openbsd
16 years agoIn pf_state_insert(), if the first pf_state_key_attach() fails, the
david [Thu, 10 Jul 2008 05:44:54 +0000 (05:44 +0000)]
In pf_state_insert(), if the first pf_state_key_attach() fails, the
state key is freed by pf_state_key_attach().  But in the case of NAT,
there are two state keys allocated, so we must free the second key
manually.  Fixes a pf_state_key_pl leak seen in certain cases with
pfsync or with pf state-insert errors.
ok mcbride@ henning@

16 years agoMake cams with Sonix chipset work by avoiding excessive alternate
mglocker [Thu, 10 Jul 2008 04:49:12 +0000 (04:49 +0000)]
Make cams with Sonix chipset work by avoiding excessive alternate
interface switching, which kind of crashed the device.

16 years agoadd X11 ports to default TCP baddynamic list
djm [Thu, 10 Jul 2008 02:19:28 +0000 (02:19 +0000)]
add X11 ports to default TCP baddynamic list

16 years agoDie -r, die! Eliminate stray -r in usage text. Spotted by James Hartley.
krw [Thu, 10 Jul 2008 00:21:39 +0000 (00:21 +0000)]
Die -r, die! Eliminate stray -r in usage text. Spotted by James Hartley.

16 years agothe DRM_{READ,WRITE}$NUMBER() macros currently use
oga [Thu, 10 Jul 2008 00:10:22 +0000 (00:10 +0000)]
the DRM_{READ,WRITE}$NUMBER() macros currently use
bus_space_{read,write}(). A bug in the code this was based on means that
this is also used for scatter gather maps, which are bus_dma memory, not
pci memory.  This obviously fails. Pull them out into a function and
Special case scatter gather to use the mapped virtual address to write
instead. Makes writeback test pass on pci and pci express radeon cards.

Tested by, among others ckuethe@ and sthen@.

16 years agoRead 'vendor' flag from ASCII disklabel since we print it. Fixes
krw [Wed, 9 Jul 2008 23:59:44 +0000 (23:59 +0000)]
Read 'vendor' flag from ASCII disklabel since we print it. Fixes
PR #5872.

ok millert@ deraadt@

16 years agoFixup AGP_MODE{GET,SET}_RATE() macros. the agp 2.0 spec defines that the lowest
oga [Wed, 9 Jul 2008 23:33:07 +0000 (23:33 +0000)]
Fixup AGP_MODE{GET,SET}_RATE() macros. the agp 2.0 spec defines that the lowest
three bits are for the current agp rate. not the lowest two. Otherwise we
preclude agp 4x mode.

It's a bit different for agp 3.0, but we don't have the bits for that yet.

Tested by several as part of a larger diff.

16 years agosync
deraadt [Wed, 9 Jul 2008 22:37:09 +0000 (22:37 +0000)]
sync

16 years agoautomatically populate net.inet.(tcp|udp).baddynamic with the
djm [Wed, 9 Jul 2008 20:23:47 +0000 (20:23 +0000)]
automatically populate net.inet.(tcp|udp).baddynamic with the
contents of /etc/services so as to avoid randomly allocating
source ports that correspond to well-known services. Auto-filling
of the baddynamic tables is performed before reading sysctl.conf,
so it is still possible to add or subtract ports, or override the
autofilling entirely there.

Note that this requires a new kernel and /sbin/sysctl.

feedback markus@ ok markus@ deraadt@ millert@

16 years agoexpand the net.inet.(tcp|udp).baddynamic dynamic source port
djm [Wed, 9 Jul 2008 20:20:45 +0000 (20:20 +0000)]
expand the net.inet.(tcp|udp).baddynamic dynamic source port
skipping bitmasks to cover the entire 65536 port space - previously
they covered 512-1024 only.

sysctl needs to be updated to cope with this change; please
"make includes" before rebuilding it.

feedback millert@ ok millert@ deraadt@ markus@

16 years agolet the kernel choose the random port. code from djm@. ok markus@.
jakob [Wed, 9 Jul 2008 20:10:30 +0000 (20:10 +0000)]
let the kernel choose the random port. code from djm@. ok markus@.

16 years agomissing new-line character.
sobrado [Wed, 9 Jul 2008 20:08:13 +0000 (20:08 +0000)]
missing new-line character.

16 years agoan ellipsis is not an optional argument.
sobrado [Wed, 9 Jul 2008 19:58:28 +0000 (19:58 +0000)]
an ellipsis is not an optional argument.

16 years agodocumentation tweaks.
sobrado [Wed, 9 Jul 2008 19:41:56 +0000 (19:41 +0000)]
documentation tweaks.

16 years agoreintroduce our own shuffle ID generator. ok djm@ millert@
jakob [Wed, 9 Jul 2008 19:04:55 +0000 (19:04 +0000)]
reintroduce our own shuffle ID generator. ok djm@ millert@

16 years agomerge and resolve conflicts introduced by 9.4.2-P1
jakob [Wed, 9 Jul 2008 19:02:52 +0000 (19:02 +0000)]
merge and resolve conflicts introduced by 9.4.2-P1

16 years agoalways verify that the received dns response id matches our request.
reyk [Wed, 9 Jul 2008 17:24:14 +0000 (17:24 +0000)]
always verify that the received dns response id matches our request.

16 years agoUse OpenBSD's knuth shuffle algorithm of random values from bind to
reyk [Wed, 9 Jul 2008 17:16:51 +0000 (17:16 +0000)]
Use OpenBSD's knuth shuffle algorithm of random values from bind to
produce the DNS request ids instead of a simple per-request
arc4random().  This ensure randomness but also satisfies the
non-repeating property we need.

ok deraadt@

16 years agoremove an unused variable that is a leftover from the old alg.
reyk [Wed, 9 Jul 2008 17:07:32 +0000 (17:07 +0000)]
remove an unused variable that is a leftover from the old alg.

ok deraadt@

16 years agoalso set the protocol, either TCP or UDP, in the NAT lookup. this
reyk [Wed, 9 Jul 2008 14:57:01 +0000 (14:57 +0000)]
also set the protocol, either TCP or UDP, in the NAT lookup.  this
unbreaks NAT lookups with UDP; tested as a transparent DNS relay.

16 years agoonly use SOCK_STREAM for TCP in bindany(), otherwise SOCK_DGRAM for UDP.
reyk [Wed, 9 Jul 2008 14:06:44 +0000 (14:06 +0000)]
only use SOCK_STREAM for TCP in bindany(), otherwise SOCK_DGRAM for UDP.

16 years agoIncorporate Otto's yacc skeleton fix.
millert [Wed, 9 Jul 2008 13:56:10 +0000 (13:56 +0000)]
Incorporate Otto's yacc skeleton fix.

16 years agoupdate the relay dns code to open a new udp socket to send the
reyk [Wed, 9 Jul 2008 10:50:34 +0000 (10:50 +0000)]
update the relay dns code to open a new udp socket to send the
forwarded dns request to the server instead of sending from the server
socket.  this will fix the limitation the the dns relay had to listen
to the "0.0.0.0" address, and will also enable relayd to use the udp
source port randomization. relayd will know randomize the source port
(on OpenBSD) and DNS request identifier for the clients behind it.

ok pyr@

16 years agoRemove an XXX comment about using the same pool for mfs and ffs, since
thib [Wed, 9 Jul 2008 09:30:51 +0000 (09:30 +0000)]
Remove an XXX comment about using the same pool for mfs and ffs, since
mfs is using the ffs code and its the same object it would make no sense
having two seperate pools...

ok art@

16 years agoSimplify the match function a bit by just returning the result from
brad [Wed, 9 Jul 2008 02:47:03 +0000 (02:47 +0000)]
Simplify the match function a bit by just returning the result from
pci_matchbyid(). This is the only driver that will match against
this hardware so it is not necessary to return a higher priority
than what pci_matchbyid() returns.

ok mbalmer@

16 years agosort options in manual page; display a better usage.
sobrado [Tue, 8 Jul 2008 21:39:52 +0000 (21:39 +0000)]
sort options in manual page; display a better usage.

16 years agofix double "usage:"
sobrado [Tue, 8 Jul 2008 21:30:15 +0000 (21:30 +0000)]
fix double "usage:"

16 years ago- add support for recursive transfers (but not for floppies), e.g.
martynas [Tue, 8 Jul 2008 21:07:57 +0000 (21:07 +0000)]
- add support for recursive transfers (but not for floppies), e.g.
'mget -cr 4.*' would recursively fetch (-r), and resume the previous
transfers (-c) of 4.X release directories
uses local matching (fnmatch), but only for recursive transfers.
current behavior is not changed in any way.
- while here, ifndef SMALL debugging stuff, this saves some space,
for floppies
- some debugging code was enabled for non-debugging mode, checks
assume debug is set to zero, but it's not initially set
- all "Confirm with" prompts are forced, remove redundant argument
- fix usage: -C and -c are not available for SMALL

discussed a year ago w/ pyr@
looks good to millert@
previous version looked good to pyr@
man page tweaks & ok jmc@

16 years agoFix an venerable bug: if we're reducing a rule that has an empty
otto [Tue, 8 Jul 2008 15:06:50 +0000 (15:06 +0000)]
Fix an venerable bug: if we're reducing a rule that has an empty
right hand side and the yacc stackpointer is pointing at the very
end of the allocated stack, we end up accessing the stack out of
bounds by the implicit $$ = $1 action.  Detected by my new malloc,
experienced by sturm@ on sparc64; ok deraadt@

16 years agobasic manpage for vmt(4)
dlg [Tue, 8 Jul 2008 13:41:20 +0000 (13:41 +0000)]
basic manpage for vmt(4)

16 years agoUse correct format string specifier for int argument. Don't try to print it
claudio [Tue, 8 Jul 2008 13:14:58 +0000 (13:14 +0000)]
Use correct format string specifier for int argument. Don't try to print it
as string. Fixes crash seen by Peter Bristow. "obviously ok" henning@

16 years agomissed in last commit, to make things compile cleaner
joris [Tue, 8 Jul 2008 12:54:50 +0000 (12:54 +0000)]
missed in last commit, to make things compile cleaner

16 years agomake these compile cleanly
joris [Tue, 8 Jul 2008 12:54:13 +0000 (12:54 +0000)]
make these compile cleanly

16 years agouse cf->fd in cvs_remote_sendfile() instead of using a local
joris [Tue, 8 Jul 2008 12:29:58 +0000 (12:29 +0000)]
use cf->fd in cvs_remote_sendfile() instead of using a local
variable that is only used once, which is in this case.

16 years agobump version to 4.4
joris [Tue, 8 Jul 2008 12:25:47 +0000 (12:25 +0000)]
bump version to 4.4

16 years agothis makes the code in sdattach less dense by collapsing the flags we pass
dlg [Tue, 8 Jul 2008 12:17:48 +0000 (12:17 +0000)]
this makes the code in sdattach less dense by collapsing the flags we pass
to all the scsi midlayer code into a local variable.

ok krw@ marco@

16 years agoTypo in summarize output.
ajacoutot [Tue, 8 Jul 2008 11:04:28 +0000 (11:04 +0000)]
Typo in summarize output.

ok sthen@

16 years agoRemove the IDE_PCI_CLASS_OVERRIDE quirk flag for the Intel AHCI controllers
brad [Tue, 8 Jul 2008 05:59:39 +0000 (05:59 +0000)]
Remove the IDE_PCI_CLASS_OVERRIDE quirk flag for the Intel AHCI controllers
as they're of the subclass SATA and the SiI3512 SATA controller as it is of
the subclass MISC.

ok jsg@

16 years agovmt(4) is a kernel level implementation of the vmware tools.
dlg [Tue, 8 Jul 2008 05:21:59 +0000 (05:21 +0000)]
vmt(4) is a kernel level implementation of the vmware tools.

it only provides the hosts machines clock as a timedelta sensor so far.

getting it into the tree so people can work on it as suggested by fgsch@

16 years agoUpdate 2300 firmware from 3.01.20 (Dec 19, 2002) to 3.03.19 (Nov
krw [Tue, 8 Jul 2008 01:46:24 +0000 (01:46 +0000)]
Update 2300 firmware from 3.01.20 (Dec 19, 2002) to 3.03.19 (Nov
30, 2005). From mjacob via FreeBSD. No known code dependency, and
prepares ground for major isp update.

Tested on a 2312 by dlg@.

ok dlg@ deraadt@

16 years agoComment correction, actually -> actual
brad [Mon, 7 Jul 2008 23:41:58 +0000 (23:41 +0000)]
Comment correction, actually -> actual

ok henning@

16 years ago/*NOTREACHED*/ for lint warning:
stevesk [Mon, 7 Jul 2008 23:32:51 +0000 (23:32 +0000)]
/*NOTREACHED*/ for lint warning:
  warning: function key_equal falls off bottom without returning value
ok djm@

16 years agoMatch on XMITS (which is a PCI-X Schizo variant found on Sun Fire
kettenis [Mon, 7 Jul 2008 23:22:27 +0000 (23:22 +0000)]
Match on XMITS (which is a PCI-X Schizo variant found on Sun Fire
E2900/E4900/E6900/E20K/E25K systems).  There's a fairly good chance it'll
just work.

16 years agosync
deraadt [Mon, 7 Jul 2008 22:08:12 +0000 (22:08 +0000)]
sync

16 years ago- Improve negotation and fix some more DPRINTFs.
mglocker [Mon, 7 Jul 2008 18:07:51 +0000 (18:07 +0000)]
- Improve negotation and fix some more DPRINTFs.

16 years agobcachestats got bumbed too u_int64_t's and I forgot about
thib [Mon, 7 Jul 2008 17:16:27 +0000 (17:16 +0000)]
bcachestats got bumbed too u_int64_t's and I forgot about
munging systat to handle the field with;

pointed out by mickey

ok canacar@

16 years agoThe current drm_locked_task*() code sometimes tries to sleep in an
oga [Mon, 7 Jul 2008 16:29:57 +0000 (16:29 +0000)]
The current drm_locked_task*() code sometimes tries to sleep in an
interrupt handler.

This is bad and wrong. So change it so that if we can't immediately grab
the hardware lock, to just leave the task flagged so that we can run it
when we release the lock. The linux implementation uses a similar
scheme.

Tested by guenther@, landry@ and bernd@. Also tested by many a while
ago as part of a larger diff.

16 years agoRework the ntpd question to default to no, and ask for the NTP server only if user...
jdixon [Mon, 7 Jul 2008 15:53:15 +0000 (15:53 +0000)]
Rework the ntpd question to default to no, and ask for the NTP server only if user chooses yes.  Easier than the 3-part question and requires far fewer keystrokes to enable the 'default'.

ok krw@

16 years agoExtend sbbc(4) to provide a console driver for the v1280.
kettenis [Mon, 7 Jul 2008 14:46:18 +0000 (14:46 +0000)]
Extend sbbc(4) to provide a console driver for the v1280.

16 years agoFix typo in comment.
kettenis [Mon, 7 Jul 2008 14:38:42 +0000 (14:38 +0000)]
Fix typo in comment.

16 years agoUpdate to tzdata2008d from elsie.nci.nih.gov.
millert [Mon, 7 Jul 2008 14:24:14 +0000 (14:24 +0000)]
Update to tzdata2008d from elsie.nci.nih.gov.

16 years agoAdd an entry to report the bus clock on bernd's atom laptop.
jsg [Mon, 7 Jul 2008 13:41:59 +0000 (13:41 +0000)]
Add an entry to report the bus clock on bernd's atom laptop.
Intel don't publish the EST voltage tables, and they don't
even publish the MSRs for a shipping processor so we
can figure out how to do this in the backwards highest/lowest
way cleanly.

The mapping might look like the Core * one, but who really knows
for sure outside of a few guys at Intel.  Other machines with
Atom processors and a different bus clock will have to be added
one by one until this stupidity changes.

Tested by bernd, ok gwk

16 years agoUse bus_dmamap_load_mbuf() instead of bus_dmamap_load() when
thib [Mon, 7 Jul 2008 12:53:43 +0000 (12:53 +0000)]
Use bus_dmamap_load_mbuf() instead of bus_dmamap_load() when
we want too load an mbuf; remove a printf that fires in case
we can't load the mbuf (We do error handling and recovery).

ok brad@, dlg@

16 years agoShorten the sensor description and reserve room for optional characters
mbalmer [Mon, 7 Jul 2008 08:01:47 +0000 (08:01 +0000)]
Shorten the sensor description and reserve room for optional characters
only when they are needed.  Makes locations like -123d45m,-123d45m fit.

16 years agoAdd AGP and hostbridge support for the Intel 82945GME chipset.
bernd [Mon, 7 Jul 2008 07:54:48 +0000 (07:54 +0000)]
Add AGP and hostbridge support for the Intel 82945GME chipset.

ok jsg@

16 years agosync
bernd [Mon, 7 Jul 2008 07:40:54 +0000 (07:40 +0000)]
sync

16 years agoAdd Intel 82945GME Hostbridge and Video.
bernd [Mon, 7 Jul 2008 07:38:40 +0000 (07:38 +0000)]
Add Intel 82945GME Hostbridge and Video.

ok jsg@

16 years agoApplication-specified thread stacks didn't work with libpthread
guenther [Mon, 7 Jul 2008 04:59:22 +0000 (04:59 +0000)]
Application-specified thread stacks didn't work with libpthread
because it treated the supplied memory as holding an internal data
structure instead of as the stack space itself

ok kurt@, "looks ok" otto@, tested on hppa by kettenis@

16 years agotweak the alignment of the rx buffers so the headers in the frame are
dlg [Mon, 7 Jul 2008 00:42:34 +0000 (00:42 +0000)]
tweak the alignment of the rx buffers so the headers in the frame are
better aligned for fast access. i didnt bench, so maybe this does nothing.

16 years agowe don't need arg after the debug3() was removed. from lint.
stevesk [Mon, 7 Jul 2008 00:31:41 +0000 (00:31 +0000)]
we don't need arg after the debug3() was removed.  from lint.
ok djm@

16 years agooops, fix getpwent() use of wrong buffers
deraadt [Sun, 6 Jul 2008 22:19:39 +0000 (22:19 +0000)]
oops, fix getpwent() use of wrong buffers
found by alexander.schrijver

16 years agoAdd the position to the sensor description.
mbalmer [Sun, 6 Jul 2008 21:03:13 +0000 (21:03 +0000)]
Add the position to the sensor description.

discussed with otto, sthen, ckuethe.  ok otto

16 years agoremove #ifndef nolint goo
thib [Sun, 6 Jul 2008 16:54:48 +0000 (16:54 +0000)]
remove #ifndef nolint goo

ok blambert@

16 years agoDon't change the size of the partition being newfs'd when sectorsize,
krw [Sun, 6 Jul 2008 15:03:36 +0000 (15:03 +0000)]
Don't change the size of the partition being newfs'd when sectorsize,
either from the disklabel or via -S, is not 512 bytes. Disklabel
partition sizes are sectors and not 512 byte blocks. One less 'block
is 512 bytes' assumption. 2^32 or so to go.

ok otto@ millert@

16 years ago- no need for .Pp before or after .Sh
jmc [Sun, 6 Jul 2008 14:34:14 +0000 (14:34 +0000)]
- no need for .Pp before or after .Sh
- no comma for last SEE ALSO item

16 years agoMake 'dhcp' the initial default when configuring the network. This
krw [Sun, 6 Jul 2008 14:31:19 +0000 (14:31 +0000)]
Make 'dhcp' the initial default when configuring the network. This
makes one more answer <cr>'able during install without changing any
existing behaviour.

Change 'NTP server?' default answer from 'none' to 'default'.

Suggested by thib@, supported by marco@, jsing@, millert@

16 years agoWhen the device gets close switch back from alternate interface to
mglocker [Sun, 6 Jul 2008 13:50:36 +0000 (13:50 +0000)]
When the device gets close switch back from alternate interface to
default interface.  Nice side effect;  Turns off the cams LED again.

16 years agospacing.
sobrado [Sun, 6 Jul 2008 13:49:35 +0000 (13:49 +0000)]
spacing.

16 years agosync synopsis and usage.
sobrado [Sun, 6 Jul 2008 13:42:35 +0000 (13:42 +0000)]
sync synopsis and usage.

16 years agospacing.
sobrado [Sun, 6 Jul 2008 13:36:09 +0000 (13:36 +0000)]
spacing.

16 years agoZap redundant prototype for nfsrv_zapsock();
thib [Sun, 6 Jul 2008 13:33:54 +0000 (13:33 +0000)]
Zap redundant prototype for nfsrv_zapsock();

16 years agopciide_match() is a bit convoluted, we match anything claiming
jsg [Sun, 6 Jul 2008 10:38:52 +0000 (10:38 +0000)]
pciide_match() is a bit convoluted, we match anything claiming
to be MASS_STORAGE subclass IDE regardless of id, and other
ids depending on the subclass and a known id or a known id
and the override flag.

We really want to check the subclass so we don't attach
pciide to the wrong function of a chip, which happened
last time around before the additional subclasses were
checked in the non override case and the overrides were pulled.

A bunch of the override flags were pulled after some common
subclasses (SATA/RAID) were added.  So if we have a known
id and a SATA/RAID subclass we would match.  Turns out some
Silicon Image SATA devices claim to be of subclass MISC
so add that as well.  Unbreaks one of hennings machines
with 3114 SATA and likely quite a few other machines.

'seems safe enough to me' miod@

16 years agoUse UT_{READ|WRITE}_VENDOR_DEVICE instead of self defined values.
mbalmer [Sun, 6 Jul 2008 10:00:47 +0000 (10:00 +0000)]
Use UT_{READ|WRITE}_VENDOR_DEVICE instead of self defined values.

16 years agoAdjust SYNOPSIS.
kettenis [Sun, 6 Jul 2008 09:08:21 +0000 (09:08 +0000)]
Adjust SYNOPSIS.

16 years agossm(4)
kettenis [Sun, 6 Jul 2008 09:05:04 +0000 (09:05 +0000)]
ssm(4)

16 years agoEnable ssm(4) and remove hacks that worked around the fact that we didn't
kettenis [Sun, 6 Jul 2008 08:53:38 +0000 (08:53 +0000)]
Enable ssm(4) and remove hacks that worked around the fact that we didn't
have it.

16 years agoAdd ssm(4), a driver for the scalable shared memory device found on
kettenis [Sun, 6 Jul 2008 08:51:44 +0000 (08:51 +0000)]
Add ssm(4), a driver for the scalable shared memory device found on
Serengeti and Starcat systems.

16 years agoBump verson number now that the bootloader works on the v1280.
kettenis [Sun, 6 Jul 2008 08:47:02 +0000 (08:47 +0000)]
Bump verson number now that the bootloader works on the v1280.

16 years agosbbc(4)
kettenis [Sun, 6 Jul 2008 07:28:32 +0000 (07:28 +0000)]
sbbc(4)

16 years agoAdd sbbc(4), a driver for the BootBus controller on Serengeti and StarCat
kettenis [Sun, 6 Jul 2008 07:27:43 +0000 (07:27 +0000)]
Add sbbc(4), a driver for the BootBus controller on Serengeti and StarCat
systems that provides time of day services and (in the future) console
services.

16 years agoAdd missing include such that RAMDISK kernels compile again.
kettenis [Sun, 6 Jul 2008 07:25:04 +0000 (07:25 +0000)]
Add missing include such that RAMDISK kernels compile again.

16 years agosync
deraadt [Sun, 6 Jul 2008 04:57:57 +0000 (04:57 +0000)]
sync

16 years agoCopy OBP breakpoint vector from the PROM trap table into our own trap table.
kettenis [Sat, 5 Jul 2008 23:08:08 +0000 (23:08 +0000)]
Copy OBP breakpoint vector from the PROM trap table into our own trap table.
This makes the kernel properly enter the prom upon halt on the v1280.

16 years agoAdd function to read %tba.
kettenis [Sat, 5 Jul 2008 23:06:06 +0000 (23:06 +0000)]
Add function to read %tba.

16 years agoApparantly the "chain" OpenFirmware interface always fails so don't bother
kettenis [Sat, 5 Jul 2008 23:03:04 +0000 (23:03 +0000)]
Apparantly the "chain" OpenFirmware interface always fails so don't bother
with it.  Makes the bootloader work on the v1280 where the firmware breaks
into the firmware debugger instead of returning failure.

16 years agoRound up memory allocations to page size. Apparently sane firmware versions do
kettenis [Sat, 5 Jul 2008 22:59:14 +0000 (22:59 +0000)]
Round up memory allocations to page size.  Apparently sane firmware versions do
this automatically, but the braindead firmware on the v1280 doesn't do this
and makes the allocation fail.

16 years agoSpecify the correct number of arguments in prom call that sets the trap table
kettenis [Sat, 5 Jul 2008 22:17:21 +0000 (22:17 +0000)]
Specify the correct number of arguments in prom call that sets the trap table
(1 for sun4u, 2 for sun4v).

16 years agoCalculate the size of kernel text correctly if it spans multiple prom mapping
kettenis [Sat, 5 Jul 2008 21:20:48 +0000 (21:20 +0000)]
Calculate the size of kernel text correctly if it spans multiple prom mapping
entries.  This is necessary on the v1280 where the firmware mixes 4MB and 8KB
mappings to map kernel text.

tested by miod@ and nick@

16 years agoLink bootloader at 1MB (same a the Solaris bootloader) instead of 8MB (which
kettenis [Sat, 5 Jul 2008 21:17:28 +0000 (21:17 +0000)]
Link bootloader at 1MB (same a the Solaris bootloader) instead of 8MB (which
the firmware on the v1280 doesn't like).

tested by many

16 years agosynchronize usage of atactl commands with manual page.
sobrado [Sat, 5 Jul 2008 21:13:47 +0000 (21:13 +0000)]
synchronize usage of atactl commands with manual page.

ok millert@, (for the change) jmc@

16 years agoeach utility must have its own usage and an unique set of options.
sobrado [Sat, 5 Jul 2008 21:00:38 +0000 (21:00 +0000)]
each utility must have its own usage and an unique set of options.

diff written by millert based on a previous one by me;
millert's diff has a much better style and adds support for different
sets of options for each utility.

ok millert@

16 years agoeach utility must have its own usage and its own set of options;
sobrado [Sat, 5 Jul 2008 20:59:42 +0000 (20:59 +0000)]
each utility must have its own usage and its own set of options;
b64encode and b64decode are equivalent to running uuencode and uudecode
respectively with the -m flag specified, so this flag should not be
available in these utilities; while here, fix synopsis.

based on millert's diff for compress/gzip.

ok millert@

16 years agoThe firmware on the v1280 changes %wstate behind our back. Work around this
kettenis [Sat, 5 Jul 2008 20:53:33 +0000 (20:53 +0000)]
The firmware on the v1280 changes %wstate behind our back.  Work around this
problem by adopting the same encoding used by Solaris for the kernel windows.
Note that this involves rearranging the trap vector tables, both fur sun4u and
for sun4v.

16 years agoDon't play games with %cleanwin when spinning up secondary CPUs. The firmware
kettenis [Sat, 5 Jul 2008 19:30:44 +0000 (19:30 +0000)]
Don't play games with %cleanwin when spinning up secondary CPUs.  The firmware
on the v1280 doesn't like it if we change it behind its back.

16 years agoWhen waking an nfs daemon, use wakeup_one vice wakeup, as
blambert [Sat, 5 Jul 2008 18:28:08 +0000 (18:28 +0000)]
When waking an nfs daemon, use wakeup_one vice wakeup, as
a) we're attempting to wake a specific process which
b) sleeps on a unique address
which means that there's no need to continue traversing the sleep
queue once the process has been found and awakened.

"looks good too me" thib@

16 years agoregen
kettenis [Sat, 5 Jul 2008 18:22:26 +0000 (18:22 +0000)]
regen

16 years agoAdd Sun SBBC.
kettenis [Sat, 5 Jul 2008 18:21:47 +0000 (18:21 +0000)]
Add Sun SBBC.

16 years agoRemove R_GETONEREP and the code that checks if its set
thib [Sat, 5 Jul 2008 17:34:26 +0000 (17:34 +0000)]
Remove R_GETONEREP and the code that checks if its set
(its never set).

"please kill it" blambert@

16 years agoin pf_state_key_attach(), when there is already an existing state key that
david [Sat, 5 Jul 2008 16:57:50 +0000 (16:57 +0000)]
in pf_state_key_attach(), when there is already an existing state key that
can be used and we return the state key back to the pool, don't insert
state items into the tailq using that garbage state key.
this makes things much happier for me.
ok henning@

16 years agoWhen SCSIDEBUG'ing, print data about to be written (for commands
krw [Sat, 5 Jul 2008 16:07:01 +0000 (16:07 +0000)]
When SCSIDEBUG'ing, print data about to be written (for commands
flagged SCSI_DATA_OUT) and data that has been returned (for commands
flagged SCSI_DATA_IN). This is better than just printing the data
buffer before the command is issued since that does not include any
data that has been read. e.g. INQUIRY data, as thib@ and I discovered.

ok marco@