[owc] Weekly src changes ending 2006-12-03
OWC
auto at squish.net
Mon Dec 4 07:11:21 GMT 2006
OpenBSD src changes summary for 2006-11-26 to 2006-12-03 inclusive
==================================================================
bin/ksh bin/ps
distrib/notes distrib/sets
etc/changelist etc/rc
gnu/usr.bin/gcc gnu/usr.bin/perl
lib/libc lib/libevent
lib/libkvm lib/libpthread
libexec/spamd libexec/spamlogd
regress/sbin sbin/bioctl
sbin/brconfig sbin/dhclient
sbin/ipsecctl sbin/isakmpd
sbin/nmeaattach sbin/pfctl
sbin/pflogd sbin/route
share/man sys/arch/alpha/alpha
sys/arch/alpha/tc sys/arch/amd64/amd64
sys/arch/amd64/conf sys/arch/arm/arm
sys/arch/arm/xscale sys/arch/armish/conf
sys/arch/hp300/dev sys/arch/hp300/hp300
sys/arch/hp300/include sys/arch/hppa/dev
sys/arch/hppa/hppa sys/arch/hppa/spmath
sys/arch/hppa64/hppa64 sys/arch/i386/conf
sys/arch/i386/i386 sys/arch/i386/include
sys/arch/i386/pci sys/arch/luna88k/dev
sys/arch/luna88k/luna88k sys/arch/m68k/conf
sys/arch/m68k/fpsp sys/arch/m68k/include
sys/arch/m68k/m68k sys/arch/m88k/include
sys/arch/mac68k/dev sys/arch/mac68k/include
sys/arch/macppc/conf sys/arch/macppc/dev
sys/arch/macppc/macppc sys/arch/macppc/pci
sys/arch/mips64/include sys/arch/mips64/mips64
sys/arch/mvme68k/dev sys/arch/mvme68k/include
sys/arch/mvme68k/mvme68k sys/arch/mvme88k/dev
sys/arch/mvme88k/mvme88k sys/arch/powerpc/include
sys/arch/powerpc/powerpc sys/arch/sh/include
sys/arch/sh/sh sys/arch/sparc/conf
sys/arch/sparc/dev sys/arch/sparc/include
sys/arch/sparc/sparc sys/arch/sparc64/conf
sys/arch/sparc64/dev sys/arch/sparc64/include
sys/arch/sparc64/sparc64 sys/arch/vax/dec
sys/arch/vax/include sys/arch/vax/vax
sys/arch/vax/vsa sys/arch/zaurus/conf
sys/arch/zaurus/zaurus sys/conf
sys/dev sys/dev/acpi
sys/dev/atapiscsi sys/dev/cardbus
sys/dev/eisa sys/dev/i2o
sys/dev/ic sys/dev/isa
sys/dev/mii sys/dev/pci
sys/dev/pcmcia sys/dev/rasops
sys/dev/sbus sys/dev/sdmmc
sys/dev/usb sys/dev/vesa
sys/dev/wscons sys/kern
sys/lib/libz sys/miscfs/procfs
sys/miscfs/specfs sys/msdosfs
sys/net sys/net80211
sys/netbt sys/netinet
sys/netinet6 sys/nfs
sys/scsi sys/sys
sys/uvm usr.bin
usr.bin/calendar usr.bin/compress
usr.bin/cvs usr.bin/dc
usr.bin/fmt usr.bin/gzsig
usr.bin/nc usr.bin/pkg-config
usr.bin/systat usr.bin/tip
usr.bin/top usr.sbin/bgpctl
usr.sbin/bgpd usr.sbin/dhcrelay
usr.sbin/dvmrpctl usr.sbin/dvmrpd
usr.sbin/ftp-proxy usr.sbin/gpioctl
usr.sbin/ifstated usr.sbin/ntpd
usr.sbin/ospfctl usr.sbin/ospfd
usr.sbin/pkg_add usr.sbin/ripctl
usr.sbin/ripd usr.sbin/sasyncd
usr.sbin/sdio
== bin =============================================================== 01/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/bin
ksh
~ ksh.1
> typos; from alf schlichting (jmc@)
ps
~ keyword.c ~ print.c
~ ps.1
> Do not test for processes being swapped out since this can't happen
> anymore. (miod@)
== distrib =========================================================== 02/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/distrib
notes
~ i386/hardware ~ macppc/hardware
> Add all available watchdog timers to i386/hardware, pwdog to
> macppc/hardware.
> reminded by miod. (mbalmer@)
~ armish/prep
> fix the installation guide of the Thecus boxes by
> providing two working methods;
> ok deraadt@ (robert@)
~ i386/prep
> Update the sections about 1024-cylinder hard disk limits on install.
> Originally from nick@, with some tweaks by jmc at . > ok jmc@ nick@ (tom@)
~ armish/prep
> Mention that the jumper must be moved to J4 on Thecus N2100. (drahn@)
sets
~ lists/base/md.alpha ~ lists/base/md.amd64
~ lists/base/md.armish ~ lists/base/md.aviion
~ lists/base/md.cats ~ lists/base/md.hp300
~ lists/base/md.hppa ~ lists/base/md.hppa64
~ lists/base/md.i386 ~ lists/base/md.landisk
~ lists/base/md.luna88k ~ lists/base/md.mac68k
~ lists/base/md.macppc ~ lists/base/md.mvme68k
~ lists/base/md.mvme88k ~ lists/base/md.mvmeppc
~ lists/base/md.sgi ~ lists/base/md.sparc
~ lists/base/md.sparc64 ~ lists/base/md.vax
~ lists/base/md.zaurus ~ lists/base/mi
~ lists/comp/mi ~ lists/man/mi
> sync (deraadt@)
~ lists/man/mi
> sync (deraadt@)
~ lists/man/mi
> sync (deraadt@)
~ lists/base/md.alpha ~ lists/base/md.amd64
~ lists/base/md.armish ~ lists/base/md.aviion
~ lists/base/md.cats ~ lists/base/md.hp300
~ lists/base/md.hppa ~ lists/base/md.hppa64
~ lists/base/md.i386 ~ lists/base/md.landisk
~ lists/base/md.luna88k ~ lists/base/md.mac68k
~ lists/base/md.macppc ~ lists/base/md.mvme68k
~ lists/base/md.mvme88k ~ lists/base/md.mvmeppc
~ lists/base/md.sgi ~ lists/base/md.sparc
~ lists/base/md.sparc64 ~ lists/base/md.vax
~ lists/base/md.zaurus ~ lists/comp/mi
~ lists/man/mi
> sync (deraadt@)
~ lists/base/md.alpha ~ lists/base/md.amd64
~ lists/base/md.armish ~ lists/base/md.aviion
~ lists/base/md.cats ~ lists/base/md.hp300
~ lists/base/md.hppa ~ lists/base/md.hppa64
~ lists/base/md.i386 ~ lists/base/md.landisk
~ lists/base/md.luna88k ~ lists/base/md.mac68k
~ lists/base/md.macppc ~ lists/base/md.mvme68k
~ lists/base/md.mvme88k ~ lists/base/md.mvmeppc
~ lists/base/md.sgi ~ lists/base/md.sparc
~ lists/base/md.sparc64 ~ lists/base/md.vax
~ lists/base/md.zaurus
> sync (deraadt@)
~ lists/base/md.alpha ~ lists/base/md.amd64
~ lists/base/md.armish ~ lists/base/md.aviion
~ lists/base/md.cats ~ lists/base/md.hp300
~ lists/base/md.hppa ~ lists/base/md.hppa64
~ lists/base/md.i386 ~ lists/base/md.landisk
~ lists/base/md.luna88k ~ lists/base/md.mac68k
~ lists/base/md.macppc ~ lists/base/md.mvme68k
~ lists/base/md.mvme88k ~ lists/base/md.mvmeppc
~ lists/base/md.sgi ~ lists/base/md.sparc
~ lists/base/md.sparc64 ~ lists/base/md.vax
~ lists/base/md.zaurus
> sync (deraadt@)
~ lists/man/mi
> sync (deraadt@)
~ lists/base/md.alpha ~ lists/base/md.amd64
~ lists/base/md.armish ~ lists/base/md.aviion
~ lists/base/md.cats ~ lists/base/md.hp300
~ lists/base/md.hppa ~ lists/base/md.hppa64
~ lists/base/md.i386 ~ lists/base/md.landisk
~ lists/base/md.luna88k ~ lists/base/md.mac68k
~ lists/base/md.macppc ~ lists/base/md.mvme68k
~ lists/base/md.mvme88k ~ lists/base/md.mvmeppc
~ lists/base/md.sgi ~ lists/base/md.sparc
~ lists/base/md.sparc64 ~ lists/base/md.vax
~ lists/base/md.zaurus ~ lists/base/mi
~ lists/man/mi
> sync (deraadt@)
~ lists/base/mi
> sync (deraadt@)
== etc =============================================================== 03/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/etc
changelist
~ changelist
> iskampds public key local.pub moved from /etc/isakmpd/private to
> /etc/isamkpd. Noticed by ajacoutot at lphp.orghanks!
> ok mcbride (hshoexer@)
rc
~ rc
> Move isakmpd's auto-generated public key, local.pub, into /etc/isakmpd
> This makes it readable by unprivileged uses, simplifying configuration,
> and there is no reason for it to be secret.
> ok msf deraadt hshoexer (mcbride@)
== gnu =============================================================== 04/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/gnu
usr.bin/gcc
~ gcc/config/sh/openbsd.h ~ gcc/config/sh/sh-protos.h
~ gcc/config/sh/sh.c ~ gcc/config/sh/sh.h
> Turn OVERRIDE_OPTIONS macro into a function, so that we can put
> preprocessor directives in it; then do our changes only if either
> OPENBSD_NATIVE or OPENBSD_CROSS is defined; then disable -fregmove
> for now.
> ok otto@ (miod@)
usr.bin/perl
~ mg.c
> When using setproctitle() don't fall back into the non-setproctitle()
> code. Fixes an environment corruption problem when $0 is modified.
> From Alexander Bluhm; this is perl bug #41008 (millert@)
== lib =============================================================== 05/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/lib
libc
~ stdio/vfprintf.c
> try first mmap() allocation at pagesize, no need to re-mmap at every power
> of 2 from 16 to pagesize(); ok otto (deraadt@)
libevent
~ poll.c
> fail without leaking memory when poll set extension fails.
> From Niels Provos via the libevent SVN
> ok deraadt@ (brad@)
~ evbuffer.c ~ event.3
> allow both read and write callbacks for bufferevents to be NULL.
> From Niels Provos via the libevent SVN
> ok deraadt@ (brad@)
~ buffer.c
> fix a potential problem in multiple handling of va_list; use va_copy
> instead.
> From Niels Provos via the libevent SVN, by Alejo.
> ok deraadt@ (brad@)
libkvm
~ kvm_proc.c
> Sync with struct proc change. (miod@)
~ kvm_mips64.c
> Fix cut'n'paste abuse in error message. (miod@)
libpthread
~ uthread/uthread_fd.c
> Normally pipes created by threaded apps are left non-blocking after being
> closed so that a threaded child process can still read it without blocking.
> However, leaving stdin/out/err non-blocking when closed is bad because it
> can be shared with non-threaded apps that can't deal with a non-blocking
> file descriptor (i.e. cat). Therefore special case stdin/out/err pipes so
> that they are reset to blocking upon a close(). Tested by robert@, jolan@
> and myself with multiple OOo builds on mp systems where the problem was
> seen more frequently. (kurt@)
== libexec =========================================================== 06/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/libexec
spamd
~ spamd.c
> Make spamd error out for a greylisted session only after the DATA
> command not after the RCPT command, as this helps people out when
> they are faced with retarded sender verification schemes - tested
> out by a number of people on tech@, and running at u of a for a
> while. Supposedly this makes verizon sender verification happy.
> Suggested as a stable candidate by people on tech@, I'm ok with
> that. (beck@)
spamlogd
~ Makefile
> repair missing DPADD requests (deraadt@)
== regress =========================================================== 07/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/regress
sbin
~ pfctl/Makefile + pfctl/pf93.in
+ pfctl/pf93.ok
> very basic regress test for rtable select code (henning@)
~ ipsecctl/ike10.ok ~ ipsecctl/ike30.ok
~ ipsecctl/ike42.ok ~ ipsecctl/ike43.ok
~ ipsecctl/ike54.ok ~ ipsecctl/ikedel10.ok
~ ipsecctl/ikedel30.ok ~ ipsecctl/ikedel42.ok
~ ipsecctl/ikedel43.ok
> sync: proto/port in lid/rid/connection (markus@)
~ ipsecctl/ikedel1.ok ~ ipsecctl/ikedel10.ok
~ ipsecctl/ikedel11.ok ~ ipsecctl/ikedel12.ok
~ ipsecctl/ikedel13.ok ~ ipsecctl/ikedel14.ok
~ ipsecctl/ikedel15.ok ~ ipsecctl/ikedel16.ok
~ ipsecctl/ikedel17.ok ~ ipsecctl/ikedel18.ok
~ ipsecctl/ikedel19.ok ~ ipsecctl/ikedel2.ok
~ ipsecctl/ikedel20.ok ~ ipsecctl/ikedel21.ok
~ ipsecctl/ikedel22.ok ~ ipsecctl/ikedel23.ok
~ ipsecctl/ikedel29.ok ~ ipsecctl/ikedel3.ok
~ ipsecctl/ikedel30.ok ~ ipsecctl/ikedel31.ok
~ ipsecctl/ikedel32.ok ~ ipsecctl/ikedel33.ok
~ ipsecctl/ikedel34.ok ~ ipsecctl/ikedel35.ok
~ ipsecctl/ikedel36.ok ~ ipsecctl/ikedel37.ok
~ ipsecctl/ikedel38.ok ~ ipsecctl/ikedel39.ok
~ ipsecctl/ikedel4.ok ~ ipsecctl/ikedel40.ok
~ ipsecctl/ikedel41.ok ~ ipsecctl/ikedel42.ok
~ ipsecctl/ikedel43.ok ~ ipsecctl/ikedel46.ok
~ ipsecctl/ikedel47.ok ~ ipsecctl/ikedel5.ok
~ ipsecctl/ikedel6.ok ~ ipsecctl/ikedel7.ok
~ ipsecctl/ikedel8.ok ~ ipsecctl/ikedel9.ok
> sync: rmv to unregister ipsec connections (markus@)
~ ipsecctl/ike10.ok ~ ipsecctl/ike30.ok
~ ipsecctl/ike42.ok ~ ipsecctl/ike43.ok
~ ipsecctl/ike54.ok
> wrong rid for protocol (markus@)
== sbin ============================================================== 08/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/sbin
bioctl
~ Makefile
> repair missing DPADD requests (deraadt@)
brconfig
~ brconfig.c ~ brconfig.8
> Add Rapid Spanning Tree Protocol support (802.1d-2004) based on work
> by Andrew Thompson (thompsa at freebsd.orgThe local changes include
> adoption to our bridge code, reduced stack usage and many other bits.
> If stp is enabled, RSTP will now be used by default.
> Thanks for help from Andrew.
> This code has been in snaps for while now, commit encouraged by deraadt@
> (reyk@)
dhclient
~ dhcpd.h
> remove invalid comment (stevesk@)
~ bpf.c ~ dhcpd.h
> Recognize when we are sending to a unicast destination IP address and
> instead of using BPF, which currently will always set the destination
> MAC to broadcast, send using a SOCK_RAW socket and sendmsg() so the
> frame has a unicast destination MAC.
> Fixes an issue when using a bootp forwarder where unicast DHCPREQUESTs
> are dropped at the router/gateway until we reach T2/REBINDING and
> change the destination IP to broadcast.
> ok henning@ (but doesn't like using a raw socket for this) (stevesk@)
~ dhclient.c
> correct comment (stevesk@)
ipsecctl
~ ike.c ~ ipsecctl.c
~ ipsecctl.h
> handle multiple SAs with different same src/dst but different port;
> store IKE connection string and phase2 IDs in the ipsec rule;
> cleanup internal API: pass rules around instead of rule members;
> report Brian Candler; fix with hshoexer, msf; ok hshoexer (markus@)
~ ike.c
> use rmv to unregister ipsec connections; ok hshoexer, ho (markus@)
~ ike.c
> typo: wrong rid for protocol (markus@)
isakmpd
~ isakmpd.8 ~ ui.c
> do not re-add existing entries; ok hshoexer (markus@)
~ isakmpd.8
> Document the new location of local.pub, and clarify the fact that local.key
> contains the entire keypair.
> ok deraadt jmc (mcbride@)
~ isakmpd.8
> zap trailing spaces; (jmc@)
~ isakmpd.8
> no need to document generation of local.key 3 times;
> spotted by mcbride, ok hshoexer; (jmc@)
~ isakmpd.8 ~ ui.c
> new ui command 'rmv': removes an entry from a list, thus reversing an
> 'add' operation; ok ho, hshoexer, jmc
> eVS: ----------------------------------------------------------------------
> (markus@)
nmeaattach
~ nmeaattach.8
> Document the new timestamping behaviour here, too, and note that enabling
> PPS is a device specific operation that might need special software (e.g.
> from our ports tree).
> gpsd is such an application and it has just been enabled to nicely play
> with
> the new timestamping. (mbalmer@)
~ nmeaattach.c
> Remove an unused variable.
> From Kevin Steves. (mbalmer@)
~ nmeaattach.c
> Forgot this in my last committ: Remove not-needed header files.
> Also from Kevin Steves. (mbalmer@)
pfctl
~ pfctl_altq.c
> fix servicecurve check; no point in checking the same sc three times, it
> was obviously intended to check all three. has been wrong since the
> beginning, 4 years... noticed by Earl Lapus <earl.lapus at gmail.com > Vasil Dimov <vd at FreeBSD.orgailed me then, ok mcbride (henning@)
pflogd
~ Makefile
> repair missing DPADD requests (deraadt@)
route
~ route.8 ~ route.c
> -T in synopsis and briefly in manpage (henning@)
~ route.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
~ route.8
> add -T to synopsis; (jmc@)
~ route.8
> tweak; ok henning (jmc@)
== share ============================================================= 09/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/share
man
~ man4/axe.4
> add the Linksys USB1000 (brad@)
~ man7/ports.7
> there's no longer a `special' main packages in MULTI_PACKAGES ports
> (espie@)
~ man5/bsd.port.mk.5
> document new MULTI_PACKAGES (espie@)
~ man5/bsd.port.mk.5 ~ man7/ports.7
> - kill trailing space
> - new sentence, new line (jmc@)
~ man4/Makefile + man4/usbf.4
> Introduction to Universal Serial Bus function support. (uwe@)
~ man4/Makefile ~ man4/usb.4
+ man4/uberry.4
> smallest usb driver ever: enable config on the Blackberry which prevents
> usb from idling, thus it can get enough power to charge in all cases
> (deraadt@)
~ man4/uberry.4
> tweaks; ok deraadt (jmc@)
~ man4/usbf.4
> - remove reference to non-existent usbf(9); noticed by jmc
> - add author information (uwe@)
~ man4/Makefile + man4/cdcef.4
> document the CDC Ethernet function (uwe@)
~ man4/man4.vax/lkkbd.4
> KB_PT (miod@)
~ man9/Makefile + man9/bio_register.9
> document bio_register and bio_unregister
> with help from jmc@ (dlg@)
~ man4/malo.4
> Comment in WEP support. (mglocker@)
~ man4/malo.4
> do not use .Pp before or after .Sh; (jmc@)
~ man4/ami.4
> useless Xr (deraadt@)
~ man4/Makefile ~ man4/pci.4
+ man4/ips.4
> It's really nice to start your day with writting a man page. (grange@)
~ man5/pf.conf.5
> mention rtable shitz now that it is enabled in the forwarding path
> (henning@)
~ man4/ddb.4
> Support "trace /p <pid>" to show the stack trace of any process.
> (Keep in mind that $radix is 0x10 by default.); ok by many (uwe@)
~ man4/ips.4
> full mnth name for .Dd; (jmc@)
~ man5/pf.conf.5
> bad space; (jmc@)
~ man4/nmea.4
> Document the changes I committed earlier: If we use a device with PPS
> output,
> the timestamp is copied from the tty timestamp (presuming tty timestamping
> has
> been enabled) or taken when the initial '$' character is received
> otherwise. (mbalmer@)
~ man5/pf.conf.5
> stateles -> stateless; from stuart henderson (jmc@)
~ man4/nmea.4
> tweaks; (jmc@)
~ man4/malo.4
> Enable monitor mode.
> ok claudio@ (mglocker@)
~ man4/usb.4 ~ man4/zyd.4
> Update for ZD1211B support. (jsg@)
~ man4/bcw.4
> Mention CardBus attachment in synopsis. (jsg@)
~ man4/zyd.4
> more ZD1211B devices (jsg@)
~ man5/pf.conf.5
> Correct the explanation of NAT evaluation order. binat is always first,
> then rdr on inbound packets or nat on outbound packets. This is _not_
> necessarily the same order in which the rules are defined in the ruleset.
> ok jmc dhartmei henning (camield@)
~ man4/pci.4 ~ man4/cardbus.4
> Mention bcw, pointed out by jmc. (jsg@)
~ man5/bsd.port.mk.5
> give a more thorough list of subpackage-dependent variable,
> rephrase description of package/subpackage to not give the impression
> one explicitly invokes the other, since it doesn't... (espie@)
~ man4/ueagle.4
> add link to external firmware package. (damien@)
== sys =============================================================== 10/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/sys
arch/alpha/alpha
~ api_up1000.c ~ dec_1000a.c
~ dec_2100_a50.c ~ dec_3000_300.c
~ dec_3000_500.c ~ dec_550.c
~ dec_6600.c ~ dec_axppci_33.c
~ dec_eb164.c ~ dec_eb64plus.c
~ dec_kn20aa.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
~ pmap.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
~ vm_machdep.c
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/alpha/tc
~ cfb.c ~ sfb.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
arch/amd64/amd64
~ acpi_machdep.c
> Only establish acpi interrupt if ACPI_ENABLE. (kettenis@)
~ fpu.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
~ vm_machdep.c
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/amd64/conf
~ GENERIC
> Enable malo(4) for amd64. Tested with a Netgear WG311v3 PCI. Further
> testing needed.
> ok deraadt@ (mglocker@)
~ GENERIC
> leave name simple (deraadt@)
~ GENERIC
> move uberry to a nicer place (deraadt@)
~ GENERIC ~ GENERIC.MP
> Added support to GENERIC for disabled acpi (jordan@)
~ GENERIC
> Add commented out ips here as well. (grange@)
arch/arm/arm
~ db_trace.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
~ vm_machdep.c
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
~ mem.c
> Use unsigned numbers for comparisons in mmmmap(). (miod@)
arch/arm/xscale
~ pxa27x_udc.c
> add needs-flag to allow pxaudc to compile without usbf (to get the
> previous behaviour of just disabling the USB device controller) (uwe@)
arch/armish/conf
~ GENERIC
> disable a gpio that is not needed (deraadt@)
arch/hp300/dev
~ mb89352.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ diofb.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ diofb.c
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
arch/hp300/hp300
~ autoconf.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
arch/hp300/include
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/hppa/dev
~ sti_sgc.c
> replace a bunch of ifs w/ an array scan and add a few more (mickey@)
arch/hppa/hppa
~ autoconf.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
~ vm_machdep.c
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
~ vm_machdep.c
> cpu_swapin is nomore (mickey@)
arch/hppa/spmath
- setovfl.c ~ Makefile
~ dbl_float.h ~ sgl_float.h
> unused file and functions (mickey@)
arch/hppa64/hppa64
~ autoconf.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
~ vm_machdep.c
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
~ vm_machdep.c
> cpu_swapin is nomore (mickey@)
arch/i386/conf
~ GENERIC
> Activate pwdog(4) after successful testing on i386 and macppc.
> ok deraadt (mbalmer@)
~ GENERIC
> leave name simple (deraadt@)
~ GENERIC
> enable uberry driver (deraadt@)
~ GENERIC
> move uberry to a nicer place (deraadt@)
~ GENERIC
> IBM ServeRAID controllers driver.
> Way far from complete but enough to fdisk and disklabel logical drives.
> (grange@)
~ GENERIC
> Enable kvm86.
> ok deraadt (gwk@)
~ files.i386
> commit vesabios for matthieu@
> ok deraadt, "slap it in" miod@ (gwk@)
~ GENERIC ~ GENERIC.MP
> Add disabled acpi to GENERIC and GENERIC.MP. This is required by the new
> and shiny ioapic code ketennis wrote.
> ok ketennis, jordan, deraadt (marco@)
~ GENERIC
> Fix the major bug in ips which prevents GENERIC to attach it. (grange@)
~ Makefile.i386
> Nuke all commons but one, and enable --warn-common in LINKFLAGS. (miod@)
arch/i386/i386
~ esm.c
> use splclock() instead of splsoftclock() when setting the timeout period
> from the watchdog framework; this is consistent with other watchdogs we
> have.
> discussed with and ok dlg (mbalmer@)
~ locore.s
> Save misc info (ebx) register from cpuid with eax=1, so we can use it
> later for some identifications.
> "slap it in" deraadt (dim@)
~ kvm86call.S
> Prevent panics when booting GENERIC.MP with a vesabios kernel. (gwk@)
~ kvm86.c
> First attempt at making kvm86 MP safe, place mutex lock around calls
> to kvm86_bioscall at IPL_IPI this might be incorrect, seems to fix
> some issues with vt switching, however X still bombs on the second attempt
> with
> disappointing frequency.
> ok deraadt@ (gwk@)
~ mainbus.c
> commit vesabios for matthieu@
> ok deraadt, "slap it in" miod@ (gwk@)
~ machdep.c ~ mainbus.c
> Only use est and other non-acpi setperf mechanisms, if acpi is not
> attached. Also moves their initialization and printing of
> information until after the primary cpu has attached to mainbus.
> prodded by deraadt (dim@)
~ machdep.c
> Don't display cpu brand string, features, etc. a second time, when booting
> an MP
> kernel on a single cpu machine.
> discussed with kettenis@ (dim@)
~ machdep.c
> Always call cyrix3_get_bus_clock in cyrix3_cpu_setup, since we might
> need the bus clock later. Add bus clock case for Core cpu's. Rewrite
> p[34]_update_cpuspeed to make of already detected bus clocks. (dim@)
~ db_trace.c
> Support "trace /p <pid>" to show the stack trace of any process.
> (Keep in mind that $radix is 0x10 by default.); ok by many (uwe@)
~ est.c
> Let est not complain about certain strange msr values on Pentium 4's. We
> can't
> use the feature on them anyway, since we don't have enough info from Intel.
> (dim@)
~ acpi_machdep.c
> Only establish acpi interrupt if ACPI_ENABLE. (kettenis@)
~ vm_machdep.c
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
~ autoconf.c
> Don't print masks if we're in apic mode. (kettenis@)
~ machdep.c ~ mainbus.c
~ mpbios.c ~ mtrr.c
~ pctr.c ~ procfs_machdep.c
> Remove all the extern cpu_{id,model,whatever} declarations from most
> stuff in arch/i386/i386. This should prevent more screwups like the
> one I did before in ichpcib.c...
> ok dlg@ kettenis@ (dim@)
~ autoconf.c ~ db_interface.c
~ pmapae.c
> Nuke all commons but one, and enable --warn-common in LINKFLAGS. (miod@)
~ kvm86call.S
> Remove the macros SET_CURPCB, GET_CURPCB and GET_CPUINFO, they are only
> used
> in a few places, and the MP version of GET_CPUINFO is broken. No regression
> in GENERIC, however GENERIC.MP still has issues. (gwk@)
arch/i386/include
~ cpu.h
> Save misc info (ebx) register from cpuid with eax=1, so we can use it
> later for some identifications.
> "slap it in" deraadt (dim@)
~ frame.h
> commit vesabios for matthieu@
> ok deraadt, "slap it in" miod@ (gwk@)
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
~ cpu.h ~ param.h
> Remove all the extern cpu_{id,model,whatever} declarations from most
> stuff in arch/i386/i386. This should prevent more screwups like the
> one I did before in ichpcib.c...
> ok dlg@ kettenis@ (dim@)
~ db_machdep.h
> Nuke all commons but one, and enable --warn-common in LINKFLAGS. (miod@)
~ param.h
> Oops, didn't mean to increase the default msgbufsize for i386. Noticed by
> brad, thanks. (dim@)
arch/i386/pci
~ ichpcib.c
> Only use the ichpcib speedstep feature if we're running on a (Mobile)
> Pentium 4,
> since Celerons don't support it.
> prodded by gwk@ (dim@)
~ ichpcib.c
> acpi may steal the timer and speedstep functionality, in which case we
> cannot io map the device. rest of the attach() functionality is still
> safely intact. we simply do not have to warn about mapping failing.
> ok kettenis (deraadt@)
~ ichpcib.c
> Fix cpu_model string screwup. (dim@)
~ ichpcib.c
> Remove all the extern cpu_{id,model,whatever} declarations from most
> stuff in arch/i386/i386. This should prevent more screwups like the
> one I did before in ichpcib.c...
> ok dlg@ kettenis@ (dim@)
~ pcibios.c ~ pcibiosvar.h
> Nuke all commons but one, and enable --warn-common in LINKFLAGS. (miod@)
arch/luna88k/dev
~ mb89352.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ lunafb.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ omrasops.c
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
arch/luna88k/luna88k
~ autoconf.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
arch/m68k/conf
~ files.m68k
> Remove MAPPEDCOPY. It does not work and I am not convinced it is worth
> fixing. (miod@)
arch/m68k/fpsp
~ res_func.sa
> s/completly/completely/
> ok brad (mk@)
arch/m68k/include
~ cpu.h
> Remove MAPPEDCOPY. It does not work and I am not convinced it is worth
> fixing. (miod@)
arch/m68k/m68k
- mappedcopy.c ~ copy.s
> Remove MAPPEDCOPY. It does not work and I am not convinced it is worth
> fixing. (miod@)
arch/m88k/include
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/mac68k/dev
~ ncr5380.c ~ sbc_obio.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ macfb.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
arch/mac68k/include
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/macppc/conf
~ GENERIC
> Activate pwdog(4) after successful testing on i386 and macppc.
> ok deraadt (mbalmer@)
~ GENERIC
> fwscsi doesnt exist, so dont have a commented entry for it left around to
> confuse people. (dlg@)
arch/macppc/dev
~ mesh.c
> Cleanup some KNF nits, if we can't grab a SCB dont return at splbio.
> ok sometime ago dlg@ (gwk@)
~ mesh.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
arch/macppc/macppc
~ cpu.c
> For the G5 processor setperf function, don't force perflevel to 50 or 100
> perserve the illusion of a much finer granularity of performance control.
> ok kettenis@ (gwk@)
arch/macppc/pci
~ vgafb.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
arch/mips64/include
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/mips64/mips64
~ vm_machdep.c
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/mvme68k/dev
~ sshdma.c ~ vsdma.c
~ wdsc.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
arch/mvme68k/include
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/mvme68k/mvme68k
~ autoconf.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
arch/mvme88k/dev
~ vs.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ vs.c
> Unbreak. (miod@)
arch/mvme88k/mvme88k
~ autoconf.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
arch/powerpc/include
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/powerpc/powerpc
~ vm_machdep.c
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/sh/include
~ ptrace.h
> Add PT_STEP. (kettenis@)
~ proc.h
> Implement PT_STEP. (kettenis@)
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/sh/sh
~ locore_c.c ~ process_machdep.c
> Implement PT_STEP. (kettenis@)
arch/sparc/conf
~ files.sparc
> zx needs rasops8 code for emulation modes, not rasops32. (miod@)
~ files.sparc
> Clean frame buffer attachment code:
> - There is no need to check for buses config(8) will not let us attach to
> - Better P4 bus logic, which does not need to abuse device flags
> - Do not bother trying to print a meaningful device description when it is
> not connected to sbus. (miod@)
~ files.sparc
> Change cgeight to run the console in the monochrome overlay plane (as
> cgtwelve
> does), and only use the 24 bit color planes when running X.
> Not tested, but can't be worse than before - the code it replaces was
> busted
> anyway. (miod@)
arch/sparc/dev
~ si.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ fb.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ fd.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
~ p9000.c ~ p9100.c
~ zx.c
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
~ p9000.c ~ p9100.c
~ zx.c
> Use ri_devcmap to get corrected color values, instead of directly using
> WSCOL_xxxx values. (miod@)
~ cgsix.c
> Use the unpack_attr emulops instead of doing an inline rasops_unpack_attr.
> (miod@)
~ fb.c
> Fix ri_devcmap[] to allow WSCOL_BLACK and WSCOL_WHITE to not have to be
> different for sparc{,64} systems. (miod@)
~ if_ie.c
> Revert r1.35 (ether_input to ether_input_mbuf conversion), this causes
> unaligned accesses on some packets. (miod@)
~ cgtwo.c
> Enable all color planes on attach and wait for retrace to program the
> colormap, this finally makes color work. (miod@)
~ xd.c ~ xy.c
> Initialize more fields of the proto disklabel before invoking
> readdisklabel(),
> to pass its recent sanity checks. (miod@)
~ fb.c
> Do the ri_devcmap reprogramming dance in fbwscons_init(), so that it
> applies
> to non-console frame buffers as well. (miod@)
~ bwtwo.c ~ cgeight.c
~ cgfour.c ~ cgsix.c
~ cgthree.c ~ cgtwo.c
> Clean frame buffer attachment code:
> - There is no need to check for buses config(8) will not let us attach to
> - Better P4 bus logic, which does not need to abuse device flags
> - Do not bother trying to print a meaningful device description when it is
> not connected to sbus. (miod@)
~ zx.c
> Instead of invoking rasops_init with a wrong depth value to have some
> values computed better, prefer invoking it with the real thing and
> recomputing what needs to be afterwards. No functional change, but this
> allows fb.c to assume a few things about its callers. (miod@)
~ fb.c
> In fb_setsize(), remove dead code in the SUN4 case. (miod@)
~ cgeight.c
> Change cgeight to run the console in the monochrome overlay plane (as
> cgtwelve
> does), and only use the 24 bit color planes when running X.
> Not tested, but can't be worse than before - the code it replaces was
> busted
> anyway. (miod@)
arch/sparc/include
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
~ fbvar.h
> Clean frame buffer attachment code:
> - There is no need to check for buses config(8) will not let us attach to
> - Better P4 bus logic, which does not need to abuse device flags
> - Do not bother trying to print a meaningful device description when it is
> not connected to sbus. (miod@)
arch/sparc/sparc
~ autoconf.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
arch/sparc64/conf
~ RAMDISK
> enable mpi (dlg@)
~ files.sparc64
> On frame buffers which endianness differs from the host CPU, we can not
> afford
> using ovbcopy() in the erasecols emulop, as the ovbcopy implementation
> might
> do larger-than-byte loads and store, which will not have the expected
> result
> if the source and destination addresses are not similarly aligned.
> Instead, roll our own byte-only ovbcopy() in this case. This is made
> dependent
> on a config(8) attribute to avoid bloating platforms which do not need
> this,
> thus frame buffers which may set RI_BSWAP in ri_flg need to depend on this
> attribute.
> Problem spotted by matthieu@ using tcsh on sparc64 console. (miod@)
arch/sparc64/dev
~ fb.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ fd.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
~ creator.c
> Use the unpack_attr emulops instead of doing an inline rasops_unpack_attr.
> (miod@)
~ fb.c
> Fix ri_devcmap[] to allow WSCOL_BLACK and WSCOL_WHITE to not have to be
> different for sparc{,64} systems. (miod@)
~ fb.c
> Do the ri_devcmap reprogramming dance in fbwscons_init(), so that it
> applies
> to non-console frame buffers as well. (miod@)
arch/sparc64/include
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/sparc64/sparc64
~ autoconf.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
~ db_trace.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
arch/vax/dec
~ wskbdmap_lk201.c
> Portuguese LK-201 layout. (miod@)
arch/vax/include
~ cpu.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/vax/vax
~ autoconf.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
~ vm_machdep.c
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
arch/vax/vsa
~ ncr.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ gpx.c ~ lcg.c
~ lcspx.c ~ smg.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ gpx.c ~ smg.c
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
arch/zaurus/conf
~ GENERIC
> usb client support can be re-enabled (deraadt@)
~ RAMDISK
> enable usbf/cdcef, can be used during install (somehow) (deraadt@)
arch/zaurus/zaurus
~ zaurus_lcd.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
conf
~ files
> On frame buffers which endianness differs from the host CPU, we can not
> afford
> using ovbcopy() in the erasecols emulop, as the ovbcopy implementation
> might
> do larger-than-byte loads and store, which will not have the expected
> result
> if the source and destination addresses are not similarly aligned.
> Instead, roll our own byte-only ovbcopy() in this case. This is made
> dependent
> on a config(8) attribute to avoid bloating platforms which do not need
> this,
> thus frame buffers which may set RI_BSWAP in ri_flg need to depend on this
> attribute.
> Problem spotted by matthieu@ using tcsh on sparc64 console. (miod@)
dev
~ audio.c
> Compile the mixer tuning via keyboard code if NWSKBD_HOTKEY if nonzero
> rather than NWSKBD being nonzero. (miod@)
~ ccd.c
> The historical reasons to spoof the 'c' partition of ccdN as being FFS
> are no longer operative. Spoof the label created for ccdN with an
> UNUSED 'c' partition like every other device's spoofed label.
> Problem pointed out by millert@ in relation to pending disklabel(8)
> changes.
> ok mickey@ deraadt@ (krw@)
dev/acpi
~ acpiprt.c
> Handle objrefs in _PRT.
> ok jordan@ (kettenis@)
~ dsdt.c ~ dsdt.h
> Added support for parsing _CRS resources
> Post-parsing fixup for forward name references (jordan@)
~ acpi.c ~ acpiprt.c
> Fixes for acpiprt routing
> Post-parse fixup for namerefs
> ok kettenis@ (jordan@)
~ acpiprt.c
> Add check for acpi pointer. This prevents some ugly panics.
> ok ketennis. (marco@)
~ dsdt.h
> Make AML_CRSLEN return the correct length for small resources.
> ok jordan@ (kettenis@)
~ acpiprt.c
> Make normal pic interrupt routing work too. (kettenis@)
~ acpi.c
> Don't spam dmesg. (kettenis@)
~ acpiprt.c ~ dsdt.c
> Post-process PCI config space opregions
> Fixed acpiprt for stinky presario v3000
> ok marco@ (jordan@)
~ acpi.c
> Fixed clearing arg for acpi_init_pic
> ok kettenis@ (jordan@)
~ acpibtn.c
> Do not consider unrecognized buttons as lid sensors. ok marco@ (miod@)
~ acpiac.c ~ acpibat.c
~ acpidev.h ~ acpiec.c
~ acpitz.c ~ dsdt.c
> After a long debate at the hackathon it was decided that the acpi
> implementation
> is not concurrent if the sensors thread goes away. This makes acpi
> completly
> lockless.
> help and ok dlg (marco@)
~ acpidebug.c
> Unused global variable. (miod@)
dev/atapiscsi
~ atapiscsi.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
dev/cardbus
~ if_re_cardbus.c
> move printing of the interrupt string into re_attach(). (brad@)
~ if_malo_cardbus.c
> Add another ID to match for the 88W8310 chip.
> ok deraadt@ (mglocker@)
dev/eisa
~ aha1742.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
dev/i2o
~ ioprbs.c ~ iopsp.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ iop.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
dev/ic
~ mpi.c
> use scsi_detach_target when a device dissapears, rather than using
> config_detach and cleaning the midlayer up ourselves. (dlg@)
~ rtl81x9reg.h
> correct hardware revision mask. (jsg@)
~ bcw.c ~ bcwvar.h
~ malo.c ~ rt2560.c
~ rt2661.c
> do not have each net80211 driver define its own rates structures. if they
> use
> the standard rates, use some defined by net80211 itself. kernel shrinks a
> bit
> ok jsg mglocker (deraadt@)
~ malo.c
> It's sizeof(struct ieee80211_frame) not sizeof(struct ieee80211_frame *).
> Spotted by damien@ (mglocker@)
~ acx.c ~ acx111.c
~ acx100.c ~ atw.c
~ rtw.c
> Make use of ieee80211_std_rateset (jsg@)
~ acx100.c ~ if_wi.c
~ if_wi_hostap.c
> avoid extra inclusions; ok jsg (deraadt@)
~ malo.c
> Replace the firmware reset command by a register write which also
> causes a device reset. This enables resetting also on amd64,
> and therefore reinitializing after the 'up' 'down' dance works
> now there.
> ok claudio@ (mglocker@)
~ malo.c
> Enable WEP.
> Tested by jsg@ (mglocker@)
~ malo.c
> Instead debug message for WEP key setting, just complain if WEP key
> setting fails. Always return with error in malo_init() if someting
> fails.
> ok claudio@ (mglocker@)
~ mpi.c
> remove dead code (dlg@)
~ mpi.c
> unhandled ioctls return ENOTTY, not 0 (dlg@)
~ re.c ~ revar.h
> move printing of the interrupt string into re_attach(). (brad@)
~ re.c
> identify the chipset. (brad@)
~ aac.c ~ adv.c
~ adw.c ~ aic6360.c
~ aic79xx_openbsd.c ~ aic7xxx_openbsd.c
~ ami.c ~ bha.c
~ cac.c ~ ciss.c
~ gdt_common.c ~ isp_openbsd.c
~ mfi.c ~ mpi.c
~ ncr53c9x.c ~ oosiop.c
~ osiop.c ~ siop.c
~ twe.c ~ uha.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ adv.c
> Unbreak. No cookie for dlg. (grange@)
~ sti.c ~ vga.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ sti.c
> Use uvm_km_alloc() instead of uvm_km_alloc1(), so that the latter can
> change within uvm without breaking code. From art@ (miod@)
~ malo.c
> Enable monitor mode.
> ok claudio@ (mglocker@)
~ sti.c ~ vga.c
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
~ pcdisplay_subr.c ~ pcdisplayvar.h
~ vga.c
> Change the getchar wsdisplay_accessops function to not return a
> display-dependent value, but instead fill a structure with the chaarcter
> and a valid attribute, suitable for use with unpack_attr.
> Adapt the wsmoused code to these changes, and remove all knowledge of
> the text-mode style pc video attributes in it.
> This will eventually allow wsmoused to be used on non-pcdisplay devices.
> (miod@)
~ bcw.c ~ bcwvar.h
> From Jon Simola <simola at mecha.com > New bcw_powercontrol_crystal_on() function, and fix the license for the
> pci wrapper. (mglocker@)
~ malo.c
> Enable short slots, which should increase / stableize throughput
> performance a bit.
> ok claudio@ (mglocker@)
~ malo.c
> Yay, forgot to add IEEE80211_C_SHPREAMBLE and IEEE80211_C_SHSLOT
> capapility.
> Spotted by damien@ (mglocker@)
~ malo.c
> fix wep key index in firmware command.
> ok mglocker@ (damien@)
~ re.c ~ rtl81x9reg.h
> rename 8100E id to spin 1 and add the 8100E second spin chip id
> chip id from Linux r8169 driver
> from brad@ (todd@)
~ vga.c
> In alloc_attr(), make sure xxansitopc[] indicies are within bounds. (miod@)
~ malo.c
> Print the number of unknown commands in malo_cmd_string(). Also don't
> print an empty newline in malo_hexdump(). OK mglocker@ (claudio@)
~ malo.c
> First bits of endian cleanup more still needed but commit early, commit
> often.
> OK mglocker@ (claudio@)
~ rt2560.c ~ rt2661.c
> fix handling of the SIOCS80211CHANNEL ioctl in monitor mode:
> don't call xxx_set_chan() if the interface is not up&running.
> patch from Steffen Schuetz (st dot sch at gmx dot net) with
> minor modifications by me.
> closes kernel/5313 (damien@)
dev/isa
~ aha.c ~ seagate.c
~ wds.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ ega.c ~ pcdisplay.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ fd.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
~ ega.c ~ pcdisplay.c
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
~ ega.c
> In alloc_attr(), make sure xxansitopc[] indicies are within bounds. (miod@)
dev/mii
~ miidevs.h
> regen (brad@)
~ eephy.c
> sync with new miidevs entries. (brad@)
~ miidevs
> rename Marvell Gig PHY E1000_5/6 to E1116/E1118.
> From Pyun YongHyeon <pyunh at gmail dot com> (brad@)
~ eephyreg.h
> some additional registers.
> From Pyun YongHyeon <pyunh at gmail dot com> (brad@)
~ mii_physubr.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
dev/pci
~ if_bge.c ~ if_ipw.c
~ if_iwi.c ~ if_wpi.c
> do not have each net80211 driver define its own rates structures. if they
> use
> the standard rates, use some defined by net80211 itself. kernel shrinks a
> bit
> ok jsg mglocker (deraadt@)
~ if_bnx.c
> commented out entry for the BCM5709. (brad@)
~ pcidevs.h ~ pcidevs_data.h
> regen (dim@)
~ agpvar.h ~ pcidevs
> Add some Intel 946, 963 and 965 entries. (dim@)
~ pcidevs.h ~ pcidevs_data.h
> regen again. :) (dim@)
~ if_bge.c
> revert the Jumbo diff that was commited as part of a wireles drivers
> commit. (brad@)
~ if_wi_pci.c
> avoid extra inclusions; ok jsg (deraadt@)
~ files.pci + ips.c
> IBM ServeRAID controllers driver.
> Way far from complete but enough to fdisk and disklabel logical drives.
> (grange@)
~ ips.c
> Properly enable enterrupts. Now I finally can newfs the whole drive.
> (grange@)
~ vga_pci.c ~ vga_pcivar.h
> commit vesabios for matthieu@
> ok deraadt, "slap it in" miod@ (gwk@)
~ pcidevs
> Add some Agere Ethernet ids. (jsg@)
~ pcidevs.h ~ pcidevs_data.h
> regen (jsg@)
~ ips.c
> Replace DMA operations with something less ugly. Still a lot of work
> remains. (grange@)
~ ips.c
> Pretend to implement SCSI SYNCHRONIZE_CACHE command. (grange@)
~ if_ixgb.c
> style changes and cleaning. no op. (brad@)
~ if_ixgb.c
> - ixgb_dma_alloc(): Uncomment the line setting the DMA tag to NUL upon
> failure.
> - ixgb_dma_free(): Return if the DMA tag is NUL. (brad@)
~ if_ixgb.c ~ if_ixgb.h
> Pre-allocate the TX DMA maps intead of creating and destroying a DMA map
> per packet sent. (brad@)
~ ips.c
> Add some debug facility. (grange@)
~ ips.c
> Keep free ccbs in a queue. (grange@)
~ ips.c
> The default debug level is ``error'' for now. (grange@)
~ amdpm.c
> acpi may steal power management functionality, in which case we
> cannot io map the device. Bail out without warning. If we want
> to retain the rgn we could re-enable io and be very very careful.
> ok deraadt@ (kettenis@)
~ ips.c
> Try to detect stalled xfers with timeouts. (grange@)
~ ips.c
> Better debug messages. (grange@)
~ if_re_pci.c
> move printing of the interrupt string into re_attach(). (brad@)
~ pcidevs
> add IDs for SiS550/1/2 audio and SiS530/1/2 vga (robert@)
~ pcidevs.h ~ pcidevs_data.h
> regen (robert@)
~ ips.c
> I just don't know why I decided to set block count to 512 unconditionaly
> on every scatter-gather transfer. (grange@)
~ arc.c ~ iha_pci.c
~ trm_pci.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ ips.c
> oops, i missed ips when giving scsibus an attach args for drivers to fill
> in.
> dedicated to grange, the most talented man at h2k6 (dlg@)
~ ips.c
> Oh, scsi layer already has timeout stuff for us. (grange@)
~ iha_pci.c
> oops, i missed a closing brace with the scsibus_attach_args fixup.
> found by ian@ (dlg@)
~ if_malo_pci.c
> Add another ID to match for the 88W8310 chip.
> ok deraadt@ (mglocker@)
~ tga.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ ips.c
> Add some constants. (grange@)
~ ips.c
> ips firmware runs on a little-endian cpu so do endian conversion
> in i/o access macros. (grange@)
~ ips.c
> (grange@)
~ ips.c
> Limit the number of openings to 1 for now, there're still bugs.
> Thanks to miod for the hint. (grange@)
~ tga.c
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
~ if_bcw_pci.c
> From Jon Simola <simola at mecha.com > New bcw_powercontrol_crystal_on() function, and fix the license for the
> pci wrapper. (mglocker@)
~ if_vic.c
> re-add a tiny little #ifdef VIC_DEBUG, used to compare the vic(4)
> features and capabilities on different vmware platforms. (reyk@)
~ if_vic.c
> always assume full duplex state if the interface is up... what does
> full duplex mean for a virtual interface? (reyk@)
~ if_sis.c
> Set sis_stopped=1 early on attach and reset it later in sis_init().
> With mickey's help. (grange@)
~ if_sis.c
> Re-enable RX engine every time otherwise it occasionally
> stops under unknown circumstances.
> Tested by many people, ok brad a long time ago. (grange@)
dev/pcmcia
~ cfxga.c ~ cfxgareg.h
> Much better register description and values. No functional change yet - but
> this code does not look like a bingo game chart anymore. (miod@)
~ cfxga.c ~ cfxgareg.h
> Start reworking the raster ops code, and be smarter when monitoring the
> FIFO
> for cpuu-to-display operation. This is much faster already. (miod@)
~ if_wi_pcmcia.c
> avoid extra inclusions; ok jsg (deraadt@)
~ cfxga.c
> Rewrite the blitter functions, providing an accelerated putchar in the
> process; and fix synchronization points, so that the card no longer gets
> stuck and in need of a reset after being flooded with rop requests. (miod@)
~ cfxga.c
> Always update cfxga_scr emulops field when creating a new screen, in case
> the card was removed and then reinserted.
> XXX It's still ugly, a better way to handle this is in the works. (miod@)
~ cfxga.c
> Store the wsscreen_descr structure in the softc. (miod@)
~ cfxga.c
> Do not perform frame buffer actions on non-active screens. (miod@)
~ cfxga.c
> Separate solid fill operations from regular raster ops; this also brings
> us accelerated copyrows and copycols. (miod@)
~ cfxga.c
> Do not use a complete bitmap as a backing store for multiple screens, but
> remember character cells instead; this is noticeably faster. (miod@)
~ cfxga.c ~ cfxgareg.h
> Support for different video modes (by playing with wsconfcfg).
> 8 bit modes are disabled until I find why the LUT writes are ignored,
> and Zaurus-like 640x480x16 is still the default. (miod@)
~ cfxga.c ~ cfxgareg.h
~ files.pcmcia
> Reorder cfxga initialization sequence for clarity, and minor tweaks; no
> functional changes. (miod@)
~ cfxga.c
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ cfxga.c
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
~ cfxga.c
> Change the getchar wsdisplay_accessops function to not return a
> display-dependent value, but instead fill a structure with the chaarcter
> and a valid attribute, suitable for use with unpack_attr.
> Adapt the wsmoused code to these changes, and remove all knowledge of
> the text-mode style pc video attributes in it.
> This will eventually allow wsmoused to be used on non-pcdisplay devices.
> (miod@)
dev/rasops
~ rasops.c ~ rasops.h
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
~ files.rasops ~ rasops.c
~ rasops8.c
> On frame buffers which endianness differs from the host CPU, we can not
> afford
> using ovbcopy() in the erasecols emulop, as the ovbcopy implementation
> might
> do larger-than-byte loads and store, which will not have the expected
> result
> if the source and destination addresses are not similarly aligned.
> Instead, roll our own byte-only ovbcopy() in this case. This is made
> dependent
> on a config(8) attribute to avoid bloating platforms which do not need
> this,
> thus frame buffers which may set RI_BSWAP in ri_flg need to depend on this
> attribute.
> Problem spotted by matthieu@ using tcsh on sparc64 console. (miod@)
~ rasops.c
> Oops, prevent a warning if rasops_bswap is not required. (miod@)
dev/sbus
~ zx.c
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
~ files.sbus
> zx needs rasops8 code for emulation modes, not rasops32. (miod@)
~ zx.c
> Use ri_devcmap to get corrected color values, instead of directly using
> WSCOL_xxxx values. (miod@)
~ cgsix.c
> Use the unpack_attr emulops instead of doing an inline rasops_unpack_attr.
> (miod@)
~ zx.c
> Instead of invoking rasops_init with a wrong depth value to have some
> values computed better, prefer invoking it with the real thing and
> recomputing what needs to be afterwards. No functional change, but this
> allows fb.c to assume a few things about its callers. (miod@)
dev/sdmmc
~ sdmmc_scsi.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ sdmmc.c ~ sdmmcvar.h
> add ioctl hack to support the sdio(8) userland tool (uwe@)
~ sdhc.c
> Show debug messages depending on current debug level, which may be
> changed with ddb or sdio(8). (uwe@)
~ sdhc.c
> oops, don't always enable debugging (uwe@)
~ sdhc.c
> make interrupt handler stay quiet in case of shared interrupts (uwe@)
~ sdmmc.c ~ sdmmcvar.h
~ sdmmc_mem.c
> move CSD/CID decoding to sdmmc_mem.c (uwe@)
dev/usb
+ usbf.h
> missing file; sorry niall! (uwe@)
~ if_ral.c ~ if_rum.c
~ if_uath.c ~ if_zyd.c
> do not have each net80211 driver define its own rates structures. if they
> use
> the standard rates, use some defined by net80211 itself. kernel shrinks a
> bit
> ok jsg mglocker (deraadt@)
~ uhidev.c
> cover up a bug (resulting in noise in dmesg) that we still do not
> understand, happens on the sun x4100 virtual console; ok dlg (deraadt@)
~ if_atu.c
> Make use of ieee80211_std_rateset (jsg@)
~ usbdevs.h ~ usbdevs_data.h
> sync (deraadt@)
~ usbdevs
> learn about RIM products (deraadt@)
~ if_wi_usb.c
> avoid extra inclusions; ok jsg (deraadt@)
~ files.usb + uberry.c
> smallest usb driver ever: enable config on the Blackberry which prevents
> usb from idling, thus it can get enough power to charge in all cases
> (deraadt@)
~ if_zyd.c ~ if_zydreg.h
> Initial untested bits for ZD1211B support, different init
> sequences for radios not yet done.
> "commit what you have" damien@ (jsg@)
~ files.usb
> add needs-flag to allow pxaudc to compile without usbf (to get the
> previous behaviour of just disabling the USB device controller) (uwe@)
~ usbdevs_data.h ~ usbdevs.h
> sync (drahn@)
~ usbdevs
> yet another cdce device. (drahn@)
~ if_cdce.c
> another motorola device. (drahn@)
~ umass_scsi.c ~ usscanner.c
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
~ usbdevs
> Linksys WUSBF54G (jsg@)
~ usbdevs.h ~ usbdevs_data.h
> regen (jsg@)
~ if_zyd.c
> match Linksys WUSBF54G (jsg@)
~ if_zyd.c ~ if_zydreg.h
> Remaining bits required for ZD1211B support.
> Seems to work fine with Linksys and 3Com adapters. (jsg@)
~ uberry.c
> Detach correctly; ok deraadt@ (miod@)
~ umass_scsi.c
> unitialized variable, busted by dlg, fixed by dlg (deraadt@)
~ usbdevs.h ~ usbdevs_data.h
> regen (jsg@)
~ if_zyd.c
> more ZD1211B devices (jsg@)
~ usbdevs
> Add a bunch more ZD1211B devices mostly from http://zd1211.ath.cx/ (jsg@)
~ if_zyd.c
> fix setting of MAC register ZYD_MAC_RETRY for ZD1211B.
> maximum retry count must be set for each EDCA access category. (damien@)
~ if_zyd.c ~ if_zydreg.h
> fix PHY initialization for ZD1211B.
> we were iterating from 0 to N(zyd_def_phy), but N(zyd_def_phy) is not
> equal to N(zyd_def_phyB). we were probably writting garbage to the PHY.
> (damien@)
~ if_zyd.c ~ if_zydreg.h
> cosmetic (damien@)
~ if_zyd.c
> Add a "break" which makes AL7230B initialize now. Still doesn't work
> afterwards, but ...
> Spotted by claudio@, ok deraadt@ jsg@ (mglocker@)
~ if_uath.c ~ if_uathvar.h
> The device doesn't always detach gracefully from the bus after a firmware
> upload. We thus force a port reset and a re-exploration on the parent hub
> after firmware upload.
> In uath_newstate(), if we're called from a process context, we don't need
> to schedule a USB task. This fixes a problem in uath_stop(): when moving
> to the S_INIT state, the task was scheduled after the device was reset.
> Don't abort the RX firmware command pipe in uath_stop(), otherwise we will
> never receive commands ACKs from the firmware anymore. We now survive to
> an ifconfig down up. (damien@)
~ if_ral.c ~ if_rum.c
> don't schedule a USB task in {ural,rum}_newstate() if we're called from
> a process context. this avoids potential races in {ural,rum}_stop().
> (damien@)
~ if_ral.c ~ if_rum.c
> fix handling of the SIOCS80211CHANNEL ioctl in monitor mode:
> don't call xxx_set_chan() if the interface is not up&running.
> patch from Steffen Schuetz (st dot sch at gmx dot net) with
> minor modifications by me.
> closes kernel/5313 (damien@)
~ if_uath.c
> don't include <net80211/ieee80211_amrr.h>
> rate control is handled by firmware. (damien@)
dev/vesa
+ files.vesa + vesabios.c
+ vesabiosreg.h + vesabiosvar.h
+ vesafb.c + vesafbvar.h
> commit vesabios for matthieu@
> ok deraadt, "slap it in" miod@ (gwk@)
~ vesabios.c
> typo (matthieu@)
~ vesafb.c
> Make a debugging printf conditionnal to VESABIOSVERBOSE.
> ok miod@, gwk@, deraadt@ (matthieu@)
dev/wscons
~ wsconsio.h
> commit vesabios for matthieu@
> ok deraadt, "slap it in" miod@ (gwk@)
~ wsdisplay.c ~ wsdisplayvar.h
> Add a new member to struct wsemuldisplaydev_attach_args, for a frame buffer
> driver to be able to tell how many wscons screens to attach to it,
> instead of WSDISPLAY_DEFAULTSCREENS which is a global setting. (miod@)
~ wsdisplayvar.h
> Add an unpack_attr function to struct wsdisplay_emulops, to match the
> existing alloc_attr function. This allows rasops_unpack_attr to be kept
> private to rasops, yet available to the screen drivers. (miod@)
~ wsdisplay.c ~ wsdisplayvar.h
~ wsmoused.h
> Change the getchar wsdisplay_accessops function to not return a
> display-dependent value, but instead fill a structure with the chaarcter
> and a valid attribute, suitable for use with unpack_attr.
> Adapt the wsmoused code to these changes, and remove all knowledge of
> the text-mode style pc video attributes in it.
> This will eventually allow wsmoused to be used on non-pcdisplay devices.
> (miod@)
~ wsdisplay.c
> No need to include pcdisplay.h here anymore. (miod@)
~ wsdisplayvar.h ~ wsemul_sun.c
> Fix ri_devcmap[] to allow WSCOL_BLACK and WSCOL_WHITE to not have to be
> different for sparc{,64} systems. (miod@)
~ wsdisplay.c ~ wsmoused.h
> Use a better strategy to make the mouse pointer visible in inverse_char();
> also check the return value of alloc_attr() for failure.
> No functional change. (miod@)
kern
~ kern_malloc.c
> Make malloc() print out a warning message when returning NULL due to
> M_CANFAIL, idea from miod@, okay deraadt@ (pedro@)
~ tty_nmea.c
> Let nmea(4) use tty timestamps if they are enabled. Using a GPS receiver
> that provides a PPS signal, more precision and less jitter can be expected.
> With my test setting (a Garmin GPS 18 LVC hooked up to a i386 machine) I
> observed a skew of 0.03 to 0.05 seconds between the tty timestamp and the
> soft
> timestamp.
> "put it in" deraadt (mbalmer@)
~ init_main.c ~ kern_fork.c
~ kern_kthread.c ~ kern_physio.c
~ kern_synch.c ~ kern_sysctl.c
~ sched_bsd.c ~ sys_process.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
~ uipc_mbuf.c
> We don't use mb_map anymore since a long time already. Remove it. (miod@)
~ tty_nmea.c
> Update the sensor description based on the GPS operation mode. Initial
> diff received from matthieu.
> ok matthieu (mbalmer@)
~ kern_rwlock.c
> s/completly/completely/
> ok brad (mk@)
~ kern_event.c
> spell kqueue, not kqeue
> ok jmc (camield@)
lib/libz
~ zutil.h
> make it compile w/ debug defined; miod@ ok (mickey@)
miscfs/procfs
~ procfs_cmdline.c ~ procfs_ctl.c
~ procfs_status.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
miscfs/specfs
~ spec_vnops.c
> Correct the calculation of block address for specfs block i/o on
> devices with sectorsizes other than 512. e.g. cd's. Fixes PR #5235
> from Paul Stoeber with a slightly tweaked diff. NetBSD did the same
> with their r1.59 in 2001, closing their PR#3261 and PR#14026.
> tweak suggestions and ok pedro@ (krw@)
msdosfs
~ msdosfs_vfsops.c
> Fix malloc() panic in msdosfs_mountfs(), okay thib@ deraadt@
> Found using fuzz generator written by lmh at info-pull.comedro@)
~ msdosfs_vfsops.c
> detect potential multiplicative overflow before allocation, and return via
> the error path; ok pedro (deraadt@)
net
~ if.h
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
~ if_bridge.c
> no need for #if NPF > 0 inside #if NPF > 0
> From: Genadijus Paleckis <lsd at nnt.lt > but the really bad description of the diff made this way more complicated
> then needed. pls plz, when sending in diffs, describe properly what they
> do and why! (henning@)
~ if_bridge.c ~ if_bridge.h
~ bridgestp.c ~ if.c
> Add Rapid Spanning Tree Protocol support (802.1d-2004) based on work
> by Andrew Thompson (thompsa at freebsd.orgThe local changes include
> adoption to our bridge code, reduced stack usage and many other bits.
> If stp is enabled, RSTP will now be used by default.
> Thanks for help from Andrew.
> This code has been in snaps for while now, commit encouraged by deraadt@
> (reyk@)
net80211
~ ieee80211.c ~ ieee80211_node.h
> do not have each net80211 driver define its own rates structures. if they
> use
> the standard rates, use some defined by net80211 itself. kernel shrinks a
> bit
> ok jsg mglocker (deraadt@)
~ ieee80211_amrr.c ~ ieee80211_regdomain.c
~ ieee80211_rssadapt.c
> avoid extra inclusions; ok jsg (deraadt@)
netbt
~ bluetooth.h
> spelling: queue, not qeueue (camield@)
netinet
~ udp_usrreq.c
> The current implementation to handle control options is broken as soon as
> SO_REUSEADDR and multicast/broadcast comes into play. Model the
> ip_savecontrol
> handling after the rip_input version which works.
> Problem found by Simon Kelley. OK henning@ michele@ (claudio@)
~ ip_input.c
> make use of multiple routing tables.
> hook up looking up routes in alternate tables to the packet forwarding
> path.
> alternate routing tables are mintained with route(8), table selection via
> pf.
> mostly hacked on a train ride with ryan some time ago, ok mcbride claudio
> (henning@)
~ ip_output.c
> rangecheck ttl on IP_TTL, collected dust in my tree (henning@)
netinet6
~ ip6_forward.c ~ ip6_input.c
> make use of multiple routing tables.
> hook up looking up routes in alternate tables to the packet forwarding
> path.
> alternate routing tables are mintained with route(8), table selection via
> pf.
> mostly hacked on a train ride with ryan some time ago, ok mcbride claudio
> (henning@)
nfs
~ nfs_bio.c ~ nfs_syscalls.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
scsi
~ scsiconf.c ~ scsiconf.h
> provide scsi_detach_bus, _target, and _lun to wrap up config_detach for
> scsi devices. the midlayer keeps some state for each device that is
> attached which needs to be cleaned up on detach, hence this wrapper. (dlg@)
~ scsiconf.h
> no argument names in prototypes pls (dlg@)
~ scsiconf.c
> if there are no luns on a target, then say there were no devices,
> otherwise return any error we find during detach of the luns. (dlg@)
~ scsiconf.c
> hook scsibus up to bio.
> ok deraadt@ krw@ an earlier diff was ok marco@ too (dlg@)
~ ch.c ~ scsi_all.h
~ scsi_base.c ~ st.c
> rework asc/ascq handling to get rid of "magic" 0x values in the code,
> changes
> asc/ascq handling to deal with the pair together rather than seperately,
> since
> modern scsi has long since polluted the "categories" of asc enough that you
> pretty much need to always look as ascq.
> ok krw@ deraadt@, dlg@ (beck@)
~ scsiconf.c
> add bio code to do hotplug of devices on the scsibus.
> thumbs up deraadt@ (dlg@)
~ cd.c ~ scsi_all.h
~ sd.c
> get rid of the rest of the asc/ascq magic codes in scsi
> ok marco@, deraadt@ (beck@)
~ cd.c ~ ch.c
~ scsi_all.h ~ scsi_base.c
~ sd.c ~ st.c
> more magic removal
> ok miod@, deraadt@ (beck@)
~ scsiconf.h
> remove a comment that was describing a behaviour that was changed over nine
> years ago. (dlg@)
~ cd.c ~ ch.c
~ safte.c ~ scsiconf.c
~ scsiconf.h ~ sd.c
~ ses.c ~ ss.c
~ ss_mustek.c ~ ss_scanjet.c
~ ssvar.h ~ st.c
~ uk.c
> rename scsibus_attach_args to scsi_attach_args. this can help avoid
> confusing when trying to attach scsibus to a hba, since it is really meant
> for attaching scsi devices to scsibus.
> ok deraadt@ marco@ (dlg@)
~ scsiconf.c ~ scsiconf.h
> give scsi controllers a real attach args to fill in when attaching scsibus.
> ok miod@ marco@ deraadt@ (dlg@)
sys
~ scsiio.h
> add bio code to do hotplug of devices on the scsibus.
> thumbs up deraadt@ (dlg@)
~ proc.h
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
~ sensors.h
> Reflect reality in a comment. (mbalmer@)
~ sockio.h
> Add Rapid Spanning Tree Protocol support (802.1d-2004) based on work
> by Andrew Thompson (thompsa at freebsd.orgThe local changes include
> adoption to our bridge code, reduced stack usage and many other bits.
> If stp is enabled, RSTP will now be used by default.
> Thanks for help from Andrew.
> This code has been in snaps for while now, commit encouraged by deraadt@
> (reyk@)
uvm
~ uvm_extern.h ~ uvm_km.c
> Add an alignment parameter to uvm_km_alloc1(), and change all callers to
> pass zero; this will be used shortly. From art@ (miod@)
~ uvm_extern.h ~ uvm_glue.c
~ uvm_meter.c
> Kernel stack can be swapped. This means that stuff that's on the stack
> should never be referenced outside the context of the process to which
> this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
> doesn't follow the rules here. Instead of trying to track down all
> offenders and fix this hairy situation, it makes much more sense
> to not swap kernel stacks.
> From art@, tested by many some time ago. (miod@)
~ uvm_extern.h
> Remove cpu_swapin() and cpu_swapout(), they are no longer necessary (except
> for cpu_swapin() on hppa* which is kept). (miod@)
~ uvm_extern.h ~ uvm_km.c
> We don't use mb_map anymore since a long time already. Remove it. (miod@)
== usr.bin =========================================================== 11/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/usr.bin
usr.bin
~ Makefile
> enter pkg-config (deraadt@)
calendar
~ calendars/calendar.birthday
> add more dudes; from Chad Dougherty <crd at andrew.cmu.edumickey@)
compress
~ main.c
> Fix two bugs: stdin is processed and the current working dir is not
> readable and having a file named '-'. PR 5301 and 5302. Diff from Vadim
> Vygonets, with some changes from me. ok millert@ (otto@)
cvs
~ Makefile ~ cmd.c
~ cvs.h
> client-side of the annotate command is usable, build it too. (xsa@)
~ client.c ~ remote.h
~ server.c
> hook up annotate server bits. (xsa@)
~ annotate.c
> don't process the file if its status is FILE_UNLINK. (xsa@)
~ client.c
> sort sent requests; (xsa@)
~ cvs.1
> we will never support pserver, so do not bother documenting
> login/logout commands, nor mark them as supported. (xsa@)
~ cvs.c
> snprintf() -> cvs_path_cat() (xsa@)
~ repository.c ~ util.c
> snprintf() -> cvs_path_cat() (xsa@)
dc
~ Makefile
> repair missing DPADD requests (deraadt@)
fmt
~ fmt.1 ~ fmt.c
> - sync synopsis and usage()
> - sort options
> - avoid line split (jmc@)
~ fmt.1
> various improvements; (jmc@)
~ fmt.1
> - clean up the !}fmt section
> - better Xr (jmc@)
~ fmt.c
> trim usage(); ok millert deraadt (jmc@)
gzsig
~ Makefile
> repair missing DPADD requests (deraadt@)
nc
~ nc.1
> tweaks; from claus assmann (pr #5317); (jmc@)
pkg-config
+ Makefile + pkg-config
> Yet Another pkg-config program, this time as a perl script.
> Not all flags and options are implemented yet, but it works well enough
> to build a bunch of ports and xenocara (which requires pkg-config).
> It is not yet activated in the build; I'm committing to allow for in-tree
> development (including a man page).
> ok espie, matthieu (ckuethe@)
~ pkg-config
> Don't search in non-openbsd-default paths - that's what PKG_CONFIG_PATH
> is for. (ckuethe@)
~ Makefile
> forgot about obj dirs. thanks matthieu (ckuethe@)
~ pkg-config
> When a package is not found, exit with an error code. Allows things like
> neon and subversion to build. (ckuethe@)
~ Makefile
> better man page based on another script Makefile (deraadt@)
~ pkg-config
> stylistic clean-ups.
> look more like BSD style, open files correctly, remove lots of
> extra parenthesis, do not depend on $_ (espie@)
~ pkg-config
> $OpenBSD$ tag (ckuethe@)
~ Makefile + pkg-config.1
> Initial version of a manpage. More options will be described as they are
> implemented. (ckuethe@)
~ pkg-config.1
> Move a sentence and talk a bit about dependency resolution. (ckuethe@)
~ pkg-config.1
> Nicer descriptions of --atleast*version and and --exists (ckuethe@)
~ pkg-config
> Sync usage with manpage (ckuethe@)
~ pkg-config ~ pkg-config.1
> Allow "pkg-config --modversion" to work. Reported by naddy@ (ckuethe@)
~ pkg-config ~ pkg-config.1
> Add the --silence-errors option. Found by naddy@ (ckuethe@)
~ pkg-config
> Remove a debug print i accidentally left behind (ckuethe@)
~ pkg-config.1
> tweaks; (jmc@)
~ pkg-config.1
> - sort options (discussed with ckuethe)
> - there is no benefit in using Fl when options take `--'; use Cm instead
> (jmc@)
~ pkg-config.1
> typo; (jmc@)
~ pkg-config.1
> - tweak text on which options are default
> - merge --help and --usage
> - .Ex
> ok ckuethe (jmc@)
~ pkg-config
> make things vaguely declared where they belong (espie@)
~ pkg-config
> rely on PkgConfig.pm, and so really parse the *.pc files.
> handle --define-variable
> handle multiple dependencies in the correct order, so that linking will
> work with complex requirements.
> okay ckuethe@, matthieu@ (espie@)
~ pkg-config
> typo in error message (espie@)
systat
~ if.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
~ pigs.c
> Do not test for processes being swapped out since this can't happen
> anymore. (miod@)
tip
~ tip.c
> Back out revision 1.31, restoring the ":dc: is always set" behaviour.
> Fixes hangs seen with uplcom(4) and serial console.
> testing pedro jolan
> ok deraadt (mcbride@)
top
~ machine.c
> Do not test for processes being swapped out since this can't happen
> anymore. (miod@)
== usr.sbin ========================================================== 12/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/usr.sbin
bgpctl
~ Makefile
> repair missing DPADD requests (deraadt@)
~ bgpctl.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
bgpd
~ bgpd.c ~ bgpd.conf.5
~ bgpd.h ~ kroute.c
~ parse.y ~ printconf.c
> allow bgpd to work on alternate routing tables, claudio ok, jmc manpage
> help (henning@)
~ kroute.c
> mib we pass to sysctl when fetching the routing table has 7 entries now,
> not 6 any more (rtableid added). need to tell sysctl so. (henning@)
~ session.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
dhcrelay
~ dhcpd.h
> remove incorrect comment (stevesk@)
dvmrpctl
~ dvmrpctl.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
dvmrpd
~ Makefile
> repair missing DPADD requests (deraadt@)
~ kroute.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
~ dvmrpd.h ~ interface.c
~ rde.c ~ rde.h
~ rde_mfc.c ~ rde_srt.c
~ report.c
> introduce rde_check_route that selects which routes must be accepted and
> which discarded.
> Still not complete, just a step forward.
> ok norby@ henning@ (michele@)
ftp-proxy
~ Makefile
> repair missing DPADD requests (deraadt@)
~ filter.c
> use flags S/SA for consistency with the rest of pf (camield@)
~ filter.c
> forgot to update the comment as well (camield@)
gpioctl
~ Makefile
> for macppc too; from ghz at darkdata.orgeraadt@)
ifstated
~ Makefile
> repair missing DPADD requests (deraadt@)
~ ifstated.c
> handle the new link states as UP (reyk@)
ntpd
~ ntp.c
> Allow sensors in a sensors-only configuration to set the time at startup.
> (ckuethe@)
ospfctl
~ ospfctl.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
ospfd
~ Makefile
> repair missing DPADD requests (deraadt@)
~ kroute.c
> mib we pass to sysctl when fetching the routing table has 7 entries now,
> not 6 any more (rtableid added). need to tell sysctl so. (henning@)
~ interface.c ~ kroute.c
~ ospfe.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
pkg_add
~ Makefile
> +OpenBSD::PkgConfig (espie@)
+ OpenBSD/PkgConfig.pm
> handle pkg-config files. (espie@)
~ OpenBSD/PkgConfig.pm
> let compress filter out stuff, add method add_bases, make sure write_fh
> writes to fh. (espie@)
~ OpenBSD/PkgConfig.pm
> add compress_list, allows for empty lines consisting of spaces... (espie@)
~ OpenBSD/PkgConfig.pm
> adjust (C), noticed by matthieu.
> Note that this lives here because pkg_create is going to use it to
> normalize *.pc files at some point... (espie@)
~ OpenBSD/PkgConfig.pm
> some *.pc files have variable= value, with a space. (espie@)
~ OpenBSD/PkgConfig.pm
> apply some normalization to some properties while reading or writing,
> for instance, parse Requirements correctly, and simplify Libs before
> writing.
> We normalize Libs before writing because it's more efficient. (espie@)
~ OpenBSD/PkgConfig.pm
> Fixes parsing of variables/delimiters, so that we classify like pkgconfig.
> Augment the Requires parser to deal correctly with comparisons. (espie@)
~ OpenBSD/PkgConfig.pm
> handles continuation lines and comments. (espie@)
~ OpenBSD/PackingElement.pm
> allows us to mark *.pc files as PkgConfig.
> This will be useful for some `normalization' work (remove ports directory,
> and replace /usr/local with DEPBASE or some such, so that building without
> installation works). (espie@)
ripctl
~ ripctl.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
~ ripctl.8
> remove "show summary" and "detail" keywords that aren't supported
> ok henning@ (michele@)
ripd
~ Makefile
> repair missing DPADD requests (deraadt@)
~ auth.c
> use correct function name in error message; ok henning@ (stevesk@)
~ kroute.c
> mib we pass to sysctl when fetching the routing table has 7 entries now,
> not 6 any more (rtableid added). need to tell sysctl so. (henning@)
~ interface.c ~ kroute.c
~ ripe.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
sasyncd
~ carp.c
> add additional link states to report the half duplex / full duplex
> state, if known by the driver. this is required to check the full
> duplex state without depending on the ifmedia ioctl which can't be
> called in the kernel without process context.
> ok henning@, brad@ (reyk@)
sdio
+ sdio.8 + sdio.c
+ Makefile
> my small tool for debugging SD/MMC cards from the command line,
> not hooked up to the build; ok deraadt (uwe@)
~ sdio.c
> avoid code duplication, include <dev/sdmmc/sdmmcvar.h> (uwe@)
===============================================================================
More information about the owc
mailing list