UPDATING revision 225757
1Updating Information for FreeBSD current users 2 3This file is maintained and copyrighted by M. Warner Losh <imp@freebsd.org>. 4See end of file for further details. For commonly done items, please see the 5COMMON ITEMS: section later in the file. These instructions assume that you 6basically know what you are doing. If not, then please consult the FreeBSD 7handbook. 8 9Items affecting the ports and packages system can be found in 10/usr/ports/UPDATING. Please read that file before running portupgrade. 11 12NOTE TO PEOPLE WHO THINK THAT FreeBSD 10.x IS SLOW: 13 FreeBSD 10.x has many debugging features turned on, in both the kernel 14 and userland. These features attempt to detect incorrect use of 15 system primitives, and encourage loud failure through extra sanity 16 checking and fail stop semantics. They also substantially impact 17 system performance. If you want to do performance measurement, 18 benchmarking, and optimization, you'll want to turn them off. This 19 includes various WITNESS- related kernel options, INVARIANTS, malloc 20 debugging flags in userland, and various verbose features in the 21 kernel. Many developers choose to disable these features on build 22 machines to maximize performance. (To disable malloc debugging, run 23 ln -s aj /etc/malloc.conf.) 24 2520110923: 26 The stable/9 branch created in subversion. This corresponds to the 27 RELENG_9 branch in CVS. 28 2920110913: 30 This commit modifies vfs_register() so that it uses a hash 31 calculation to set vfc_typenum, which is enabled by default. 32 The first time a system is booted after this change, the 33 vfc_typenum values will change for all file systems. The 34 main effect of this is a change to the NFS server file handles 35 for file systems that use vfc_typenum in their fsid, such as ZFS. 36 It will, however, prevent vfc_typenum from changing when file 37 systems are loaded in a different order for subsequent reboots. 38 To disable this, you can set vfs.typenumhash=0 in /boot/loader.conf 39 until you are ready to remount all NFS clients after a reboot. 40 4120110828: 42 Bump the shared library version numbers for libraries that 43 do not use symbol versioning, have changed the ABI compared 44 to stable/8 and which shared library version was not bumped. 45 Done as part of 9.0-RELEASE cycle. 46 4720110815: 48 During the merge of Capsicum features, the fget(9) KPI was modified. 49 This may require the rebuilding of out-of-tree device drivers -- 50 issues have been reported specifically with the nVidia device driver. 51 __FreeBSD_version is bumped to 900041. 52 53 Also, there is a period between 20110811 and 20110814 where the 54 special devices /dev/{stdin,stdout,stderr} did not work correctly. 55 Building world from a kernel during that window may not work. 56 5720110628: 58 The packet filter (pf) code has been updated to OpenBSD 4.5. 59 You need to update userland tools to be in sync with kernel. 60 This update breaks backward compatibility with earlier pfsync(4) 61 versions. Care must be taken when updating redundant firewall setups. 62 6320110608: 64 The following sysctls and tunables are retired on x86 platforms: 65 machdep.hlt_cpus 66 machdep.hlt_logical_cpus 67 The following sysctl is retired: 68 machdep.hyperthreading_allowed 69 The sysctls were supposed to provide a way to dynamically offline and 70 online selected CPUs on x86 platforms, but the implementation has not 71 been reliable especially with SCHED_ULE scheduler. 72 machdep.hyperthreading_allowed tunable is still available to ignore 73 hyperthreading CPUs at OS level. 74 Individual CPUs can be disabled using hint.lapic.X.disabled tunable, 75 where X is an APIC ID of a CPU. Be advised, though, that disabling 76 CPUs in non-uniform fashion will result in non-uniform topology and 77 may lead to sub-optimal system performance with SCHED_ULE, which is 78 a default scheduler. 79 8020110607: 81 cpumask_t type is retired and cpuset_t is used in order to describe 82 a mask of CPUs. 83 8420110531: 85 Changes to ifconfig(8) for dynamic address family detection mandate 86 that you are running a kernel of 20110525 or later. Make sure to 87 follow the update procedure to boot a new kernel before installing 88 world. 89 9020110513: 91 Support for sun4v architecture is officially dropped 92 9320110503: 94 Several KPI breaking changes have been committed to the mii(4) layer, 95 the PHY drivers and consequently some Ethernet drivers using mii(4). 96 This means that miibus.ko and the modules of the affected Ethernet 97 drivers need to be recompiled. 98 99 Note to kernel developers: Given that the OUI bit reversion problem 100 was fixed as part of these changes all mii(4) commits related to OUIs, 101 i.e. to sys/dev/mii/miidevs, PHY driver probing and vendor specific 102 handling, no longer can be merged verbatim to stable/8 and previous 103 branches. 104 10520110430: 106 Users of the Atheros AR71xx SoC code now need to add 'device ar71xx_pci' 107 into their kernel configurations along with 'device pci'. 108 10920110427: 110 The default NFS client is now the new NFS client, so fstype "newnfs" 111 is now "nfs" and the regular/old NFS client is now fstype "oldnfs". 112 Although mounts via fstype "nfs" will usually work without userland 113 changes, it is recommended that the mount(8) and mount_nfs(8) 114 commands be rebuilt from sources and that a link to mount_nfs called 115 mount_oldnfs be created. The new client is compiled into the 116 kernel with "options NFSCL" and this is needed for diskless root 117 file systems. The GENERIC kernel configs have been changed to use 118 NFSCL and NFSD (the new server) instead of NFSCLIENT and NFSSERVER. 119 To use the regular/old client, you can "mount -t oldnfs ...". For 120 a diskless root file system, you must also include a line like: 121 122 vfs.root.mountfrom="oldnfs:" 123 124 in the boot/loader.conf on the root fs on the NFS server to make 125 a diskless root fs use the old client. 126 12720110424: 128 The GENERIC kernels for all architectures now default to the new 129 CAM-based ATA stack. It means that all legacy ATA drivers were 130 removed and replaced by respective CAM drivers. If you are using 131 ATA device names in /etc/fstab or other places, make sure to update 132 them respectively (adX -> adaY, acdX -> cdY, afdX -> daY, astX -> saY, 133 where 'Y's are the sequential numbers starting from zero for each type 134 in order of detection, unless configured otherwise with tunables, 135 see cam(4)). There will be symbolic links created in /dev/ to map 136 old adX devices to the respective adaY. They should provide basic 137 compatibility for file systems mounting in most cases, but they do 138 not support old user-level APIs and do not have respective providers 139 in GEOM. Consider using updated management tools with new device names. 140 141 It is possible to load devices ahci, ata, siis and mvs as modules, 142 but option ATA_CAM should remain in kernel configuration to make ata 143 module work as CAM driver supporting legacy ATA controllers. Device ata 144 still can be used in modular fashion (atacore + ...). Modules atadisk 145 and atapi* are not used and won't affect operation in ATA_CAM mode. 146 Note that to use CAM-based ATA kernel should include CAM devices 147 scbus, pass, da (or explicitly ada), cd and optionally others. All of 148 them are parts of the cam module. 149 150 ataraid(4) functionality is now supported by the RAID GEOM class. 151 To use it you can load geom_raid kernel module and use graid(8) tool 152 for management. Instead of /dev/arX device names, use /dev/raid/rX. 153 154 No kernel config options or code have been removed, so if a problem 155 arises, please report it and optionally revert to the old ATA stack. 156 In order to do it you can remove from the kernel config: 157 options ATA_CAM 158 device ahci 159 device mvs 160 device siis 161 , and instead add back: 162 device atadisk # ATA disk drives 163 device ataraid # ATA RAID drives 164 device atapicd # ATAPI CDROM drives 165 device atapifd # ATAPI floppy drives 166 device atapist # ATAPI tape drives 167 16820110423: 169 The default NFS server has been changed to the new server, which 170 was referred to as the experimental server. If you need to switch 171 back to the old NFS server, you must now put the "-o" option on 172 both the mountd and nfsd commands. This can be done using the 173 mountd_flags and nfs_server_flags rc.conf variables until an 174 update to the rc scripts is committed, which is coming soon. 175 17620110418: 177 The GNU Objective-C runtime library (libobjc), and other Objective-C 178 related components have been removed from the base system. If you 179 require an Objective-C library, please use one of the available ports. 180 18120110331: 182 ath(4) has been split into bus- and device- modules. if_ath contains 183 the HAL, the TX rate control and the network device code. if_ath_pci 184 contains the PCI bus glue. For Atheros MIPS embedded systems, if_ath_ahb 185 contains the AHB glue. Users need to load both if_ath_pci and if_ath 186 in order to use ath on everything else. 187 188 TO REPEAT: if_ath_ahb is not needed for normal users. Normal users only 189 need to load if_ath and if_ath_pci for ath(4) operation. 190 19120110314: 192 As part of the replacement of sysinstall, the process of building 193 release media has changed significantly. For details, please re-read 194 release(7), which has been updated to reflect the new build process. 195 19620110218: 197 GNU binutils 2.17.50 (as of 2007-07-03) has been merged to -HEAD. This 198 is the last available version under GPLv2. It brings a number of new 199 features, such as support for newer x86 CPU's (with SSE-3, SSSE-3, SSE 200 4.1 and SSE 4.2), better support for powerpc64, a number of new 201 directives, and lots of other small improvements. See the ChangeLog 202 file in contrib/binutils for the full details. 203 20420110218: 205 IPsec's HMAC_SHA256-512 support has been fixed to be RFC4868 206 compliant, and will now use half of hash for authentication. 207 This will break interoperability with all stacks (including all 208 actual FreeBSD versions) who implement 209 draft-ietf-ipsec-ciph-sha-256-00 (they use 96 bits of hash for 210 authentication). 211 The only workaround with such peers is to use another HMAC 212 algorithm for IPsec ("phase 2") authentication. 213 21420110207: 215 Remove the uio_yield prototype and symbol. This function has 216 been misnamed since it was introduced and should not be 217 globally exposed with this name. The equivalent functionality 218 is now available using kern_yield(curthread->td_user_pri). 219 The function remains undocumented. 220 22120110112: 222 A SYSCTL_[ADD_]UQUAD was added for unsigned uint64_t pointers, 223 symmetric with the existing SYSCTL_[ADD_]QUAD. Type checking 224 for scalar sysctls is defined but disabled. Code that needs 225 UQUAD to pass the type checking that must compile on older 226 systems where the define is not present can check against 227 __FreeBSD_version >= 900030. 228 229 The system dialog(1) has been replaced with a new version previously 230 in ports as devel/cdialog. dialog(1) is mostly command-line compatible 231 with the previous version, but the libdialog associated with it has 232 a largely incompatible API. As such, the original version of libdialog 233 will be kept temporarily as libodialog, until its base system consumers 234 are replaced or updated. Bump __FreeBSD_version to 900030. 235 23620110103: 237 If you are trying to run make universe on a -stable system, and you get 238 the following warning: 239 "Makefile", line 356: "Target architecture for i386/conf/GENERIC 240 unknown. config(8) likely too old." 241 or something similar to it, then you must upgrade your -stable system 242 to 8.2-Release or newer (really, any time after r210146 7/15/2010 in 243 stable/8) or build the config from the latest stable/8 branch and 244 install it on your system. 245 246 Prior to this date, building a current universe on 8-stable system from 247 between 7/15/2010 and 1/2/2011 would result in a weird shell parsing 248 error in the first kernel build phase. A new config on those old 249 systems will fix that problem for older versions of -current. 250 25120101228: 252 The TCP stack has been modified to allow Khelp modules to interact with 253 it via helper hook points and store per-connection data in the TCP 254 control block. Bump __FreeBSD_version to 900029. User space tools that 255 rely on the size of struct tcpcb in tcp_var.h (e.g. sockstat) need to 256 be recompiled. 257 25820101114: 259 Generic IEEE 802.3 annex 31B full duplex flow control support has been 260 added to mii(4) and bge(4), bce(4), msk(4), nfe(4) and stge(4) along 261 with brgphy(4), e1000phy(4) as well as ip1000phy() have been converted 262 to take advantage of it instead of using custom implementations. This 263 means that these drivers now no longer unconditionally advertise 264 support for flow control but only do so if flow control is a selected 265 media option. This was implemented in the generic support that way in 266 order to allow flow control to be switched on and off via ifconfig(8) 267 with the PHY specific default to typically off in order to protect 268 from unwanted effects. Consequently, if you used flow control with 269 one of the above mentioned drivers you now need to explicitly enable 270 it, for example via: 271 ifconfig bge0 media auto mediaopt flowcontrol 272 273 Along with the above mentioned changes generic support for setting 274 1000baseT master mode also has been added and brgphy(4), ciphy(4), 275 e1000phy(4) as well as ip1000phy(4) have been converted to take 276 advantage of it. This means that these drivers now no longer take the 277 link0 parameter for selecting master mode but the master media option 278 has to be used instead, for example like in the following: 279 ifconfig bge0 media 1000baseT mediaopt full-duplex,master 280 281 Selection of master mode now is also available with all other PHY 282 drivers supporting 1000baseT. 283 28420101111: 285 The TCP stack has received a significant update to add support for 286 modularised congestion control and generally improve the clarity of 287 congestion control decisions. Bump __FreeBSD_version to 900025. User 288 space tools that rely on the size of struct tcpcb in tcp_var.h (e.g. 289 sockstat) need to be recompiled. 290 29120101002: 292 The man(1) utility has been replaced by a new version that no longer 293 uses /etc/manpath.config. Please consult man.conf(5) for how to 294 migrate local entries to the new format. 295 29620100928: 297 The copyright strings printed by login(1) and sshd(8) at the time of a 298 new connection have been removed to follow other operating systems and 299 upstream sshd. 300 30120100915: 302 A workaround for a fixed ld bug has been removed in kernel code, 303 so make sure that your system ld is built from sources after 304 revision 210245 from 2010-07-19 (r211583 if building head kernel 305 on stable/8, r211584 for stable/7; both from 2010-08-21). 306 A symptom of incorrect ld version is different addresses for 307 set_pcpu section and __start_set_pcpu symbol in kernel and/or modules. 308 30920100913: 310 The $ipv6_prefer variable in rc.conf(5) has been split into 311 $ip6addrctl_policy and $ipv6_activate_all_interfaces. 312 313 The $ip6addrctl_policy is a variable to choose a pre-defined 314 address selection policy set by ip6addrctl(8). A value 315 "ipv4_prefer", "ipv6_prefer" or "AUTO" can be specified. The 316 default is "AUTO". 317 318 The $ipv6_activate_all_interfaces specifies whether IFDISABLED 319 flag (see an entry of 20090926) is set on an interface with no 320 corresponding $ifconfig_IF_ipv6 line. The default is "NO" for 321 security reason. If you want IPv6 link-local address on all 322 interfaces by default, set this to "YES". 323 324 The old ipv6_prefer="YES" is equivalent to 325 ipv6_activate_all_interfaces="YES" and 326 ip6addrctl_policy="ipv6_prefer". 327 32820100913: 329 DTrace has grown support for userland tracing. Due to this, DTrace is 330 now i386 and amd64 only. 331 dtruss(1) is now installed by default on those systems and a new 332 kernel module is needed for userland tracing: fasttrap. 333 No changes to your kernel config file are necessary to enable 334 userland tracing, but you might consider adding 'STRIP=' and 335 'CFLAGS+=-fno-omit-frame-pointer' to your make.conf if you want 336 to have informative userland stack traces in DTrace (ustack). 337 33820100725: 339 The acpi_aiboost(4) driver has been removed in favor of the new 340 aibs(4) driver. You should update your kernel configuration file. 341 34220100722: 343 BSD grep has been imported to the base system and it is built by 344 default. It is completely BSD licensed, highly GNU-compatible, uses 345 less memory than its GNU counterpart and has a small codebase. 346 However, it is slower than its GNU counterpart, which is mostly 347 noticeable for larger searches, for smaller ones it is measurable 348 but not significant. The reason is complex, the most important factor 349 is that we lack a modern and efficient regex library and GNU 350 overcomes this by optimizing the searches internally. Future work 351 on improving the regex performance is planned, for the meantime, 352 users that need better performance, can build GNU grep instead by 353 setting the WITH_GNU_GREP knob. 354 35520100713: 356 Due to the import of powerpc64 support, all existing powerpc kernel 357 configuration files must be updated with a machine directive like this: 358 machine powerpc powerpc 359 360 In addition, an updated config(8) is required to build powerpc kernels 361 after this change. 362 36320100713: 364 A new version of ZFS (version 15) has been merged to -HEAD. 365 This version uses a python library for the following subcommands: 366 zfs allow, zfs unallow, zfs groupspace, zfs userspace. 367 For full functionality of these commands the following port must 368 be installed: sysutils/py-zfs 369 37020100429: 371 'vm_page's are now hashed by physical address to an array of mutexes. 372 Currently this is only used to serialize access to hold_count. Over 373 time the page queue mutex will be peeled away. This changes the size 374 of pmap on every architecture. And requires all callers of vm_page_hold 375 and vm_page_unhold to be updated. 376 37720100402: 378 WITH_CTF can now be specified in src.conf (not recommended, there 379 are some problems with static executables), make.conf (would also 380 affect ports which do not use GNU make and do not override the 381 compile targets) or in the kernel config (via "makeoptions 382 WITH_CTF=yes"). 383 When WITH_CTF was specified there before this was silently ignored, 384 so make sure that WITH_CTF is not used in places which could lead 385 to unwanted behavior. 386 38720100311: 388 The kernel option COMPAT_IA32 has been replaced with COMPAT_FREEBSD32 389 to allow 32-bit compatibility on non-x86 platforms. All kernel 390 configurations on amd64 and ia64 platforms using these options must 391 be modified accordingly. 392 39320100113: 394 The utmp user accounting database has been replaced with utmpx, 395 the user accounting interface standardized by POSIX. 396 Unfortunately the semantics of utmp and utmpx don't match, 397 making it practically impossible to support both interfaces. 398 The user accounting database is used by tools like finger(1), 399 last(1), talk(1), w(1) and ac(8). 400 401 All applications in the base system use utmpx. This means only 402 local binaries (e.g. from the ports tree) may still use these 403 utmp database files. These applications must be rebuilt to make 404 use of utmpx. 405 406 After the system has been upgraded, it is safe to remove the old 407 log files (/var/run/utmp, /var/log/lastlog and /var/log/wtmp*), 408 assuming their contents is of no importance anymore. Old wtmp 409 databases can only be used by last(1) and ac(8) after they have 410 been converted to the new format using wtmpcvt(1). 411 41220100108: 413 Introduce the kernel thread "deadlock resolver" (which can be enabled 414 via the DEADLKRES option, see NOTES for more details) and the 415 sleepq_type() function for sleepqueues. 416 41720091202: 418 The rc.firewall and rc.firewall6 were unified, and 419 rc.firewall6 and rc.d/ip6fw were removed. 420 According to the removal of rc.d/ip6fw, ipv6_firewall_* rc 421 variables are obsoleted. Instead, the following new rc 422 variables are added to rc.d/ipfw: 423 424 firewall_client_net_ipv6, firewall_simple_iif_ipv6, 425 firewall_simple_inet_ipv6, firewall_simple_oif_ipv6, 426 firewall_simple_onet_ipv6, firewall_trusted_ipv6 427 428 The meanings correspond to the relevant IPv4 variables. 429 43020091125: 431 8.0-RELEASE. 432 43320091113: 434 The default terminal emulation for syscons(4) has been changed 435 from cons25 to xterm on all platforms except pc98. This means 436 that the /etc/ttys file needs to be updated to ensure correct 437 operation of applications on the console. 438 439 The terminal emulation style can be toggled per window by using 440 vidcontrol(1)'s -T flag. The TEKEN_CONS25 kernel configuration 441 options can be used to change the compile-time default back to 442 cons25. 443 444 To prevent graphical artifacts, make sure the TERM environment 445 variable is set to match the terminal emulation that is being 446 performed by syscons(4). 447 44820091109: 449 The layout of the structure ieee80211req_scan_result has changed. 450 Applications that require wireless scan results (e.g. ifconfig(8)) 451 from net80211 need to be recompiled. 452 453 Applications such as wpa_supplicant(8) may require a full world 454 build without using NO_CLEAN in order to get synchronized with the 455 new structure. 456 45720091025: 458 The iwn(4) driver has been updated to support the 5000 and 5150 series. 459 There's one kernel module for each firmware. Adding "device iwnfw" 460 to the kernel configuration file means including all three firmware 461 images inside the kernel. If you want to include just the one for 462 your wireless card, use the the devices iwn4965fw, iwn5000fw or 463 iwn5150fw. 464 46520090926: 466 The rc.d/network_ipv6, IPv6 configuration script has been integrated 467 into rc.d/netif. The changes are the following: 468 469 1. To use IPv6, simply define $ifconfig_IF_ipv6 like $ifconfig_IF 470 for IPv4. For aliases, $ifconfig_IF_aliasN should be used. 471 Note that both variables need the "inet6" keyword at the head. 472 473 Do not set $ipv6_network_interfaces manually if you do not 474 understand what you are doing. It is not needed in most cases. 475 476 $ipv6_ifconfig_IF and $ipv6_ifconfig_IF_aliasN still work, but 477 they are obsolete. 478 479 2. $ipv6_enable is obsolete. Use $ipv6_prefer and 480 "inet6 accept_rtadv" keyword in ifconfig(8) instead. 481 482 If you define $ipv6_enable=YES, it means $ipv6_prefer=YES and 483 all configured interfaces have "inet6 accept_rtadv" in the 484 $ifconfig_IF_ipv6. These are for backward compatibility. 485 486 3. A new variable $ipv6_prefer has been added. If NO, IPv6 487 functionality of interfaces with no corresponding 488 $ifconfig_IF_ipv6 is disabled by using "inet6 ifdisabled" flag, 489 and the default address selection policy of ip6addrctl(8) 490 is the IPv4-preferred one (see rc.d/ip6addrctl for more details). 491 Note that if you want to configure IPv6 functionality on the 492 disabled interfaces after boot, first you need to clear the flag by 493 using ifconfig(8) like: 494 495 ifconfig em0 inet6 -ifdisabled 496 497 If YES, the default address selection policy is set as 498 IPv6-preferred. 499 500 The default value of $ipv6_prefer is NO. 501 502 4. If your system need to receive Router Advertisement messages, 503 define "inet6 accept_rtadv" in $ifconfig_IF_ipv6. The rc(8) 504 scripts automatically invoke rtsol(8) when the interface becomes 505 UP. The Router Advertisement messages are used for SLAAC 506 (State-Less Address AutoConfiguration). 507 50820090922: 509 802.11s D3.03 support was committed. This is incompatible with the 510 previous code, which was based on D3.0. 511 51220090912: 513 A sysctl variable net.inet6.ip6.accept_rtadv now sets the default value 514 of a per-interface flag ND6_IFF_ACCEPT_RTADV, not a global knob to 515 control whether accepting Router Advertisement messages or not. 516 Also, a per-interface flag ND6_IFF_AUTO_LINKLOCAL has been added and 517 a sysctl variable net.inet6.ip6.auto_linklocal is its default value. 518 The ifconfig(8) utility now supports these flags. 519 52020090910: 521 ZFS snapshots are now mounted with MNT_IGNORE flag. Use -v option for 522 mount(8) and -a option for df(1) to see them. 523 52420090825: 525 The old tunable hw.bus.devctl_disable has been superseded by 526 hw.bus.devctl_queue. hw.bus.devctl_disable=1 in loader.conf should be 527 replaced by hw.bus.devctl_queue=0. The default for this new tunable 528 is 1000. 529 53020090813: 531 Remove the option STOP_NMI. The default action is now to use NMI only 532 for KDB via the newly introduced function stop_cpus_hard() and 533 maintain stop_cpus() to just use a normal IPI_STOP on ia32 and amd64. 534 53520090803: 536 The stable/8 branch created in subversion. This corresponds to the 537 RELENG_8 branch in CVS. 538 53920090719: 540 Bump the shared library version numbers for all libraries that do not 541 use symbol versioning as part of the 8.0-RELEASE cycle. Bump 542 __FreeBSD_version to 800105. 543 54420090714: 545 Due to changes in the implementation of virtual network stack support, 546 all network-related kernel modules must be recompiled. As this change 547 breaks the ABI, bump __FreeBSD_version to 800104. 548 54920090713: 550 The TOE interface to the TCP syncache has been modified to remove 551 struct tcpopt (<netinet/tcp_var.h>) from the ABI of the network stack. 552 The cxgb driver is the only TOE consumer affected by this change, and 553 needs to be recompiled along with the kernel. As this change breaks 554 the ABI, bump __FreeBSD_version to 800103. 555 55620090712: 557 Padding has been added to struct tcpcb, sackhint and tcpstat in 558 <netinet/tcp_var.h> to facilitate future MFCs and bug fixes whilst 559 maintaining the ABI. However, this change breaks the ABI, so bump 560 __FreeBSD_version to 800102. User space tools that rely on the size of 561 any of these structs (e.g. sockstat) need to be recompiled. 562 56320090630: 564 The NFS_LEGACYRPC option has been removed along with the old kernel 565 RPC implementation that this option selected. Kernel configurations 566 may need to be adjusted. 567 56820090629: 569 The network interface device nodes at /dev/net/<interface> have been 570 removed. All ioctl operations can be performed the normal way using 571 routing sockets. The kqueue functionality can generally be replaced 572 with routing sockets. 573 57420090628: 575 The documentation from the FreeBSD Documentation Project (Handbook, 576 FAQ, etc.) is now installed via packages by sysinstall(8) and under 577 the /usr/local/share/doc/freebsd directory instead of /usr/share/doc. 578 57920090624: 580 The ABI of various structures related to the SYSV IPC API have been 581 changed. As a result, the COMPAT_FREEBSD[456] and COMPAT_43 kernel 582 options now all require COMPAT_FREEBSD7. Bump __FreeBSD_version to 583 800100. 584 58520090622: 586 Layout of struct vnet has changed as routing related variables were 587 moved to their own Vimage module. Modules need to be recompiled. Bump 588 __FreeBSD_version to 800099. 589 59020090619: 591 NGROUPS_MAX and NGROUPS have been increased from 16 to 1023 and 1024 592 respectively. As long as no more than 16 groups per process are used, 593 no changes should be visible. When more than 16 groups are used, old 594 binaries may fail if they call getgroups() or getgrouplist() with 595 statically sized storage. Recompiling will work around this, but 596 applications should be modified to use dynamically allocated storage 597 for group arrays as POSIX.1-2008 does not cap an implementation's 598 number of supported groups at NGROUPS_MAX+1 as previous versions did. 599 600 NFS and portalfs mounts may also be affected as the list of groups is 601 truncated to 16. Users of NFS who use more than 16 groups, should 602 take care that negative group permissions are not used on the exported 603 file systems as they will not be reliable unless a GSSAPI based 604 authentication method is used. 605 60620090616: 607 The compiling option ADAPTIVE_LOCKMGRS has been introduced. This 608 option compiles in the support for adaptive spinning for lockmgrs 609 which want to enable it. The lockinit() function now accepts the flag 610 LK_ADAPTIVE in order to make the lock object subject to adaptive 611 spinning when both held in write and read mode. 612 61320090613: 614 The layout of the structure returned by IEEE80211_IOC_STA_INFO has 615 changed. User applications that use this ioctl need to be rebuilt. 616 61720090611: 618 The layout of struct thread has changed. Kernel and modules need to 619 be rebuilt. 620 62120090608: 622 The layout of structs ifnet, domain, protosw and vnet_net has changed. 623 Kernel modules need to be rebuilt. Bump __FreeBSD_version to 800097. 624 62520090602: 626 window(1) has been removed from the base system. It can now be 627 installed from ports. The port is called misc/window. 628 62920090601: 630 The way we are storing and accessing `routing table' entries has 631 changed. Programs reading the FIB, like netstat, need to be 632 re-compiled. 633 63420090601: 635 A new netisr implementation has been added for FreeBSD 8. Network 636 file system modules, such as igmp, ipdivert, and others, should be 637 rebuilt. 638 Bump __FreeBSD_version to 800096. 639 64020090530: 641 Remove the tunable/sysctl debug.mpsafevfs as its initial purpose is no 642 more valid. 643 64420090530: 645 Add VOP_ACCESSX(9). File system modules need to be rebuilt. 646 Bump __FreeBSD_version to 800094. 647 64820090529: 649 Add mnt_xflag field to 'struct mount'. File system modules need to be 650 rebuilt. 651 Bump __FreeBSD_version to 800093. 652 65320090528: 654 The compiling option ADAPTIVE_SX has been retired while it has been 655 introduced the option NO_ADAPTIVE_SX which handles the reversed logic. 656 The KPI for sx_init_flags() changes as accepting flags: 657 SX_ADAPTIVESPIN flag has been retired while the SX_NOADAPTIVE flag has 658 been introduced in order to handle the reversed logic. 659 Bump __FreeBSD_version to 800092. 660 66120090527: 662 Add support for hierarchical jails. Remove global securelevel. 663 Bump __FreeBSD_version to 800091. 664 66520090523: 666 The layout of struct vnet_net has changed, therefore modules 667 need to be rebuilt. 668 Bump __FreeBSD_version to 800090. 669 67020090523: 671 The newly imported zic(8) produces a new format in the output. Please 672 run tzsetup(8) to install the newly created data to /etc/localtime. 673 67420090520: 675 The sysctl tree for the usb stack has renamed from hw.usb2.* to 676 hw.usb.* and is now consistent again with previous releases. 677 67820090520: 679 802.11 monitor mode support was revised and driver api's were changed. 680 Drivers dependent on net80211 now support DLT_IEEE802_11_RADIO instead 681 of DLT_IEEE802_11. No user-visible data structures were changed but 682 applications that use DLT_IEEE802_11 may require changes. 683 Bump __FreeBSD_version to 800088. 684 68520090430: 686 The layout of the following structs has changed: sysctl_oid, 687 socket, ifnet, inpcbinfo, tcpcb, syncache_head, vnet_inet, 688 vnet_inet6 and vnet_ipfw. Most modules need to be rebuild or 689 panics may be experienced. World rebuild is required for 690 correctly checking networking state from userland. 691 Bump __FreeBSD_version to 800085. 692 69320090429: 694 MLDv2 and Source-Specific Multicast (SSM) have been merged 695 to the IPv6 stack. VIMAGE hooks are in but not yet used. 696 The implementation of SSM within FreeBSD's IPv6 stack closely 697 follows the IPv4 implementation. 698 699 For kernel developers: 700 701 * The most important changes are that the ip6_output() and 702 ip6_input() paths no longer take the IN6_MULTI_LOCK, 703 and this lock has been downgraded to a non-recursive mutex. 704 705 * As with the changes to the IPv4 stack to support SSM, filtering 706 of inbound multicast traffic must now be performed by transport 707 protocols within the IPv6 stack. This does not apply to TCP and 708 SCTP, however, it does apply to UDP in IPv6 and raw IPv6. 709 710 * The KPIs used by IPv6 multicast are similar to those used by 711 the IPv4 stack, with the following differences: 712 * im6o_mc_filter() is analogous to imo_multicast_filter(). 713 * The legacy KAME entry points in6_joingroup and in6_leavegroup() 714 are shimmed to in6_mc_join() and in6_mc_leave() respectively. 715 * IN6_LOOKUP_MULTI() has been deprecated and removed. 716 * IPv6 relies on MLD for the DAD mechanism. KAME's internal KPIs 717 for MLDv1 have an additional 'timer' argument which is used to 718 jitter the initial membership report for the solicited-node 719 multicast membership on-link. 720 * This is not strictly needed for MLDv2, which already jitters 721 its report transmissions. However, the 'timer' argument is 722 preserved in case MLDv1 is active on the interface. 723 724 * The KAME linked-list based IPv6 membership implementation has 725 been refactored to use a vector similar to that used by the IPv4 726 stack. 727 Code which maintains a list of its own multicast memberships 728 internally, e.g. carp, has been updated to reflect the new 729 semantics. 730 731 * There is a known Lock Order Reversal (LOR) due to in6_setscope() 732 acquiring the IF_AFDATA_LOCK and being called within ip6_output(). 733 Whilst MLDv2 tries to avoid this otherwise benign LOR, it is an 734 implementation constraint which needs to be addressed in HEAD. 735 736 For application developers: 737 738 * The changes are broadly similar to those made for the IPv4 739 stack. 740 741 * The use of IPv4 and IPv6 multicast socket options on the same 742 socket, using mapped addresses, HAS NOT been tested or supported. 743 744 * There are a number of issues with the implementation of various 745 IPv6 multicast APIs which need to be resolved in the API surface 746 before the implementation is fully compatible with KAME userland 747 use, and these are mostly to do with interface index treatment. 748 749 * The literature available discusses the use of either the delta / ASM 750 API with setsockopt(2)/getsockopt(2), or the full-state / ASM API 751 using setsourcefilter(3)/getsourcefilter(3). For more information 752 please refer to RFC 3768, 'Socket Interface Extensions for 753 Multicast Source Filters'. 754 755 * Applications which use the published RFC 3678 APIs should be fine. 756 757 For systems administrators: 758 759 * The mtest(8) utility has been refactored to support IPv6, in 760 addition to IPv4. Interface addresses are no longer accepted 761 as arguments, their names must be used instead. The utility 762 will map the interface name to its first IPv4 address as 763 returned by getifaddrs(3). 764 765 * The ifmcstat(8) utility has also been updated to print the MLDv2 766 endpoint state and source filter lists via sysctl(3). 767 768 * The net.inet6.ip6.mcast.loop sysctl may be tuned to 0 to disable 769 loopback of IPv6 multicast datagrams by default; it defaults to 1 770 to preserve the existing behaviour. Disabling multicast loopback is 771 recommended for optimal system performance. 772 773 * The IPv6 MROUTING code has been changed to examine this sysctl 774 instead of attempting to perform a group lookup before looping 775 back forwarded datagrams. 776 777 Bump __FreeBSD_version to 800084. 778 77920090422: 780 Implement low-level Bluetooth HCI API. 781 Bump __FreeBSD_version to 800083. 782 78320090419: 784 The layout of struct malloc_type, used by modules to register new 785 memory allocation types, has changed. Most modules will need to 786 be rebuilt or panics may be experienced. 787 Bump __FreeBSD_version to 800081. 788 78920090415: 790 Anticipate overflowing inp_flags - add inp_flags2. 791 This changes most offsets in inpcb, so checking v4 connection 792 state will require a world rebuild. 793 Bump __FreeBSD_version to 800080. 794 79520090415: 796 Add an llentry to struct route and struct route_in6. Modules 797 embedding a struct route will need to be recompiled. 798 Bump __FreeBSD_version to 800079. 799 80020090414: 801 The size of rt_metrics_lite and by extension rtentry has changed. 802 Networking administration apps will need to be recompiled. 803 The route command now supports show as an alias for get, weighting 804 of routes, sticky and nostick flags to alter the behavior of stateful 805 load balancing. 806 Bump __FreeBSD_version to 800078. 807 80820090408: 809 Do not use Giant for kbdmux(4) locking. This is wrong and 810 apparently causing more problems than it solves. This will 811 re-open the issue where interrupt handlers may race with 812 kbdmux(4) in polling mode. Typical symptoms include (but 813 not limited to) duplicated and/or missing characters when 814 low level console functions (such as gets) are used while 815 interrupts are enabled (for example geli password prompt, 816 mountroot prompt etc.). Disabling kbdmux(4) may help. 817 81820090407: 819 The size of structs vnet_net, vnet_inet and vnet_ipfw has changed; 820 kernel modules referencing any of the above need to be recompiled. 821 Bump __FreeBSD_version to 800075. 822 82320090320: 824 GEOM_PART has become the default partition slicer for storage devices, 825 replacing GEOM_MBR, GEOM_BSD, GEOM_PC98 and GEOM_GPT slicers. It 826 introduces some changes: 827 828 MSDOS/EBR: the devices created from MSDOS extended partition entries 829 (EBR) can be named differently than with GEOM_MBR and are now symlinks 830 to devices with offset-based names. fstabs may need to be modified. 831 832 BSD: the "geometry does not match label" warning is harmless in most 833 cases but it points to problems in file system misalignment with 834 disk geometry. The "c" partition is now implicit, covers the whole 835 top-level drive and cannot be (mis)used by users. 836 837 General: Kernel dumps are now not allowed to be written to devices 838 whose partition types indicate they are meant to be used for file 839 systems (or, in case of MSDOS partitions, as something else than 840 the "386BSD" type). 841 842 Most of these changes date approximately from 200812. 843 84420090319: 845 The uscanner(4) driver has been removed from the kernel. This follows 846 Linux removing theirs in 2.6 and making libusb the default interface 847 (supported by sane). 848 84920090319: 850 The multicast forwarding code has been cleaned up. netstat(1) 851 only relies on KVM now for printing bandwidth upcall meters. 852 The IPv4 and IPv6 modules are split into ip_mroute_mod and 853 ip6_mroute_mod respectively. The config(5) options for statically 854 compiling this code remain the same, i.e. 'options MROUTING'. 855 85620090315: 857 Support for the IFF_NEEDSGIANT network interface flag has been 858 removed, which means that non-MPSAFE network device drivers are no 859 longer supported. In particular, if_ar, if_sr, and network device 860 drivers from the old (legacy) USB stack can no longer be built or 861 used. 862 86320090313: 864 POSIX.1 Native Language Support (NLS) has been enabled in libc and 865 a bunch of new language catalog files have also been added. 866 This means that some common libc messages are now localized and 867 they depend on the LC_MESSAGES environmental variable. 868 86920090313: 870 The k8temp(4) driver has been renamed to amdtemp(4) since 871 support for Family 10 and Family 11 CPU families was added. 872 87320090309: 874 IGMPv3 and Source-Specific Multicast (SSM) have been merged 875 to the IPv4 stack. VIMAGE hooks are in but not yet used. 876 877 For kernel developers, the most important changes are that the 878 ip_output() and ip_input() paths no longer take the IN_MULTI_LOCK(), 879 and this lock has been downgraded to a non-recursive mutex. 880 881 Transport protocols (UDP, Raw IP) are now responsible for filtering 882 inbound multicast traffic according to group membership and source 883 filters. The imo_multicast_filter() KPI exists for this purpose. 884 Transports which do not use multicast (SCTP, TCP) already reject 885 multicast by default. Forwarding and receive performance may improve 886 as a mutex acquisition is no longer needed in the ip_input() 887 low-level input path. in_addmulti() and in_delmulti() are shimmed 888 to new KPIs which exist to support SSM in-kernel. 889 890 For application developers, it is recommended that loopback of 891 multicast datagrams be disabled for best performance, as this 892 will still cause the lock to be taken for each looped-back 893 datagram transmission. The net.inet.ip.mcast.loop sysctl may 894 be tuned to 0 to disable loopback by default; it defaults to 1 895 to preserve the existing behaviour. 896 897 For systems administrators, to obtain best performance with 898 multicast reception and multiple groups, it is always recommended 899 that a card with a suitably precise hash filter is used. Hash 900 collisions will still result in the lock being taken within the 901 transport protocol input path to check group membership. 902 903 If deploying FreeBSD in an environment with IGMP snooping switches, 904 it is recommended that the net.inet.igmp.sendlocal sysctl remain 905 enabled; this forces 224.0.0.0/24 group membership to be announced 906 via IGMP. 907 908 The size of 'struct igmpstat' has changed; netstat needs to be 909 recompiled to reflect this. 910 Bump __FreeBSD_version to 800070. 911 91220090309: 913 libusb20.so.1 is now installed as libusb.so.1 and the ports system 914 updated to use it. This requires a buildworld/installworld in order to 915 update the library and dependencies (usbconfig, etc). Its advisable to 916 rebuild all ports which uses libusb. More specific directions are given 917 in the ports collection UPDATING file. Any /etc/libmap.conf entries for 918 libusb are no longer required and can be removed. 919 92020090302: 921 A workaround is committed to allow the creation of System V shared 922 memory segment of size > 2 GB on the 64-bit architectures. 923 Due to a limitation of the existing ABI, the shm_segsz member 924 of the struct shmid_ds, returned by shmctl(IPC_STAT) call is 925 wrong for large segments. Note that limits must be explicitly 926 raised to allow such segments to be created. 927 92820090301: 929 The layout of struct ifnet has changed, requiring a rebuild of all 930 network device driver modules. 931 93220090227: 933 The /dev handling for the new USB stack has changed, a 934 buildworld/installworld is required for libusb20. 935 93620090223: 937 The new USB2 stack has now been permanently moved in and all kernel and 938 module names reverted to their previous values (eg, usb, ehci, ohci, 939 ums, ...). The old usb stack can be compiled in by prefixing the name 940 with the letter 'o', the old usb modules have been removed. 941 Updating entry 20090216 for xorg and 20090215 for libmap may still 942 apply. 943 94420090217: 945 The rc.conf(5) option if_up_delay has been renamed to 946 defaultroute_delay to better reflect its purpose. If you have 947 customized this setting in /etc/rc.conf you need to update it to 948 use the new name. 949 95020090216: 951 xorg 7.4 wants to configure its input devices via hald which does not 952 yet work with USB2. If the keyboard/mouse does not work in xorg then 953 add 954 Option "AllowEmptyInput" "off" 955 to your ServerLayout section. This will cause X to use the configured 956 kbd and mouse sections from your xorg.conf. 957 95820090215: 959 The GENERIC kernels for all architectures now default to the new USB2 960 stack. No kernel config options or code have been removed so if a 961 problem arises please report it and optionally revert to the old USB 962 stack. If you are loading USB kernel modules or have a custom kernel 963 that includes GENERIC then ensure that usb names are also changed over, 964 eg uftdi -> usb2_serial_ftdi. 965 966 Older programs linked against the ports libusb 0.1 need to be 967 redirected to the new stack's libusb20. /etc/libmap.conf can 968 be used for this: 969 # Map old usb library to new one for usb2 stack 970 libusb-0.1.so.8 libusb20.so.1 971 97220090209: 973 All USB ethernet devices now attach as interfaces under the name ueN 974 (eg. ue0). This is to provide a predictable name as vendors often 975 change usb chipsets in a product without notice. 976 97720090203: 978 The ichsmb(4) driver has been changed to require SMBus slave 979 addresses be left-justified (xxxxxxx0b) rather than right-justified. 980 All of the other SMBus controller drivers require left-justified 981 slave addresses, so this change makes all the drivers provide the 982 same interface. 983 98420090201: 985 INET6 statistics (struct ip6stat) was updated. 986 netstat(1) needs to be recompiled. 987 98820090119: 989 NTFS has been removed from GENERIC kernel on amd64 to match 990 GENERIC on i386. Should not cause any issues since mount_ntfs(8) 991 will load ntfs.ko module automatically when NTFS support is 992 actually needed, unless ntfs.ko is not installed or security 993 level prohibits loading kernel modules. If either is the case, 994 "options NTFS" has to be added into kernel config. 995 99620090115: 997 TCP Appropriate Byte Counting (RFC 3465) support added to kernel. 998 New field in struct tcpcb breaks ABI, so bump __FreeBSD_version to 999 800061. User space tools that rely on the size of struct tcpcb in 1000 tcp_var.h (e.g. sockstat) need to be recompiled. 1001 100220081225: 1003 ng_tty(4) module updated to match the new TTY subsystem. 1004 Due to API change, user-level applications must be updated. 1005 New API support added to mpd5 CVS and expected to be present 1006 in next mpd5.3 release. 1007 100820081219: 1009 With __FreeBSD_version 800060 the makefs tool is part of 1010 the base system (it was a port). 1011 101220081216: 1013 The afdata and ifnet locks have been changed from mutexes to 1014 rwlocks, network modules will need to be re-compiled. 1015 101620081214: 1017 __FreeBSD_version 800059 incorporates the new arp-v2 rewrite. 1018 RTF_CLONING, RTF_LLINFO and RTF_WASCLONED flags are eliminated. 1019 The new code reduced struct rtentry{} by 16 bytes on 32-bit 1020 architecture and 40 bytes on 64-bit architecture. The userland 1021 applications "arp" and "ndp" have been updated accordingly. 1022 The output from "netstat -r" shows only routing entries and 1023 none of the L2 information. 1024 102520081130: 1026 __FreeBSD_version 800057 marks the switchover from the 1027 binary ath hal to source code. Users must add the line: 1028 1029 options AH_SUPPORT_AR5416 1030 1031 to their kernel config files when specifying: 1032 1033 device ath_hal 1034 1035 The ath_hal module no longer exists; the code is now compiled 1036 together with the driver in the ath module. It is now 1037 possible to tailor chip support (i.e. reduce the set of chips 1038 and thereby the code size); consult ath_hal(4) for details. 1039 104020081121: 1041 __FreeBSD_version 800054 adds memory barriers to 1042 <machine/atomic.h>, new interfaces to ifnet to facilitate 1043 multiple hardware transmit queues for cards that support 1044 them, and a lock-less ring-buffer implementation to 1045 enable drivers to more efficiently manage queueing of 1046 packets. 1047 104820081117: 1049 A new version of ZFS (version 13) has been merged to -HEAD. 1050 This version has zpool attribute "listsnapshots" off by 1051 default, which means "zfs list" does not show snapshots, 1052 and is the same as Solaris behavior. 1053 105420081028: 1055 dummynet(4) ABI has changed. ipfw(8) needs to be recompiled. 1056 105720081009: 1058 The uhci, ohci, ehci and slhci USB Host controller drivers have 1059 been put into separate modules. If you load the usb module 1060 separately through loader.conf you will need to load the 1061 appropriate *hci module as well. E.g. for a UHCI-based USB 2.0 1062 controller add the following to loader.conf: 1063 1064 uhci_load="YES" 1065 ehci_load="YES" 1066 106720081009: 1068 The ABI used by the PMC toolset has changed. Please keep 1069 userland (libpmc(3)) and the kernel module (hwpmc(4)) in 1070 sync. 1071 107220081009: 1073 atapci kernel module now includes only generic PCI ATA 1074 driver. AHCI driver moved to ataahci kernel module. 1075 All vendor-specific code moved into separate kernel modules: 1076 ataacard, ataacerlabs, ataadaptec, ataamd, ataati, atacenatek, 1077 atacypress, atacyrix, atahighpoint, ataintel, ataite, atajmicron, 1078 atamarvell, atamicron, atanational, atanetcell, atanvidia, 1079 atapromise, ataserverworks, atasiliconimage, atasis, atavia 1080 108120080820: 1082 The TTY subsystem of the kernel has been replaced by a new 1083 implementation, which provides better scalability and an 1084 improved driver model. Most common drivers have been migrated to 1085 the new TTY subsystem, while others have not. The following 1086 drivers have not yet been ported to the new TTY layer: 1087 1088 PCI/ISA: 1089 cy, digi, rc, rp, sio 1090 1091 USB: 1092 ubser, ucycom 1093 1094 Line disciplines: 1095 ng_h4, ng_tty, ppp, sl, snp 1096 1097 Adding these drivers to your kernel configuration file shall 1098 cause compilation to fail. 1099 110020080818: 1101 ntpd has been upgraded to 4.2.4p5. 1102 110320080801: 1104 OpenSSH has been upgraded to 5.1p1. 1105 1106 For many years, FreeBSD's version of OpenSSH preferred DSA 1107 over RSA for host and user authentication keys. With this 1108 upgrade, we've switched to the vendor's default of RSA over 1109 DSA. This may cause upgraded clients to warn about unknown 1110 host keys even for previously known hosts. Users should 1111 follow the usual procedure for verifying host keys before 1112 accepting the RSA key. 1113 1114 This can be circumvented by setting the "HostKeyAlgorithms" 1115 option to "ssh-dss,ssh-rsa" in ~/.ssh/config or on the ssh 1116 command line. 1117 1118 Please note that the sequence of keys offered for 1119 authentication has been changed as well. You may want to 1120 specify IdentityFile in a different order to revert this 1121 behavior. 1122 112320080713: 1124 The sio(4) driver has been removed from the i386 and amd64 1125 kernel configuration files. This means uart(4) is now the 1126 default serial port driver on those platforms as well. 1127 1128 To prevent collisions with the sio(4) driver, the uart(4) driver 1129 uses different names for its device nodes. This means the 1130 onboard serial port will now most likely be called "ttyu0" 1131 instead of "ttyd0". You may need to reconfigure applications to 1132 use the new device names. 1133 1134 When using the serial port as a boot console, be sure to update 1135 /boot/device.hints and /etc/ttys before booting the new kernel. 1136 If you forget to do so, you can still manually specify the hints 1137 at the loader prompt: 1138 1139 set hint.uart.0.at="isa" 1140 set hint.uart.0.port="0x3F8" 1141 set hint.uart.0.flags="0x10" 1142 set hint.uart.0.irq="4" 1143 boot -s 1144 114520080609: 1146 The gpt(8) utility has been removed. Use gpart(8) to partition 1147 disks instead. 1148 114920080603: 1150 The version that Linuxulator emulates was changed from 2.4.2 1151 to 2.6.16. If you experience any problems with Linux binaries 1152 please try to set sysctl compat.linux.osrelease to 2.4.2 and 1153 if it fixes the problem contact emulation mailing list. 1154 115520080525: 1156 ISDN4BSD (I4B) was removed from the src tree. You may need to 1157 update a your kernel configuration and remove relevant entries. 1158 115920080509: 1160 I have checked in code to support multiple routing tables. 1161 See the man pages setfib(1) and setfib(2). 1162 This is a hopefully backwards compatible version, 1163 but to make use of it you need to compile your kernel 1164 with options ROUTETABLES=2 (or more up to 16). 1165 116620080420: 1167 The 802.11 wireless support was redone to enable multi-bss 1168 operation on devices that are capable. The underlying device 1169 is no longer used directly but instead wlanX devices are 1170 cloned with ifconfig. This requires changes to rc.conf files. 1171 For example, change: 1172 ifconfig_ath0="WPA DHCP" 1173 to 1174 wlans_ath0=wlan0 1175 ifconfig_wlan0="WPA DHCP" 1176 see rc.conf(5) for more details. In addition, mergemaster of 1177 /etc/rc.d is highly recommended. Simultaneous update of userland 1178 and kernel wouldn't hurt either. 1179 1180 As part of the multi-bss changes the wlan_scan_ap and wlan_scan_sta 1181 modules were merged into the base wlan module. All references 1182 to these modules (e.g. in kernel config files) must be removed. 1183 118420080408: 1185 psm(4) has gained write(2) support in native operation level. 1186 Arbitrary commands can be written to /dev/psm%d and status can 1187 be read back from it. Therefore, an application is responsible 1188 for status validation and error recovery. It is a no-op in 1189 other operation levels. 1190 119120080312: 1192 Support for KSE threading has been removed from the kernel. To 1193 run legacy applications linked against KSE libmap.conf may 1194 be used. The following libmap.conf may be used to ensure 1195 compatibility with any prior release: 1196 1197 libpthread.so.1 libthr.so.1 1198 libpthread.so.2 libthr.so.2 1199 libkse.so.3 libthr.so.3 1200 120120080301: 1202 The layout of struct vmspace has changed. This affects libkvm 1203 and any executables that link against libkvm and use the 1204 kvm_getprocs() function. In particular, but not exclusively, 1205 it affects ps(1), fstat(1), pkill(1), systat(1), top(1) and w(1). 1206 The effects are minimal, but it's advisable to upgrade world 1207 nonetheless. 1208 120920080229: 1210 The latest em driver no longer has support in it for the 1211 82575 adapter, this is now moved to the igb driver. The 1212 split was done to make new features that are incompatible 1213 with older hardware easier to do. 1214 121520080220: 1216 The new geom_lvm(4) geom class has been renamed to geom_linux_lvm(4), 1217 likewise the kernel option is now GEOM_LINUX_LVM. 1218 121920080211: 1220 The default NFS mount mode has changed from UDP to TCP for 1221 increased reliability. If you rely on (insecurely) NFS 1222 mounting across a firewall you may need to update your 1223 firewall rules. 1224 122520080208: 1226 Belatedly note the addition of m_collapse for compacting 1227 mbuf chains. 1228 122920080126: 1230 The fts(3) structures have been changed to use adequate 1231 integer types for their members and so to be able to cope 1232 with huge file trees. The old fts(3) ABI is preserved 1233 through symbol versioning in libc, so third-party binaries 1234 using fts(3) should still work, although they will not take 1235 advantage of the extended types. At the same time, some 1236 third-party software might fail to build after this change 1237 due to unportable assumptions made in its source code about 1238 fts(3) structure members. Such software should be fixed 1239 by its vendor or, in the worst case, in the ports tree. 1240 FreeBSD_version 800015 marks this change for the unlikely 1241 case that a portable fix is impossible. 1242 124320080123: 1244 To upgrade to -current after this date, you must be running 1245 FreeBSD not older than 6.0-RELEASE. Upgrading to -current 1246 from 5.x now requires a stop over at RELENG_6 or RELENG_7 systems. 1247 124820071128: 1249 The ADAPTIVE_GIANT kernel option has been retired because its 1250 functionality is the default now. 1251 125220071118: 1253 The AT keyboard emulation of sunkbd(4) has been turned on 1254 by default. In order to make the special symbols of the Sun 1255 keyboards driven by sunkbd(4) work under X these now have 1256 to be configured the same way as Sun USB keyboards driven 1257 by ukbd(4) (which also does AT keyboard emulation), f.e.: 1258 1259 Option "XkbLayout" "us" 1260 Option "XkbRules" "xorg" 1261 Option "XkbSymbols" "pc(pc105)+sun_vndr/usb(sun_usb)+us" 1262 126320071024: 1264 It has been decided that it is desirable to provide ABI 1265 backwards compatibility to the FreeBSD 4/5/6 versions of the 1266 PCIOCGETCONF, PCIOCREAD and PCIOCWRITE IOCTLs, which was 1267 broken with the introduction of PCI domain support (see the 1268 20070930 entry). Unfortunately, this required the ABI of 1269 PCIOCGETCONF to be broken again in order to be able to 1270 provide backwards compatibility to the old version of that 1271 IOCTL. Thus consumers of PCIOCGETCONF have to be recompiled 1272 again. As for prominent ports this affects neither pciutils 1273 nor xorg-server this time, the hal port needs to be rebuilt 1274 however. 1275 127620071020: 1277 The misnamed kthread_create() and friends have been renamed 1278 to kproc_create() etc. Many of the callers already 1279 used kproc_start().. 1280 I will return kthread_create() and friends in a while 1281 with implementations that actually create threads, not procs. 1282 Renaming corresponds with version 800002. 1283 128420071010: 1285 RELENG_7 branched. 1286 1287COMMON ITEMS: 1288 1289 General Notes 1290 ------------- 1291 Avoid using make -j when upgrading. While generally safe, there are 1292 sometimes problems using -j to upgrade. If your upgrade fails with 1293 -j, please try again without -j. From time to time in the past there 1294 have been problems using -j with buildworld and/or installworld. This 1295 is especially true when upgrading between "distant" versions (eg one 1296 that cross a major release boundary or several minor releases, or when 1297 several months have passed on the -current branch). 1298 1299 Sometimes, obscure build problems are the result of environment 1300 poisoning. This can happen because the make utility reads its 1301 environment when searching for values for global variables. To run 1302 your build attempts in an "environmental clean room", prefix all make 1303 commands with 'env -i '. See the env(1) manual page for more details. 1304 1305 When upgrading from one major version to another it is generally best 1306 to upgrade to the latest code in the currently installed branch first, 1307 then do an upgrade to the new branch. This is the best-tested upgrade 1308 path, and has the highest probability of being successful. Please try 1309 this approach before reporting problems with a major version upgrade. 1310 1311 ZFS notes 1312 --------- 1313 When upgrading the boot ZFS pool to a new version, always follow 1314 these two steps: 1315 1316 1.) recompile and reinstall the ZFS boot loader and boot block 1317 (this is part of "make buildworld" and "make installworld") 1318 1319 2.) update the ZFS boot block on your boot drive 1320 1321 The following example updates the ZFS boot block on the first 1322 partition (freebsd-boot) of a GPT partitioned drive ad0: 1323 "gpart bootcode -p /boot/gptzfsboot -i 1 ad0" 1324 1325 Non-boot pools do not need these updates. 1326 1327 To build a kernel 1328 ----------------- 1329 If you are updating from a prior version of FreeBSD (even one just 1330 a few days old), you should follow this procedure. It is the most 1331 failsafe as it uses a /usr/obj tree with a fresh mini-buildworld, 1332 1333 make kernel-toolchain 1334 make -DALWAYS_CHECK_MAKE buildkernel KERNCONF=YOUR_KERNEL_HERE 1335 make -DALWAYS_CHECK_MAKE installkernel KERNCONF=YOUR_KERNEL_HERE 1336 1337 To test a kernel once 1338 --------------------- 1339 If you just want to boot a kernel once (because you are not sure 1340 if it works, or if you want to boot a known bad kernel to provide 1341 debugging information) run 1342 make installkernel KERNCONF=YOUR_KERNEL_HERE KODIR=/boot/testkernel 1343 nextboot -k testkernel 1344 1345 To just build a kernel when you know that it won't mess you up 1346 -------------------------------------------------------------- 1347 This assumes you are already running a CURRENT system. Replace 1348 ${arch} with the architecture of your machine (e.g. "i386", 1349 "arm", "amd64", "ia64", "pc98", "sparc64", "powerpc", "mips", etc). 1350 1351 cd src/sys/${arch}/conf 1352 config KERNEL_NAME_HERE 1353 cd ../compile/KERNEL_NAME_HERE 1354 make depend 1355 make 1356 make install 1357 1358 If this fails, go to the "To build a kernel" section. 1359 1360 To rebuild everything and install it on the current system. 1361 ----------------------------------------------------------- 1362 # Note: sometimes if you are running current you gotta do more than 1363 # is listed here if you are upgrading from a really old current. 1364 1365 <make sure you have good level 0 dumps> 1366 make buildworld 1367 make kernel KERNCONF=YOUR_KERNEL_HERE 1368 [1] 1369 <reboot in single user> [3] 1370 mergemaster -p [5] 1371 make installworld 1372 mergemaster -i [4] 1373 make delete-old [6] 1374 <reboot> 1375 1376 1377 To cross-install current onto a separate partition 1378 -------------------------------------------------- 1379 # In this approach we use a separate partition to hold 1380 # current's root, 'usr', and 'var' directories. A partition 1381 # holding "/", "/usr" and "/var" should be about 2GB in 1382 # size. 1383 1384 <make sure you have good level 0 dumps> 1385 <boot into -stable> 1386 make buildworld 1387 make buildkernel KERNCONF=YOUR_KERNEL_HERE 1388 <maybe newfs current's root partition> 1389 <mount current's root partition on directory ${CURRENT_ROOT}> 1390 make installworld DESTDIR=${CURRENT_ROOT} 1391 make distribution DESTDIR=${CURRENT_ROOT} # if newfs'd 1392 make installkernel KERNCONF=YOUR_KERNEL_HERE DESTDIR=${CURRENT_ROOT} 1393 cp /etc/fstab ${CURRENT_ROOT}/etc/fstab # if newfs'd 1394 <edit ${CURRENT_ROOT}/etc/fstab to mount "/" from the correct partition> 1395 <reboot into current> 1396 <do a "native" rebuild/install as described in the previous section> 1397 <maybe install compatibility libraries from ports/misc/compat*> 1398 <reboot> 1399 1400 1401 To upgrade in-place from 8.x-stable to current 1402 ---------------------------------------------- 1403 <make sure you have good level 0 dumps> 1404 make buildworld [9] 1405 make kernel KERNCONF=YOUR_KERNEL_HERE [8] 1406 [1] 1407 <reboot in single user> [3] 1408 mergemaster -p [5] 1409 make installworld 1410 mergemaster -i [4] 1411 make delete-old [6] 1412 <reboot> 1413 1414 Make sure that you've read the UPDATING file to understand the 1415 tweaks to various things you need. At this point in the life 1416 cycle of current, things change often and you are on your own 1417 to cope. The defaults can also change, so please read ALL of 1418 the UPDATING entries. 1419 1420 Also, if you are tracking -current, you must be subscribed to 1421 freebsd-current@freebsd.org. Make sure that before you update 1422 your sources that you have read and understood all the recent 1423 messages there. If in doubt, please track -stable which has 1424 much fewer pitfalls. 1425 1426 [1] If you have third party modules, such as vmware, you 1427 should disable them at this point so they don't crash your 1428 system on reboot. 1429 1430 [3] From the bootblocks, boot -s, and then do 1431 fsck -p 1432 mount -u / 1433 mount -a 1434 cd src 1435 adjkerntz -i # if CMOS is wall time 1436 Also, when doing a major release upgrade, it is required that 1437 you boot into single user mode to do the installworld. 1438 1439 [4] Note: This step is non-optional. Failure to do this step 1440 can result in a significant reduction in the functionality of the 1441 system. Attempting to do it by hand is not recommended and those 1442 that pursue this avenue should read this file carefully, as well 1443 as the archives of freebsd-current and freebsd-hackers mailing lists 1444 for potential gotchas. The -U option is also useful to consider. 1445 See mergemaster(8) for more information. 1446 1447 [5] Usually this step is a noop. However, from time to time 1448 you may need to do this if you get unknown user in the following 1449 step. It never hurts to do it all the time. You may need to 1450 install a new mergemaster (cd src/usr.sbin/mergemaster && make 1451 install) after the buildworld before this step if you last updated 1452 from current before 20020224 or from -stable before 20020408. 1453 1454 [6] This only deletes old files and directories. Old libraries 1455 can be deleted by "make delete-old-libs", but you have to make 1456 sure that no program is using those libraries anymore. 1457 1458 [8] In order to have a kernel that can run the 4.x binaries needed to 1459 do an installworld, you must include the COMPAT_FREEBSD4 option in 1460 your kernel. Failure to do so may leave you with a system that is 1461 hard to boot to recover. A similar kernel option COMPAT_FREEBSD5 is 1462 required to run the 5.x binaries on more recent kernels. And so on 1463 for COMPAT_FREEBSD6 and COMPAT_FREEBSD7. 1464 1465 Make sure that you merge any new devices from GENERIC since the 1466 last time you updated your kernel config file. 1467 1468 [9] When checking out sources, you must include the -P flag to have 1469 cvs prune empty directories. 1470 1471 If CPUTYPE is defined in your /etc/make.conf, make sure to use the 1472 "?=" instead of the "=" assignment operator, so that buildworld can 1473 override the CPUTYPE if it needs to. 1474 1475 MAKEOBJDIRPREFIX must be defined in an environment variable, and 1476 not on the command line, or in /etc/make.conf. buildworld will 1477 warn if it is improperly defined. 1478FORMAT: 1479 1480This file contains a list, in reverse chronological order, of major 1481breakages in tracking -current. It is not guaranteed to be a complete 1482list of such breakages, and only contains entries since October 10, 2007. 1483If you need to see UPDATING entries from before that date, you will need 1484to fetch an UPDATING file from an older FreeBSD release. 1485 1486Copyright information: 1487 1488Copyright 1998-2009 M. Warner Losh. All Rights Reserved. 1489 1490Redistribution, publication, translation and use, with or without 1491modification, in full or in part, in any form or format of this 1492document are permitted without further permission from the author. 1493 1494THIS DOCUMENT IS PROVIDED BY WARNER LOSH ``AS IS'' AND ANY EXPRESS OR 1495IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 1496WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 1497DISCLAIMED. IN NO EVENT SHALL WARNER LOSH BE LIABLE FOR ANY DIRECT, 1498INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 1499(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 1500SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 1501HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, 1502STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING 1503IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 1504POSSIBILITY OF SUCH DAMAGE. 1505 1506Contact Warner Losh if you have any questions about your use of 1507this document. 1508 1509$FreeBSD: head/UPDATING 225757 2011-09-26 02:27:04Z kensmith $ 1510