[owc] Weekly src changes ending 2007-06-24
OWC
auto at squish.net
Mon Jun 25 07:00:01 BST 2007
OpenBSD src changes summary for 2007-06-17 to 2007-06-24 inclusive
==================================================================
distrib/alpha distrib/amd64
distrib/armish distrib/cats
distrib/hppa distrib/i386
distrib/landisk distrib/luna88k
distrib/mac68k distrib/macppc
distrib/miniroot distrib/mvme68k
distrib/mvme88k distrib/mvmeppc
distrib/notes distrib/ramdisk
distrib/sets distrib/sgi
distrib/sparc distrib/sparc64
distrib/vax distrib/zaurus
etc/etc.alpha/MAKEDEV etc/etc.alpha/MAKEDEV.md
etc/ospfd.conf lib/libc
lib/libkvm libexec/ftpd
libexec/tftp-proxy regress/sbin
sbin/disklabel sbin/ifconfig
sbin/pfctl sbin/ping6
sbin/scan_ffs share/man
sys/altq sys/arch/alpha/alpha
sys/arch/alpha/conf sys/arch/alpha/include
sys/arch/amd64/amd64 sys/arch/amd64/include
sys/arch/amd64/stand/etc sys/arch/amd64/stand/libsa
sys/arch/arm/arm sys/arch/arm/include
sys/arch/armish/include sys/arch/armish/stand/boot
sys/arch/aviion/aviion sys/arch/aviion/include
sys/arch/cats/include sys/arch/hp300/dev
sys/arch/hp300/hp300 sys/arch/hp300/include
sys/arch/hppa/conf sys/arch/hppa/dev
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/i386/stand/etc
sys/arch/i386/stand/libsa sys/arch/landisk/include
sys/arch/landisk/landisk sys/arch/luna88k/include
sys/arch/luna88k/luna88k sys/arch/m88k/m88k
sys/arch/mac68k/include sys/arch/mac68k/mac68k
sys/arch/macppc/dev sys/arch/macppc/include
sys/arch/macppc/macppc sys/arch/macppc/stand
sys/arch/mips64/conf sys/arch/mips64/include
sys/arch/mips64/mips64 sys/arch/mvme68k/include
sys/arch/mvme68k/mvme68k sys/arch/mvme68k/stand/installboot
sys/arch/mvme68k/stand/libsa sys/arch/mvme68k/stand/prtvid
sys/arch/mvme68k/stand/wrtvid sys/arch/mvme88k/include
sys/arch/mvme88k/mvme88k sys/arch/mvme88k/stand/installboot
sys/arch/mvme88k/stand/libsa sys/arch/mvme88k/stand/wrtvid
sys/arch/mvmeppc/include sys/arch/mvmeppc/mvmeppc
sys/arch/mvmeppc/stand/installboot sys/arch/powerpc/powerpc
sys/arch/sgi/conf sys/arch/sgi/include
sys/arch/sgi/localbus sys/arch/sgi/pci
sys/arch/sgi/sgi sys/arch/sh/include
sys/arch/sh/sh sys/arch/solbourne/include
sys/arch/sparc/dev sys/arch/sparc/include
sys/arch/sparc/sparc sys/arch/sparc64/dev
sys/arch/sparc64/include sys/arch/sparc64/sparc64
sys/arch/vax/include sys/arch/vax/mscp
sys/arch/vax/vax sys/arch/vax/vsa
sys/arch/zaurus/dev sys/arch/zaurus/include
sys/dev sys/dev/acpi
sys/dev/ata sys/dev/dec
sys/dev/i2c sys/dev/ic
sys/dev/isa sys/dev/onewire
sys/dev/pci sys/dev/pcmcia
sys/dev/raidframe sys/dev/sdmmc
sys/dev/tc sys/dev/usb
sys/kern sys/miscfs/fifofs
sys/miscfs/portal sys/miscfs/procfs
sys/miscfs/specfs sys/net
sys/net80211 sys/netbt
sys/netinet sys/nfs
sys/scsi sys/stand
sys/sys sys/ufs/ext2fs
sys/ufs/ffs sys/ufs/ufs
sys/uvm usr.bin/at
usr.bin/cvs usr.bin/ftp
usr.bin/mail usr.bin/netstat
usr.bin/ssh usr.bin/touch
usr.sbin/bgpd usr.sbin/hoststated
usr.sbin/ospfd usr.sbin/pkg_add
usr.sbin/tcpdump usr.sbin/ztsscale
== distrib =========================================================== 01/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/distrib
alpha
~ bsd.rd/Makefile ~ common/Makefile.inc
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
amd64
~ common/Makefile.inc
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
armish
~ ramdisk/Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
cats
~ ramdisk/Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
hppa
~ ramdisk/Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
i386
~ common/Makefile.inc
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
landisk
~ miniroot/Makefile ~ ramdisk/Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
luna88k
~ ramdisk/Makefile ~ ramdisk/Makefile.inc
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
mac68k
~ ramdisk/Makefile.inc
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
macppc
~ ramdisk/Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
miniroot
~ Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
mvme68k
~ ramdisk/Makefile ~ ramdisk/Makefile.inc
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
mvme88k
~ ramdisk/Makefile ~ ramdisk/Makefile.inc
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
mvmeppc
~ ramdisk/Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
notes
~ armish/prep
> Firmware version warning; several re-wordings.
> Improved by deraadt at . (ian@)
~ alpha/hardware
> PCI malo(4) works
> from Janjaap van Velthooven (martin@)
ramdisk
~ Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
sets
~ lists/base/md.sgi ~ lists/comp/md.sgi
> sync (deraadt@)
~ lists/man/mi
> sync (deraadt@)
~ lists/man/mi
> sync (deraadt@)
sgi
~ ramdisk/Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
sparc
~ miniroot/Makefile.inc ~ ramdisk/Makefile.inc
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
sparc64
~ bsd.rd/Makefile ~ common/Makefile.inc
~ ramdisk/Makefile ~ ramdiskB/Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
vax
~ ramdisk/Makefile.inc
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
zaurus
~ ramdisk/Makefile
> Stop using disklabel -r. The disklabel code does not know all the
> semantics
> for pushing disklabels onto the disk (and besides that, it is buggy and
> unreadable crap); thanks for help from krw (deraadt@)
== etc =============================================================== 02/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/etc
etc.alpha/MAKEDEV
~ etc.alpha/MAKEDEV
> Regen (miod@)
etc.alpha/MAKEDEV.md
~ etc.alpha/MAKEDEV.md
> alpha kernel have wsmux but not the actual devices; PR #5469 from janjaap
> (miod@)
ospfd.conf
~ ospfd.conf
> add rtlabel to external-tag example (reyk@)
== lib =============================================================== 03/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/lib
libc
~ stdio/fgetln.3
> The `len' argument is guaranteed to be > 0 upon successful completion.
> ok jmc@ ray@ millert@ (moritz@)
libkvm
~ kvm_proc.c
> Bring back Mickey's UVM anon change. Testing by thib@, beck@ and
> ckuethe@ for a while. Okay beck@, "it is good timing" deraadt at . (pedro@)
== libexec =========================================================== 04/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/libexec
ftpd
~ ftpd.c
> Make sure perror_reply() is only called when errno is set. When
> changing perror_reply() -> reply(), increase error message detail
> for ftp client. Move free() statement to prevent errno clobbering.
> OK moritz (ray@)
~ ftpd.c
> Adjust spacing, combine malloc + strlcpy + strlcat into a single
> asprintf call.
> OK moritz (ray@)
tftp-proxy
~ filter.c
> default rule.flags and rule.flagset to 0 instead of null
> from max laier (jcs@)
== regress =========================================================== 05/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/regress
sbin
~ pfctl/pf80.in ~ pfctl/pf80.loaded
~ pfctl/pf80.ok ~ pfctl/pf80.optimized
~ pfctl/pfail51.in ~ pfctl/pfail51.ok
> Allow "log" for nat rules without "pass".
> OK henning@, ``passt scho'' markus@ (mpf@)
== sbin ============================================================== 06/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/sbin
disklabel
~ Makefile ~ disklabel.c
~ editor.c
> On Sun machines, disklabels coming from the kernel will now have a d_flags
> bit D_VENDOR if they face the Sun cylinder / start-at-zero semantics.
> Other
> disklabels found on these architectures do not have these issues, and do
> not need to be constrained. checked by otto (deraadt@)
~ Makefile
> typo (deraadt@)
~ disklabel.c
> -r will die. Until then, make sure any labels it dredges up are
> converted to v1 format before being used.
> ok deraadt@ otto@ (krw@)
~ disklabel.c
> If disklabel(8) is run against unreadable media (e.g. a blank cd)
> don't say "can't read master boot record". Just display an empty
> disklabel. Really yucky code that needs a lot of tough love, but
> marginally less confusing until then.
> ok deraadt@ (krw@)
~ editor.c
> if a 4.2BSD partition falls partly within the area defined by the
> 'b' command, subtract the overlapping space from the free space.
> fixes weird free space calculation on hppa, where converted lif
> labels start at sector 1, but the 'b' area skips the first cylinder.
> ok millert@ (otto@)
~ disklabel.c
> expand size/offset to 16 characters, and drop the Cyl comments which
> noone cares to see anymore (for those architectures where this is
> important, disklabel(8) continues to enfoce the cyl-locked policy)
> ok otto (deraadt@)
ifconfig
~ ifconfig.8 ~ ifconfig.c
> Allow IPv6 addresses to use the CIDR notation too, no need for separate
> prefixlen specification when using this form.
> man page bits by jmc.
> ok henning@, ``looks sane'' djm at . (pyr@)
pfctl
~ parse.y
> Allow "log" for nat rules without "pass".
> OK henning@, ``passt scho'' markus@ (mpf@)
~ parse.y
> always initialize logif, from max laier, ok ryan (henning@)
ping6
~ ping6.8
> sync description of "-m" option to reality. mcbride@ ok (itojun@)
scan_ffs
~ scan_ffs.c
> daddr_t -> daddr64_t; ok pedro@ thib@ (otto@)
== share ============================================================= 07/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/share
man
~ man5/bsd.port.mk.5
> explicit ties to the package system. (espie@)
~ man7/hier.7
> addd /usr/xobj/; from Martynas Venckus (jmc@)
~ man4/ubt.4
> Update history section (uwe@)
~ man4/Makefile + man4/sbt.4
> Man page for SDIO Bleutooth adapter driver, reminded by Alexey Suslikov
> (uwe@)
~ man4/man4.hppa/Makefile + man4/man4.hppa/ssio.4
> Add ssio(4). (kettenis@)
~ man4/com.4
> Update for com at ssio. (kettenis@)
~ man4/pciide.4
> Add support for VIA CX700/VX700 IDE.
> Tested by jcs@ on a model 2 oqo. (jsg@)
~ man4/man4.i386/npx.4
> "an NPX", not "a NPX"; (jmc@)
~ man4/man4.i386/cpu.4 ~ man4/man4.i386/intro.4
~ man4/man4.amd64/intro.4 ~ man4/man4.hppa/intro.4
> do not try to have cpu.4 list everything - the bus pages already do that;
> instead, make sure cpu.4 pages point to intro.4;
> prompted by a diff from Stefan Sperling (jmc@)
~ man4/man4.hppa/ssio.4
> tweaks; (jmc@)
~ man4/com.4
> space required before punctuation; (jmc@)
~ man4/usb.4
> Reflect recent changes (UPACKED -> __packed).
> ok jmc. (mbalmer@)
~ man4/pci.4
> add ssio(4); (jmc@)
~ man4/lpt.4
> Update for lpt at ssio. (kettenis@)
~ man9/disk.9 ~ man9/disklabel.9
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ man8/man8.alpha/MAKEDEV.8
> Regen (miod@)
~ man4/umsm.4
> add novatel es620 here too
> reminded by jsg@ (jcs@)
~ man4/man4.hppa/gsc.4
> sync the lpt.4 entry; (jmc@)
~ man4/man4.hppa/gsc.4 ~ man4/man4.sgi/intro.4
> sync the com(4) entries; (jmc@)
~ man4/man4.aviion/intro.4
> .Tn -> .Sx: section headers are not tradenames; (jmc@)
~ man4/man4.alpha/intro.4 ~ man4/man4.amd64/intro.4
~ man4/man4.armish/intro.4 ~ man4/man4.cats/intro.4
~ man4/man4.hp300/intro.4 ~ man4/man4.hppa/intro.4
~ man4/man4.i386/intro.4 ~ man4/man4.landisk/intro.4
~ man4/man4.luna88k/intro.4 ~ man4/man4.mac68k/intro.4
~ man4/man4.macppc/intro.4 ~ man4/man4.mvme68k/intro.4
~ man4/man4.mvme88k/intro.4 ~ man4/man4.sgi/intro.4
~ man4/man4.sparc/intro.4 ~ man4/man4.sparc64/intro.4
~ man4/man4.vax/intro.4 ~ man4/man4.zaurus/intro.4
> !.Tn is spawning behind my back!
> .Tn -> .Sx: section headers are not tradenames; (jmc@)
~ man4/man4.sparc/daadio.4
> uppercase "i/o" (jmc@)
~ man4/man4.hppa/ssio.4
> list supported devices, similarly to gsc(4); ok kettenis (jmc@)
~ man4/man4.hppa/gsc.4
> end sentences with full stops, not commas; (jmc@)
~ man4/man4.aviion/intro.4 ~ man4/man4.hp300/intro.4
~ man4/man4.mac68k/intro.4 ~ man4/man4.mvme68k/intro.4
~ man4/man4.mvme88k/intro.4 ~ man4/man4.sparc/intro.4
~ man4/man4.vax/intro.4 ~ man4/man4.zaurus/intro.4
> as for the rest of man4, use the .Nd lines to describe devices;
> whilst here, comment out entries for the following devices which have
> no man pages (miod can tell me if this is a problem):
> hp300: hpibbus
> mac68k: sbc
> mvme68k: flash ipic mc memc pcc pcctwo sram vme vs wdsc zs
> sparc: fd (jmc@)
~ man4/man4.mac68k/intro.4
> sbc goes back in, as it's shared w/ ncrscsi(4);
> thanks martin for pointing this out. (jmc@)
~ man4/man4.hp300/intro.4 ~ man4/man4.hp300/nhpib.4
> ..and hpibbus goes back in, as it's shared w/ nhpib; thanks to
> jsg for spotting that one;
> whilst here, zap a silly comma in the .Nd (jmc@)
~ man4/man4.sparc/intro.4
> fd -> fdc; from miod (jmc@)
~ man4/man4.mvme68k/intro.4
> relist these commented out entries, but w/o the man section number until
> pages get written; after some discussion w/ miod (jmc@)
~ man4/ral.4
> more supported models; from brad and jsg (jmc@)
~ man5/bsd.port.mk.5
> document REGRESS_LOG* (steven@)
~ man9/sensor_attach.9
> document changes to the sensor_task api. task_register now returns a
> pointer to the sensor_task it just created, and task_unregister takes it
> to remove that task from the system. (dlg@)
~ man9/sensor_attach.9
> tweak the description of sensor_task_unregister; (jmc@)
~ man4/san.4
> Add a BUGS section, ok jmc@ (canacar@)
== sys =============================================================== 08/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/sys
altq
~ altq_rio.c ~ altq_rmclass.h
> de-register
> ok henning@ (jasper@)
arch/alpha/alpha
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ interrupt.c
> Some alpha (or SRM) designs use level 3 for i/o interrupts, instead of the
> more commonly encountered level 4. Do not complain in splassert() in this
> case
> (this is similar to the older vsbus vax machines workaround). (miod@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ conf.c
> alpha kernel have wsmux but not the actual devices; PR #5469 from janjaap
> (miod@)
arch/alpha/conf
~ GENERIC
> PCI malo(4) works
> from Janjaap van Velthooven (martin@)
arch/alpha/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> Eliminate a few straggling 'labelsector' field declarations from
> cpu_disklabel's. 'labelsector' is not used anywhere according to grep.
> 'go for it' deraadt@ (krw@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/amd64/amd64
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> this XXX has no place here anymore. extended partition label writing
> might even work now (ie. if it works, it is because finding the label
> location is now MI code shared for the read & write paths) (deraadt@)
~ disksubr.c
> simplify error return case from writedisklabel() (deraadt@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/amd64/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/amd64/stand/etc
~ genassym.cf
> (error in commit messages to other similar MI files; see position in
> ChangeLog to see other files)
> avoid pulling in machine/disklabel.h when sys/disklabel.h is a better
> choice (deraadt@)
arch/amd64/stand/libsa
~ diskprobe.c
> Calculate the disklabel checksum *after* filling in all the fields in
> the disklabel.
> ok deraadt@ (krw@)
arch/arm/arm
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> this XXX has no place here anymore. extended partition label writing
> might even work now (ie. if it works, it is because finding the label
> location is now MI code shared for the read & write paths) (deraadt@)
~ disksubr.c
> simplify error return case from writedisklabel(), too (deraadt@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/arm/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/armish/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
arch/armish/stand/boot
~ wd.c
> amd64/stand (deraadt@)
arch/aviion/aviion
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> this XXX has no place here anymore. extended partition label writing
> might even work now (ie. if it works, it is because finding the label
> location is now MI code shared for the read & write paths) (deraadt@)
~ disksubr.c
> simplify error return case from writedisklabel() (deraadt@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/aviion/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/cats/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
arch/hp300/dev
~ hd.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ hd.c
> These drivers were abusing b_cylinder to store device-specific information
> in strategy(), and were reusing it in start().
> I first considered introducing a b_rawblkno field in struct buf, as has
> been
> done in NetBSD, to stop this abuse.
> However, it does not cost more to simply move the device-specific
> ``was-b_cylinder'' computation to the start() routine. Plus we get type
> fixes
> (daddr64_t) for free, although this does not really matter for these
> pedro-sized devices. Thus we do not need to grow struct buf for these
> devices which really ought to live in your Attic (and my machineroom, of
> course).
> hp300 HP-IB tested, vax mfm not, but the logic is the same. (miod@)
~ hd.c
> Brace got lost in rev 1.51. (miod@)
arch/hp300/hp300
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/hp300/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> delete comment above cpu_disklabel that is not needed (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/hppa/conf
~ GENERIC ~ files.hppa
> Add ssio(4), a driver for the castrated National Semiconductor PC87560
> Sucky SuperIO chip. For now it only provides support for the
> i8259-compatible
> PICs on the chip, but that support is necessary to make USB interrupts
> work. (kettenis@)
~ GENERIC ~ files.hppa
> com at ssio (kettenis@)
~ GENERIC ~ files.hppa
> lpt at ssio (kettenis@)
arch/hppa/dev
~ apic.c ~ elroy.c
~ elroyvar.h
> Get interrupt information from PDC and use it to program the IO SAPIC.
> Needed to make edge triggered interrupts work. (kettenis@)
~ apic.c
> Lookup interrupt routing info based on elroy(4) bus number.
> Fix and simplify debug code that prints the interrupt routing info while
> I'm there. (kettenis@)
+ ssio.c
> Add ssio(4), a driver for the castrated National Semiconductor PC87560
> Sucky SuperIO chip. For now it only provides support for the
> i8259-compatible
> PICs on the chip, but that support is necessary to make USB interrupts
> work. (kettenis@)
~ ssio.c + com_ssio.c
+ ssiovar.h
> com at ssio (kettenis@)
~ ssio.c
> lpt at ssio (kettenis@)
+ lpt_ssio.c
> Add file such that a kernel with lpt at ssio actually compiles. (kettenis@)
~ com_ssio.c
> Attach as console when appropriate. (kettenis@)
arch/hppa/hppa
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> simplify error return case from writedisklabel() (deraadt@)
~ disksubr.c
> Don't use cpu_disklabel as local storage. Use local variables and discard
> the unneeded info when done with it. Another empty struct cpu_disklabel.
> ok deraadt@ (krw@)
~ vm_machdep.c
> In vunmapbuf(), explicitely remove mappings before invoking uvm_km_free().
> Even if the latter would end up removing the mappings by itself, it would
> do so using pmap_remove() because phys_map is not intrsafe; but some
> platforms use pmap_kenter_pa() in vmapbuf(). By removing the mappings
> ourselves, we can ensure the remove function used matches the enter
> function
> which has been used.
> Discussed and theoretical ok art@ (miod@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ machdep.c
> Crank initial storage for extents a bit. We really should allocate extents
> using malloc once vm is initialised, but this will do for now.
> ok deraadt@ (kettenis@)
arch/hppa/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> Eliminate a few straggling 'labelsector' field declarations from
> cpu_disklabel's. 'labelsector' is not used anywhere according to grep.
> 'go for it' deraadt@ (krw@)
~ disklabel.h
> Don't use cpu_disklabel as local storage. Use local variables and discard
> the unneeded info when done with it. Another empty struct cpu_disklabel.
> ok deraadt@ (krw@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/hppa64/hppa64
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> simplify error return case from writedisklabel() (deraadt@)
~ disksubr.c
> Don't use cpu_disklabel as local storage. Use local variables and discard
> the unneeded info when done with it. Another empty struct cpu_disklabel.
> ok deraadt@ (krw@)
~ vm_machdep.c
> In vunmapbuf(), explicitely remove mappings before invoking uvm_km_free().
> Even if the latter would end up removing the mappings by itself, it would
> do so using pmap_remove() because phys_map is not intrsafe; but some
> platforms use pmap_kenter_pa() in vmapbuf(). By removing the mappings
> ourselves, we can ensure the remove function used matches the enter
> function
> which has been used.
> Discussed and theoretical ok art@ (miod@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/hppa64/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> Eliminate a few straggling 'labelsector' field declarations from
> cpu_disklabel's. 'labelsector' is not used anywhere according to grep.
> 'go for it' deraadt@ (krw@)
~ disklabel.h
> Don't use cpu_disklabel as local storage. Use local variables and discard
> the unneeded info when done with it. Another empty struct cpu_disklabel.
> ok deraadt@ (krw@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/i386/conf
~ RAMDISKC
> net80211 growth means vr(4) goes away for now (deraadt@)
arch/i386/i386
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> this XXX has no place here anymore. extended partition label writing
> might even work now (ie. if it works, it is because finding the label
> location is now MI code shared for the read & write paths) (deraadt@)
~ disksubr.c
> simplify error return case from writedisklabel() (deraadt@)
~ pmap.c
> Mark PDE entries with PG_U and PG_M. We never need that information from
> the
> mmu, it slightly speeds up tlb misses and according to an errata from AMD
> it can actually work around a bug in the mmu.
> toby@ ok (art@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/i386/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/i386/stand/etc
~ genassym.cf
> amd64/stand (deraadt@)
arch/i386/stand/libsa
~ diskprobe.c
> Calculate the disklabel checksum *after* filling in all the fields in
> the disklabel.
> ok deraadt@ (krw@)
arch/landisk/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
arch/landisk/landisk
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> this XXX has no place here anymore. extended partition label writing
> might even work now (ie. if it works, it is because finding the label
> location is now MI code shared for the read & write paths) (deraadt@)
~ disksubr.c
> simplify error return case from writedisklabel() (deraadt@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/luna88k/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> Empty another cpu_disklabel. In this case read in the boot
> block/native label as needed rather than keeping it in cpu_disklabel.
> Also a number of assorted minor tweaks to reduce the delta to sparc64
> disksubr.c.
> 'get this in' deraadt@ (krw@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/luna88k/luna88k
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> using same words in similar code sequences is good (deraadt@)
~ disksubr.c
> Empty another cpu_disklabel. In this case read in the boot
> block/native label as needed rather than keeping it in cpu_disklabel.
> Also a number of assorted minor tweaks to reduce the delta to sparc64
> disksubr.c.
> 'get this in' deraadt@ (krw@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/m88k/m88k
~ vm_machdep.c
> In vunmapbuf(), explicitely remove mappings before invoking uvm_km_free().
> Even if the latter would end up removing the mappings by itself, it would
> do so using pmap_remove() because phys_map is not intrsafe; but some
> platforms use pmap_kenter_pa() in vmapbuf(). By removing the mappings
> ourselves, we can ensure the remove function used matches the enter
> function
> which has been used.
> Discussed and theoretical ok art@ (miod@)
arch/mac68k/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/mac68k/mac68k
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> make it compile (martin@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/macppc/dev
~ asms.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
arch/macppc/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> split out dpme disklabel reading, simplifying things significantly
> and making it look a lot more like other architectures; tested by
> myself and drahn and gwk (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/macppc/macppc
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> split out dpme disklabel reading, simplifying things significantly
> and making it look a lot more like other architectures; tested by
> myself and drahn and gwk (deraadt@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/macppc/stand
~ Locore.c
> Enable workaround for decrementer exception problem. (drahn@)
~ main.c
> When bootfile was made global so that macppc could modify it, it
> clashed with a global variable already used by the bootblocks.
> Therefore rename bootfile here to kernelfile.
> ok drahn@ (tom@)
arch/mips64/conf
~ files.mips64
> this is the sgi disksubr, not some mythical "all mips64 architectures can
> use it" balony (deraadt@)
~ files.mips64
> Extent sgi bus_dma to cope with different views of memory: non-contiguous
> for the cpu, contiguous from different bases for devices. This allows
> memory above 256MB to be used with bus_dma (and we had really been lucky
> with the first few large-memory builds).
> Information about memory accesses taken from Linux. (miod@)
arch/mips64/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> Eliminate a few straggling 'labelsector' field declarations from
> cpu_disklabel's. 'labelsector' is not used anywhere according to grep.
> 'go for it' deraadt@ (krw@)
~ cpu.h
> Use a shorter form to load XKPHYS constants in .S code, shaves a few text
> bytes, no functional change. (miod@)
- disklabel.h
> this is the sgi disklabel, mips64 is not the same (deraadt@)
arch/mips64/mips64
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> Coerce this into compiling. (miod@)
~ disksubr.c
> simplify error return case from writedisklabel() (deraadt@)
~ interrupt.c
> Disable instruction reordering around cpl assignments. (miod@)
~ cache_r10k.S ~ cache_r5k.S
> Use a shorter form to load XKPHYS constants in .S code, shaves a few text
> bytes, no functional change. (miod@)
- disksubr.c
> this is the sgi disksubr, not some mythical "all mips64 architectures can
> use it" balony (deraadt@)
~ pmap.c
> Protect pool operations with splvm. (miod@)
~ interrupt.c
> Make sure IPL_CLOCK blocks device interrupts. (miod@)
~ vm_machdep.c
> In vunmapbuf(), explicitely remove mappings before invoking uvm_km_free().
> Even if the latter would end up removing the mappings by itself, it would
> do so using pmap_remove() because phys_map is not intrsafe; but some
> platforms use pmap_kenter_pa() in vmapbuf(). By removing the mappings
> ourselves, we can ensure the remove function used matches the enter
> function
> which has been used.
> Discussed and theoretical ok art@ (miod@)
~ interrupt.c
> Do not hardcode imask[] size when intializing it. (miod@)
- busdma.c
> Extent sgi bus_dma to cope with different views of memory: non-contiguous
> for the cpu, contiguous from different bases for devices. This allows
> memory above 256MB to be used with bus_dma (and we had really been lucky
> with the first few large-memory builds).
> Information about memory accesses taken from Linux. (miod@)
arch/mvme68k/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> move comment to right place (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/mvme68k/mvme68k
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> include not needed (deraadt@)
~ vm_machdep.c
> In vunmapbuf(), explicitely remove mappings before invoking uvm_km_free().
> Even if the latter would end up removing the mappings by itself, it would
> do so using pmap_remove() because phys_map is not intrsafe; but some
> platforms use pmap_kenter_pa() in vmapbuf(). By removing the mappings
> ourselves, we can ensure the remove function used matches the enter
> function
> which has been used.
> Discussed and theoretical ok art@ (miod@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/mvme68k/stand/installboot
~ installboot.c
> amd64/stand (deraadt@)
arch/mvme68k/stand/libsa
~ bugdev.c
> amd64/stand (deraadt@)
arch/mvme68k/stand/prtvid
~ prtvid.c
> amd64/stand (deraadt@)
arch/mvme68k/stand/wrtvid
~ wrtvid.c
> amd64/stand (deraadt@)
arch/mvme88k/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> move comment to right place (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/mvme88k/mvme88k
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> include not needed (deraadt@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/mvme88k/stand/installboot
~ installboot.c
> amd64/stand (deraadt@)
arch/mvme88k/stand/libsa
~ bugdev.c
> amd64/stand (deraadt@)
arch/mvme88k/stand/wrtvid
- disklabel.h ~ wrtvid.c
> amd64/stand (deraadt@)
arch/mvmeppc/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/mvmeppc/mvmeppc
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> this XXX has no place here anymore. extended partition label writing
> might even work now (ie. if it works, it is because finding the label
> location is now MI code shared for the read & write paths) (deraadt@)
~ disksubr.c
> simplify error return case from writedisklabel() (deraadt@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/mvmeppc/stand/installboot
~ installboot.c
> amd64/stand (deraadt@)
arch/powerpc/powerpc
~ vm_machdep.c
> In vunmapbuf(), explicitely remove mappings before invoking uvm_km_free().
> Even if the latter would end up removing the mappings by itself, it would
> do so using pmap_remove() because phys_map is not intrsafe; but some
> platforms use pmap_kenter_pa() in vmapbuf(). By removing the mappings
> ourselves, we can ensure the remove function used matches the enter
> function
> which has been used.
> Discussed and theoretical ok art@ (miod@)
arch/sgi/conf
~ files.sgi
> this is the sgi disksubr, not some mythical "all mips64 architectures can
> use it" balony (deraadt@)
~ files.sgi
> Extent sgi bus_dma to cope with different views of memory: non-contiguous
> for the cpu, contiguous from different bases for devices. This allows
> memory above 256MB to be used with bus_dma (and we had really been lucky
> with the first few large-memory builds).
> Information about memory accesses taken from Linux. (miod@)
arch/sgi/include
~ disklabel.h
> this is the sgi disklabel, mips64 is not the same (deraadt@)
~ intr.h
> Make sure IPL_CLOCK blocks device interrupts. (miod@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ bus.h
> Extent sgi bus_dma to cope with different views of memory: non-contiguous
> for the cpu, contiguous from different bases for devices. This allows
> memory above 256MB to be used with bus_dma (and we had really been lucky
> with the first few large-memory builds).
> Information about memory accesses taken from Linux. (miod@)
arch/sgi/localbus
~ macebus.c
> Disable instruction reordering around cpl assignments. (miod@)
~ macebus.c
> Make sure IPL_CLOCK blocks device interrupts. (miod@)
~ macebus.c
> Do not hardcode imask[] size when intializing it. (miod@)
~ crimebus.h ~ macebus.c
> Extent sgi bus_dma to cope with different views of memory: non-contiguous
> for the cpu, contiguous from different bases for devices. This allows
> memory above 256MB to be used with bus_dma (and we had really been lucky
> with the first few large-memory builds).
> Information about memory accesses taken from Linux. (miod@)
arch/sgi/pci
~ macepcibridge.c
> Extent sgi bus_dma to cope with different views of memory: non-contiguous
> for the cpu, contiguous from different bases for devices. This allows
> memory above 256MB to be used with bus_dma (and we had really been lucky
> with the first few large-memory builds).
> Information about memory accesses taken from Linux. (miod@)
arch/sgi/sgi
~ machdep.c
> Disable memory above 256MB for now, it triggers bugs (probably related to
> cache operations). (miod@)
~ genassym.cf
> Use a shorter form to load XKPHYS constants in .S code, shaves a few text
> bytes, no functional change. (miod@)
+ disksubr.c
> this is the sgi disksubr, not some mythical "all mips64 architectures can
> use it" balony (deraadt@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ machdep.c
> Enable memory above 256MB again now that bus_dma has been fixed. (miod@)
+ bus_dma.c
> Extent sgi bus_dma to cope with different views of memory: non-contiguous
> for the cpu, contiguous from different bases for devices. This allows
> memory above 256MB to be used with bus_dma (and we had really been lucky
> with the first few large-memory builds).
> Information about memory accesses taken from Linux. (miod@)
arch/sh/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ intr.h
> Attach interrupt counters to intc interrupt sources too, so that clock
> interrupts get counted; from mickey (miod@)
arch/sh/sh
~ interrupt.c
> Attach interrupt counters to intc interrupt sources too, so that clock
> interrupts get counted; from mickey (miod@)
~ pmap.c
> Simple optimizations:
> - in __pmap_pv_enter, only walk the pv list to search for writeable
> mappings
> if we are ading a readonly mapping.
> - im pmap_protect, do not invoke pmap_extract(), instead directly extract
> the paddr from the pte we have already computed a few lines above. (miod@)
~ clock.c
> Fix microtime to not lose clocik ticks, gives us among other things, real
> ping times on slow links; from mickey (miod@)
arch/solbourne/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
arch/sparc/dev
~ xd.c ~ xy.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it.. except
> these two drivers. knock out that code for now to let it compile while
> miod gets his fix ready for these drivers. (deraadt@)
~ presto.c ~ fd.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ xd.c ~ xy.c
> Fix xd and xy disklabel handling. (miod@)
arch/sparc/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/sparc/sparc
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> msg is already gauranteed to be NULL (deraadt@)
~ disksubr.c
> using same words in similar code sequences is good (deraadt@)
~ disksubr.c
> more comments that are different (deraadt@)
~ vm_machdep.c
> In vunmapbuf(), explicitely remove mappings before invoking uvm_km_free().
> Even if the latter would end up removing the mappings by itself, it would
> do so using pmap_remove() because phys_map is not intrsafe; but some
> platforms use pmap_kenter_pa() in vmapbuf(). By removing the mappings
> ourselves, we can ensure the remove function used matches the enter
> function
> which has been used.
> Discussed and theoretical ok art@ (miod@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ disksubr.c
> Fix xd and xy disklabel handling. (miod@)
arch/sparc64/dev
~ fd.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ pcf8591_ofw.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
arch/sparc64/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/sparc64/sparc64
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disksubr.c
> msg is already gauranteed to be NULL (deraadt@)
~ disksubr.c
> using same words in similar code sequences is good (deraadt@)
~ vm_machdep.c
> In vunmapbuf(), explicitely remove mappings before invoking uvm_km_free().
> Even if the latter would end up removing the mappings by itself, it would
> do so using pmap_remove() because phys_map is not intrsafe; but some
> platforms use pmap_kenter_pa() in vmapbuf(). By removing the mappings
> ourselves, we can ensure the remove function used matches the enter
> function
> which has been used.
> Discussed and theoretical ok art@ (miod@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/vax/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/vax/mscp
~ mscp_disk.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/vax/vax
~ disksubr.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ vm_machdep.c
> In vunmapbuf(), explicitely remove mappings before invoking uvm_km_free().
> Even if the latter would end up removing the mappings by itself, it would
> do so using pmap_remove() because phys_map is not intrsafe; but some
> platforms use pmap_kenter_pa() in vmapbuf(). By removing the mappings
> ourselves, we can ensure the remove function used matches the enter
> function
> which has been used.
> Discussed and theoretical ok art@ (miod@)
~ disksubr.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/vax/vsa
~ hdc9224.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ hdc9224.c
> These drivers were abusing b_cylinder to store device-specific information
> in strategy(), and were reusing it in start().
> I first considered introducing a b_rawblkno field in struct buf, as has
> been
> done in NetBSD, to stop this abuse.
> However, it does not cost more to simply move the device-specific
> ``was-b_cylinder'' computation to the start() routine. Plus we get type
> fixes
> (daddr64_t) for free, although this does not really matter for these
> pedro-sized devices. Thus we do not need to grow struct buf for these
> devices which really ought to live in your Attic (and my machineroom, of
> course).
> hp300 HP-IB tested, vax mfm not, but the logic is the same. (miod@)
arch/zaurus/dev
~ zaurus_flash.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
arch/zaurus/include
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
dev
~ cons.c
> ansify/de-register
> ok miod@ (jasper@)
~ rnd.c
> ansify/de-register
> ok miod@ (jasper@)
~ ramdisk.c
> Implement DIOCRLDINFO, DIOCGPDINFO and DIOCGPART. Tweak DIOCGDINFO to
> avoid unnecessary rdgetdisklabel() call.
> requested by & ok deraadt@ (krw@)
~ ramdisk.c
> avoid modification race in DIOCRLDINFO; ok krw miod (deraadt@)
~ vnd.c ~ ramdisk.c
~ flashvar.h ~ flash.c
~ ccd.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ softraid.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
dev/acpi
~ acpiac.c
> Accept 0x00 as notify type since some notebooks (MSI-S260)
> incorrectly use it when signaling events. ok marco@ (canacar@)
dev/ata
~ wd.c
> avoid modification race in DIOCRLDINFO; ok krw miod (deraadt@)
~ wd.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
dev/dec
~ if_le_dec.c
> ansify/de-register
> ok miod@ (jasper@)
~ if_le_dec.c
> fix pasto's
> from janjaap at stack.nl (jasper@)
dev/i2c
~ ad741x.c ~ adm1021.c
~ adm1024.c ~ adm1025.c
~ adm1026.c ~ adm1030.c
~ adm1031.c ~ adt7460.c
~ asb100.c ~ ds1631.c
~ fcu.c ~ fintek.c
~ gl518sm.c ~ lm87.c
~ maxim6690.c ~ pca9554.c
~ tsl2560.c ~ w83l784r.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
dev/ic
~ sti.c
> Make sure to pass valid extended initialization structure pointers in
> sti_init(), for recent sti proms require them. Also, return a meaningful
> error value.
> ok kettenis@ mickey@ (miod@)
~ sti.c
> We can not consider the first sti region as always mapped, since on pci
> cards it is no longer the rom image. This lets sti at pci work on non-dino
> bridges.
> Found by and ok kettenis@ (miod@)
~ osiop.c
> Better recovery when osiop goes nuts and spits
> osiop0: osiop_select while connected?
> in loops. Before returning from the interrupt handler, whack the chip.
> Unfortunately, this causes the few i/o which were queued to time out,
> but this is better than spinning and eventually panicing. (miod@)
~ com.c
> ansify/de-register
> no binary change (jasper@)
~ ami.c ~ ciss.c
~ lm78.c ~ lm78var.h
~ mfi.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
dev/isa
~ if_le.c ~ if_le_isa.c
~ if_le_isapnp.c ~ if_levar.h
> ansify/de-register
> ok miod@ (jasper@)
~ mcd.c
> avoid modification race in DIOCRLDINFO; ok krw miod (deraadt@)
~ mcd.c ~ fd.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ it.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
dev/onewire
~ owsbm.c ~ owtemp.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
dev/pci
~ sti_pci.c
> Repair a couple of printfs. (miod@)
~ bktr/bktr_os.c
> remove newline in dmesg that slipped in in last commit
> go ahead robert@ (jakemsr@)
~ pcidevs
> another geforce variant (martin@)
~ pcidevs.h ~ pcidevs_data.h
> regen (martin@)
~ azalia.c
> Handle interrupts for the record stream as well. Recording now works.
> Tested by ajacoutot@, steven@ and Will Backman. (deanna@)
~ azalia_codec.c
> Add a dac group init, mixer table and mixer init for the Sigmatel
> codec ID 83847661 found in some Sony VAIO FE and SZ laptops.
> No regressions noticed by jasper@, steven@ or Will Backman. I've been
> using it for months. (deanna@)
~ pcidevs
> Add a few VIA devices found in jcs' model 2 oqo. (jsg@)
~ pcidevs.h ~ pcidevs_data.h
> regen (jsg@)
~ pciide.c
> Add support for VIA CX700/VX700 IDE.
> Tested by jcs@ on a model 2 oqo. (jsg@)
~ if_bge.c
> rework how the IFF_PROMISC and IFF_ALLMULTI flags are dealt with, and how
> the multicast filter is programmed.
> IFF_ALLMULTI is for use by the driver and only the driver, meaning that we
> have to clear it when its not needed anymore.
> now that the ethernet layer counts the number of multicast address ranges
> we can check that early to determine if ALLMULTI is needed, rather than
> doing the stupid goto allmulti dance as we iterate over the multicast
> address list.
> the imperfect multicast filter is a 16 byte wide bitfield, so we can use
> the "setbit" macro to build it in memory, and then write it to the hardware
> as a bus_space region. this simplifies the code a lot and avoids confusing
> bitshifts on u_int32_ts to get the bits in the right place.
> tested by krw on amd64, naddy on alpha, deraadt on sparc64, and beck on
> various bits.
> ok beck (dlg@)
~ pcidevs.h ~ pcidevs_data.h
> sync (todd@)
~ pcidevs
> according to claudio, entries should be ordered by PCI ID, fix two not in
> order
> ok claudio@, mk@ .. noticed by brad (todd@)
~ arc.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
~ pciide_natsemi_reg.h
> Fix typo in (unused) #define. (kettenis@)
dev/pcmcia
~ if_malo.c ~ if_maloreg.h
~ if_malovar.h
> Add first shoot of SCAN command. (mglocker@)
~ if_malo.c
> Pass the device a better rate set. (mglocker@)
~ if_malo.c ~ if_maloreg.h
~ if_malovar.h
> Add background SCAN command. (mglocker@)
dev/raidframe
~ rf_openbsdkintf.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ rf_openbsdkintf.c
> Remove raidmakedisklabel(). The only invocation was commented out in
> 2000, presumably because we spoof a label when one is not found. This
> removes the last (?) code that assigned a file type other than
> FS_UNUSED to the RAW_PART partition. (krw@)
dev/sdmmc
~ sbt.c
> better debug output (uwe@)
~ sbt.c
> Pass the correct flag to sbt_start() in sbt_start_cmd() (uwe@)
dev/tc
~ if_le_ioasic.c ~ if_le_tc.c
> ansify/de-register
> ok miod@ (jasper@)
~ if_le_ioasic.c
> fix pasto's
> from janjaap at stack.nl (jasper@)
dev/usb
~ ukbdmap.c ~ usb.h
~ usb_port.h ~ makemap.awk
> No longer include usb_port.h, all macros have been converted to their C
> counterparts.
> Revert usb_port.h to it's original form with all the macro definitions to
> ease the import of new drivers from other BSDs.
> after discussion with deraadt; feedback from mk, miod, jsg, dlg.
> ok dlg. (mbalmer@)
~ umodem.c
> Use the right interface id when iterating over the control interface
> endpoints.
> bug found and patch provided by Marc Winiger <mw at msys.ch>.
> ok dlg. (mbalmer@)
~ usbdi.c ~ usbdi.h
> KNF, especially re-indent prototypes. No binary change.
> ok dlg. (mbalmer@)
~ usbdi_util.c
> KNF after macro removals, no binary change.
> ok dlg. (mbalmer@)
~ usbf.c ~ usbf.h
> KNF after the recent macro expansion, mostly prototype indendation.
> No binary change.
> ok uwe. (mbalmer@)
~ usbdevs
> novatel ES620 CDMA modem found in the oqo model 2 (jcs@)
~ usbdevs.h ~ usbdevs_data.h
> regen (jcs@)
~ umsm.c
> attach to the novatel ES620 found in the oqo model 2 w/sprint card (jcs@)
~ usbdevs.h ~ usbdevs_data.h
> Regen. (grange@)
~ usbdevs
> Add vendor/product HUMAX/PVR-SMART for jmc at . (grange@)
kern
~ subr_disk.c
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ vfs_bio.c
> de-register
> ok thib@ (jasper@)
~ subr_disk.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ vfs_cache.c
> english, okay jmc@ (pedro@)
~ kern_sensors.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
miscfs/fifofs
~ fifo_vnops.c
> ansify/de-register.
> no binary change (jasper@)
miscfs/portal
~ portal_vfsops.c ~ portal_vnops.c
> ansify/de-register.
> no binary change (jasper@)
miscfs/procfs
~ procfs_cmdline.c ~ procfs_ctl.c
~ procfs_linux.c ~ procfs_note.c
~ procfs_status.c ~ procfs_subr.c
~ procfs_vfsops.c ~ procfs_vnops.c
> ansify/de-register.
> no binary change (jasper@)
~ procfs.h ~ procfs_subr.c
> declare "pid" as pid_t instead of long, and remove unneeded cast to pid_t
> ok pedro@ thib@ (jasper@)
miscfs/specfs
~ spec_vnops.c
> ansify/de-register.
> no binary change (jasper@)
net
~ if.c
> add back missing bcopy & break, got lost when removing arcnet.
> spotted by Mike Belopuhov <mkb at crypt.org.ru> (henning@)
~ pf.c
> Allow "log" for nat rules without "pass".
> OK henning@, ``passt scho'' markus@ (mpf@)
~ if_pfsync.c ~ pf.c
~ pfvar.h
> reimplement interface bound states in a non-retarded way.
> previously, we had a set of state tables attached to each interface. so for
> every packet we had to do a lookup in the tables for the interface, and
> afterwards in the global tables.
> since we split state keys and states now, use only the global tables, and
> put the actual states in a tail queue attached to the state key. sort the
> list so that ifbound states come before global ones. on lookup, we only
> have to compare the interface pointer on the actual states and use the
> first one where either the interface matches or the state is not interface
> bound. thus, if you don't actually use ifbound states, and there is only
> one state per state key, the overhead is close to zero, where we had extra
> lookups before. in addition to a much cleaner design (that'll allow for
> more
> goodies later) this gives us ~12.5% more forwarding performance.
> mostly hacked at c2k7, lots of help, testing and ok mcbride & markus
> (henning@)
~ pf_ioctl.c
> force logif to zero if no logging is asked for
> check the logif when changing a rule
> from max laier, ok ryan (henning@)
~ pfkeyv2.c
> export the flow/filter information attached to the SA, too; ok hshoexer@
> (markus@)
~ if_pfsync.c ~ pf.c
~ pf_ioctl.c ~ pfvar.h
> Save some bytes and make code more readable by removing junk union and
> unused ifname (this information is in struct pf_state_sync now).
> Also a bit of KNF on the pf_state struct.
> ok mpf@ henning@ (mcbride@)
net80211
~ ieee80211.h
> - remove two unused structures
> - add three new information elements identifiers:
> IEEE80211_ELEMID_QBSS_LOAD : QBSS Load
> IEEE80211_ELEMID_EDCA_PARAMS: EDCA Parameter Set
> IEEE80211_ELEMID_QOS_CAP : QoS Capability (damien@)
~ ieee80211.h ~ ieee80211_node.h
> keep track of the Tx/Rx sequence numbers for each TID in the
> ieee80211_node structure.
> add a flag to indicate whether a STA is a QSTA or not. (damien@)
~ ieee80211_var.h
> Add a new capability flag IEEE80211_C_QOS indicating that the driver
> supports QoS.
> Add a new flag IEEE80211_F_QOS indicating whether QoS is enabled or not.
> Add a callback to notify drivers that AC parameters have changed in
> the QBSS. Drivers implementing QoS should define this callback and
> update their EDCA settings accordingly. (damien@)
~ ieee80211_input.c
> split ieee80211_recv_mgmt() into smaller functions:
> - ieee80211_recv_probe_resp
> - ieee80211_recv_probe_req
> - ieee80211_recv_auth
> - ieee80211_recv_assoc_req
> - ieee80211_recv_assoc_resp
> - ieee80211_recv_deauth
> - ieee80211_recv_disassoc
> "reads ok" jsg@ claudio@ mglocker@ (damien@)
~ ieee80211.h
> s/IEEE80211_ELEMID_EDCA_PARAMS/IEEE80211_ELEMID_EDCAPARMS/ for consistency.
> (damien@)
~ ieee80211_input.c
> add code to parse EDCA parameter set information elements.
> add myself to the copyright list. (damien@)
~ ieee80211_var.h
> add fields for EDCA. (damien@)
~ ieee80211.h
> cleanup the list of Organizationally Unique Identifiers (OUI). (damien@)
~ ieee80211_input.c
> parse vendor-specific Microsoft WMM OUI.
> unfortunately, most APs that advertise themselves as being
> 802.11e capable still use the Microsoft OUI instead of the
> standard IEEE 802.11 information element. (damien@)
~ ieee80211_input.c ~ ieee80211_var.h
> add (currently unused) code to parse RSN IEs. (damien@)
~ ieee80211_output.c
> add two new functions:
> ieee80211_up_to_ac(): returns the access category to be used for
> transmitting a frame with a given user priority.
> ieee80211_classify(): returns the user-priority of an mbuf based
> on VLAN 802.1D user-priority (if any) or IP TOS precedence field.
> indent prototypes and add myself to the copyright list while i'm here.
> (damien@)
~ ieee80211_output.c ~ ieee80211_proto.h
> add EDCA Parameter Set and QoS capability IEs to appropriate frames
> when QoS is supported and activated. (damien@)
~ ieee80211.h
> cleanup CAPINFO values. (damien@)
netbt
~ hci.h ~ hci_link.c
~ hci_event.c
> Abort all pending ACL connections when we get an HCI command parameter
> error in response to a HCI_CMD_CREATE_CON command instead of waiting
> that the user aborts due to boredom. Unfortunately, the command status
> event does not include a hint to tells us which connection failed. (uwe@)
~ bt_input.c ~ bt_proto.c
~ bt_var.h ~ hci.h
~ hci_unit.c
> Remove some unused leftover code; original request from jasper@ (uwe@)
netinet
~ ip_carp.c
> use a masked destination network instead of the interface ip for the
> destination of the route belonging to the carp interface.
> ok mpf@ claudio@ (reyk@)
nfs
~ nfs_vfsops.c
> READDIRPLUS is v3 only, bail out if the mount isnt a v3.
> Prevents weird things from happening.
> ok tedu@,pedro@ (thib@)
~ nfs_var.h ~ nfsnode.h
> remove a whole bunch of duplicate function
> prototypes.
> (survived build/release on macppc/amd64).
> ok pedro@ (thib@)
scsi
~ cd.c ~ sd.c
> avoid modification race in DIOCRLDINFO; ok krw miod (deraadt@)
~ cd.c ~ sd.c
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ scsi_base.c ~ scsi_disk.h
~ scsiconf.h ~ sd.c
~ sdvar.h
> Implement disk sizes > 2^32-1. Code modelled on NetBSD.
> Tested, tweaked and ok otto@ (krw@)
~ safte.c ~ ses.c
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
stand
~ boot/boot.c
> When bootfile was made global so that macppc could modify it, it
> clashed with a global variable already used by the bootblocks.
> Therefore rename bootfile here to kernelfile.
> ok drahn@ (tom@)
sys
~ disklabel.h
> significantly simplified disklabel infrastructure. MBR handling becomes MI
> to support hotplug media on most architectures. disklabel setup and
> verification done using new helper functions. Disklabels must *always*
> have a correct checksum now. Same code paths are used to learn on-disk
> location disklabels, to avoid new errors sneaking in. Tested on almost all
> cases, testing help from todd, kettenis, krw, otto, dlg, robert, gwk, drahn
> (deraadt@)
~ malloc.h
> de-register
> ok thib@ (jasper@)
~ disk.h ~ disklabel.h
> b_cylinder does not need to be set on the callpath down into drivers.
> cpu_disklabel can go away, since nothing anymore needs to use it; ok miod
> (deraadt@)
~ sensors.h
> rework sensor tasks to use the kernels generic workq rather than a special
> kernel thread of its own. the api has changed (which will be fixed in the
> manpage shortly) so all the users of sensor tasks that i can find have
> been fixed too.
> noone tested, so its going in to force people to run with it.
> "put it in" deraadt@ (dlg@)
ufs/ext2fs
~ ext2fs_alloc.c ~ ext2fs_bmap.c
~ ext2fs_bswap.c ~ ext2fs_inode.c
~ ext2fs_lookup.c ~ ext2fs_readwrite.c
~ ext2fs_subr.c ~ ext2fs_vfsops.c
~ ext2fs_vnops.c
> ansify/de-register.
> no binary change (jasper@)
~ ext2fs_alloc.c
> declare "uid" as uid_t instead of u_int, and fix format string accordingly
> ok pedro@ thib@ (jasper@)
ufs/ffs
~ ffs_alloc.c
> Update ffs[1/2] time last written for cg's in
> clusteralloc();
> From mickey; ok pedro@ (thib@)
ufs/ufs
~ ufs_vnops.c
> remove an XXX comment;
> Basically the reason why we dont do VOP_ABORTOP()'s in NFS
> is that sys_rename() actually handles the garbage collection
> of the pathname buffers, the calls in ufs are actually fluff
> too, but will be delt with later.
> ok pedro@ (thib@)
uvm
~ uvm.h ~ uvm_amap.c
~ uvm_amap.h ~ uvm_amap_i.h
~ uvm_anon.c ~ uvm_anon.h
~ uvm_fault.c ~ uvm_init.c
~ uvm_loan.c ~ uvm_map.c
~ uvm_mmap.c ~ uvm_page.c
~ uvm_pdaemon.c ~ uvm_swap.c
> Bring back Mickey's UVM anon change. Testing by thib@, beck@ and
> ckuethe@ for a while. Okay beck@, "it is good timing" deraadt at . (pedro@)
== usr.bin =========================================================== 09/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/usr.bin
at
~ at.c
> Fix typo introduced in seconds parsing. From Tim van der Molen. (millert@)
cvs
~ Makefile ~ checkout.c
~ cmd.c ~ commit.c
~ cvs.h ~ tag.c
~ update.c + history.c
> first stab at history stuff for opencvs, currently only writes
> to CVSROOT/history but cannot parse it yet with the 'history' command.
> "Commit it." ray@ (joris@)
~ history.c
> Add missing `his' command alias. OK joris at . (xsa@)
ftp
~ ruserpass.c
> Add missing return value checks for strdup() andd fgetc().
> From Nicholas Marriott. Reminded by ray at .
> ok ray@ jaredy@ (moritz@)
mail
~ vars.c
> Grammar-o.
> OK jmc (ray@)
netstat
~ inet.c
> "off" isn't a pointer, it's a ulong. According to printf(3), %#lx
> is equivalent to %p.
> OK markus and itojun (ray@)
~ netstat.h
> net80211_ifstats() is called in main.c but there was no prototype.
> OK markus and itojun (ray@)
~ net80211.c
> err(3) was being called but was missing an include, and end: was
> unused.
> OK markus and itojun (ray@)
~ show.c
> Remove unused variable. (ray@)
ssh
~ atomicio.c
> if the fd passed to atomicio/atomiciov() is non blocking, then poll() to
> avoid a spin if it is not yet ready for reading/writing; ok dtucker@ (djm@)
touch
~ touch.c
> Fix typo introduced in seconds parsing. From Tim van der Molen. (millert@)
== usr.sbin ========================================================== 10/10 ==
http://www.openbsd.org/cgi-bin/cvsweb/src/usr.sbin
bgpd
~ rde.c ~ session.c
> Ignore SIGHUP in bgpd's children to allow the use of pkill -1 bgpd for
> reloading.
> ``yeah that is good'' henning@ (pyr@)
hoststated
~ hoststated.h
> we're going to need more room for flags (again).
> promote the field to u_int32_t.
> no impact on hoststatectl. (pyr@)
~ hce.c ~ hoststated.c
~ pfe.c ~ relay.c
> Do not fatal out with ``pipe closed'' when a short read occurs on one
> of our socket pairs. Instead disable listening on the pipe, terminate the
> event loop, and let the parent process's SIGCHLD handler do a clean
> shutdown.
> from an ospfd diff by claudio, ok claudio@ (pyr@)
~ hce.c ~ pfe.c
> don't go through the effort of creating a SIGHUP handler through
> libevent just to ignore it, use SIG_IGN instead.
> this syncs hoststated with bgpd and (soon) ospfd. (pyr@)
ospfd
~ ospfd.c ~ ospfe.c
~ rde.c
> Allow ospfd reconfiguration through SIGHUP, add SIGHUP handlers in
> children like in bgpd so that ``pkill -1 ospfd'' works as expected.
> ok claudio@ (pyr@)
~ kroute.c ~ name2id.c
~ ospfd.conf.5 ~ ospfd.h
~ parse.y ~ printconf.c
~ rde.c ~ rde.h
~ rde_spf.c
> add support for mapping route labels to AS-external route tags.
> manpage bits from jmc@
> ok norby@ claudio@ (reyk@)
pkg_add
~ OpenBSD/PackingElement.pm
> add a bit of scaffolding for signatures (espie@)
~ pkg_create.1 ~ pkg_add.1
> document @exec/@unexec variations in more detail.
> Point users at pkg_create for packing-list details (suggested by
> ajacoutot@) (espie@)
~ pkg_add ~ OpenBSD/PackingElement.pm
> more scaffolding we don't really use yet. (espie@)
~ OpenBSD/PackingElement.pm
> fix format for timestamps. Use a non-ambiguous subset of ISO8601. (espie@)
~ pkg_info
> only display header if we actually display some information.
> Allows saying pkg_info -a -M, and not having 100s of empty lines
> corresponding to all packages without a MESSAGE. (espie@)
~ OpenBSD/Add.pm ~ OpenBSD/Delete.pm
~ OpenBSD/PackingElement.pm ~ OpenBSD/Replace.pm
> add tag no shadow, to allow us to specify that we do NOT want tempfiles
> into some directories.
> a few minor style tweaks, such as using a nice join(map()) to build some
> strings. (espie@)
~ OpenBSD/Dependencies.pm
> more uniform: all deps routines should be methods, even if they don't
> use the data. (espie@)
~ OpenBSD/Dependencies.pm
> reorganize lookup_library so I can reuse the exact same algorithm to
> look up different things (in particular, the incremental lookup through
> dependencies). (espie@)
~ OpenBSD/Dependencies.pm
> .. and zap old code (espie@)
~ OpenBSD/Dependencies.pm
> sometimes, wanting to share code gives insight: disentangle the code
> that walks dependencies entirely. It is far less related to the main
> solver than I thought, it can be built when needed, and own its own
> state. Much cleaner design. (espie@)
~ OpenBSD/Dependencies.pm ~ OpenBSD/PackingList.pm
> first draft of a tag solver, needs more flesh to be useful (especially:
> transform the lib cache into a tags/lib cache) (espie@)
~ pkg_info ~ pkg_mklocatedb
~ OpenBSD/CollisionReport.pm ~ OpenBSD/PackingList.pm
~ OpenBSD/SharedItems.pm ~ OpenBSD/SharedLibs.pm
> sprinkle code to deal gracefully with absent packing-lists.
> makes it easier for people to recover after a disk crash that
> leaves /var/db/pkg half-empty. (espie@)
~ pkg_add.1 ~ pkg_info.1
> mention https protocol, for which support was added recently.
> ok espie@ (steven@)
~ OpenBSD/PackageRepository.pm
> setup a local sigalarm to send a HUP to the ftp process if it takes longer
> than 30 seconds to stop... 30 seconds chosen as a `nice' average value.
> Note that this is a `work-around' for some specificities of ftp(1), which
> usually doesn't react to SIGINT, but reacts instantly to SIGHUP... (espie@)
tcpdump
~ addrtoname.c
> Print leading zeros in MAC addresses. Initial diff from Stanislav
> Kruchinin, idea to use ether_ntoa from Claudio Jeker.
> OK canacar, idea OK henning, jasper, and tedu. (ray@)
ztsscale
~ ztsscale.c
> do not try to reopen the wsmouse device when
> the calibration was not accurate enough and
> keep the device open all the time since there is
> no need to close and reopen it if we need to retry. (robert@)
===============================================================================
More information about the owc
mailing list