freebsd-dev/UPDATING

1076 lines
42 KiB
Plaintext
Raw Normal View History

Updating Information for FreeBSD current users
This file is maintained and copyrighted by M. Warner Losh <imp@freebsd.org>.
See end of file for further details. For commonly done items, please see the
COMMON ITEMS: section later in the file. These instructions assume that you
basically know what you are doing. If not, then please consult the FreeBSD
handbook.
Items affecting the ports and packages system can be found in
/usr/ports/UPDATING. Please read that file before running portupgrade.
NOTE TO PEOPLE WHO THINK THAT FreeBSD 9.x IS SLOW:
FreeBSD 9.x has many debugging features turned on, in both the kernel
and userland. These features attempt to detect incorrect use of
system primitives, and encourage loud failure through extra sanity
checking and fail stop semantics. They also substantially impact
system performance. If you want to do performance measurement,
benchmarking, and optimization, you'll want to turn them off. This
includes various WITNESS- related kernel options, INVARIANTS, malloc
debugging flags in userland, and various verbose features in the
kernel. Many developers choose to disable these features on build
machines to maximize performance. (To disable malloc debugging, run
ln -s aj /etc/malloc.conf.)
20091113:
The default terminal emulation for syscons(4) has been changed
from cons25 to xterm on all platforms except pc98. This means
that the /etc/ttys file needs to be updated to ensure correct
operation of applications on the console.
The terminal emulation style can be toggled per window by using
vidcontrol(1)'s -T flag. The TEKEN_CONS25 kernel configuration
options can be used to change the compile-time default back to
cons25.
To prevent graphical artifacts, make sure the TERM environment
variable is set to match the terminal emulation that is being
performed by syscons(4).
20091109:
The layout of the structure ieee80211req_scan_result has changed.
Applications that require wireless scan results (e.g. ifconfig(8))
from net80211 need to be recompiled.
20091025:
The iwn(4) driver has been updated to support the 5000 and 5150 series.
There's one kernel module for each firmware. Adding "device iwnfw"
to the kernel configuration file means including all three firmware
images inside the kernel. If you want to include just the one for
your wireless card, use the the devices iwn4965fw, iwn5000fw or
iwn5150fw.
Fix several logic bugs in the previous IPv6 variable change and re-add $ipv6_enable support for backward compatibility. From UPDATING: 1. To use IPv6, simply define $ifconfig_IF_ipv6 like $ifconfig_IF for IPv4. For aliases, $ifconfig_IF_aliasN should be used. Note that both variables need the "inet6" keyword at the head. Do not set $ipv6_network_interfaces manually if you do not understand what you are doing. It is not needed in most cases. $ipv6_ifconfig_IF and $ipv6_ifconfig_IF_aliasN still work, but they are obsolete. 2. $ipv6_enable is obsolete. Use $ipv6_prefer and/or "inet6 accept_rtadv" keyword in ifconfig(8) instead. If you define $ipv6_enable=YES, it means $ipv6_prefer=YES and all configured interfaces have "inet6 accept_rtadv" in the $ifconfig_IF_ipv6. These are for backward compatibility. 3. A new variable $ipv6_prefer has been added. If NO, IPv6 functionality of interfaces with no corresponding $ifconfig_IF_ipv6 is disabled by using "inet6 ifdisabled" flag, and the default address selection policy of ip6addrctl(8) is the IPv4-preferred one (see rc.d/ip6addrctl for more details). Note that if you want to configure IPv6 functionality on the disabled interfaces after boot, first you need to clear the flag by using ifconfig(8) like: ifconfig em0 inet6 -ifdisabled If YES, the default address selection policy is set as IPv6-preferred. The default value of $ipv6_prefer is NO. 4. If your system need to receive Router Advertisement messages, define "inet6 accept_rtadv" in $ifconfig_IF_ipv6. The rc(8) scripts automatically invoke rtsol(8) when the interface becomes UP. The Router Advertisement messages are used for SLAAC (State-Less Address AutoConfiguration).
2009-09-26 18:59:00 +00:00
20090926:
The rc.d/network_ipv6, IPv6 configuration script has been integrated
into rc.d/netif. The changes are the following:
1. To use IPv6, simply define $ifconfig_IF_ipv6 like $ifconfig_IF
for IPv4. For aliases, $ifconfig_IF_aliasN should be used.
Note that both variables need the "inet6" keyword at the head.
Do not set $ipv6_network_interfaces manually if you do not
understand what you are doing. It is not needed in most cases.
$ipv6_ifconfig_IF and $ipv6_ifconfig_IF_aliasN still work, but
they are obsolete.
2. $ipv6_enable is obsolete. Use $ipv6_prefer and
"inet6 accept_rtadv" keyword in ifconfig(8) instead.
If you define $ipv6_enable=YES, it means $ipv6_prefer=YES and
all configured interfaces have "inet6 accept_rtadv" in the
$ifconfig_IF_ipv6. These are for backward compatibility.
3. A new variable $ipv6_prefer has been added. If NO, IPv6
functionality of interfaces with no corresponding
$ifconfig_IF_ipv6 is disabled by using "inet6 ifdisabled" flag,
and the default address selection policy of ip6addrctl(8)
is the IPv4-preferred one (see rc.d/ip6addrctl for more details).
Note that if you want to configure IPv6 functionality on the
disabled interfaces after boot, first you need to clear the flag by
using ifconfig(8) like:
ifconfig em0 inet6 -ifdisabled
If YES, the default address selection policy is set as
IPv6-preferred.
The default value of $ipv6_prefer is NO.
4. If your system need to receive Router Advertisement messages,
define "inet6 accept_rtadv" in $ifconfig_IF_ipv6. The rc(8)
scripts automatically invoke rtsol(8) when the interface becomes
UP. The Router Advertisement messages are used for SLAAC
(State-Less Address AutoConfiguration).
20090922:
802.11s D3.03 support was committed. This is incompatible with the
previous code, which was based on D3.0.
Fix several logic bugs in the previous IPv6 variable change and re-add $ipv6_enable support for backward compatibility. From UPDATING: 1. To use IPv6, simply define $ifconfig_IF_ipv6 like $ifconfig_IF for IPv4. For aliases, $ifconfig_IF_aliasN should be used. Note that both variables need the "inet6" keyword at the head. Do not set $ipv6_network_interfaces manually if you do not understand what you are doing. It is not needed in most cases. $ipv6_ifconfig_IF and $ipv6_ifconfig_IF_aliasN still work, but they are obsolete. 2. $ipv6_enable is obsolete. Use $ipv6_prefer and/or "inet6 accept_rtadv" keyword in ifconfig(8) instead. If you define $ipv6_enable=YES, it means $ipv6_prefer=YES and all configured interfaces have "inet6 accept_rtadv" in the $ifconfig_IF_ipv6. These are for backward compatibility. 3. A new variable $ipv6_prefer has been added. If NO, IPv6 functionality of interfaces with no corresponding $ifconfig_IF_ipv6 is disabled by using "inet6 ifdisabled" flag, and the default address selection policy of ip6addrctl(8) is the IPv4-preferred one (see rc.d/ip6addrctl for more details). Note that if you want to configure IPv6 functionality on the disabled interfaces after boot, first you need to clear the flag by using ifconfig(8) like: ifconfig em0 inet6 -ifdisabled If YES, the default address selection policy is set as IPv6-preferred. The default value of $ipv6_prefer is NO. 4. If your system need to receive Router Advertisement messages, define "inet6 accept_rtadv" in $ifconfig_IF_ipv6. The rc(8) scripts automatically invoke rtsol(8) when the interface becomes UP. The Router Advertisement messages are used for SLAAC (State-Less Address AutoConfiguration).
2009-09-26 18:59:00 +00:00
20090912:
A sysctl variable net.inet6.ip6.accept_rtadv now sets the default value
of a per-interface flag ND6_IFF_ACCEPT_RTADV, not a global knob to
control whether accepting Router Advertisement messages or not.
Also, a per-interface flag ND6_IFF_AUTO_LINKLOCAL has been added and
a sysctl variable net.inet6.ip6.auto_linklocal is its default value.
The ifconfig(8) utility now supports these flags.
20090910:
ZFS snapshots are now mounted with MNT_IGNORE flag. Use -v option for
mount(8) and -a option for df(1) to see them.
20090825:
The old tunable hw.bus.devctl_disable has been superseded by
hw.bus.devctl_queue. hw.bus.devctl_disable=1 in loader.conf should be
replaced by hw.bus.devctl_queue=0. The default for this new tunable
is 1000.
2009-08-13 17:09:45 +00:00
20090813:
Remove the option STOP_NMI. The default action is now to use NMI only
for KDB via the newly introduced function stop_cpus_hard() and
maintain stop_cpus() to just use a normal IPI_STOP on ia32 and amd64.
20090803:
The stable/8 branch created in subversion. This corresponds to the
RELENG_8 branch in CVS.
2009-08-13 17:09:45 +00:00
20090719:
Bump the shared library version numbers for all libraries that do not
use symbol versioning as part of the 8.0-RELEASE cycle. Bump
__FreeBSD_version to 800105.
Build on Jeff Roberson's linker-set based dynamic per-CPU allocator (DPCPU), as suggested by Peter Wemm, and implement a new per-virtual network stack memory allocator. Modify vnet to use the allocator instead of monolithic global container structures (vinet, ...). This change solves many binary compatibility problems associated with VIMAGE, and restores ELF symbols for virtualized global variables. Each virtualized global variable exists as a "reference copy", and also once per virtual network stack. Virtualized global variables are tagged at compile-time, placing the in a special linker set, which is loaded into a contiguous region of kernel memory. Virtualized global variables in the base kernel are linked as normal, but those in modules are copied and relocated to a reserved portion of the kernel's vnet region with the help of a the kernel linker. Virtualized global variables exist in per-vnet memory set up when the network stack instance is created, and are initialized statically from the reference copy. Run-time access occurs via an accessor macro, which converts from the current vnet and requested symbol to a per-vnet address. When "options VIMAGE" is not compiled into the kernel, normal global ELF symbols will be used instead and indirection is avoided. This change restores static initialization for network stack global variables, restores support for non-global symbols and types, eliminates the need for many subsystem constructors, eliminates large per-subsystem structures that caused many binary compatibility issues both for monitoring applications (netstat) and kernel modules, removes the per-function INIT_VNET_*() macros throughout the stack, eliminates the need for vnet_symmap ksym(2) munging, and eliminates duplicate definitions of virtualized globals under VIMAGE_GLOBALS. Bump __FreeBSD_version and update UPDATING. Portions submitted by: bz Reviewed by: bz, zec Discussed with: gnn, jamie, jeff, jhb, julian, sam Suggested by: peter Approved by: re (kensmith)
2009-07-14 22:48:30 +00:00
20090714:
Due to changes in the implementation of virtual network stack support,
all network-related kernel modules must be recompiled. As this change
breaks the ABI, bump __FreeBSD_version to 800104.
Build on Jeff Roberson's linker-set based dynamic per-CPU allocator (DPCPU), as suggested by Peter Wemm, and implement a new per-virtual network stack memory allocator. Modify vnet to use the allocator instead of monolithic global container structures (vinet, ...). This change solves many binary compatibility problems associated with VIMAGE, and restores ELF symbols for virtualized global variables. Each virtualized global variable exists as a "reference copy", and also once per virtual network stack. Virtualized global variables are tagged at compile-time, placing the in a special linker set, which is loaded into a contiguous region of kernel memory. Virtualized global variables in the base kernel are linked as normal, but those in modules are copied and relocated to a reserved portion of the kernel's vnet region with the help of a the kernel linker. Virtualized global variables exist in per-vnet memory set up when the network stack instance is created, and are initialized statically from the reference copy. Run-time access occurs via an accessor macro, which converts from the current vnet and requested symbol to a per-vnet address. When "options VIMAGE" is not compiled into the kernel, normal global ELF symbols will be used instead and indirection is avoided. This change restores static initialization for network stack global variables, restores support for non-global symbols and types, eliminates the need for many subsystem constructors, eliminates large per-subsystem structures that caused many binary compatibility issues both for monitoring applications (netstat) and kernel modules, removes the per-function INIT_VNET_*() macros throughout the stack, eliminates the need for vnet_symmap ksym(2) munging, and eliminates duplicate definitions of virtualized globals under VIMAGE_GLOBALS. Bump __FreeBSD_version and update UPDATING. Portions submitted by: bz Reviewed by: bz, zec Discussed with: gnn, jamie, jeff, jhb, julian, sam Suggested by: peter Approved by: re (kensmith)
2009-07-14 22:48:30 +00:00
20090713:
The TOE interface to the TCP syncache has been modified to remove
struct tcpopt (<netinet/tcp_var.h>) from the ABI of the network stack.
The cxgb driver is the only TOE consumer affected by this change, and
needs to be recompiled along with the kernel. As this change breaks
the ABI, bump __FreeBSD_version to 800103.
20090712:
Padding has been added to struct tcpcb, sackhint and tcpstat in
<netinet/tcp_var.h> to facilitate future MFCs and bug fixes whilst
maintainig the ABI. However, this change breaks the ABI, so bump
__FreeBSD_version to 800102. User space tools that rely on the size of
any of these structs (e.g. sockstat) need to be recompiled.
20090630:
The NFS_LEGACYRPC option has been removed along with the old kernel
RPC implementation that this option selected. Kernel configurations
may need to be adjusted.
20090629:
The network interface device nodes at /dev/net/<interface> have been
removed. All ioctl operations can be performed the normal way using
routing sockets. The kqueue functionality can generally be replaced
with routing sockets.
20090628:
The documentation from the FreeBSD Documentation Project (Handbook,
FAQ, etc.) is now installed via packages by sysinstall(8) and under
the /usr/local/share/doc/freebsd directory instead of /usr/share/doc.
20090624:
The ABI of various structures related to the SYSV IPC API have been
changed. As a result, the COMPAT_FREEBSD[456] and COMPAT_43 kernel
options now all require COMPAT_FREEBSD7. Bump __FreeBSD_version to
800100.
20090622:
Layout of struct vnet has changed as routing related variables were
moved to their own Vimage module. Modules need to be recompiled. Bump
__FreeBSD_version to 800099.
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
20090619:
NGROUPS_MAX and NGROUPS have been increased from 16 to 1023 and 1024
respectively. As long as no more than 16 groups per process are used,
no changes should be visible. When more than 16 groups are used, old
binaries may fail if they call getgroups() or getgrouplist() with
statically sized storage. Recompiling will work around this, but
applications should be modified to use dynamically allocated storage
for group arrays as POSIX.1-2008 does not cap an implementation's
number of supported groups at NGROUPS_MAX+1 as previous versions did.
NFS and portalfs mounts may also be affected as the list of groups is
truncated to 16. Users of NFS who use more than 16 groups, should
take care that negative group permissions are not used on the exported
file systems as they will not be reliable unless a GSSAPI based
authentication method is used.
20090616:
The compiling option ADAPTIVE_LOCKMGRS has been introduced. This
option compiles in the support for adaptive spinning for lockmgrs
which want to enable it. The lockinit() function now accepts the flag
LK_ADAPTIVE in order to make the lock object subject to adaptive
spinning when both held in write and read mode.
20090613:
The layout of the structure returned by IEEE80211_IOC_STA_INFO has
changed. User applications that use this ioctl need to be rebuilt.
20090611:
The layout of struct thread has changed. Kernel and modules need to
be rebuilt.
20090608:
The layout of structs ifnet, domain, protosw and vnet_net has changed.
Kernel modules need to be rebuilt. Bump __FreeBSD_version to 800097.
20090602:
window(1) has been removed from the base system. It can now be
installed from ports. The port is called misc/window.
20090601:
The way we are storing and accessing `routing table' entries has
changed. Programs reading the FIB, like netstat, need to be
re-compiled.
20090601:
A new netisr implementation has been added for FreeBSD 8. Network
file system modules, such as igmp, ipdivert, and others, should be
rebuilt.
Bump __FreeBSD_version to 800096.
20090530:
Remove the tunable/sysctl debug.mpsafevfs as its initial purpose is no
more valid.
20090530:
Add VOP_ACCESSX(9). File system modules need to be rebuilt.
Bump __FreeBSD_version to 800094.
20090529:
Add mnt_xflag field to 'struct mount'. File system modules need to be
rebuilt.
Bump __FreeBSD_version to 800093.
20090528:
The compiling option ADAPTIVE_SX has been retired while it has been
introduced the option NO_ADAPTIVE_SX which handles the reversed logic.
The KPI for sx_init_flags() changes as accepting flags:
SX_ADAPTIVESPIN flag has been retired while the SX_NOADAPTIVE flag has
been introduced in order to handle the reversed logic.
Bump __FreeBSD_version to 800092.
20090527:
Add support for hierarchical jails. Remove global securelevel.
Bump __FreeBSD_version to 800091.
20090523:
The layout of struct vnet_net has changed, therefore modules
need to be rebuilt.
Bump __FreeBSD_version to 800090.
20090523:
The newly imported zic(8) produces a new format in the output. Please
run tzsetup(8) to install the newly created data to /etc/localtime.
20090520:
The sysctl tree for the usb stack has renamed from hw.usb2.* to
hw.usb.* and is now consistent again with previous releases.
2009-05-20 20:05:56 +00:00
20090520:
802.11 monitor mode support was revised and driver api's were changed.
Drivers dependent on net80211 now support DLT_IEEE802_11_RADIO instead
of DLT_IEEE802_11. No user-visible data structures were changed but
applications that use DLT_IEEE802_11 may require changes.
2009-05-20 20:05:56 +00:00
Bump __FreeBSD_version to 800088.
Permit buiding kernels with options VIMAGE, restricted to only a single active network stack instance. Turning on options VIMAGE at compile time yields the following changes relative to default kernel build: 1) V_ accessor macros for virtualized variables resolve to structure fields via base pointers, instead of being resolved as fields in global structs or plain global variables. As an example, V_ifnet becomes: options VIMAGE: ((struct vnet_net *) vnet_net)->_ifnet default build: vnet_net_0._ifnet options VIMAGE_GLOBALS: ifnet 2) INIT_VNET_* macros will declare and set up base pointers to be used by V_ accessor macros, instead of resolving to whitespace: INIT_VNET_NET(ifp->if_vnet); becomes struct vnet_net *vnet_net = (ifp->if_vnet)->mod_data[VNET_MOD_NET]; 3) Memory for vnet modules registered via vnet_mod_register() is now allocated at run time in sys/kern/kern_vimage.c, instead of per vnet module structs being declared as globals. If required, vnet modules can now request the framework to provide them with allocated bzeroed memory by filling in the vmi_size field in their vmi_modinfo structures. 4) structs socket, ifnet, inpcbinfo, tcpcb and syncache_head are extended to hold a pointer to the parent vnet. options VIMAGE builds will fill in those fields as required. 5) curvnet is introduced as a new global variable in options VIMAGE builds, always pointing to the default and only struct vnet. 6) struct sysctl_oid has been extended with additional two fields to store major and minor virtualization module identifiers, oid_v_subs and oid_v_mod. SYSCTL_V_* family of macros will fill in those fields accordingly, and store the offset in the appropriate vnet container struct in oid_arg1. In sysctl handlers dealing with virtualized sysctls, the SYSCTL_RESOLVE_V_ARG1() macro will compute the address of the target variable and make it available in arg1 variable for further processing. Unused fields in structs vnet_inet, vnet_inet6 and vnet_ipfw have been deleted. Reviewed by: bz, rwatson Approved by: julian (mentor)
2009-04-30 13:36:26 +00:00
20090430:
The layout of the following structs has changed: sysctl_oid,
socket, ifnet, inpcbinfo, tcpcb, syncache_head, vnet_inet,
vnet_inet6 and vnet_ipfw. Most modules need to be rebuild or
panics may be experienced. World rebuild is required for
correctly checking networking state from userland.
Bump __FreeBSD_version to 800085.
Bite the bullet, and make the IPv6 SSM and MLDv2 mega-commit: import from p4 bms_netdev. Summary of changes: * Connect netinet6/in6_mcast.c to build. The legacy KAME KPIs are mostly preserved. * Eliminate now dead code from ip6_output.c. Don't do mbuf bingo, we are not going to do RFC 2292 style CMSG tricks for multicast options as they are not required by any current IPv6 normative reference. * Refactor transports (UDP, raw_ip6) to do own mcast filtering. SCTP, TCP unaffected by this change. * Add ip6_msource, in6_msource structs to in6_var.h. * Hookup mld_ifinfo state to in6_ifextra, allocate from domifattach path. * Eliminate IN6_LOOKUP_MULTI(), it is no longer referenced. Kernel consumers which need this should use in6m_lookup(). * Refactor IPv6 socket group memberships to use a vector (like IPv4). * Update ifmcstat(8) for IPv6 SSM. * Add witness lock order for IN6_MULTI_LOCK. * Move IN6_MULTI_LOCK out of lower ip6_output()/ip6_input() paths. * Introduce IP6STAT_ADD/SUB/INC/DEC as per rwatson's IPv4 cleanup. * Update carp(4) for new IPv6 SSM KPIs. * Virtualize ip6_mrouter socket. Changes mostly localized to IPv6 MROUTING. * Don't do a local group lookup in MROUTING. * Kill unused KAME prototypes in6_purgemkludge(), in6_restoremkludge(). * Preserve KAME DAD timer jitter behaviour in MLDv1 compatibility mode. * Bump __FreeBSD_version to 800084. * Update UPDATING. NOTE WELL: * This code hasn't been tested against real MLDv2 queriers (yet), although the on-wire protocol has been verified in Wireshark. * There are a few unresolved issues in the socket layer APIs to do with scope ID propagation. * There is a LOR present in ip6_output()'s use of in6_setscope() which needs to be resolved. See comments in mld6.c. This is believed to be benign and can't be avoided for the moment without re-introducing an indirect netisr. This work was mostly derived from the IGMPv3 implementation, and has been sponsored by a third party.
2009-04-29 19:19:13 +00:00
20090429:
MLDv2 and Source-Specific Multicast (SSM) have been merged
to the IPv6 stack. VIMAGE hooks are in but not yet used.
The implementation of SSM within FreeBSD's IPv6 stack closely
follows the IPv4 implementation.
For kernel developers:
* The most important changes are that the ip6_output() and
ip6_input() paths no longer take the IN6_MULTI_LOCK,
and this lock has been downgraded to a non-recursive mutex.
* As with the changes to the IPv4 stack to support SSM, filtering
of inbound multicast traffic must now be performed by transport
protocols within the IPv6 stack. This does not apply to TCP and
SCTP, however, it does apply to UDP in IPv6 and raw IPv6.
* The KPIs used by IPv6 multicast are similar to those used by
the IPv4 stack, with the following differences:
* im6o_mc_filter() is analogous to imo_multicast_filter().
* The legacy KAME entry points in6_joingroup and in6_leavegroup()
are shimmed to in6_mc_join() and in6_mc_leave() respectively.
* IN6_LOOKUP_MULTI() has been deprecated and removed.
* IPv6 relies on MLD for the DAD mechanism. KAME's internal KPIs
for MLDv1 have an additional 'timer' argument which is used to
jitter the initial membership report for the solicited-node
multicast membership on-link.
* This is not strictly needed for MLDv2, which already jitters
its report transmissions. However, the 'timer' argument is
preserved in case MLDv1 is active on the interface.
* The KAME linked-list based IPv6 membership implementation has
been refactored to use a vector similar to that used by the IPv4
stack.
Code which maintains a list of its own multicast memberships
internally, e.g. carp, has been updated to reflect the new
semantics.
* There is a known Lock Order Reversal (LOR) due to in6_setscope()
acquiring the IF_AFDATA_LOCK and being called within ip6_output().
Whilst MLDv2 tries to avoid this otherwise benign LOR, it is an
implementation constraint which needs to be addressed in HEAD.
For application developers:
* The changes are broadly similar to those made for the IPv4
stack.
* The use of IPv4 and IPv6 multicast socket options on the same
socket, using mapped addresses, HAS NOT been tested or supported.
* There are a number of issues with the implementation of various
IPv6 multicast APIs which need to be resolved in the API surface
before the implementation is fully compatible with KAME userland
use, and these are mostly to do with interface index treatment.
* The literature available discusses the use of either the delta / ASM
API with setsockopt(2)/getsockopt(2), or the full-state / ASM API
using setsourcefilter(3)/getsourcefilter(3). For more information
please refer to RFC 3768, 'Socket Interface Extensions for
Multicast Source Filters'.
* Applications which use the published RFC 3678 APIs should be fine.
For systems administrators:
* The mtest(8) utility has been refactored to support IPv6, in
addition to IPv4. Interface addresses are no longer accepted
as arguments, their names must be used instead. The utility
will map the interface name to its first IPv4 address as
returned by getifaddrs(3).
* The ifmcstat(8) utility has also been updated to print the MLDv2
endpoint state and source filter lists via sysctl(3).
* The net.inet6.ip6.mcast.loop sysctl may be tuned to 0 to disable
loopback of IPv6 multicast datagrams by default; it defaults to 1
to preserve the existing behaviour. Disabling multicast loopback is
recommended for optimal system performance.
* The IPv6 MROUTING code has been changed to examine this sysctl
instead of attempting to perform a group lookup before looping
back forwarded datagrams.
Bump __FreeBSD_version to 800084.
20090422:
Implement low-level Bluetooth HCI API.
Bump __FreeBSD_version to 800083.
20090419:
The layout of struct malloc_type, used by modules to register new
memory allocation types, has changed. Most modules will need to
be rebuilt or panics may be experienced.
Bump __FreeBSD_version to 800081.
20090415:
Anticipate overflowing inp_flags - add inp_flags2.
This changes most offsets in inpcb, so checking v4 connection
state will require a world rebuild.
Bump __FreeBSD_version to 800080.
20090415:
Add an llentry to struct route and struct route_in6. Modules
embedding a struct route will need to be recompiled.
Bump __FreeBSD_version to 800079.
20090414:
The size of rt_metrics_lite and by extension rtentry has changed.
Networking administration apps will need to be recompiled.
The route command now supports show as an alias for get, weighting
of routes, sticky and nostick flags to alter the behavior of stateful
load balancing.
Bump __FreeBSD_version to 800078.
2009-04-12 19:42:25 +00:00
20090408:
Do not use Giant for kbdmux(4) locking. This is wrong and
apparently causing more problems than it solves. This will
re-open the issue where interrupt handlers may race with
kbdmux(4) in polling mode. Typical symptoms include (but
not limited to) duplicated and/or missing characters when
low level console functions (such as gets) are used while
interrupts are enabled (for example geli password prompt,
mountroot prompt etc.). Disabling kbdmux(4) may help.
20090407:
The size of structs vnet_net, vnet_inet and vnet_ipfw has changed;
kernel modules referencing any of the above need to be recompiled.
Bump __FreeBSD_version to 800075.
20090320:
GEOM_PART has become the default partition slicer for storage devices,
replacing GEOM_MBR, GEOM_BSD, GEOM_PC98 and GEOM_GPT slicers. It
introduces some changes:
MSDOS/EBR: the devices created from MSDOS extended partition entries
(EBR) can be named differently than with GEOM_MBR and are now symlinks
to devices with offset-based names. fstabs may need to be modified.
BSD: the "geometry does not match label" warning is harmless in most
cases but it points to problems in file system misalignment with
disk geometry. The "c" partition is now implicit, covers the whole
top-level drive and cannot be (mis)used by users.
General: Kernel dumps are now not allowed to be written to devices
whose partition types indicate they are meant to be used for file
systems (or, in case of MSDOS partitions, as something else than
the "386BSD" type).
Most of these changes date approximately from 200812.
20090319:
The uscanner(4) driver has been removed from the kernel. This follows
Linux removing theirs in 2.6 and making libusb the default interface
(supported by sane).
Introduce a number of changes to the MROUTING code. This is purely a forwarding plane cleanup; no control plane code is involved. Summary: * Split IPv4 and IPv6 MROUTING support. The static compile-time kernel option remains the same, however, the modules may now be built for IPv4 and IPv6 separately as ip_mroute_mod and ip6_mroute_mod. * Clean up the IPv4 multicast forwarding code to use BSD queue and hash table constructs. Don't build our own timer abstractions when ratecheck() and timevalclear() etc will do. * Expose the multicast forwarding cache (MFC) and virtual interface table (VIF) as sysctls, to reduce netstat's dependence on libkvm for this information for running kernels. * bandwidth meters however still require libkvm. * Make the MFC hash table size a boot/load-time tunable ULONG, net.inet.ip.mfchashsize (defaults to 256). * Remove unused members from struct vif and struct mfc. * Kill RSVP support, as no current RSVP implementation uses it. These stubs could be moved to raw_ip.c. * Don't share locks or initialization between IPv4 and IPv6. * Don't use a static struct route_in6 in ip6_mroute.c. The v6 code is still using a cached struct route_in6, this is moved to mif6 for the time being. * More cleanup remains to be merged from ip_mroute.c to ip6_mroute.c. v4 path tested using ports/net/mcast-tools. v6 changes are mostly mechanical locking and *have not* been tested. As these changes partially break some kernel ABIs, they will not be MFCed. There is a lot more work to be done here. Reviewed by: Pavlin Radoslavov
2009-03-19 01:43:03 +00:00
20090319:
The multicast forwarding code has been cleaned up. netstat(1)
only relies on KVM now for printing bandwidth upcall meters.
The IPv4 and IPv6 modules are split into ip_mroute_mod and
ip6_mroute_mod respectively. The config(5) options for statically
compiling this code remain the same, i.e. 'options MROUTING'.
20090315:
Support for the IFF_NEEDSGIANT network interface flag has been
removed, which means that non-MPSAFE network device drivers are no
longer supported. In particular, if_ar, if_sr, and network device
drivers from the old (legacy) USB stack can no longer be built or
used.
20090313:
POSIX.1 Native Language Support (NLS) has been enabled in libc and
a bunch of new language catalog files have also been added.
This means that some common libc messages are now localized and
they depend on the LC_MESSAGES environmental variable.
20090313:
The k8temp(4) driver has been renamed to amdtemp(4) since
support for K10 and K11 CPU families was added.
20090309:
IGMPv3 and Source-Specific Multicast (SSM) have been merged
to the IPv4 stack. VIMAGE hooks are in but not yet used.
For kernel developers, the most important changes are that the
ip_output() and ip_input() paths no longer take the IN_MULTI_LOCK(),
and this lock has been downgraded to a non-recursive mutex.
Transport protocols (UDP, Raw IP) are now responsible for filtering
inbound multicast traffic according to group membership and source
filters. The imo_multicast_filter() KPI exists for this purpose.
Transports which do not use multicast (SCTP, TCP) already reject
multicast by default. Forwarding and receive performance may improve
as a mutex acquisition is no longer needed in the ip_input()
low-level input path. in_addmulti() and in_delmulti() are shimmed
to new KPIs which exist to support SSM in-kernel.
For application developers, it is recommended that loopback of
multicast datagrams be disabled for best performance, as this
will still cause the lock to be taken for each looped-back
datagram transmission. The net.inet.ip.mcast.loop sysctl may
be tuned to 0 to disable loopback by default; it defaults to 1
to preserve the existing behaviour.
For systems administrators, to obtain best performance with
multicast reception and multiple groups, it is always recommended
that a card with a suitably precise hash filter is used. Hash
collisions will still result in the lock being taken within the
transport protocol input path to check group membership.
If deploying FreeBSD in an environment with IGMP snooping switches,
it is recommended that the net.inet.igmp.sendlocal sysctl remain
enabled; this forces 224.0.0.0/24 group membership to be announced
via IGMP.
The size of 'struct igmpstat' has changed; netstat needs to be
recompiled to reflect this.
Bump __FreeBSD_version to 800070.
20090309:
libusb20.so.1 is now installed as libusb.so.1 and the ports system
updated to use it. This requires a buildworld/installworld in order to
update the library and dependencies (usbconfig, etc). Its advisable to
rebuild all ports which uses libusb. More specific directions are given
in the ports collection UPDATING file. Any /etc/libmap.conf entries for
2009-03-09 22:43:00 +00:00
libusb are no longer required and can be removed.
20090302:
A workaround is committed to allow the creation of System V shared
memory segment of size > 2 GB on the 64-bit architectures.
Due to a limitation of the existing ABI, the shm_segsz member
of the struct shmid_ds, returned by shmctl(IPC_STAT) call is
2009-05-23 09:24:07 +00:00
wrong for large segments. Note that limits must be explicitly
raised to allow such segments to be created.
20090301:
The layout of struct ifnet has changed, requiring a rebuild of all
network device driver modules.
2009-02-27 17:32:49 +00:00
20090227:
The /dev handling for the new USB stack has changed, a
buildworld/installworld is required for libusb20.
20090223:
The new USB2 stack has now been permanently moved in and all kernel and
module names reverted to their previous values (eg, usb, ehci, ohci,
2009-02-23 19:28:29 +00:00
ums, ...). The old usb stack can be compiled in by prefixing the name
with the letter 'o', the old usb modules have been removed.
Updating entry 20090216 for xorg and 20090215 for libmap may still
apply.
20090217:
The rc.conf(5) option if_up_delay has been renamed to
defaultroute_delay to better reflect its purpose. If you have
customized this setting in /etc/rc.conf you need to update it to
use the new name.
20090216:
xorg 7.4 wants to configure its input devices via hald which does not
yet work with USB2. If the keyboard/mouse does not work in xorg then
add
Option "AllowEmptyInput" "off"
to your ServerLayout section. This will cause X to use the configured
2009-02-17 10:49:36 +00:00
kbd and mouse sections from your xorg.conf.
20090215:
The GENERIC kernels for all architectures now default to the new USB2
stack. No kernel config options or code have been removed so if a
problem arises please report it and optionally revert to the old USB
stack. If you are loading USB kernel modules or have a custom kernel
that includes GENERIC then ensure that usb names are also changed over,
2009-02-17 10:49:36 +00:00
eg uftdi -> usb2_serial_ftdi.
Older programs linked against the ports libusb 0.1 need to be
redirected to the new stack's libusb20. /etc/libmap.conf can
be used for this:
# Map old usb library to new one for usb2 stack
libusb-0.1.so.8 libusb20.so.1
20090209:
All USB ethernet devices now attach as interfaces under the name ueN
(eg. ue0). This is to provide a predictable name as vendors often
change usb chipsets in a product without notice.
20090203:
The ichsmb(4) driver has been changed to require SMBus slave
addresses be left-justified (xxxxxxx0b) rather than right-justified.
All of the other SMBus controller drivers require left-justified
slave addresses, so this change makes all the drivers provide the
same interface.
20090201:
INET6 statistics (struct ip6stat) was updated.
netstat(1) needs to be recompiled.
20090119:
NTFS has been removed from GENERIC kernel on amd64 to match
GENERIC on i386. Should not cause any issues since mount_ntfs(8)
will load ntfs.ko module automatically when NTFS support is
actually needed, unless ntfs.ko is not installed or security
level prohibits loading kernel modules. If either is the case,
"options NTFS" has to be added into kernel config.
20090115:
TCP Appropriate Byte Counting (RFC 3465) support added to kernel.
New field in struct tcpcb breaks ABI, so bump __FreeBSD_version to
800061. User space tools that rely on the size of struct tcpcb in
tcp_var.h (e.g. sockstat) need to be recompiled.
20081225:
ng_tty(4) module updated to match the new TTY subsystem.
Due to API change, user-level applications must be updated.
New API support added to mpd5 CVS and expected to be present
in next mpd5.3 release.
20081219:
2008-12-19 23:12:14 +00:00
With __FreeBSD_version 800060 the makefs tool is part of
the base system (it was a port).
20081216:
2009-02-17 10:50:18 +00:00
The afdata and ifnet locks have been changed from mutexes to
rwlocks, network modules will need to be re-compiled.
20081214:
__FreeBSD_version 800059 incorporates the new arp-v2 rewrite.
RTF_CLONING, RTF_LLINFO and RTF_WASCLONED flags are eliminated.
2009-02-17 10:50:18 +00:00
The new code reduced struct rtentry{} by 16 bytes on 32-bit
architecture and 40 bytes on 64-bit architecture. The userland
applications "arp" and "ndp" have been updated accordingly.
The output from "netstat -r" shows only routing entries and
none of the L2 information.
20081130:
__FreeBSD_version 800057 marks the switchover from the
binary ath hal to source code. Users must add the line:
2008-12-01 23:09:58 +00:00
options AH_SUPPORT_AR5416
to their kernel config files when specifying:
device ath_hal
The ath_hal module no longer exists; the code is now compiled
together with the driver in the ath module. It is now
possible to tailor chip support (i.e. reduce the set of chips
and thereby the code size); consult ath_hal(4) for details.
20081121:
__FreeBSD_version 800054 adds memory barriers to
<machine/atomic.h>, new interfaces to ifnet to facilitate
multiple hardware transmit queues for cards that support
them, and a lock-less ring-buffer implementation to
enable drivers to more efficiently manage queueing of
packets.
20081117:
A new version of ZFS (version 13) has been merged to -HEAD.
This version has zpool attribute "listsnapshots" off by
default, which means "zfs list" does not show snapshots,
and is the same as Solaris behavior.
20081028:
dummynet(4) ABI has changed. ipfw(8) needs to be recompiled.
20081009:
The uhci, ohci, ehci and slhci USB Host controller drivers have
been put into separate modules. If you load the usb module
separately through loader.conf you will need to load the
appropriate *hci module as well. E.g. for a UHCI-based USB 2.0
controller add the following to loader.conf:
uhci_load="YES"
ehci_load="YES"
20081009:
The ABI used by the PMC toolset has changed. Please keep
userland (libpmc(3)) and the kernel module (hwpmc(4)) in
sync.
20081009:
atapci kernel module now includes only generic PCI ATA
driver. AHCI driver moved to ataahci kernel module.
All vendor-specific code moved into separate kernel modules:
ataacard, ataacerlabs, ataadaptec, ataamd, ataati, atacenatek,
atacypress, atacyrix, atahighpoint, ataintel, ataite, atajmicron,
atamarvell, atamicron, atanational, atanetcell, atanvidia,
atapromise, ataserverworks, atasiliconimage, atasis, atavia
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
20080820:
The TTY subsystem of the kernel has been replaced by a new
implementation, which provides better scalability and an
improved driver model. Most common drivers have been migrated to
the new TTY subsystem, while others have not. The following
drivers have not yet been ported to the new TTY layer:
PCI/ISA:
cy, digi, rc, rp, sio
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
USB:
ubser, ucycom
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
Line disciplines:
ng_h4, ng_tty, ppp, sl, snp
Adding these drivers to your kernel configuration file shall
cause compilation to fail.
2008-09-03 08:30:17 +00:00
20080818:
ntpd has been upgraded to 4.2.4p5.
20080801:
OpenSSH has been upgraded to 5.1p1.
For many years, FreeBSD's version of OpenSSH preferred DSA
over RSA for host and user authentication keys. With this
upgrade, we've switched to the vendor's default of RSA over
DSA. This may cause upgraded clients to warn about unknown
host keys even for previously known hosts. Users should
follow the usual procedure for verifying host keys before
accepting the RSA key.
This can be circumvented by setting the "HostKeyAlgorithms"
option to "ssh-dss,ssh-rsa" in ~/.ssh/config or on the ssh
command line.
Please note that the sequence of keys offered for
authentication has been changed as well. You may want to
specify IdentityFile in a different order to revert this
behavior.
20080713:
The sio(4) driver has been removed from the i386 and amd64
kernel configuration files. This means uart(4) is now the
default serial port driver on those platforms as well.
2008-07-25 09:30:53 +00:00
To prevent collisions with the sio(4) driver, the uart(4) driver
uses different names for its device nodes. This means the
onboard serial port will now most likely be called "ttyu0"
instead of "ttyd0". You may need to reconfigure applications to
use the new device names.
When using the serial port as a boot console, be sure to update
/boot/device.hints and /etc/ttys before booting the new kernel.
If you forget to do so, you can still manually specify the hints
at the loader prompt:
set hint.uart.0.at="isa"
set hint.uart.0.port="0x3F8"
set hint.uart.0.flags="0x10"
set hint.uart.0.irq="4"
boot -s
2008-06-09 21:33:57 +00:00
20080609:
The gpt(8) utility has been removed. Use gpart(8) to partition
disks instead.
20080603:
The version that Linuxulator emulates was changed from 2.4.2
to 2.6.16. If you experience any problems with Linux binaries
2008-07-25 09:30:53 +00:00
please try to set sysctl compat.linux.osrelease to 2.4.2 and
if it fixes the problem contact emulation mailing list.
20080525:
ISDN4BSD (I4B) was removed from the src tree. You may need to
update a your kernel configuration and remove relevant entries.
20080509:
I have checked in code to support multiple routing tables.
2008-07-25 09:13:18 +00:00
See the man pages setfib(1) and setfib(2).
This is a hopefully backwards compatible version,
but to make use of it you need to compile your kernel
with options ROUTETABLES=2 (or more up to 16).
2008-04-20 21:25:37 +00:00
20080420:
The 802.11 wireless support was redone to enable multi-bss
operation on devices that are capable. The underlying device
is no longer used directly but instead wlanX devices are
cloned with ifconfig. This requires changes to rc.conf files.
For example, change:
ifconfig_ath0="WPA DHCP"
to
wlans_ath0=wlan0
ifconfig_wlan0="WPA DHCP"
2008-04-29 19:55:18 +00:00
see rc.conf(5) for more details. In addition, mergemaster of
/etc/rc.d is highly recommended. Simultaneous update of userland
and kernel wouldn't hurt either.
2008-04-20 21:25:37 +00:00
As part of the multi-bss changes the wlan_scan_ap and wlan_scan_sta
modules were merged into the base wlan module. All references
to these modules (e.g. in kernel config files) must be removed.
20080408:
psm(4) has gained write(2) support in native operation level.
Arbitrary commands can be written to /dev/psm%d and status can
be read back from it. Therefore, an application is responsible
for status validation and error recovery. It is a no-op in
other operation levels.
20080312:
Support for KSE threading has been removed from the kernel. To
run legacy applications linked against KSE libmap.conf may
be used. The following libmap.conf may be used to ensure
compatibility with any prior release:
libpthread.so.1 libthr.so.1
libpthread.so.2 libthr.so.2
libkse.so.3 libthr.so.3
20080301:
The layout of struct vmspace has changed. This affects libkvm
and any executables that link against libkvm and use the
kvm_getprocs() function. In particular, but not exclusively,
it affects ps(1), fstat(1), pkill(1), systat(1), top(1) and w(1).
The effects are minimal, but it's advisable to upgrade world
nonetheless.
20080229:
The latest em driver no longer has support in it for the
82575 adapter, this is now moved to the igb driver. The
split was done to make new features that are incompatible
with older hardware easier to do.
20080220:
The new geom_lvm(4) geom class has been renamed to geom_linux_lvm(4),
likewise the kernel option is now GEOM_LINUX_LVM.
20080211:
The default NFS mount mode has changed from UDP to TCP for
increased reliability. If you rely on (insecurely) NFS
mounting across a firewall you may need to update your
firewall rules.
2008-02-08 21:24:58 +00:00
20080208:
Belatedly note the addition of m_collapse for compacting
mbuf chains.
Our fts(3) API, as inherited from 4.4BSD, suffers from integer fields in FTS and FTSENT structs being too narrow. In addition, the narrow types creep from there into fts.c. As a result, fts(3) consumers, e.g., find(1) or rm(1), can't handle file trees an ordinary user can create, which can have security implications. To fix the historic implementation of fts(3), OpenBSD and NetBSD have already changed <fts.h> in somewhat incompatible ways, so we are free to do so, too. This change is a superset of changes from the other BSDs with a few more improvements. It doesn't touch fts(3) functionality; it just extends integer types used by it to match modern reality and the C standard. Here are its points: o For C object sizes, use size_t unless it's 100% certain that the object will be really small. (Note that fts(3) can construct pathnames _much_ longer than PATH_MAX for its consumers.) o Avoid the short types because on modern platforms using them results in larger and slower code. Change shorts to ints as follows: - For variables than count simple, limited things like states, use plain vanilla `int' as it's the type of choice in C. - For a limited number of bit flags use `unsigned' because signed bit-wise operations are implementation-defined, i.e., unportable, in C. o For things that should be at least 64 bits wide, use long long and not int64_t, as the latter is an optional type. See FTSENT.fts_number aka FTS.fts_bignum. Extending fts_number `to satisfy future needs' is pointless because there is fts_pointer, which can be used to link to arbitrary data from an FTSENT. However, there already are fts(3) consumers that require fts_number, or fts_bignum, have at least 64 bits in it, so we must allow for them. o For the tree depth, use `long'. This is a trade-off between making this field too wide and allowing for 64-bit inode numbers and/or chain-mounted filesystems. On the one hand, `long' is almost enough for 32-bit filesystems on a 32-bit platform (our ino_t is uint32_t now). On the other hand, platforms with a 64-bit (or wider) `long' will be ready for 64-bit inode numbers, as well as for several 32-bit filesystems mounted one under another. Note that fts_level has to be signed because -1 is a magic value for it, FTS_ROOTPARENTLEVEL. o For the `nlinks' local var in fts_build(), use `long'. The logic in fts_build() requires that `nlinks' be signed, but our nlink_t currently is uint16_t. Therefore let's make the signed var wide enough to be able to represent 2^16-1 in pure C99, and even 2^32-1 on a 64-bit platform. Perhaps the logic should be changed just to use nlink_t, but it can be done later w/o breaking fts(3) ABI any more because `nlinks' is just a local var. This commit also inludes supporting stuff for the fts change: o Preserve the old versions of fts(3) functions through libc symbol versioning because the old versions appeared in all our former releases. o Bump __FreeBSD_version just in case. There is a small chance that some ill-written 3-rd party apps may fail to build or work correctly if compiled after this change. o Update the fts(3) manpage accordingly. In particular, remove references to fts_bignum, which was a FreeBSD-specific hack to work around the too narrow types of FTSENT members. Now fts_number is at least 64 bits wide (long long) and fts_bignum is an undocumented alias for fts_number kept around for compatibility reasons. According to Google Code Search, the only big consumers of fts_bignum are in our own source tree, so they can be fixed easily to use fts_number. o Mention the change in src/UPDATING. PR: bin/104458 Approved by: re (quite a while ago) Discussed with: deischen (the symbol versioning part) Reviewed by: -arch (mostly silence); das (generally OK, but we didn't agree on some types used; assuming that no objections on -arch let me to stick to my opinion)
2008-01-26 17:09:40 +00:00
20080126:
The fts(3) structures have been changed to use adequate
integer types for their members and so to be able to cope
with huge file trees. The old fts(3) ABI is preserved
through symbol versioning in libc, so third-party binaries
using fts(3) should still work, although they will not take
advantage of the extended types. At the same time, some
third-party software might fail to build after this change
due to unportable assumptions made in its source code about
fts(3) structure members. Such software should be fixed
by its vendor or, in the worst case, in the ports tree.
FreeBSD_version 800015 marks this change for the unlikely
case that a portable fix is impossible.
20080123:
To upgrade to -current after this date, you must be running
FreeBSD not older than 6.0-RELEASE. Upgrading to -current
from 5.x now requires a stop over at RELENG_6 or RELENG_7 systems.
20071128:
The ADAPTIVE_GIANT kernel option has been retired because its
functionality is the default now.
20071118:
The AT keyboard emulation of sunkbd(4) has been turned on
by default. In order to make the special symbols of the Sun
keyboards driven by sunkbd(4) work under X these now have
to be configured the same way as Sun USB keyboards driven
by ukbd(4) (which also does AT keyboard emulation), f.e.:
Option "XkbLayout" "us"
Option "XkbRules" "xorg"
Option "XkbSymbols" "pc(pc105)+sun_vndr/usb(sun_usb)+us"
20071024:
It has been decided that it is desirable to provide ABI
backwards compatibility to the FreeBSD 4/5/6 versions of the
PCIOCGETCONF, PCIOCREAD and PCIOCWRITE IOCTLs, which was
broken with the introduction of PCI domain support (see the
20070930 entry). Unfortunately, this required the ABI of
PCIOCGETCONF to be broken again in order to be able to
provide backwards compatibility to the old version of that
IOCTL. Thus consumers of PCIOCGETCONF have to be recompiled
again. As for prominent ports this affects neither pciutils
nor xorg-server this time, the hal port needs to be rebuilt
however.
20071020:
The misnamed kthread_create() and friends have been renamed
to kproc_create() etc. Many of the callers already
used kproc_start()..
I will return kthread_create() and friends in a while
with implementations that actually create threads, not procs.
Renaming corresponds with version 800002.
20071010:
RELENG_7 branched.
COMMON ITEMS:
General Notes
-------------
Avoid using make -j when upgrading. While generally safe, there are
sometimes problems using -j to upgrade. If your upgrade fails with
-j, please try again wtihout -j. From time to time in the past there
have been problems using -j with buildworld and/or installworld. This
is especially true when upgrading between "distant" versions (eg one
that cross a major release boundary or several minor releases, or when
several months have passed on the -current branch).
Sometimes, obscure build problems are the result of environment
poisoning. This can happen because the make utility reads its
environment when searching for values for global variables. To run
your build attempts in an "environmental clean room", prefix all make
commands with 'env -i '. See the env(1) manual page for more details.
When upgrading from one major version to another it is generally best
to upgrade to the latest code in the currently installed branch first,
then do an upgrade to the new branch. This is the best-tested upgrade
path, and has the highest probability of being successful. Please try
this approach before reporting problems with a major version upgrade.
To build a kernel
-----------------
If you are updating from a prior version of FreeBSD (even one just
a few days old), you should follow this procedure. It is the most
failsafe as it uses a /usr/obj tree with a fresh mini-buildworld,
make kernel-toolchain
make -DALWAYS_CHECK_MAKE buildkernel KERNCONF=YOUR_KERNEL_HERE
make -DALWAYS_CHECK_MAKE installkernel KERNCONF=YOUR_KERNEL_HERE
To test a kernel once
---------------------
If you just want to boot a kernel once (because you are not sure
if it works, or if you want to boot a known bad kernel to provide
debugging information) run
make installkernel KERNCONF=YOUR_KERNEL_HERE KODIR=/boot/testkernel
nextboot -k testkernel
To just build a kernel when you know that it won't mess you up
--------------------------------------------------------------
This assumes you are already running a CURRENT system. Replace
${arch} with the architecture of your machine (e.g. "i386",
"arm", "amd64", "ia64", "pc98", "sparc64", "powerpc", "mips", etc).
cd src/sys/${arch}/conf
config KERNEL_NAME_HERE
cd ../compile/KERNEL_NAME_HERE
make depend
make
2004-06-22 10:33:58 +00:00
make install
If this fails, go to the "To build a kernel" section.
To rebuild everything and install it on the current system.
-----------------------------------------------------------
# Note: sometimes if you are running current you gotta do more than
# is listed here if you are upgrading from a really old current.
<make sure you have good level 0 dumps>
make buildworld
make kernel KERNCONF=YOUR_KERNEL_HERE
[1]
<reboot in single user> [3]
mergemaster -p [5]
make installworld
make delete-old
mergemaster [4]
<reboot>
To cross-install current onto a separate partition
--------------------------------------------------
# In this approach we use a separate partition to hold
# current's root, 'usr', and 'var' directories. A partition
# holding "/", "/usr" and "/var" should be about 2GB in
# size.
<make sure you have good level 0 dumps>
<boot into -stable>
make buildworld
make buildkernel KERNCONF=YOUR_KERNEL_HERE
<maybe newfs current's root partition>
<mount current's root partition on directory ${CURRENT_ROOT}>
make installworld DESTDIR=${CURRENT_ROOT}
make distribution DESTDIR=${CURRENT_ROOT} # if newfs'd
make installkernel KERNCONF=YOUR_KERNEL_HERE DESTDIR=${CURRENT_ROOT}
cp /etc/fstab ${CURRENT_ROOT}/etc/fstab # if newfs'd
<edit ${CURRENT_ROOT}/etc/fstab to mount "/" from the correct partition>
<reboot into current>
<do a "native" rebuild/install as described in the previous section>
<maybe install compatibility libraries from ports/misc/compat*>
<reboot>
To upgrade in-place from 8.x-stable to current
----------------------------------------------
<make sure you have good level 0 dumps>
make buildworld [9]
make kernel KERNCONF=YOUR_KERNEL_HERE [8]
[1]
<reboot in single user> [3]
mergemaster -p [5]
make installworld
make delete-old
mergemaster -i [4]
<reboot>
2000-08-28 03:54:51 +00:00
Make sure that you've read the UPDATING file to understand the
tweaks to various things you need. At this point in the life
cycle of current, things change often and you are on your own
to cope. The defaults can also change, so please read ALL of
the UPDATING entries.
Also, if you are tracking -current, you must be subscribed to
freebsd-current@freebsd.org. Make sure that before you update
your sources that you have read and understood all the recent
messages there. If in doubt, please track -stable which has
much fewer pitfalls.
[1] If you have third party modules, such as vmware, you
should disable them at this point so they don't crash your
system on reboot.
[3] From the bootblocks, boot -s, and then do
fsck -p
mount -u /
mount -a
cd src
adjkerntz -i # if CMOS is wall time
2002-04-25 02:14:48 +00:00
Also, when doing a major release upgrade, it is required that
you boot into single user mode to do the installworld.
[4] Note: This step is non-optional. Failure to do this step
can result in a significant reduction in the functionality of the
system. Attempting to do it by hand is not recommended and those
that pursue this avenue should read this file carefully, as well
as the archives of freebsd-current and freebsd-hackers mailing lists
for potential gotchas.
[5] Usually this step is a noop. However, from time to time
you may need to do this if you get unknown user in the following
step. It never hurts to do it all the time. You may need to
install a new mergemaster (cd src/usr.sbin/mergemaster && make
install) after the buildworld before this step if you last updated
from current before 20020224 or from -stable before 20020408.
[8] In order to have a kernel that can run the 4.x binaries needed to
do an installworld, you must include the COMPAT_FREEBSD4 option in
your kernel. Failure to do so may leave you with a system that is
hard to boot to recover. A similar kernel option COMPAT_FREEBSD5 is
required to run the 5.x binaries on more recent kernels. And so on
for COMPAT_FREEBSD6 and COMPAT_FREEBSD7.
Make sure that you merge any new devices from GENERIC since the
last time you updated your kernel config file.
[9] When checking out sources, you must include the -P flag to have
cvs prune empty directories.
If CPUTYPE is defined in your /etc/make.conf, make sure to use the
"?=" instead of the "=" assignment operator, so that buildworld can
override the CPUTYPE if it needs to.
MAKEOBJDIRPREFIX must be defined in an environment variable, and
not on the command line, or in /etc/make.conf. buildworld will
warn if it is improperly defined.
FORMAT:
This file contains a list, in reverse chronological order, of major
breakages in tracking -current. Not all things will be listed here,
and it only starts on October 16, 2004. Updating files can found in
previous releases if your system is older than this.
Copyright information:
Copyright 1998-2009 M. Warner Losh. All Rights Reserved.
Redistribution, publication, translation and use, with or without
modification, in full or in part, in any form or format of this
document are permitted without further permission from the author.
THIS DOCUMENT IS PROVIDED BY WARNER LOSH ``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 WARNER LOSH 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.
Contact Warner Losh if you have any questions about your use of
this document.
1999-08-28 01:35:59 +00:00
$FreeBSD$