_rcv.sb_state |= SBS_RCVATMARK; sohasoutofband(so); tp->t_oobflags &= ~(TCPOOB_HAVEDATA | TCPOOB_HADDATA); } SOCKBUF_UNLOCK(&so->so_rcv); /* * Remove out of band data so doesn't get presented to user. * This can happen independent of advancing the URG pointer, * but if two URG's are pending at once, some out-of-band * data may creep in... ick. */ if (th->th_urp <= (u_long)tlen && !(so->so_options & SO_OOBINLINE)) { /* hdr drop is delayed */ tcp_pulloutofband(so, th, m, drop_hdrlen); } } else { /* * If no out of band data is expected, * pull receive urgent pointer along * with the receive window. */ if (SEQ_GT(tp->rcv_nxt, tp->rcv_up)) tp->rcv_up = tp->rcv_nxt; } dodata: /* XXX */ INP_WLOCK_ASSERT(tp->t_inpcb); /* * Process the segment text, merging it into the TCP sequencing queue, * and arranging for acknowledgment of receipt if necessary. * This process logically involves adjusting tp->rcv_wnd as data * is presented to the user (this happens in tcp_usrreq.c, * case PRU_RCVD). If a FIN has already been received on this * connection then we just ignore the text. */ if ((tlen || (thflags & TH_FIN)) && TCPS_HAVERCVDFIN(tp->t_state) == 0) { tcp_seq save_start = th->th_seq; m_adj(m, drop_hdrlen); /* delayed header drop */ /* * Insert segment which includes th into TCP reassembly queue * with control block tp. Set thflags to whether reassembly now * includes a segment with FIN. This handles the common case * inline (segment is the next to be received on an established * connection, and the queue is empty), avoiding linkage into * and removal from the queue and repetition of various * conversions. * Set DELACK for segments received in order, but ack * immediately when segments are out of order (so * fast retransmit can work). */ if (th->th_seq == tp->rcv_nxt && tp->t_segq == NULL && TCPS_HAVEESTABLISHED(tp->t_state)) { if (DELAY_ACK(tp, tlen)) tp->t_flags |= TF_DELACK; else tp->t_flags |= TF_ACKNOW; tp->rcv_nxt += tlen; thflags = th->th_flags & TH_FIN; TCPSTAT_INC(tcps_rcvpack); TCPSTAT_ADD(tcps_rcvbyte, tlen); ND6_HINT(tp); SOCKBUF_LOCK(&so->so_rcv); if (so->so_rcv.sb_state & SBS_CANTRCVMORE) m_freem(m); else sbappendstream_locked(&so->so_rcv, m, 0); /* NB: sorwakeup_locked() does an implicit unlock. */ sorwakeup_locked(so); } else { /* * XXX: Due to the header drop above "th" is * theoretically invalid by now. Fortunately * m_adj() doesn't actually frees any mbufs * when trimming from the head. */ thflags = tcp_reass(tp, th, &tlen, m); tp->t_flags |= TF_ACKNOW; } if (tlen > 0 && (tp->t_flags & TF_SACK_PERMIT)) tcp_update_sack_list(tp, save_start, save_start + tlen); #if 0 /* * Note the amount of data that peer has sent into * our window, in order to estimate the sender's * buffer size. * XXX: Unused. */ if (SEQ_GT(tp->rcv_adv, tp->rcv_nxt)) len = so->so_rcv.sb_hiwat - (tp->rcv_adv - tp->rcv_nxt); else len = so->so_rcv.sb_hiwat; #endif } else { m_freem(m); thflags &= ~TH_FIN; } /* * If FIN is received ACK the FIN and let the user know * that the connection is closing. */ if (thflags & TH_FIN) { if (TCPS_HAVERCVDFIN(tp->t_state) == 0) { socantrcvmore(so); /* * If connection is half-synchronized * (ie NEEDSYN flag on) then delay ACK, * so it may be piggybacked when SYN is sent. * Otherwise, since we received a FIN then no * more input can be expected, send ACK now. */ if (tp->t_flags & TF_NEEDSYN) tp->t_flags |= TF_DELACK; else tp->t_flags |= TF_ACKNOW; tp->rcv_nxt++; } switch (tp->t_state) { /* * In SYN_RECEIVED and ESTABLISHED STATES * enter the CLOSE_WAIT state. */ case TCPS_SYN_RECEIVED: tp->t_starttime = ticks; /* FALLTHROUGH */ case TCPS_ESTABLISHED: tcp_state_change(tp, TCPS_CLOSE_WAIT); break; /* * If still in FIN_WAIT_1 STATE FIN has not been acked so * enter the CLOSING state. */ case TCPS_FIN_WAIT_1: tcp_state_change(tp, TCPS_CLOSING); break; /* * In FIN_WAIT_2 state enter the TIME_WAIT state, * starting the time-wait timer, turning off the other * standard timers. */ case TCPS_FIN_WAIT_2: INP_INFO_WLOCK_ASSERT(&V_tcbinfo); KASSERT(ti_locked == TI_WLOCKED, ("%s: dodata " "TCP_FIN_WAIT_2 ti_locked: %d", __func__, ti_locked)); tcp_twstart(tp); INP_INFO_WUNLOCK(&V_tcbinfo); return; } } if (ti_locked == TI_WLOCKED) INP_INFO_WUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif /* * Return any desired output. */ if (needoutput || (tp->t_flags & TF_ACKNOW)) (void) tcp_output(tp); check_delack: KASSERT(ti_locked == TI_UNLOCKED, ("%s: check_delack ti_locked %d", __func__, ti_locked)); INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(tp->t_inpcb); if (tp->t_flags & TF_DELACK) { tp->t_flags &= ~TF_DELACK; tcp_timer_activate(tp, TT_DELACK, tcp_delacktime); } INP_WUNLOCK(tp->t_inpcb); return; dropafterack: /* * Generate an ACK dropping incoming segment if it occupies * sequence space, where the ACK reflects our state. * * We can now skip the test for the RST flag since all * paths to this code happen after packets containing * RST have been dropped. * * In the SYN-RECEIVED state, don't send an ACK unless the * segment we received passes the SYN-RECEIVED ACK test. * If it fails send a RST. This breaks the loop in the * "LAND" DoS attack, and also prevents an ACK storm * between two listening ports that have been sent forged * SYN segments, each with the source address of the other. */ if (tp->t_state == TCPS_SYN_RECEIVED && (thflags & TH_ACK) && (SEQ_GT(tp->snd_una, th->th_ack) || SEQ_GT(th->th_ack, tp->snd_max)) ) { rstreason = BANDLIM_RST_OPENPORT; goto dropwithreset; } #ifdef TCPDEBUG if (so->so_options & SO_DEBUG) tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif if (ti_locked == TI_WLOCKED) INP_INFO_WUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; tp->t_flags |= TF_ACKNOW; (void) tcp_output(tp); INP_WUNLOCK(tp->t_inpcb); m_freem(m); return; dropwithreset: if (ti_locked == TI_WLOCKED) INP_INFO_WUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; if (tp != NULL) { tcp_dropwithreset(m, th, tp, tlen, rstreason); INP_WUNLOCK(tp->t_inpcb); } else tcp_dropwithreset(m, th, NULL, tlen, rstreason); return; drop: if (ti_locked == TI_WLOCKED) { INP_INFO_WUNLOCK(&V_tcbinfo); ti_locked = TI_UNLOCKED; } #ifdef INVARIANTS else INP_INFO_UNLOCK_ASSERT(&V_tcbinfo); #endif /* * Drop space held by incoming segment and return. */ #ifdef TCPDEBUG if (tp == NULL || (tp->t_inpcb->inp_socket->so_options & SO_DEBUG)) tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen, &tcp_savetcp, 0); #endif if (tp != NULL) INP_WUNLOCK(tp->t_inpcb); m_freem(m); } /* * Issue RST and make ACK acceptable to originator of segment. * The mbuf must still include the original packet header. * tp may be NULL. */ static void tcp_dropwithreset(struct mbuf *m, struct tcphdr *th, struct tcpcb *tp, int tlen, int rstreason) { #ifdef INET struct ip *ip; #endif #ifdef INET6 struct ip6_hdr *ip6; #endif if (tp != NULL) { INP_WLOCK_ASSERT(tp->t_inpcb); } /* Don't bother if destination was broadcast/multicast. */ if ((th->th_flags & TH_RST) || m->m_flags & (M_BCAST|M_MCAST)) goto drop; #ifdef INET6 if (mtod(m, struct ip *)->ip_v == 6) { ip6 = mtod(m, struct ip6_hdr *); if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) || IN6_IS_ADDR_MULTICAST(&ip6->ip6_src)) goto drop; /* IPv6 anycast check is done at tcp6_input() */ } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { ip = mtod(m, struct ip *); if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) || IN_MULTICAST(ntohl(ip->ip_src.s_addr)) || ip->ip_src.s_addr == htonl(INADDR_BROADCAST) || in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) goto drop; } #endif /* Perform bandwidth limiting. */ if (badport_bandlim(rstreason) < 0) goto drop; /* tcp_respond consumes the mbuf chain. */ if (th->th_flags & TH_ACK) { tcp_respond(tp, mtod(m, void *), th, m, (tcp_seq)0, th->th_ack, TH_RST); } else { if (th->th_flags & TH_SYN) tlen++; tcp_respond(tp, mtod(m, void *), th, m, th->th_seq+tlen, (tcp_seq)0, TH_RST|TH_ACK); } return; drop: m_freem(m); } /* * Parse TCP options and place in tcpopt. */ static void tcp_dooptions(struct tcpopt *to, u_char *cp, int cnt, int flags) { int opt, optlen; to->to_flags = 0; for (; cnt > 0; cnt -= optlen, cp += optlen) { opt = cp[0]; if (opt == TCPOPT_EOL) break; if (opt == TCPOPT_NOP) optlen = 1; else { if (cnt < 2) break; optlen = cp[1]; if (optlen < 2 || optlen > cnt) break; } switch (opt) { case TCPOPT_MAXSEG: if (optlen != TCPOLEN_MAXSEG) continue; if (!(flags & TO_SYN)) continue; to->to_flags |= TOF_MSS; bcopy((char *)cp + 2, (char *)&to->to_mss, sizeof(to->to_mss)); to->to_mss = ntohs(to->to_mss); break; case TCPOPT_WINDOW: if (optlen != TCPOLEN_WINDOW) continue; if (!(flags & TO_SYN)) continue; to->to_flags |= TOF_SCALE; to->to_wscale = min(cp[2], TCP_MAX_WINSHIFT); break; case TCPOPT_TIMESTAMP: if (optlen != TCPOLEN_TIMESTAMP) continue; to->to_flags |= TOF_TS; bcopy((char *)cp + 2, (char *)&to->to_tsval, sizeof(to->to_tsval)); to->to_tsval = ntohl(to->to_tsval); bcopy((char *)cp + 6, (char *)&to->to_tsecr, sizeof(to->to_tsecr)); to->to_tsecr = ntohl(to->to_tsecr); break; #ifdef TCP_SIGNATURE /* * XXX In order to reply to a host which has set the * TCP_SIGNATURE option in its initial SYN, we have to * record the fact that the option was observed here * for the syncache code to perform the correct response. */ case TCPOPT_SIGNATURE: if (optlen != TCPOLEN_SIGNATURE) continue; to->to_flags |= TOF_SIGNATURE; to->to_signature = cp + 2; break; #endif case TCPOPT_SACK_PERMITTED: if (optlen != TCPOLEN_SACK_PERMITTED) continue; if (!(flags & TO_SYN)) continue; if (!V_tcp_do_sack) continue; to->to_flags |= TOF_SACKPERM; break; case TCPOPT_SACK: if (optlen <= 2 || (optlen - 2) % TCPOLEN_SACK != 0) continue; if (flags & TO_SYN) continue; to->to_flags |= TOF_SACK; to->to_nsacks = (optlen - 2) / TCPOLEN_SACK; to->to_sacks = cp + 2; TCPSTAT_INC(tcps_sack_rcv_blocks); break; default: continue; } } } /* * Pull out of band byte out of a segment so * it doesn't appear in the user's data queue. * It is still reflected in the segment length for * sequencing purposes. */ static void tcp_pulloutofband(struct socket *so, struct tcphdr *th, struct mbuf *m, int off) { int cnt = off + th->th_urp - 1; while (cnt >= 0) { if (m->m_len > cnt) { char *cp = mtod(m, caddr_t) + cnt; struct tcpcb *tp = sototcpcb(so); INP_WLOCK_ASSERT(tp->t_inpcb); tp->t_iobc = *cp; tp->t_oobflags |= TCPOOB_HAVEDATA; bcopy(cp+1, cp, (unsigned)(m->m_len - cnt - 1)); m->m_len--; if (m->m_flags & M_PKTHDR) m->m_pkthdr.len--; return; } cnt -= m->m_len; m = m->m_next; if (m == NULL) break; } panic("tcp_pulloutofband"); } /* * Collect new round-trip time estimate * and update averages and current timeout. */ static void tcp_xmit_timer(struct tcpcb *tp, int rtt) { int delta; INP_WLOCK_ASSERT(tp->t_inpcb); TCPSTAT_INC(tcps_rttupdated); tp->t_rttupdated++; if (tp->t_srtt != 0) { /* * srtt is stored as fixed point with 5 bits after the * binary point (i.e., scaled by 8). The following magic * is equivalent to the smoothing algorithm in rfc793 with * an alpha of .875 (srtt = rtt/8 + srtt*7/8 in fixed * point). Adjust rtt to origin 0. */ delta = ((rtt - 1) << TCP_DELTA_SHIFT) - (tp->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)); if ((tp->t_srtt += delta) <= 0) tp->t_srtt = 1; /* * We accumulate a smoothed rtt variance (actually, a * smoothed mean difference), then set the retransmit * timer to smoothed rtt + 4 times the smoothed variance. * rttvar is stored as fixed point with 4 bits after the * binary point (scaled by 16). The following is * equivalent to rfc793 smoothing with an alpha of .75 * (rttvar = rttvar*3/4 + |delta| / 4). This replaces * rfc793's wired-in beta. */ if (delta < 0) delta = -delta; delta -= tp->t_rttvar >> (TCP_RTTVAR_SHIFT - TCP_DELTA_SHIFT); if ((tp->t_rttvar += delta) <= 0) tp->t_rttvar = 1; if (tp->t_rttbest > tp->t_srtt + tp->t_rttvar) tp->t_rttbest = tp->t_srtt + tp->t_rttvar; } else { /* * No rtt measurement yet - use the unsmoothed rtt. * Set the variance to half the rtt (so our first * retransmit happens at 3*rtt). */ tp->t_srtt = rtt << TCP_RTT_SHIFT; tp->t_rttvar = rtt << (TCP_RTTVAR_SHIFT - 1); tp->t_rttbest = tp->t_srtt + tp->t_rttvar; } tp->t_rtttime = 0; tp->t_rxtshift = 0; /* * the retransmit should happen at rtt + 4 * rttvar. * Because of the way we do the smoothing, srtt and rttvar * will each average +1/2 tick of bias. When we compute * the retransmit timer, we want 1/2 tick of rounding and * 1 extra tick because of +-1/2 tick uncertainty in the * firing of the timer. The bias will give us exactly the * 1.5 tick we need. But, because the bias is * statistical, we have to test that we don't drop below * the minimum feasible timer (which is 2 ticks). */ TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp), max(tp->t_rttmin, rtt + 2), TCPTV_REXMTMAX); /* * We received an ack for a packet that wasn't retransmitted; * it is probably safe to discard any error indications we've * received recently. This isn't quite right, but close enough * for now (a route might have failed after we sent a segment, * and the return path might not be symmetrical). */ tp->t_softerror = 0; } /* * Determine a reasonable value for maxseg size. * If the route is known, check route for mtu. * If none, use an mss that can be handled on the outgoing interface * without forcing IP to fragment. If no route is found, route has no mtu, * or the destination isn't local, use a default, hopefully conservative * size (usually 512 or the default IP max size, but no more than the mtu * of the interface), as we can't discover anything about intervening * gateways or networks. We also initialize the congestion/slow start * window to be a single segment if the destination isn't local. * While looking at the routing entry, we also initialize other path-dependent * parameters from pre-set or cached values in the routing entry. * * Also take into account the space needed for options that we * send regularly. Make maxseg shorter by that amount to assure * that we can send maxseg amount of data even when the options * are present. Store the upper limit of the length of options plus * data in maxopd. * * NOTE that this routine is only called when we process an incoming * segment, or an ICMP need fragmentation datagram. Outgoing SYN/ACK MSS * settings are handled in tcp_mssopt(). */ void tcp_mss_update(struct tcpcb *tp, int offer, int mtuoffer, struct hc_metrics_lite *metricptr, struct tcp_ifcap *cap) { int mss = 0; u_long maxmtu = 0; struct inpcb *inp = tp->t_inpcb; struct hc_metrics_lite metrics; int origoffer; #ifdef INET6 int isipv6 = ((inp->inp_vflag & INP_IPV6) != 0) ? 1 : 0; size_t min_protoh = isipv6 ? sizeof (struct ip6_hdr) + sizeof (struct tcphdr) : sizeof (struct tcpiphdr); #else const size_t min_protoh = sizeof(struct tcpiphdr); #endif INP_WLOCK_ASSERT(tp->t_inpcb); if (mtuoffer != -1) { KASSERT(offer == -1, ("%s: conflict", __func__)); offer = mtuoffer - min_protoh; } origoffer = offer; /* Initialize. */ #ifdef INET6 if (isipv6) { maxmtu = tcp_maxmtu6(&inp->inp_inc, cap); tp->t_maxopd = tp->t_maxseg = V_tcp_v6mssdflt; } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { maxmtu = tcp_maxmtu(&inp->inp_inc, cap); tp->t_maxopd = tp->t_maxseg = V_tcp_mssdflt; } #endif /* * No route to sender, stay with default mss and return. */ if (maxmtu == 0) { /* * In case we return early we need to initialize metrics * to a defined state as tcp_hc_get() would do for us * if there was no cache hit. */ if (metricptr != NULL) bzero(metricptr, sizeof(struct hc_metrics_lite)); return; } /* What have we got? */ switch (offer) { case 0: /* * Offer == 0 means that there was no MSS on the SYN * segment, in this case we use tcp_mssdflt as * already assigned to t_maxopd above. */ offer = tp->t_maxopd; break; case -1: /* * Offer == -1 means that we didn't receive SYN yet. */ /* FALLTHROUGH */ default: /* * Prevent DoS attack with too small MSS. Round up * to at least minmss. */ offer = max(offer, V_tcp_minmss); } /* * rmx information is now retrieved from tcp_hostcache. */ tcp_hc_get(&inp->inp_inc, &metrics); if (metricptr != NULL) bcopy(&metrics, metricptr, sizeof(struct hc_metrics_lite)); /* * If there's a discovered mtu in tcp hostcache, use it. * Else, use the link mtu. */ if (metrics.rmx_mtu) mss = min(metrics.rmx_mtu, maxmtu) - min_protoh; else { #ifdef INET6 if (isipv6) { mss = maxmtu - min_protoh; if (!V_path_mtu_discovery && !in6_localaddr(&inp->in6p_faddr)) mss = min(mss, V_tcp_v6mssdflt); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { mss = maxmtu - min_protoh; if (!V_path_mtu_discovery && !in_localaddr(inp->inp_faddr)) mss = min(mss, V_tcp_mssdflt); } #endif /* * XXX - The above conditional (mss = maxmtu - min_protoh) * probably violates the TCP spec. * The problem is that, since we don't know the * other end's MSS, we are supposed to use a conservative * default. But, if we do that, then MTU discovery will * never actually take place, because the conservative * default is much less than the MTUs typically seen * on the Internet today. For the moment, we'll sweep * this under the carpet. * * The conservative default might not actually be a problem * if the only case this occurs is when sending an initial * SYN with options and data to a host we've never talked * to before. Then, they will reply with an MSS value which * will get recorded and the new parameters should get * recomputed. For Further Study. */ } mss = min(mss, offer); /* * Sanity check: make sure that maxopd will be large * enough to allow some data on segments even if the * all the option space is used (40bytes). Otherwise * funny things may happen in tcp_output. */ mss = max(mss, 64); /* * maxopd stores the maximum length of data AND options * in a segment; maxseg is the amount of data in a normal * segment. We need to store this value (maxopd) apart * from maxseg, because now every segment carries options * and thus we normally have somewhat less data in segments. */ tp->t_maxopd = mss; /* * origoffer==-1 indicates that no segments were received yet. * In this case we just guess. */ if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP && (origoffer == -1 || (tp->t_flags & TF_RCVD_TSTMP) == TF_RCVD_TSTMP)) mss -= TCPOLEN_TSTAMP_APPA; tp->t_maxseg = mss; } void tcp_mss(struct tcpcb *tp, int offer) { int mss; u_long bufsize; struct inpcb *inp; struct socket *so; struct hc_metrics_lite metrics; struct tcp_ifcap cap; KASSERT(tp != NULL, ("%s: tp == NULL", __func__)); bzero(&cap, sizeof(cap)); tcp_mss_update(tp, offer, -1, &metrics, &cap); mss = tp->t_maxseg; inp = tp->t_inpcb; /* * If there's a pipesize, change the socket buffer to that size, * don't change if sb_hiwat is different than default (then it * has been changed on purpose with setsockopt). * Make the socket buffers an integral number of mss units; * if the mss is larger than the socket buffer, decrease the mss. */ so = inp->inp_socket; SOCKBUF_LOCK(&so->so_snd); if ((so->so_snd.sb_hiwat == V_tcp_sendspace) && metrics.rmx_sendpipe) bufsize = metrics.rmx_sendpipe; else bufsize = so->so_snd.sb_hiwat; if (bufsize < mss) mss = bufsize; else { bufsize = roundup(bufsize, mss); if (bufsize > sb_max) bufsize = sb_max; if (bufsize > so->so_snd.sb_hiwat) (void)sbreserve_locked(&so->so_snd, bufsize, so, NULL); } SOCKBUF_UNLOCK(&so->so_snd); tp->t_maxseg = mss; SOCKBUF_LOCK(&so->so_rcv); if ((so->so_rcv.sb_hiwat == V_tcp_recvspace) && metrics.rmx_recvpipe) bufsize = metrics.rmx_recvpipe; else bufsize = so->so_rcv.sb_hiwat; if (bufsize > mss) { bufsize = roundup(bufsize, mss); if (bufsize > sb_max) bufsize = sb_max; if (bufsize > so->so_rcv.sb_hiwat) (void)sbreserve_locked(&so->so_rcv, bufsize, so, NULL); } SOCKBUF_UNLOCK(&so->so_rcv); /* Check the interface for TSO capabilities. */ if (cap.ifcap & CSUM_TSO) { tp->t_flags |= TF_TSO; tp->t_tsomax = cap.tsomax; tp->t_tsomaxsegcount = cap.tsomaxsegcount; tp->t_tsomaxsegsize = cap.tsomaxsegsize; } } /* * Determine the MSS option to send on an outgoing SYN. */ int tcp_mssopt(struct in_conninfo *inc) { int mss = 0; u_long maxmtu = 0; u_long thcmtu = 0; size_t min_protoh; KASSERT(inc != NULL, ("tcp_mssopt with NULL in_conninfo pointer")); #ifdef INET6 if (inc->inc_flags & INC_ISIPV6) { mss = V_tcp_v6mssdflt; maxmtu = tcp_maxmtu6(inc, NULL); min_protoh = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { mss = V_tcp_mssdflt; maxmtu = tcp_maxmtu(inc, NULL); min_protoh = sizeof(struct tcpiphdr); } #endif #if defined(INET6) || defined(INET) thcmtu = tcp_hc_getmtu(inc); /* IPv4 and IPv6 */ #endif if (maxmtu && thcmtu) mss = min(maxmtu, thcmtu) - min_protoh; else if (maxmtu || thcmtu) mss = max(maxmtu, thcmtu) - min_protoh; return (mss); } /* * On a partial ack arrives, force the retransmission of the * next unacknowledged segment. Do not clear tp->t_dupacks. * By setting snd_nxt to ti_ack, this forces retransmission timer to * be started again. */ static void tcp_newreno_partial_ack(struct tcpcb *tp, struct tcphdr *th) { tcp_seq onxt = tp->snd_nxt; u_long ocwnd = tp->snd_cwnd; INP_WLOCK_ASSERT(tp->t_inpcb); tcp_timer_activate(tp, TT_REXMT, 0); tp->t_rtttime = 0; tp->snd_nxt = th->th_ack; /* * Set snd_cwnd to one segment beyond acknowledged offset. * (tp->snd_una has not yet been updated when this function is called.) */ tp->snd_cwnd = tp->t_maxseg + BYTES_THIS_ACK(tp, th); tp->t_flags |= TF_ACKNOW; (void) tcp_output(tp); tp->snd_cwnd = ocwnd; if (SEQ_GT(onxt, tp->snd_nxt)) tp->snd_nxt = onxt; /* * Partial window deflation. Relies on fact that tp->snd_una * not updated yet. */ if (tp->snd_cwnd > BYTES_THIS_ACK(tp, th)) tp->snd_cwnd -= BYTES_THIS_ACK(tp, th); else tp->snd_cwnd = 0; tp->snd_cwnd += tp->t_maxseg; }