[owc] Weekly src changes ending 2007-03-18
OWC
auto at squish.net
Mon Mar 19 07:03:30 GMT 2007
OpenBSD src changes summary for 2007-03-11 to 2007-03-18 inclusive
==================================================================
distrib/miniroot distrib/notes
distrib/sets etc/rc
etc/rc.conf gnu/usr.bin/gcc
gnu/usr.sbin/sendmail include/blf.h
include/ctype.h include/signal.h
lib/csu lib/libpthread
libexec/spamd regress/sbin
regress/usr.bin sbin/disklabel
sbin/fsck_ffs sbin/ipsecctl
sbin/isakmpd share/man
share/misc sys/arch/alpha/alpha
sys/arch/alpha/conf sys/arch/alpha/include
sys/arch/alpha/mcbus sys/arch/alpha/pci
sys/arch/amd64/amd64 sys/arch/amd64/conf
sys/arch/amd64/include sys/arch/arm/arm
sys/arch/arm/include sys/arch/arm/xscale
sys/arch/hp300/dev sys/arch/hp300/hp300
sys/arch/hp300/include sys/arch/hppa/hppa
sys/arch/hppa/include sys/arch/hppa64/hppa64
sys/arch/hppa64/include sys/arch/i386/conf
sys/arch/i386/i386 sys/arch/i386/include
sys/arch/landisk/include sys/arch/landisk/landisk
sys/arch/m88k/include sys/arch/m88k/m88k
sys/arch/mac68k/dev sys/arch/mac68k/include
sys/arch/mac68k/mac68k sys/arch/macppc/dev
sys/arch/macppc/macppc sys/arch/mips64/include
sys/arch/mips64/mips64 sys/arch/mvme68k/include
sys/arch/mvme68k/mvme68k sys/arch/powerpc/include
sys/arch/powerpc/powerpc sys/arch/sh/include
sys/arch/sh/sh sys/arch/solbourne/solbourne
sys/arch/sparc/dev sys/arch/sparc/include
sys/arch/sparc/sparc sys/arch/sparc64/include
sys/arch/sparc64/sparc64 sys/arch/vax/include
sys/arch/vax/mscp sys/arch/vax/vax
sys/arch/zaurus/conf sys/arch/zaurus/dev
sys/compat/hpux sys/compat/osf1
sys/compat/svr4 sys/conf
sys/ddb sys/dev
sys/dev/adb sys/dev/ata
sys/dev/ic sys/dev/pci
sys/dev/sbus sys/dev/sdmmc
sys/dev/usb sys/kern
sys/lib/libsa sys/miscfs/procfs
sys/msdosfs sys/net
sys/netinet sys/netinet6
sys/nfs sys/sys
sys/ufs/ext2fs sys/ufs/ffs
usr.bin/awk usr.bin/calendar
usr.bin/cdio usr.bin/ctags
usr.bin/cut usr.bin/diff
usr.bin/elf2olf usr.bin/less
usr.bin/locate usr.bin/logger
usr.bin/make usr.bin/mesg
usr.bin/msgs usr.bin/pctr
usr.bin/pkg-config usr.bin/rcs
usr.bin/renice usr.bin/rev
usr.bin/skey usr.bin/sort
usr.bin/telnet usr.bin/top
usr.bin/tput usr.bin/vgrind
usr.bin/vis usr.sbin/bgpctl
usr.sbin/bgpd usr.sbin/dvmrpd
usr.sbin/gpioctl usr.sbin/hoststated
usr.sbin/kvm_mkdb usr.sbin/lpr
usr.sbin/ospfd usr.sbin/pkg_add
usr.sbin/ripd usr.sbin/syslogd
usr.sbin/tcpdump usr.sbin/tokenadm
usr.sbin/tokeninit usr.sbin/vnconfig
== distrib =========================================================== 01/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/distrib
miniroot
~ install.sub
> Keep copyright year current. (krw@)
~ install.sub
> Give the upgrader an option to *not* fsck -f non-root filesystems. In
> fact make that the default. Help upgraders with big filesystems and no
> patience.
> Suggested by & ok beck at .rw@)
notes
~ Makefile + aviion/contents
+ aviion/hardware + aviion/install
+ aviion/prep + aviion/upgrade
+ aviion/whatis + aviion/xfer
> OpenBSD/aviion notes draft. (miod@)
~ alpha/hardware
> make a note about AlphaServer 1200 and 4100 support
> reminder from martin@ (robert@)
~ alpha/hardware
> sync for UP2000 (martin@)
sets
~ lists/base/md.aviion ~ lists/comp/md.aviion
> Sync (miod@)
== etc =============================================================== 02/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/etc
rc
~ rc
> Allow nmeaattach to start a sensor before starting ntpd. Many modern
> receivers can cold start in the time it takes the boot process to get
> close to starting ntpd. Even if the gps is not ready or the fix is not
> valid, at least the sensor has been created; ntpd won't have to wait
> a few minutes before scanning for the sensor. This makes using GPS as
> the sole source of time a bit easier.
> ok deraadt (ckuethe@)
rc.conf
~ rc.conf
> Allow nmeaattach to start a sensor before starting ntpd. Many modern
> receivers can cold start in the time it takes the boot process to get
> close to starting ntpd. Even if the gps is not ready or the fix is not
> valid, at least the sensor has been created; ntpd won't have to wait
> a few minutes before scanning for the sensor. This makes using GPS as
> the sole source of time a bit easier.
> ok deraadt (ckuethe@)
== gnu =============================================================== 03/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/gnu
usr.bin/gcc
~ gcc/config/alpha/alpha.c ~ gcc/config/alpha/openbsd.h
> Disable stack frame checking in the prologue under OpenBSD, unless you
> compile
> with -fstack-check, for consistency with other platforms; ok kettenis@
> (miod@)
usr.sbin/sendmail
~ sendmail/sendmail.8 ~ sendmail/mailq.8
~ sendmail/newaliases.8
> .Nm does not need an argument; from Igor Sobrado (jmc@)
== include =========================================================== 04/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/include
blf.h
~ blf.h
> do the adjustment of arbitrary 128 -> BLF_MAXUTILIZED in userland as well,
> also discussed on tech@ a while ago.
> ok pedro@ tedu@ thib@ (grunk@)
ctype.h
~ ctype.h
> allow extern inline stuff to proceed unchanged thru the mess that the
> combined GCC and ISO committees managed to make out of it.
> With this, gcc >= 4.3 can grok its way through it correctly.
> okay kettenis@, with some help figuring stuff out. (espie@)
signal.h
~ signal.h
> allow extern inline stuff to proceed unchanged thru the mess that the
> combined GCC and ISO committees managed to make out of it.
> With this, gcc >= 4.3 can grok its way through it correctly.
> okay kettenis@, with some help figuring stuff out. (espie@)
== lib =============================================================== 05/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/lib
csu
~ sh/crt0.c
> Set the DN bit in the fpscr upon startup, to prevent faulting when loading
> denormals. This is recommended by the manual, and is necessary for the gdb
> testsuite to pass; found by, discussed with, and ok kettenis@
> (be sure to make includes before recompiling csu) (miod@)
libpthread
~ include/pthread.h
> give _POSIX_THREAD_* defines a value of 1 so applications that
> conditionally compile code using these defines will work.
> okay tedu@, marc@, milllert@ (kurt@)
== libexec =========================================================== 06/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/libexec
spamd
~ grey.c
> log when we trap a host. (beck@)
~ grey.c
> Don't delete grey entries when whitelisting. They expire anyway
> and this hides the evidence. Avoids a potential race when
> using -M traps where a retry comes in between deletion and
> whitelisting.
> ok millert@ (beck@)
~ grey.c
> when using low prio MX trapping, ignore hosts connecting to the low prio
> mx for the first minute, to avoid problems if hosts legitimately try it
> after we are down.
> suggested by millert@ (beck@)
~ spamd.8
> fix -y example;
> closes user/5408 from sthen
> ok reyk (jmc@)
~ grey.c
> now that we leave the grey, avoid annoying re-adds of white entries
> by checking for them being in the db first.
> ok millert@ (beck@)
~ sync.c
> Fix dribbly HMAC leak.
> ok reyk@ (beck@)
~ grey.c
> Be sure to exit if the db_open of the second db for conversion fails.
> (beck@)
== regress =========================================================== 07/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/regress
sbin
~ ipsecctl/sa1.in ~ ipsecctl/sa1.ok
~ ipsecctl/sa13.in ~ ipsecctl/sa13.ok
~ ipsecctl/sa15.in ~ ipsecctl/sa15.ok
~ ipsecctl/sa19.in ~ ipsecctl/sa19.ok
~ ipsecctl/sa3.in ~ ipsecctl/sa3.ok
~ ipsecctl/sa7.in ~ ipsecctl/sa7.ok
> We switched to aes cbc quite some time ago, so also use the correct
> key sizes here, too. We now have to use 128 bit key instead of
> 160.
> Noticed by david@ (hshoexer@)
~ ipsecctl/Makefile + ipsecctl/ike57.in
+ ipsecctl/ike57.ok
> move autodetection of the ID type to the parser. this way the
> static flows have the correct ID, too. ok hshoexer, reyk (markus@)
usr.bin
~ rcs/Makefile
> When commiting multiple files with ci, the flags from the previous
> files affected the later ones.
> ok niallo@ xsa@ (bluhm@)
== sbin ============================================================== 08/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/sbin
disklabel
~ editor.c
> Refactor & fix computation of offset for next partition. Now the
> code is almost readable. Keep in mind that the starting offset is
> inclusive, but the ending offset is exclusive. I heard rumors that
> disklabel was misbehaving in some cases before, but espie@ was the
> only one to provide me with enough details to actually find the
> problem.
> ok millert@ tom@ (otto@)
~ editor.c
> Check for zero divisor _before_ dividing; do not use floating
> point arithmetic to round to a cylinder boundary when a simple
> integer expression can do the job as well. ok millert@ miod@ (otto@)
fsck_ffs
~ pass5.c
> Check for cg_irotor and cg_frotor being negative, okay millert@ deraadt@
> (pedro@)
~ setup.c
> there's no way we're going to be able to handle illegal block or
> fragment sizes, so just make those conditions fatal errors and tell
> the calling code to go look for alternate superblocks. ok millert@ (otto@)
ipsecctl
~ ike.c ~ ipsecctl.h
~ parse.y ~ pfkey.c
> move autodetection of the ID type to the parser. this way the
> static flows have the correct ID, too. ok hshoexer, reyk (markus@)
isakmpd
~ isakmpd.conf.5
> Fix usage of predefined lifetimes. "Default-phase-[12]-lifetime"
> just specifies the values to be used. However, the specifications
> are called "LIFE_MAIN_MODE" and "LIFE_QUICK_MODE".
> ok ho@ jmc@ (hshoexer@)
== share ============================================================= 09/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/share
man
~ man5/printcap.5
> document mc (maximum number of copies allowed); ok millert@ jmc@ (stevesk@)
~ man4/adt.4
> add ADT7475 to adt(4) and re-format.
> ok jmc@ (jasper@)
~ man4/com.4
> add ST16C654
> ok jmc@ (jasper@)
~ man5/printcap.5
> mx default has been 0 for almost 10 years; ok millert@ jmc@ (stevesk@)
~ man4/msk.4
> add D-Link DGE-550T B1
> ok kettenis@ (jasper@)
~ man4/lm.4
> add W83627EHF
> ok jmc@ (jasper@)
~ man4/sequencer.4
> s/orientated/oriented/, last instance of this in our tree (cloder@)
~ man4/nfe.4
> nfe can also be found paired with rgephy(4), ie in Gigabyte M61PM-S2
> from a submitted dmesg. (jsg@)
~ man4/ddb.4
> Disable x/q functionnality on 32bit platforms, because the examination code
> is not supposed to work on values larger than db_expr_t. (miod@)
~ man4/bnx.4
> no jumbosity for bnx
> from brad; ok reyk (jmc@)
~ man4/re.4
> Update the re(4) man page to mention VLAN tag insertion support.
> from brad; ok reyk (jmc@)
~ man4/Makefile ~ man4/usb.4
+ man4/uts.4
> add a manual page for uts(4) (robert@)
~ man4/man4.alpha/Makefile + man4/man4.alpha/mcbus.4
+ man4/man4.alpha/mcpcia.4
> add mcbus(4), mcpcia(4) manual pages and also create a link
> from mcbus(4) to mcmem(4) (robert@)
~ man4/uts.4
> tweaks; (jmc@)
~ man4/man4.alpha/mcbus.4
> tweaks; (jmc@)
~ man4/man4.alpha/mcpcia.4
> tweaks; (jmc@)
~ man4/man4.i386/glxsb.4
> glxsb(4) has had AES enabled for a while, but I forgot to update the
> man page. Noticed by Sebastian Rother < sebastian (dot) rother (at)
> jpberlin (dot) de >, thanks.
> ok jmc@ (tom@)
~ man4/pci.4
> Add glxsb(4) under "Cryptography accelerators".
> Requested by jmc@ (tom@)
~ man4/ddb.4
> document show buf|vnode|mount (mickey@)
~ man4/hifn.4 ~ man4/ubsec.4
~ man4/man4.i386/glxsb.4
> Add pci(4) to SEE ALSO, as other PCI crypto accelerators already do.
> ok jmc@ (tom@)
~ man4/umsm.4
> move Sierra Wireless AirCard 875 to known to work as I've been using
> this since January; ok fkr@ (stevesk@)
~ man4/carp.4
> Manpage bits for IP load balancing.
> Configuration is subject to change.
> With help from jmc@ (mpf@)
~ man4/man4.i386/glxsb.4
> tweaks; ok tom (jmc@)
misc
~ usb_hid_usages
> spelling error fixes; pr#5412 (deraadt@)
== sys =============================================================== 10/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/sys
arch/alpha/alpha
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ cpuconf.c + dec_kn300.c
> add DEC_KN300 support from NetBSD;
> This code makes it possible to run on some of the AlphaServers,
> namely AlphaServer 4100 and 1200.
> add mcbus(4) and mcpcia(4) to provide support for the system bus and
> the MCPCIA-to-PCI bus adapter that can be found in these systems
> allow the pci_swiz_bus code to handle variable extent names
> to be able to handle more than one mcpcia(4)
> "just commit it" deraadt@ (robert@)
arch/alpha/conf
~ GENERIC ~ files.alpha
> add DEC_KN300 support from NetBSD;
> This code makes it possible to run on some of the AlphaServers,
> namely AlphaServer 4100 and 1200.
> add mcbus(4) and mcpcia(4) to provide support for the system bus and
> the MCPCIA-to-PCI bus adapter that can be found in these systems
> allow the pci_swiz_bus code to handle variable extent names
> to be able to handle more than one mcpcia(4)
> "just commit it" deraadt@ (robert@)
~ RAMDISK ~ RAMDISKBIG
> put support for alphaservers 1200 and 4100 on ramdisk kernels
> ok deraadt@ (martin@)
arch/alpha/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ atomic.h
> add proper atomic_{set,clear}bits_int implementation
> prodded by art@ ok art@ (martin@)
arch/alpha/mcbus
+ mcbus.c + mcbusreg.h
+ mcbusvar.h + mcmem.c
> add DEC_KN300 support from NetBSD;
> This code makes it possible to run on some of the AlphaServers,
> namely AlphaServer 4100 and 1200.
> add mcbus(4) and mcpcia(4) to provide support for the system bus and
> the MCPCIA-to-PCI bus adapter that can be found in these systems
> allow the pci_swiz_bus code to handle variable extent names
> to be able to handle more than one mcpcia(4)
> "just commit it" deraadt@ (robert@)
arch/alpha/pci
~ pci_swiz_bus_io_chipdep.c ~ pci_swiz_bus_mem_chipdep.c
+ pci_kn300.c + pci_kn300.h
+ mcpcia.c + mcpcia_bus_io.c
+ mcpcia_bus_mem.c + mcpcia_dma.c
+ mcpcia_pci.c + mcpciareg.h
+ mcpciavar.h
> add DEC_KN300 support from NetBSD;
> This code makes it possible to run on some of the AlphaServers,
> namely AlphaServer 4100 and 1200.
> add mcbus(4) and mcpcia(4) to provide support for the system bus and
> the MCPCIA-to-PCI bus adapter that can be found in these systems
> allow the pci_swiz_bus code to handle variable extent names
> to be able to handle more than one mcpcia(4)
> "just commit it" deraadt@ (robert@)
arch/amd64/amd64
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/amd64/conf
~ GENERIC
> wpi(4) works fine on amd64 according to
> Stefan Sperling <stsp at stsp.in-berlin.de > so enable it (jsg@)
arch/amd64/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/arm/arm
~ ast.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/arm/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ atomic.h
> For arm pre-v6 (ie all supported machines) it is necessary to disable
> interrupts to be able to have an atomic update of a variable without a
> mutex. (drahn@)
arch/arm/xscale
~ files.pxa2x0 ~ pxa2x0reg.h
+ pxa2x0_mmc.c + pxammcvar.h
> Support the PXA27x SD/SDIO/MMC controller on Zaurus
> We use the suggested workaround for the problem E40 in the PXA27x errata
> sheet. Unfortunately this limits the bus speed to 9.75Mhz. (uwe@)
~ pxa2x0_mmc.c
> typo (deraadt@)
arch/hp300/dev
~ ct.c
> Be more strict during probe, consider failures to get device description
> and device status code fatal, and do not attach in these cases. (miod@)
arch/hp300/hp300
~ vm_machdep.c
> Reset psl to spl0 in cpu_fork(), so that kthreads start at spl0 and not
> splhigh(); from NetBSD (miod@)
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/hp300/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/hppa/hppa
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/hppa/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/hppa64/hppa64
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/hppa64/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/i386/conf
~ GENERIC
> enable uts(4) (robert@)
~ GENERIC
> we do not want uts(4) enabled until we have proper calibration
> support, so revert my previous commit (robert@)
arch/i386/i386
~ pmap.c
> tlb shootdown in pmap_change_attrs was shooting the wrong va.
> "yes man" mickey@ (I think that meant 'ok'). (art@)
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ pmap.c ~ pmapae.c
> do not steal page table pages on page allocation failure -- it can be dealt
> w/ in a more simple way as other archs do; art@ ok (mickey@)
arch/i386/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/landisk/include
- types.h
> Dangling include (miod@)
+ fpu.h
> Add <machine/fpu.h> for upcoming userland changes. (miod@)
arch/landisk/landisk
~ machdep.c
> Move UKC later in the boot process, so that it can use malloc(). (miod@)
arch/m88k/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/m88k/m88k
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/mac68k/dev
~ adb.c
> Drop the second and third arguments to adb_op_sync, for they are not used;
> while there, remove explicit global variable initializations to zero,
> and appease the gods by removing the most ludicrous spurious casts.
> No functional change (except for code shrinkage). (miod@)
~ pm_direct.c
> Indentation repairs, no code change. (miod@)
arch/mac68k/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/mac68k/mac68k
~ vm_machdep.c
> Reset psl to spl0 in cpu_fork(), so that kthreads start at spl0 and not
> splhigh(); from NetBSD (miod@)
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ locore.s ~ machdep.c
> Only invoke printstar() in get_pte() if option DEBUG. (miod@)
arch/macppc/dev
~ adb.c
> Drop the second and third arguments to adb_op_sync, for they are not used;
> while there, remove explicit global variable initializations to zero,
> and appease the gods by removing the most ludicrous spurious casts.
> No functional change (except for code shrinkage). (miod@)
~ smu.c
> Make SMU sensors and fan control work on PowerMac8,1.
> tested by Alastair Poole. (kettenis@)
arch/macppc/macppc
~ machdep.c
> Make lcsplx() call spllower() instead of duplicating its code.
> ok miod@ (kettenis@)
arch/mips64/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/mips64/mips64
~ interrupt.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/mvme68k/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/mvme68k/mvme68k
~ vm_machdep.c
> Reset psl to spl0 in cpu_fork(), so that kthreads start at spl0 and not
> splhigh(); from NetBSD (miod@)
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/powerpc/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ atomic.h
> Implement proper atomic.h for powerpc. With help from art@ and drahn at . > ok drahn@ (kettenis@)
arch/powerpc/powerpc
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/sh/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/sh/sh
~ trap.c
> It is safe to run tlb_exception() at the pre-fault ipl. (miod@)
~ sh_machdep.c
> Move UKC later in the boot process, so that it can use malloc(). (miod@)
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ db_trace.c
> Do not start lines printed by the kernel with '<' so that syslog won't get
> confused. (miod@)
arch/solbourne/solbourne
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/sparc/dev
~ cgfourteen.c ~ cgtwelve.c
~ tcx.c ~ tvtwo.c
~ zx.c
> When about to run X11 in 32 bpp mode, advertize correctly that we are
> really using 24 bpp pixels on 32 bit boundaries. (miod@)
arch/sparc/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/sparc/sparc
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/sparc64/include
~ atomic.h
> Implement proper atomic.h for sparc64.
> ok kettenis@ (art@)
~ ctlreg.h
> Add two more SFSR bits found on UltraSPARC-III cpus. (kettenis@)
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ _types.h
> Timecounters for sparc64. The cleanest and easiest timecounter
> implementation this far. Uses the %tick register (as microtime was
> using before).
> kettenis@ ok (art@)
arch/sparc64/sparc64
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ autoconf.c
> Make sure we find the OpenFirmware node for UltraSPARC-IV cpus. (kettenis@)
~ cpu.c
> UltraSPARC-IV cpus appear a bit different in OpenFirmware than older cpus.
> Make sure we fetch the right properties such that we don't crash later in
> the
> pci bus walking code when we divide by a zero cache line size. (kettenis@)
~ clock.c ~ intr.c
> Timecounters for sparc64. The cleanest and easiest timecounter
> implementation this far. Uses the %tick register (as microtime was
> using before).
> kettenis@ ok (art@)
arch/vax/include
~ cpu.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/vax/mscp
~ mscp_disk.c
> Nuke __HAVE_OLD_DISKLABEL code, this was never defined anyway. (miod@)
arch/vax/vax
~ trap.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
arch/zaurus/conf
~ GENERIC ~ files.zaurus
> Support the PXA27x SD/SDIO/MMC controller on Zaurus
> We use the suggested workaround for the problem E40 in the PXA27x errata
> sheet. Unfortunately this limits the bus speed to 9.75Mhz. (uwe@)
~ RAMDISK
> Enable pxammc and sdmmc in RAMDISK, too. (uwe@)
arch/zaurus/dev
~ zaurus_scoop.c ~ zaurus_scoopreg.h
~ zaurus_scoopvar.h
> Add an interface to scoop(4) to control CF and SD card power (uwe@)
+ scoop_mmc.c
> Support the PXA27x SD/SDIO/MMC controller on Zaurus
> We use the suggested workaround for the problem E40 in the PXA27x errata
> sheet. Unfortunately this limits the bus speed to 9.75Mhz. (uwe@)
compat/hpux
~ hpux_file.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
compat/osf1
~ osf1_prot.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
compat/svr4
~ svr4_misc.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
conf
~ newvers.sh
> unlock tree, move on towards 4.1-current (deraadt@)
~ param.c
> better desiredvnodes not based on maxusers; pedro@ deraadt@ ok (mickey@)
ddb
~ db_access.c ~ db_examine.c
> Disable x/q functionnality on 32bit platforms, because the examination code
> is not supposed to work on values larger than db_expr_t. (miod@)
dev
~ systrace.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
dev/adb
~ adb.h ~ akbd.c
~ ams.c
> Drop the second and third arguments to adb_op_sync, for they are not used;
> while there, remove explicit global variable initializations to zero,
> and appease the gods by removing the most ludicrous spurious casts.
> No functional change (except for code shrinkage). (miod@)
dev/ata
~ atascsi.c
> finish teh ata_identify struct. (dlg@)
~ atascsi.c
> parse the ata_identify to get a more correct value for the capacity of the
> disk. (dlg@)
~ atascsi.c
> ata counts sectors from 1, scsi counts from 0. adjust the value in the
> read capacity completion to compensate. (dlg@)
~ atascsi.h
> we need a resid too (dlg@)
~ atascsi.c ~ atascsi.h
> add timeouts on ata_xfers (dlg@)
dev/ic
~ ar5210.c ~ ar5211.c
~ ar5212.c ~ ar5xxx.h
> do not use bcopy to set the mac address registers. this diff should
> fix the "hal" on big endian systems.
> thanks to the madwifi-old-openhal people (reyk@)
~ ar5210.c ~ ar5210reg.h
~ ar5210var.h ~ ar5211.c
~ ar5211reg.h ~ ar5211var.h
~ ar5212.c ~ ar5212reg.h
~ ar5212var.h ~ ar5xxx.c
~ ar5xxx.h
> add some HAL functions (again) inspired by the changes from the
> madwifi-old-openhal project. this was backed out the last time because
> it was too close to the release and i didn't respect the ABI/API lock.
> (reyk@)
~ bcw.c ~ bcwreg.h
~ bcwvar.h
> Finish chip base initialization. (mglocker@)
~ bcw.c
> In bcw_init() check for real revision numbers now and fix a wrong
> register (typo). (mglocker@)
~ ciss.c
TAGGED OPENBSD_4_0
> MFC: do not freak out if cannot map physdrives into volumes and handle it
> proper everywhere - commit message from mickey, -stable patch from Reyk
> ok deraadt (ckuethe@)
~ acx.c
TAGGED OPENBSD_4_0
> In HostAP mode, ieee80211_input() will enqueue packets in if_snd without
> calling if_start(). So call ifp->if_start() at the end of the RX interrupt
> function. This solves the porblem of getting dhcp leases from the AP if the
> network is quiet. OK mglocker@ (claudio@)
~ malo.c
TAGGED OPENBSD_4_0
> Call ifp->if_start() at the end of the rx interrupt handling just in case
> ieee80211_input() enqueued a packet beforehand. malo(4) does not support
> hostap mode so this should not happen but perhaps mglocker@ gets it working
> someday. OK mglocker@ (claudio@)
~ bcw.c ~ bcwreg.h
TAGGED OPENBSD_4_0
> Set chip operation mode in the init path. Add first bits to enable
> interrupts at some point. (mglocker@)
~ bcwreg.h ~ bcwvar.h
TAGGED OPENBSD_4_0
> First super massive register cleanup run; group/sort registers,
> remove lots of obsolete registers, mark unused registers for deletion,
> mark suspect registers for closer verification, comments ... (mglocker@)
~ acx.c
TAGGED OPENBSD_4_0
> Split the beacon returned by ieee80211_beacon_alloc() in two parts and
> fill everything up to IEEE80211_ELEMID_TIM into the beacon template and
> everything after and including IEEE80211_ELEMID_TIM into the tim template.
> This fixes the beacons on 11g where two TIM fields where announced.
> acx_beacon_locate() is used to calculate the offset to a defined TLV field.
> If other dirvers need it as well we may move it up into net80211.
> OK mglocker@ (claudio@)
~ bcw.c ~ bcwreg.h
TAGGED OPENBSD_4_0
> Better (constant) register naming. (mglocker@)
~ bcw.c ~ bcwreg.h
TAGGED OPENBSD_4_0
> Remove a lot of dead code and nonexisting registers. This stuff was mostly
> copied over from bce(4) and is obsolete or needs to be rewritten.
> (mglocker@)
~ bcw.c
TAGGED OPENBSD_4_0
> Cleanup prototypes. (mglocker@)
~ fxp.c
TAGGED OPENBSD_4_0
> Add missing "goto fail;" in an error path of fxp_attach(). (claudio@)
~ bcw.c
TAGGED OPENBSD_4_0
> Split the whole chip init stuff in bcw_init() into a new function
> called bcw_chip_init(), according to the 80211Init steps in the spec.
> (mglocker@)
~ mpi.c ~ mpivar.h
TAGGED OPENBSD_4_0
> replace the VMWARE quirk that restricts the bus width to 16 targets with
> one for all SPI controllers. krw has a sun machine with a 1030 that gets
> the bus width wrong too, so since vmware emulates that type of hardware
> too, we can just limit the lot of them and forget about it. (dlg@)
~ bcw.c ~ bcwreg.h
~ bcwvar.h
TAGGED OPENBSD_4_0
> o Add bcw_80211_core_init() according to
> http://bcm-specs.sipsolutions.net/80211Init.
> o Add a rewritten version of bcw_core_reset() called bcw_80211_core_reset()
> according to http://bcm-specs.sipsolutions.net/80211CoreReset. This
> will probably replace bcw_core_reset(), but I need more verification
> first.
> o Add bcw_core_enable() and bcw_core_disable(). There seems to be
> devices which contain several 802.11 cores. The unused needs to be
> disabled. (mglocker@)
~ bcw.c
TAGGED OPENBSD_4_0
> Comment routine which does the chip access validation. (mglocker@)
dev/pci
~ pcidevs
> JMB368 has no SATA. (jsg@)
~ pcidevs.h ~ pcidevs_data.h
> regen (jsg@)
~ pcidevs
> Add a few different things found in submitted dmesgs. (jsg@)
~ pcidevs
> commit the correct version this time (jsg@)
~ pcidevs.h ~ pcidevs_data.h
> regen (jsg@)
~ pcidevs
> add Intel PCI-E quad port fiber device ID (reyk@)
~ pcidevs.h ~ pcidevs_data.h
> sync (reyk@)
~ if_em.c ~ if_em_hw.c
~ if_em_hw.h
> A new PCI id for Intel's PCIe quad port fiber adapter. (reyk@)
~ pcidevs
> Add some more NVIDIA graphics cards.
> From Eric Gillingham <sysrq at sysrq.tkjsg@)
~ pcidevs.h ~ pcidevs_data.h
> regen (jsg@)
~ mpi_pci.c
> replace the VMWARE quirk that restricts the bus width to 16 targets with
> one for all SPI controllers. krw has a sun machine with a 1030 that gets
> the bus width wrong too, so since vmware emulates that type of hardware
> too, we can just limit the lot of them and forget about it. (dlg@)
~ pcidevs
> add device IDs of the new VIA CX700 chipset
> reported by brad (reyk@)
~ pcidevs.h ~ pcidevs_data.h
> sync (reyk@)
~ ohci_pci.c ~ uhci_pci.c
> Perform legacy emulation handover for USB1 companion controllers
> immediately
> but defer the remainder of their initialisation until after the other
> devices
> on their PCI bus have attached. This ensures that any USB2 controller has
> also
> completed its initialisation before we start to initialise the USB1 parts.
> This minimises the chance that a nasty SMM implementation will trash the
> USB1
> controller's config when it performs legacy emulation handover of the USB2
> part. (pascoe@)
~ pcidevs
> add OX16PCI954K puc id (mickey@)
~ pucdata.c ~ pcidevs.h
~ pcidevs_data.h
> OX16PCI954K is a puc at cardbusissed in the previous puc at cardbusmmits)
> (mickey@)
~ ohci_pci.c ~ uhci_pci.c
> Backout pascoe@'s last USB change because it page faults at attachment
> time.
> Chris; Get your shiz fixed and tested for the next time. We have
> better todo then wasting our time by backing out untested stuff.
> OK deraadt, OK ckuethe (mglocker@)
dev/sbus
~ cgtwelve.c ~ tvtwo.c
~ zx.c
> When about to run X11 in 32 bpp mode, advertize correctly that we are
> really using 24 bpp pixels on 32 bit boundaries. (miod@)
dev/sdmmc
~ sdmmcvar.h
> Support the PXA27x SD/SDIO/MMC controller on Zaurus
> We use the suggested workaround for the problem E40 in the PXA27x errata
> sheet. Unfortunately this limits the bus speed to 9.75Mhz. (uwe@)
~ sdmmc_mem.c ~ sdmmcvar.h
> DHC automatically sends a CMD12 after multiple-block transfers, but other
> controllers don't. Put in a temporary hack for pxammc on Zaurus. (uwe@)
~ sdmmcreg.h
> Define MMC_STOP_TRANSMISSION command (CMD12) (uwe@)
dev/usb
~ umodem.c
> very ugly indent (deraadt@)
~ umass_quirks.c
> add quirk for Sony DSC-T10, which is UFI, not SCSI.
> help and ok krw@ (steven@)
~ usbdevs
> Add ITM Touchscreen from Future Technology Devices (robert@)
~ usbdevs.h ~ usbdevs_data.h
> regen (robert@)
+ uts.c
> uts(4) is a driver for USB Touchscreens;
> currently the driver only support touchscreens made by eGalax and
> Future Technology Devices (ITM)
> ok deraadt@ (robert@)
~ files.usb
> add uts(4); this was missing from the initial commit (robert@)
~ uhci.c
> Save SOF (frame timing adjustment) before the very first host controller
> reset, rather than at suspend/resume, otherwise any BIOS inserted value
> is lost immediately.
> ok dlg@ (pascoe@)
~ uhub.c ~ usb.c
> Wait the stable power delay on a USB hub only once per hub (not per port),
> and defer the delay for root hubs until the host controller event thread
> starts, permitting some concurrency. Speeds up the boot process
> dramatically
> when you have lots of host controllers.
> ok dlg@ (pascoe@)
~ ohci.c
> Perform legacy emulation handover for USB1 companion controllers
> immediately
> but defer the remainder of their initialisation until after the other
> devices
> on their PCI bus have attached. This ensures that any USB2 controller has
> also
> completed its initialisation before we start to initialise the USB1 parts.
> This minimises the chance that a nasty SMM implementation will trash the
> USB1
> controller's config when it performs legacy emulation handover of the USB2
> part. (pascoe@)
~ ohci.c ~ uhci.c
~ uhub.c ~ usb.c
> Backout pascoe@'s last USB change because it page faults at attachment
> time.
> Chris; Get your shiz fixed and tested for the next time. We have
> better todo then wasting our time by backing out untested stuff.
> OK deraadt, OK ckuethe (mglocker@)
kern
~ vfs_subr.c
> better desiredvnodes not based on maxusers; pedro@ deraadt@ ok (mickey@)
~ exec_elf.c
> Don't bother checking for OLF binaries if option SMALL_KERNEL. (miod@)
~ exec_elf.c
> Clean up after miod. (kettenis@)
~ init_main.c ~ kern_clock.c
~ kern_descrip.c ~ kern_exec.c
~ kern_exit.c ~ kern_fork.c
~ kern_kthread.c ~ kern_lock.c
~ kern_proc.c ~ kern_prot.c
~ kern_sig.c ~ kern_synch.c
~ subr_prof.c ~ sys_generic.c
~ sys_process.c ~ tty.c
~ tty_tty.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ uipc_mbuf.c
> m_prepend() works only for sizes smaller than MHLEN.
> OK beck@ deraadt@ pyr@ (claudio@)
~ uipc_mbuf2.c
> Since we're not always called with M_WAIT, check return value from
> MGETHDR and fail if it's NULL.
> ok miod@ (art@)
~ uipc_mbuf2.c
TAGGED OPENBSD_3_9
> MFC (art@), requested by deraadt
> Since we're not always called with M_WAIT, check return value from
> MGETHDR and fail if it's NULL.
> ok miod@ (henning@)
~ uipc_mbuf2.c
TAGGED OPENBSD_4_0
> MFC (art@), requested by deraadt
> Since we're not always called with M_WAIT, check return value from
> MGETHDR and fail if it's NULL.
> ok miod@ (henning@)
~ kern_synch.c
TAGGED OPENBSD_4_0
> Don't restart thrsleep after a signal. After a signal happened and we
> weren't on the sleep queues, the condition we were sleeping on might
> have changed, so we need to go back to userland and recheck that condition.
> This fixes the majority of lockups and and hanging threads in rthreads
> since it fixes a race in the semaphore code.
> ok tedu@ (art@)
lib/libsa
~ ufs.c
> Re-initialise f_seekp to 0 on a successful read_inode() return.
> Fixes a problem that prevented the booting of a kernel that, to be
> found, needs more than one block per directory iin its path to be read.
> ok miod@, deraadt@ (tsi@)
miscfs/procfs
~ procfs_ctl.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
msdosfs
~ denode.h ~ msdosfs_vnops.c
> get rid of noop vop_reallocblks; pedro@ miod@ ok (mickey@)
net
~ if.c ~ if_ethersubr.c
> Add IP load balancing support for carp(4).
> This provides a similar functionality as ARP balancing,
> but also works for traffic that comes across routers.
> IPv6 is supported as well.
> The configuration scheme will change as soon we have sth better.
> Also add support for changing the MAC address on carp(4)
> interfaces. (code from mcbride)
> Tested by pyr@ and reyk@
> OK mcbride@ (mpf@)
netinet
~ if_ether.c
> remove message that prints
> arplookup: unable to enter address for xxx.xxx.xxx.xxx
> because it is absolutely useless in a modern world with cable modems
> which leak arp packets over IP netblock zones; Anyone who used this as
> a diagnostic for real problems has other tools to use; from discussion
> with kitella, ok claudio (deraadt@)
~ if_ether.c ~ ip_carp.c
~ ip_carp.h ~ ip_icmp.c
~ ip_input.c
> Add IP load balancing support for carp(4).
> This provides a similar functionality as ARP balancing,
> but also works for traffic that comes across routers.
> IPv6 is supported as well.
> The configuration scheme will change as soon we have sth better.
> Also add support for changing the MAC address on carp(4)
> interfaces. (code from mcbride)
> Tested by pyr@ and reyk@
> OK mcbride@ (mpf@)
netinet6
~ icmp6.c ~ ip6_input.c
~ nd6_nbr.c
> Add IP load balancing support for carp(4).
> This provides a similar functionality as ARP balancing,
> but also works for traffic that comes across routers.
> IPv6 is supported as well.
> The configuration scheme will change as soon we have sth better.
> Also add support for changing the MAC address on carp(4)
> interfaces. (code from mcbride)
> Tested by pyr@ and reyk@
> OK mcbride@ (mpf@)
nfs
~ nfs_vnops.c ~ nfsnode.h
> get rid of noop vop_reallocblks; pedro@ miod@ ok (mickey@)
sys
~ resourcevar.h
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
~ errno.h
> unwrap EILSEQ
> okay millert@ (espie@)
ufs/ext2fs
~ ext2fs_alloc.c
> Check if there are still blocks left to be allocated _after_ we have
> read in the bitmap (an operation that might cause us to sleep), and
> before committing to the allocation. Fixes PR 5230, okay krw@, millert at . > (pedro@)
ufs/ffs
~ ffs_softdep.c
> Since p_flag is often manipulated in interrupts and without biglock
> it's a good idea to use atomic.h operations on it. This mechanic
> change updates all bit operations on p_flag to atomic_{set,clear}bits_int.
> Only exception is that P_OWEUPC is set by MI code before calling
> need_proftick and it's automatically cleared by ADDUPC. There's
> no reason for MD handling of that flag since everyone handles it the
> same way.
> kettenis@ ok (art@)
== usr.bin =========================================================== 11/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/usr.bin
awk
~ Makefile
> allow this to build w/ cross-tools thru providing proper hostcflags; found
> by rainor's work (mickey@)
calendar
~ calendars/calendar.history
> typo; Matthew Clarke <clamat at telus.netjasper@)
~ calendars/calendar.birthday
> typo, okay jasper@ (pedro@)
~ calendars/ru_RU.KOI8-R/calendar.primety
> one more mine mutter told me (mickey@)
~ calendars/calendar.history
> first diesel (mickey@)
cdio
~ cdio.1
> /dev/rcd0d -> /dev/rcd0c; from AndrAcs (jmc@)
ctags
~ ctags.1 ~ ctags.c
> simplify synopsis/usage() and sort options;
> from Igor Sobrado (jmc@)
cut
~ cut.1 ~ cut.c
> simplify synopsis/usage() and sort options;
> from Igor Sobrado (jmc@)
diff
~ diffreg.c
> improve -p for C++ code: classes definition often have
> public:/protected:/private: at the start of line.
> This lets the -p scanner just take note of the section and keep
> looking for the actual class definition.
> Also increase function name bufsize so it shows most of these pesky C++
> decls...
> okay otto@ (espie@)
elf2olf
~ elf2olf.1 ~ elf2olf.c
> simplify synopsis/usage() and sort options;
> from Igor Sobrado (jmc@)
less
~ less.1
> sort options; from Igor Sobrado (jmc@)
locate
~ locate/updatedb.sh ~ locate/locate.updatedb.8
> simplify synopsis/usage() and sort options. also fixes documentation/5411
> from han (jmc@)
~ bigram/locate.bigram.c ~ code/locate.code.c
~ locate/mklocatedb.sh
> It's not smart to use a space as a field separator when a space can
> be part of a bigram. So switch to tab and as a bonus detect in
> locate.code if the bigram array is too short. Problem seen by Han
> Boetes and other people with too big mps3 collections; these typically
> contain lots of file names with spaces in them. ok jasper@ (otto@)
logger
~ logger.1 ~ logger.c
> simplify synopsis/usage() and sort options;
> from Igor Sobrado (jmc@)
~ logger.1
> +.Ex -std logger
> plus a little cleanup (jmc@)
make
~ Makefile
> allow this to build w/ cross-tools thru providing proper hostcflags; found
> by rainor's work (mickey@)
mesg
~ mesg.1 ~ mesg.c
> sort options; from Igor Sobrado (jmc@)
msgs
~ msgs.1
> "days" is optional; from Igor Sobrado (jmc@)
pctr
~ pctr.1
> s/oriented/orientated/ (cloder@)
pkg-config
~ pkg-config.1
> - add missing word: from alek; ok sturm
> - sort options: from Igor Sobrado (jmc@)
rcs
~ ci.c
> When commiting multiple files with ci, the flags from the previous
> files affected the later ones.
> ok niallo@ xsa@ (bluhm@)
renice
~ renice.8 ~ renice.c
> make this page more readable; (jmc@)
rev
~ rev.c
> use __progname
> ok otto@ (jasper@)
skey
~ skey.1
> use a full reference for RFC 2289; from Igor Sobrado (jmc@)
sort
~ fsort.c ~ fsort.h
~ msort.c ~ sort.c
> some bug (in sort? or compiler) gets hit too late in the release cycle.
> please try again right after release. (deraadt@)
~ fsort.c ~ fsort.h
~ msort.c ~ sort.c
> Use separate buffers in fsort.c and msort.c. Fixes a problem with
> buffer corruptions for some very large files. Also fixes some
> TRECHEADER vs. RECHEADER mismatches. Back in after unlock. (millert@)
~ Makefile
> Work around gcc optimization bug on sh (landisk). From otto@ (millert@)
telnet
~ telnet.1 ~ main.c
> telnet.1: sort options; from Igor Sobrado
> main.c: whack most of usage(), and have different usage() depending
> whether this is invoked as telnet or tn3270;
> help otto; ok millert (jmc@)
top
~ top.1
> make the description of command line options and interactive
> commands more consistent;
> started by a diff from Mark Lumsden (jmc@)
tput
~ tput.1
> sort options and remove some unneccessary markup; from Igor Sobrado (jmc@)
vgrind
~ vgrind.1
> - simplify synopsis and sort options: from Igor Sobrado
> - some small cleanup whilst here (jmc@)
vis
~ vis.1 ~ vis.c
> `b' before `c'; from Igor Sobrado (jmc@)
== usr.sbin ========================================================== 12/12 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/usr.sbin
bgpctl
~ bgpctl.c
> Missed to remove SAFI_BOTH here. Informed about it by Steven Mestdagh.
> (claudio@)
bgpd
~ rde.c
> When running show commands that are restarted make sure that we either
> have pending imsgs to send or that we reduce the poll timout to 0. This
> fixes hanging and extrem slow performing bgpctl show rib commands.
> Found by Jon Morby additional testing by Stuart Henderson. OK henning@
> (claudio@)
~ bgpd.h ~ rde.c
~ rde_update.c
> Update multiprotocol extension code to RFC 4760. This actually removes
> a few things we actually never realy cared about.
> OK henning@ (claudio@)
~ bgpd.8
> Multiprotocol Extensions for BGP-4 is now RFC 4760 (claudio@)
~ bgpd.h ~ rde.c
~ rde_update.c ~ session.c
> A few more s/2858/4760/ and use RFC <number> everywhere. (claudio@)
dvmrpd
~ Makefile
> er,, no, we don't compile everything with -g (henning@)
gpioctl
~ gpioctl.c
> use strtonum() to avoid out of range values, way simpler than the
> 12-line song and dance that standard functions need; bret.lambert at gmail > (deraadt@)
hoststated
~ hoststated.conf.5
> hoststated.conf is not a program.
> thanks to Sebastian Reitenbach, closes pr 5409 (reyk@)
~ hoststated.conf.5 ~ hoststated.h
~ parse.y ~ relay.c
> allow to specify the IP_TTL and IP_MINTTL options for the relays to
> support the Generalized TTL Security Mechanism (GTSM) according to RFC
> 3682. this is especially useful with inbound connections and a fixed
> distance to the backend servers.
> ok pyr@ (reyk@)
~ relay.c
> fix the natlook mode. (reyk@)
~ relay.c
> close unused relay2pfe privsep sockets correctly (reyk@)
~ hoststated.h
> move some elements in the relay imsg ctl structures (just for the style)
> (reyk@)
~ hoststated.c
> remove a debug message (reyk@)
~ pfe.c
> fix a bug in the initial setup of the pfe2relay communication sockets
> which prevented the pfe to accept statistics updates and natlookups
> from any other process then the first one. in other words, this will
> show you the total relay statistics off _all_ preforked processes
> (hoststatectl show relays) and it will unbreak the natlookup mode with
> more than one running relay process. (reyk@)
kvm_mkdb
~ nlist.c
> do not use section names for locating a string tab; miod@ ok (mickey@)
lpr
~ common_source/common.c
> typos in comments; ok millert@ (stevesk@)
~ common_source/common.c
> remove unused variable; ok millert@ (stevesk@)
ospfd
~ kroute.c
> rtm_tableid is only valid for RTM_ADD, RTM_CHANGE and RTM_DELETE but not
> for stuff like RTM_IFINFO or RTM_IFANNOUNCE. So move check to a save
> location.
> Always check rtm_type before accessing some random value inside the route
> message. (claudio@)
~ kroute.c
> Don't call rtlabel_unref() twice on RTM_DELETE. kroute_remove() does it
> so don't unref before calling kroute_remove() or we end up with wrong
> reference counts. (claudio@)
~ ospfd.c
> Don't forget to set the metric in the REDIST_LABEL case. OK pyr@ (claudio@)
~ parse.y
> Make sure that the metric on redistributed prefixes is always != 0.
> If no metric was specified for it to DEFAULT_REDIST_METRIC. While this was
> checked for the simple case it was not done if only a metric type was
> specified. OK pyr@ (claudio@)
~ ospfd.c
> Exchange a strlcpy() with a strncpy(). A similar change was done in parse.y
> some time ago. The simple password needs to be nul extended and is allowed
> to fill the full buffer the perfect job for stncpy().
> Found and tested by Jon Morby. (claudio@)
pkg_add
~ OpenBSD/PackageInfo.pm
> remove some unnecessary sanity checks. /var/db/pkg is supposed to contain
> only packages, so it's not really necessary to check for file existence
> and rights. Besides, actual use of package stuff will try reading +CONTENTS
> in a guarded way, so it will error out sanely...
> Trims out hundreds of stat() under /var/db/pkg, which are largely
> responsible
> for the slow start time of pkg_* tools. (espie@)
~ pkg_create
> handle more errors during package creation.
> - if archiving discovers issues, then go into high speed mode where stuff
> doesn't get written.
> - in case of errors or fatal signals, make sure to remove a partially
> written archive. (espie@)
ripd
~ kroute.c
> Same problem as in ospfd's kroute.c don't check rtm_tableid without
> verifying
> that rtm_type is one of RTM_ADD, RTM_CHANGE or RTM_DELETE. (claudio@)
syslogd
~ privsep.c
> use warnx and not logerror in monitor, logerror is for the slave;
> ok henning@ while in lock (djm@)
tcpdump
~ ike.h
> Recognise more vendor payloads (taken from IKE Scan) (tom@)
tokenadm
~ tokenadm.8 ~ tokenadm.c
> simplify synopsis/usage() and sort options;
> from Igor Sobrado (jmc@)
tokeninit
~ tokeninit.8 ~ tokeninit.c
> simplify synopsis/usage() and sort options;
> from Igor Sobrado (jmc@)
vnconfig
~ vnconfig.c
> do the adjustment of arbitrary 128 -> BLF_MAXUTILIZED in userland as well,
> also discussed on tech@ a while ago.
> ok pedro@ tedu@ thib@ (grunk@)
===============================================================================
More information about the owc
mailing list