]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoStyle: new-speak functions, remove pointless "return".
phk [Wed, 5 Jan 2005 22:23:03 +0000 (22:23 +0000)]
Style: new-speak functions, remove pointless "return".

19 years ago- Move the function prototypes for kern_setrlimit() and kern_wait() to
jhb [Wed, 5 Jan 2005 22:19:44 +0000 (22:19 +0000)]
- Move the function prototypes for kern_setrlimit() and kern_wait() to
  sys/syscallsubr.h where all the other kern_foo() prototypes live.
- Resort kern_execve() while I'm there.

19 years agoStart each of the license/copyright comments with /*-
imp [Wed, 5 Jan 2005 22:16:58 +0000 (22:16 +0000)]
Start each of the license/copyright comments with /*-

19 years agoRemove obsolete note about the boot loader not being ready for stable.
imp [Wed, 5 Jan 2005 22:16:10 +0000 (22:16 +0000)]
Remove obsolete note about the boot loader not being ready for stable.
It has been in stable for at least 5 years now.

19 years agoMake an attempt at explaining why sx(9) locks cannot be safely obtained
keramida [Wed, 5 Jan 2005 22:04:12 +0000 (22:04 +0000)]
Make an attempt at explaining why sx(9) locks cannot be safely obtained
after acquiring a mutex(9).

PR: docs/75571
Submitted by: darrenr
Explanation by: jhb

19 years agoStart all license statements with /*-
imp [Wed, 5 Jan 2005 21:58:49 +0000 (21:58 +0000)]
Start all license statements with /*-

19 years agoClarify /*- convention.
imp [Wed, 5 Jan 2005 21:27:01 +0000 (21:27 +0000)]
Clarify /*- convention.

19 years agoRework the optimization for spinlocks on UP to be slightly less drastic and
jhb [Wed, 5 Jan 2005 21:13:27 +0000 (21:13 +0000)]
Rework the optimization for spinlocks on UP to be slightly less drastic and
turn it back on.  Specifically, the actual changes are now less intrusive
in that the _get_spin_lock() and _rel_spin_lock() macros now have their
contents changed for UP vs SMP kernels which centralizes the changes.
Also, UP kernels do not use _mtx_lock_spin() and no longer include it.  The
UP versions of the spin lock functions do not use any atomic operations,
but simple compares and stores which allow mtx_owned() to still work for
spin locks while removing the overhead of atomic operations.

Tested on: i386, alpha

19 years agoThese appear to be unused in our tree, so remove them.
imp [Wed, 5 Jan 2005 20:50:31 +0000 (20:50 +0000)]
These appear to be unused in our tree, so remove them.

19 years agoBegin all license/copyright comments with /*-
imp [Wed, 5 Jan 2005 20:17:21 +0000 (20:17 +0000)]
Begin all license/copyright comments with /*-

19 years agoPC98 will never be defined for amd64
imp [Wed, 5 Jan 2005 20:11:13 +0000 (20:11 +0000)]
PC98 will never be defined for amd64

19 years agoFix KASSERT inversion that was introduced in 1.150, resulting in instant
keramida [Wed, 5 Jan 2005 20:10:05 +0000 (20:10 +0000)]
Fix KASSERT inversion that was introduced in 1.150, resulting in instant
panic curlen != 0, which is perfectly normal.

Approved by: mux

19 years agoAdd support for Chase/Perle PCI FAST-4 4 port serial cards which appear
wilko [Wed, 5 Jan 2005 20:06:04 +0000 (20:06 +0000)]
Add support for Chase/Perle PCI FAST-4 4 port serial cards which appear
to be the same as Boca Research Turbo Serial 654 (4 serial port).
While add the 8 port variants as well.

Submitted by: sten@blinkenlights.nl
PR: kern/75793
MFC after: 1 week

19 years agoBegin all license/copyright comments with /*- or #-
imp [Wed, 5 Jan 2005 20:05:52 +0000 (20:05 +0000)]
Begin all license/copyright comments with /*- or #-

19 years agoA large majority of the source files in the tree start their license
imp [Wed, 5 Jan 2005 19:16:01 +0000 (19:16 +0000)]
A large majority of the source files in the tree start their license
and copyright statements in a comment that begins with /*-.  Document
this tradition.  A strict adherence to this rule will help resellers
that wish to publish all copyright notices, generated automatically
from the tree.  There are too many variant licenses to do it purely
by more complicated pattern matching.

19 years agoStart all license/copyright notice comments with /*-, per tradition
imp [Wed, 5 Jan 2005 19:10:48 +0000 (19:10 +0000)]
Start all license/copyright notice comments with /*-, per tradition

19 years agoChanges corresponding to the 9.1.5.2 release of twa on the 3ware website.
vkashyap [Wed, 5 Jan 2005 19:04:28 +0000 (19:04 +0000)]
Changes corresponding to the 9.1.5.2 release of twa on the 3ware website.
The main changes are:
1. Use of multiple bus dma tags.
2. Timing of CAM requests by the driver.
3, Firmware interface change relating to retrieving AEN's.
4. Removal of twa_intrhook.
5. Bundling of latest firmware with BBU capability.

Reviewed by:re
Approved by:re

19 years agoAdd forgotten -a to the usage() message.
le [Wed, 5 Jan 2005 18:41:54 +0000 (18:41 +0000)]
Add forgotten -a to the usage() message.

19 years agoTeach sdpd(8) about 32-bit and 128-bit uuid's.
emax [Wed, 5 Jan 2005 18:37:37 +0000 (18:37 +0000)]
Teach sdpd(8) about 32-bit and 128-bit uuid's.

MFC after: 3 days

19 years agoRemove an unnecessary defination of New_Chunk, which has been
delphij [Wed, 5 Jan 2005 14:07:31 +0000 (14:07 +0000)]
Remove an unnecessary defination of New_Chunk, which has been
defined in libdisk.h.

19 years agoAdd a usb category for USB PRs.
ceri [Wed, 5 Jan 2005 12:13:10 +0000 (12:13 +0000)]
Add a usb category for USB PRs.

Requested by: Craig Rodrigues on behalf of usb@
MFC After: 1 day

19 years agoFix a typo in comment.
cognet [Wed, 5 Jan 2005 11:52:40 +0000 (11:52 +0000)]
Fix a typo in comment.

Reviewed by: mux (mentor)

19 years agoMake sis_initl() take a typed argument.
phk [Wed, 5 Jan 2005 10:26:12 +0000 (10:26 +0000)]
Make sis_initl() take a typed argument.
Expect caller to lock before calling sis_stop()
Various style stuff.

19 years agoAdd locked/unlocked variants of sis_init()
phk [Wed, 5 Jan 2005 10:11:37 +0000 (10:11 +0000)]
Add locked/unlocked variants of sis_init()

19 years agoMake a locked and unlocked variant of sis_start()
phk [Wed, 5 Jan 2005 10:04:45 +0000 (10:04 +0000)]
Make a locked and unlocked variant of sis_start()

19 years agoo Add -4 and -6 flags to a man page and usage(). Bump the man page
maxim [Wed, 5 Jan 2005 09:59:38 +0000 (09:59 +0000)]
o Add -4 and -6 flags to a man page and usage().  Bump the man page
date.

19 years agoRemove trailing spaces.
krion [Wed, 5 Jan 2005 09:52:12 +0000 (09:52 +0000)]
Remove trailing spaces.

Forgotten by: me

19 years agoo Remove -t flag from getopt(3), it was killed in rev. 1.15 three
maxim [Wed, 5 Jan 2005 09:47:10 +0000 (09:47 +0000)]
o Remove -t flag from getopt(3), it was killed in rev. 1.15 three
years ago.

19 years agoDon't declare variables "register", the compiler ought to know what to do.
phk [Wed, 5 Jan 2005 09:02:05 +0000 (09:02 +0000)]
Don't declare variables "register", the compiler ought to know what to do.

19 years ago- Initial revamp of vop_table.tcl. This scripting is ugly and will be
arr [Wed, 5 Jan 2005 06:53:03 +0000 (06:53 +0000)]
- Initial revamp of vop_table.tcl.  This scripting is ugly and will be
  cleaned up; just wanted to make a place holder.  The biggest issue is
  that it is an incomplete revamp; that is... the inconsistent naming
  table generation is not done yet ... will be added soon with the clean
  up.  The purpose of this script is to handle the vop_vector changes that
  phk@ did over the recent weeks.

19 years agoAdd needed synchronization to the error handling code that was introduced
alc [Wed, 5 Jan 2005 05:32:52 +0000 (05:32 +0000)]
Add needed synchronization to the error handling code that was introduced
in revision 1.141.

Lock assertion failures reported by: Kris Kennaway

19 years agoo Use tab instead of spaces for puc(4) line.
kuriyama [Wed, 5 Jan 2005 05:25:21 +0000 (05:25 +0000)]
o Use tab instead of spaces for puc(4) line.
o Use capitalized "Ethernet" for consistency.

19 years ago- add new supported chips: 82541ER and 82546GB
brueffer [Wed, 5 Jan 2005 04:31:00 +0000 (04:31 +0000)]
- add new supported chips: 82541ER and 82546GB
- add the long supported, but still missing 82545
- change .Nd to look better (no need to mention FreeBSD there)
- bump .Dd

MFC after: 3 days

19 years agoFix comment. One of the two "Step 4" shuold be a "step 5"
julian [Wed, 5 Jan 2005 01:04:35 +0000 (01:04 +0000)]
Fix comment. One of the two "Step 4" shuold be a "step 5"

19 years agoDisable checksum offloading by default. It seems to produce corrupted packets
cognet [Wed, 5 Jan 2005 00:06:15 +0000 (00:06 +0000)]
Disable checksum offloading by default. It seems to produce corrupted packets
with some revisions of the chip (particularly when using multiple TX
descriptors).

MFC after: 1 week

19 years agoInstead of keeping track of the index into the receive ring use the already
phk [Tue, 4 Jan 2005 22:39:35 +0000 (22:39 +0000)]
Instead of keeping track of the index into the receive ring use the already
implemented "sis_nextdesc" pointer to keep a pointer instead.

19 years agoForget about the sis_list_data and sis_ring_data structures and embedd
phk [Tue, 4 Jan 2005 22:25:58 +0000 (22:25 +0000)]
Forget about the sis_list_data and sis_ring_data structures and embedd
their fields directly in the softc structure.

This is a no-op which shortens most of the affected source lines
by N * 10 characters.

19 years agoPrototype busdma callback using the typedef.
phk [Tue, 4 Jan 2005 22:14:18 +0000 (22:14 +0000)]
Prototype busdma callback using the typedef.

19 years agoKill a half dead comment. It's first half was killed in rev 1.12.
rik [Tue, 4 Jan 2005 21:59:09 +0000 (21:59 +0000)]
Kill a half dead comment. It's first half was killed in rev 1.12.

MFC after: 3 days

19 years agoo Print a correct status for unix domain sockets.
maxim [Tue, 4 Jan 2005 21:22:32 +0000 (21:22 +0000)]
o Print a correct status for unix domain sockets.
o Restore input mode when return from the command one.

PR: bin/49983
Submitted by: Volker Stolz
OK in general from: markm
MFC after: 1 month

19 years agoRename 'class' field to 'uclass' in the ng_hci_inquiry_response structure.
emax [Tue, 4 Jan 2005 20:13:48 +0000 (20:13 +0000)]
Rename 'class' field to 'uclass' in the ng_hci_inquiry_response structure.
class is a reserved word in C++

Submitted by: Markus Brueffer < markus AT brueffer DOT de >
MFC after: 3 days

19 years agoUse __unused macro instead of a bare void for main().
delphij [Tue, 4 Jan 2005 20:07:12 +0000 (20:07 +0000)]
Use __unused macro instead of a bare void for main().

Suggested by: nectar

19 years agoFix typo: NUL -> NULL
emax [Tue, 4 Jan 2005 20:00:27 +0000 (20:00 +0000)]
Fix typo: NUL -> NULL

Submitted by: Markus Brueffer < markus AT brueffer DOT de >
MFC after: 3 days

19 years agoAllow fxp(4) cards with a revision id of 0 to use the 82503 serial
mux [Tue, 4 Jan 2005 19:30:14 +0000 (19:30 +0000)]
Allow fxp(4) cards with a revision id of 0 to use the 82503 serial
interface as well.  This is not an expected revision id per the
datasheet, but unfortunately there are such cards out there with
a 82557 chipset, and they want to use the 82503.

PR: kern/75739
Reported by: Andre Albsmeier <andre.albsmeier@siemens.com>

19 years agoRegenerate.
jhb [Tue, 4 Jan 2005 18:54:40 +0000 (18:54 +0000)]
Regenerate.

19 years agoPartial sync up to the master syscalls.master file:
jhb [Tue, 4 Jan 2005 18:53:32 +0000 (18:53 +0000)]
Partial sync up to the master syscalls.master file:
- Mark mount, unmount and nmount MPSAFE.
- Add a stub for _umtx_op().
- Mark open(), link(), unlink(), and freebsd32_sigaction() MPSAFE.

Pointy hats to: several

19 years agoreturn after freeing data element, instead of falling through, and using
jmg [Tue, 4 Jan 2005 18:15:27 +0000 (18:15 +0000)]
return after freeing data element, instead of falling through, and using
the free'd element, and ultimate NULL deref of the failed allocation.

MFC after: 1 week

19 years agoUpdate the INDEX file to INDEX-6
paul [Tue, 4 Jan 2005 17:13:16 +0000 (17:13 +0000)]
Update the INDEX file to INDEX-6

19 years agoCorrect syntactical weirdness in a call to /etc/rc.d/dhclient. Fixes:
ceri [Tue, 4 Jan 2005 16:52:15 +0000 (16:52 +0000)]
Correct syntactical weirdness in a call to /etc/rc.d/dhclient. Fixes:

PR: bin/75786
Reported by: Radko Keves <rado at daemon dot sk>
Approved by: murray
MFC After: 5 days ( to RELENG_5 )

19 years agoPorts index file is now INDEX-6
paul [Tue, 4 Jan 2005 16:35:30 +0000 (16:35 +0000)]
Ports index file is now INDEX-6

19 years agoUse INDEX-6 for versions of FreeBSD > 600000
paul [Tue, 4 Jan 2005 16:18:55 +0000 (16:18 +0000)]
Use INDEX-6 for versions of FreeBSD > 600000

19 years ago- some SATA controllers were erroneously marked as SCSI, correct that
brueffer [Tue, 4 Jan 2005 15:59:00 +0000 (15:59 +0000)]
- some SATA controllers were erroneously marked as SCSI, correct that
- move those controllers into the right place
- bump .Dd

Thanks to: Justin Bennett <justin@z-axis.com>
MFC after: 3 days

19 years agoIDE_DELAY was only available as an option for the wd driver of PC98.
keramida [Tue, 4 Jan 2005 13:21:27 +0000 (13:21 +0000)]
IDE_DELAY was only available as an option for the wd driver of PC98.
Now that the driver is gone, `tuning' IDE_DELAY makes no sense.

Submitted by: Joel Dahl joel at automatvapen dot se

19 years agoCorrect typo.
ceri [Tue, 4 Jan 2005 12:44:25 +0000 (12:44 +0000)]
Correct typo.

19 years ago- Fix 'rebuild' command - it can no longer relay on retaste event
pjd [Tue, 4 Jan 2005 12:15:21 +0000 (12:15 +0000)]
- Fix 'rebuild' command - it can no longer relay on retaste event
  (we ignore it).
- Remove code used for handling spoil events, as spoiling is not possible
  anymore, because we keep consumers open for writing all the time.

MFC after: 4 days

19 years agoSpoiling is now not possible, because we keep consumers open for writing
pjd [Tue, 4 Jan 2005 12:11:49 +0000 (12:11 +0000)]
Spoiling is now not possible, because we keep consumers open for writing
all the time. Remove unused code then.

MFC after: 4 days

19 years agoFix 'redraw' of checklist box. It is occurs in situation when size of
phantom [Tue, 4 Jan 2005 11:30:24 +0000 (11:30 +0000)]
Fix 'redraw' of checklist box.  It is occurs in situation when size of
checklist box is strictly set via command line, but amount of checklist
items less than height of checklist box.  In this case bottom part
of box was not redrawn (occurs when passing focus behind of 'Cancel' button
while configuring any FreeBSD port OPTIONS)

MFC after: 3 days

19 years agoRemove old wdc driver completely.
nyan [Tue, 4 Jan 2005 11:04:03 +0000 (11:04 +0000)]
Remove old wdc driver completely.

19 years agoMinor nits in formatting continued lines
imp [Tue, 4 Jan 2005 10:22:11 +0000 (10:22 +0000)]
Minor nits in formatting continued lines

19 years agocnw as a pccard device was commented out, so uncomment it so LINT will build.
imp [Tue, 4 Jan 2005 10:21:06 +0000 (10:21 +0000)]
cnw as a pccard device was commented out, so uncomment it so LINT will build.

19 years agoSince we do not support forceful unmount of DEVFS we can do away with
phk [Tue, 4 Jan 2005 08:49:14 +0000 (08:49 +0000)]
Since we do not support forceful unmount of DEVFS we can do away with
the partially implemented vnode-readoption code in vgonechrl().

19 years agoUnsupport forceful unmounts of DEVFS.
phk [Tue, 4 Jan 2005 07:52:26 +0000 (07:52 +0000)]
Unsupport forceful unmounts of DEVFS.

After disscussing things I have decided to take the easy and
consistent 90% solution instead of aiming for the very involved 99%
solution.

If we allow forceful unmounts of DEVFS we need to decide how to handle
the devices which are in use through this filesystem at the time.

We cannot just readopt the open devices in the main /dev instance since
that would open us to security issues.

For the majority of the devices, this is relatively straightforward
as we can just pretend they got revoke(2)'ed.

Some devices get tricky:  /dev/console and /dev/tty for instance
does a sort of recursive open of the real console device.   Other devices
may be mmap'ed (kill the processes ?).

And then there are disk devices which are mounted.

The correct thing here would be to recursively unmount the filesystems
mounte from devices from our DEVFS instance (forcefully) and if
this succeeds, complete the forcefully unmount of DEVFS.  But if
one of the forceful unmounts fail we cannot complete the forceful
unmount of DEVFS, but we are likely to already have severed a lot
of stuff in the process of trying.

Event attempting this would be a lot of code for a very far out
corner-case which most people would never see or get in touch with.

It's just not worth it.

19 years agoThere's no longer a wd driver.
imp [Tue, 4 Jan 2005 07:28:57 +0000 (07:28 +0000)]
There's no longer a wd driver.

19 years agoCatchup to wd removal
imp [Tue, 4 Jan 2005 07:26:51 +0000 (07:26 +0000)]
Catchup to wd removal

19 years agoAllow usbd to handle event notificatiosn where mor ethan one
julian [Tue, 4 Jan 2005 06:45:41 +0000 (06:45 +0000)]
Allow usbd to handle event notificatiosn where mor ethan one
device is involved, (e.g. plugin a hub with multiple devices
already attached to it)

Submitted by: Naoyuki Tai <ntai@smartfruit.com>
PR: 43993
MFC in: 1 week

19 years agoRemove last vestiges of old wd driver. ata works well on pc98 and
imp [Tue, 4 Jan 2005 06:25:24 +0000 (06:25 +0000)]
Remove last vestiges of old wd driver.  ata works well on pc98 and
there was no objection on the pc98 list when I asked if it could be
removed a while ago.

19 years agomove all the card entries to files.pc98
imp [Tue, 4 Jan 2005 06:07:25 +0000 (06:07 +0000)]
move all the card entries to files.pc98
style change: regularize tabbing

19 years agoFirst step in removing OLDCARD from FreeBSD 6.0:
imp [Tue, 4 Jan 2005 05:34:28 +0000 (05:34 +0000)]
First step in removing OLDCARD from FreeBSD 6.0:
o Move card from all architectures to just pc98.  It is only needed there,
  although real issues remain with NEWCARD's support of ISA devices.

19 years ago- list the LSI MegaRAID SATA 150-(2|4|6) controllers as supported (1)
brueffer [Tue, 4 Jan 2005 04:18:21 +0000 (04:18 +0000)]
- list the LSI MegaRAID SATA 150-(2|4|6) controllers as supported (1)
- mention SATA support in a few places
- AMI -> AMI/LSI for controllers now sold my LSI
- bump .Dd

PR: 61878 (1)
Submitted by: Sten Spans <sten@blinkenlights.nl>
Discussed with: simon (sorting)
MFC after: 3 days

19 years agonologin(8) does not seem to require any command line parameters,
delphij [Tue, 4 Jan 2005 03:57:20 +0000 (03:57 +0000)]
nologin(8) does not seem to require any command line parameters,
so remove argc and argv from main() argument.

At the same time, user and tt is not likely to be changed during
execution so mark them const.

This commit should guarantee nologin to pass WARNS=6.

Tested on: i386
MFC After: 1 month

19 years agoBecause the `permission' field in conf_entry is intended to be used as
delphij [Tue, 4 Jan 2005 02:24:01 +0000 (02:24 +0000)]
Because the `permission' field in conf_entry is intended to be used as
parameter 2 in chmod(2), which is a mode_t (and in turn a __uint_16_t),
it's more likely that it should be defined as an unsigned variable.

This commit should make newsyslog WARNS=6 clean, but don't bump the knob
until I have a universe build.

MFC After: 1 month

19 years ago- list the 3Com 3c920B-EMB-WNM as supported.
brueffer [Mon, 3 Jan 2005 20:54:29 +0000 (20:54 +0000)]
- list the 3Com 3c920B-EMB-WNM as supported.
- bump .Dd

19 years agoAdd a simple regression test for mlock()/munlock() to make sure they
rwatson [Mon, 3 Jan 2005 19:56:20 +0000 (19:56 +0000)]
Add a simple regression test for mlock()/munlock() to make sure they
work on a single page as root, and fails on a single page as nobody.
Intended to help diagnose reports regarding insecure memory use with
gnupg.

19 years agoFix 'rebuild' command (we ignore retaste event now, so don't relay on it).
pjd [Mon, 3 Jan 2005 19:42:37 +0000 (19:42 +0000)]
Fix 'rebuild' command (we ignore retaste event now, so don't relay on it).

19 years agoAdd PCI and device ID's to if_xl to support:
rwatson [Mon, 3 Jan 2005 19:22:37 +0000 (19:22 +0000)]
Add PCI and device ID's to if_xl to support:

  3C920B-EMB-WNM Integrated Fast Ethernet Controller

Submitter reports that the card appears to autonegotiate properly, and
operate well with high levels of NFS traffic.

PR: 75253
Submitted by: "Oleg V. Nauman" <oleg at reis dot zp dot ua>
MFC after: 2 weeks

19 years agoThe badsect(8) utility uses atol(), which doesn't allow very good error
rwatson [Mon, 3 Jan 2005 19:03:40 +0000 (19:03 +0000)]
The badsect(8) utility uses atol(), which doesn't allow very good error
checking and only recognizes numbers in base 10.  The attached patch
checks errno after strtol() and uses a base of 0 to allow octal, or hex
sector numbers too.

PR: 73112
Submitted by: keramida
MFC after: 2 weeks

19 years agoThe ffsinfo utility uses atol() to parse numeric values out of optarg
rwatson [Mon, 3 Jan 2005 18:59:04 +0000 (18:59 +0000)]
The ffsinfo utility uses atol() to parse numeric values out of optarg
strings.  This isn't necessarily a bug, but it can be slightly inconvenient,
because atol() doesn't know how to parse hexadecimal or octal numbers and at
least one of the options of ffsinfo(8) would be easier to use if it did.

Changing atol() -> strtol() allows one to use hex masks for -l MASK, i.e.:

orion:/a/freebsd/src/sbin/ffsinfo# ./ffsinfo -l 0x3ff /

PR: 73110
Submitted by: keramida
MFC after: 2 weeks

19 years agoUse byteorder(9) functions to implement the [g,s]et[d,w][b,l]e macros so
marius [Mon, 3 Jan 2005 18:03:00 +0000 (18:03 +0000)]
Use byteorder(9) functions to implement the [g,s]et[d,w][b,l]e macros so
only one set is needed for either endianess. This also completes them for
big endian archs and fixes the compilation of libncp, netncp, etc. there.

Reviewed by: bp, rwatson
Compile tested on: i386, sparc64
MFC after: 1 week

19 years agoThere is a better way to find out whether a port is installed or not.
josef [Mon, 3 Jan 2005 14:17:23 +0000 (14:17 +0000)]
There is a better way to find out whether a port is installed or not.

19 years agoWhen allocating bio's in the swap_pager use M_WAITOK since the
phk [Mon, 3 Jan 2005 13:28:56 +0000 (13:28 +0000)]
When allocating bio's in the swap_pager use M_WAITOK since the
alternative is much worse.

19 years agoReduce diffs to work in progress before checking in serious changes.
philip [Mon, 3 Jan 2005 13:19:49 +0000 (13:19 +0000)]
Reduce diffs to work in progress before checking in serious changes.

 o Move the sysctls under debug.psm.* and hw.psm.* making them a bit
   clearer and more consistent with other drivers.

 o Remove the debug.psm_soft_timeout sysctl.  It was introduced many
   moons ago in r1.64 but never referenced anywhere.

 o Introduce hw.psm.tap_threshold and hw.psm.tap_timeout to control
   the behaviour of taps on touchpads.  People might like to fiddle
   with these if tapping seems to slow or too fast for them.

 o Add debug.psm.loglevel as a tunable so that verbosity can be set
   easily at boot-time (to watch probes and such) without having to
   compile a kernel with options PSM_DEBUG=N.

19 years agoGarbage collect unused ipx_abort().
rwatson [Mon, 3 Jan 2005 12:54:31 +0000 (12:54 +0000)]
Garbage collect unused ipx_abort().

Spell NULL right in a KASSERT() panic message.

MFC after: 1 week

19 years agoRemove unused #include.
pjd [Mon, 3 Jan 2005 12:53:10 +0000 (12:53 +0000)]
Remove unused #include.

19 years agoExempt the superuser from mac_seeotheruids checks.
rwatson [Mon, 3 Jan 2005 12:08:18 +0000 (12:08 +0000)]
Exempt the superuser from mac_seeotheruids checks.

Submitted by: bkoenig at cs dot tu-berlin dot de
PR: 72238
MFC after: 2 weeks

19 years agoDon't ignore the last line of config file (/etc/hosts, /etc/services, etc)
sobomax [Mon, 3 Jan 2005 11:07:45 +0000 (11:07 +0000)]
Don't ignore the last line of config file (/etc/hosts, /etc/services, etc)
which doesn't end in \n, since it may be very confusing. Also this should
increase consistency, since most other config files work just fine regardless
of the presence of traling \n in the last line.

MFC After: 2 weeks

19 years agoRemove comment that doesn't seem to be true and add comments describing
julian [Mon, 3 Jan 2005 08:52:26 +0000 (08:52 +0000)]
Remove comment that doesn't seem to be true and add comments describing
what is going on, to replace it. Slight formatting changes
Code here is alredy different to NetBSD.
MFC after: 1 week

19 years agoA much simpler way to work out if the required transfer can be done in 2
julian [Mon, 3 Jan 2005 08:09:05 +0000 (08:09 +0000)]
A much simpler way to work out if the required transfer can be done in 2
scatter gather segments.

MFC after: 1 week

19 years agoBack out previous commits. Installing new share/mk fixes this.
njl [Mon, 3 Jan 2005 07:27:17 +0000 (07:27 +0000)]
Back out previous commits.  Installing new share/mk fixes this.

19 years agoAdd a sysctl (net.inet.tcp.insecure_rst) which allows one to specify
silby [Mon, 3 Jan 2005 07:08:37 +0000 (07:08 +0000)]
Add a sysctl (net.inet.tcp.insecure_rst) which allows one to specify
that the RFC 793 specification for accepting RST packets should be
following.  When followed, this makes one vulnerable to the attacks
described in "slipping in the window", but it may be necessary in
some odd circumstances.

19 years agoFirst cut at RockRidge support.
kientzle [Mon, 3 Jan 2005 05:51:33 +0000 (05:51 +0000)]
First cut at RockRidge support.

Large thanks to the easy-to-read and well-documented
sys/isofs/cd9660 source code, which provided many of the
details I needed for this exercise.

19 years agoNO_MAN is now NOMAN.
njl [Mon, 3 Jan 2005 05:36:55 +0000 (05:36 +0000)]
NO_MAN is now NOMAN.

19 years agoFix kernel modules build. For some reason, this module now bails out with
njl [Mon, 3 Jan 2005 05:35:58 +0000 (05:35 +0000)]
Fix kernel modules build.  For some reason, this module now bails out with
an error that there is no aicasm.1 man page.

19 years agoCatch up with the kernel and set any global variables we are using, in
njl [Mon, 3 Jan 2005 04:14:05 +0000 (04:14 +0000)]
Catch up with the kernel and set any global variables we are using, in
particular, enabling interpreter workarounds for bad ASL.

MFC after: 1 day

19 years agoConstify arguments.
marcel [Mon, 3 Jan 2005 02:56:15 +0000 (02:56 +0000)]
Constify arguments.

19 years agoNext round of work on ISO9660 support:
kientzle [Mon, 3 Jan 2005 01:24:13 +0000 (01:24 +0000)]
Next round of work on ISO9660 support:
  * Reference-count the directory data so that
    we don't leak memory.
  * Correctly step through the directory records
    (skipping unrecognized extensions)
  * Use better defaults for file modes
  * Sort directory entries by offset of the end of the file
    rather than the beginning of the file.  This fixes a
    lot of "out-of-order" problems with zero-length files,
    in particular.
  * Style fixes, remove some debug code, add some error messages.

19 years agoRegen.
marcel [Mon, 3 Jan 2005 00:47:23 +0000 (00:47 +0000)]
Regen.

19 years agouuidgen(2) is MP safe.
marcel [Mon, 3 Jan 2005 00:45:57 +0000 (00:45 +0000)]
uuidgen(2) is MP safe.

19 years agoIf MALLOC() fails in at_pcballoc(), return ENOBUFS rather than
rwatson [Mon, 3 Jan 2005 00:16:07 +0000 (00:16 +0000)]
If MALLOC() fails in at_pcballoc(), return ENOBUFS rather than
potentially panicking.

MFC after: 1 week

19 years agoReplace a mention of the ia64_unaligned directory with one of the
marcel [Sun, 2 Jan 2005 21:40:36 +0000 (21:40 +0000)]
Replace a mention of the ia64_unaligned directory with one of the
ia64 subdirectory. All ia64 specific regression tests live under
the ia64 directory.

19 years agoRegression test for unaligned loads and stores for short, int, long,
marcel [Sun, 2 Jan 2005 21:33:40 +0000 (21:33 +0000)]
Regression test for unaligned loads and stores for short, int, long,
float, double and long double types. No post-increment tests yet.
All tests are skipped if the debug.unaligned_test sysctl variable
cannot be set to 1.