]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoFix some small errors and markup nits.
cjc [Tue, 28 Aug 2001 19:10:51 +0000 (19:10 +0000)]
Fix some small errors and markup nits.

Approved by: ru
MFC after: 2 days

22 years agoRemove reference to mknod zs0 for TurboLaser console
wilko [Tue, 28 Aug 2001 17:54:07 +0000 (17:54 +0000)]
Remove reference to mknod zs0 for TurboLaser console

22 years agoo Improve the style of a number of routines and comments in kern_prot.c,
rwatson [Tue, 28 Aug 2001 16:35:33 +0000 (16:35 +0000)]
o Improve the style of a number of routines and comments in kern_prot.c,
  with regards to redundancy, formatting, and style(9).

Submitted by: bde

22 years agoIt turns out that while Toshiba laptops don't want to route interrupts
imp [Tue, 28 Aug 2001 16:35:01 +0000 (16:35 +0000)]
It turns out that while Toshiba laptops don't want to route interrupts
multiple times, others do.  The last strategy, which was to assume
that already routed interrupts were good and just return them doesn't
work for some laptops.  So, instead, we have a new strategy: we notice
that we have an interrupt that's already routed.  We go ahead and try
to route it, none the less.  We will assume that it is correctly
routed, even if the route fails.  We still assume that other failures
in the bios32 call are because the interrupt is NOT routed.

Note: some laptops do not support the bios32 interface to PCI BIOS and
we need to call it via the INT 2A interface.  That is another windmill
to till at later.

Also correct a minor typo and minor whitespace nits.

Strong MFC candidate.

22 years agoUpdate the Promise TX2 code to DTRT and not what I guessed earlier.
sos [Tue, 28 Aug 2001 13:36:06 +0000 (13:36 +0000)]
Update the Promise TX2 code to DTRT and not what I guessed earlier.
Allow tagged queing on the Promise TX2 as it seems to work.

Cudos to promise for making the most simple to program ATA chip.

22 years ago-a is not an options (sic).
dd [Tue, 28 Aug 2001 12:48:19 +0000 (12:48 +0000)]
-a is not an options (sic).

22 years agoMention what an asterisk means in the -i display.
dd [Tue, 28 Aug 2001 12:45:47 +0000 (12:45 +0000)]
Mention what an asterisk means in the -i display.

PR: 30069
Submitted by: andrew@ugh.net.au

22 years agoNew release note: ftpd -o
dd [Tue, 28 Aug 2001 12:08:57 +0000 (12:08 +0000)]
New release note: ftpd -o

22 years agoAdd a new option, '-o', for "Write-only". Disables the RETR command,
nik [Tue, 28 Aug 2001 11:59:21 +0000 (11:59 +0000)]
Add a new option, '-o', for "Write-only".  Disables the RETR command,
preventing anyone from downloading files.  In conjunction with -A, and some
appropriate file permissions, this lets you create an anonymous FTP drop
box for people to upload files to.

The more obvious "-w" flag is already taken by NetBSD's ftpd.  "-o" was
available as an option letter in all three BSDs.

22 years agoAdd support for yet another VIA '686 combination.
sos [Tue, 28 Aug 2001 08:59:17 +0000 (08:59 +0000)]
Add support for yet another VIA '686 combination.

22 years agowi driver now includes pci.h, so we need to generate it to generate it.
imp [Tue, 28 Aug 2001 05:28:40 +0000 (05:28 +0000)]
wi driver now includes pci.h, so we need to generate it to generate it.

22 years agoMake this compile when there is no pci bus in your kernel.
imp [Tue, 28 Aug 2001 05:26:43 +0000 (05:26 +0000)]
Make this compile when there is no pci bus in your kernel.

Note: This should be multiple files, but since it is also broken in
stable, I thought I'd do a fix that could be MFC'd.

This is a MFC candidate.

22 years agoFix typos in recent comments.
rwatson [Tue, 28 Aug 2001 05:16:19 +0000 (05:16 +0000)]
Fix typos in recent comments.

Submitted by: dd

22 years agoFix typos and abuses of the markup system.
imp [Tue, 28 Aug 2001 02:51:15 +0000 (02:51 +0000)]
Fix typos and abuses of the markup system.

Submitted by: ru, gshapiro

22 years agoAdd support for the 3c656B cardbus adapter. This is one half of a
wpaul [Tue, 28 Aug 2001 00:40:18 +0000 (00:40 +0000)]
Add support for the 3c656B cardbus adapter. This is one half of a
dual function card. It needs pretty much the same flags as the 656C,
except that it seems to need both the INVERT_MII_PWR and INVERT_LED_PWR
flags set. Tested with cardbus in -current as of today.

Also added support for the 3c656, which looks to be the same as
the 656B, except it doesn't need the EEPROM_8BIT flag. I think. This
one is untested, but the added support should not break any of the
other cards.

22 years agoWhen net.inet.tcp.icmp_may_rst is enabled, report ECONNREFUSED not ENETRESET
jesper [Mon, 27 Aug 2001 22:10:07 +0000 (22:10 +0000)]
When net.inet.tcp.icmp_may_rst is enabled, report ECONNREFUSED not ENETRESET
to the application as a RST would, this way we're compatible with the most
applications.

MFC candidate.

Submitted by: Scott Renfro <scott@renfro.org>
Reviewed by: Mike Silbersack <silby@silby.com>

22 years agoNew release note: SA-01:57.
bmah [Mon, 27 Aug 2001 22:04:34 +0000 (22:04 +0000)]
New release note:  SA-01:57.

22 years agoGives syslogd(8) the ability to refresh its idea of the hostname of
cjc [Mon, 27 Aug 2001 21:37:15 +0000 (21:37 +0000)]
Gives syslogd(8) the ability to refresh its idea of the hostname of
the system on which it is running. The hostname is reloaded when
'HUPped' and a log message generated to note a change (before anyone
points it out, this is not an added security feature).

PR: bin/24444
Reviewed by: freebsd-audit
Approved by: ru
MFC after: 2 weeks

22 years agoMFS: IRQ ordering, PRVERB and more whining in pcibios_get_version on failure.
imp [Mon, 27 Aug 2001 20:44:38 +0000 (20:44 +0000)]
MFS: IRQ ordering, PRVERB and more whining in pcibios_get_version on failure.
Check return value from bios32.

[[ Yes, I was bad and committed this to stable first.  I should have done
   the commit in the other order. ]]

22 years agoUgggg. I thought I'd already committed this to -current:
imp [Mon, 27 Aug 2001 20:42:07 +0000 (20:42 +0000)]
Ugggg.  I thought I'd already committed this to -current:

If the intline is 0 or 255, then it needs an interrupt routed.  Some
Sony laptops improperly flag devices that need an interrupt with 0 :-(.

22 years agoSome cosmetic changes to sync with OpenBSD
brian [Mon, 27 Aug 2001 19:08:43 +0000 (19:08 +0000)]
Some cosmetic changes to sync with OpenBSD

22 years agoOOPS, remove local change that somehow slipped into a commit (I swear that
sobomax [Mon, 27 Aug 2001 17:48:37 +0000 (17:48 +0000)]
OOPS, remove local change that somehow slipped into a commit (I swear that
I already deleted it some time ago). This should fix problem people have with
unsefined reference to `MD_PRELOAD_COMPRESSED'.

Submitted by: Manfred Antar <null@pozo.com>

22 years agoo Attach vaccess_acl_posix1e.9 to the build.
rwatson [Mon, 27 Aug 2001 17:45:06 +0000 (17:45 +0000)]
o Attach vaccess_acl_posix1e.9 to the build.
o Cross-reference vaccess_acl_posix1e.9 from vaccess.9.

Obtained from: TrustedBSD Project

22 years agoo Introduce a man page for vaccess_acl_posix1e(), based on the existing
rwatson [Mon, 27 Aug 2001 17:44:22 +0000 (17:44 +0000)]
o Introduce a man page for vaccess_acl_posix1e(), based on the existing
  vaccess().  This man page leaves something to be desired, as it doesn't
  currently document the POSIX.1e evaluation algorithm, which will in
  the future be incorporated, or cross-referenced.

Obtained from: TrustedBSD Project

22 years agoBack out prev. change. According to bde:
ache [Mon, 27 Aug 2001 17:05:29 +0000 (17:05 +0000)]
Back out prev. change. According to bde:
_POSIX_SOURCE means POSIX.1-1990, not the current version of POSIX.

22 years agoEOVERFLOW is POSIXed, so move it out #ifndef _POSIX_SOURCE section
ache [Mon, 27 Aug 2001 16:48:34 +0000 (16:48 +0000)]
EOVERFLOW is POSIXed, so move it out #ifndef _POSIX_SOURCE section

22 years agoSystems based on the ICH2/ICH2-M chip from Intel have a defect where
jlemon [Mon, 27 Aug 2001 16:07:12 +0000 (16:07 +0000)]
Systems based on the ICH2/ICH2-M chip from Intel have a defect where
the chip can cause a PCI protocol violation in under certain scenarios.
The workaround is to rewrite the EEPROM to disable Dynamic Standby Mode.

Once the EEPROM is rewritten, the system needs to be rebooted in order
to pick up the new settings.

This has been tested on several ICH2/ICH2-M systems, found in 815E based
boards, and usually identified by the presence of the 82562 ET/EM PHY.

Thanks to: Mike Tansca, Paul Saab for samples of the problematic boards.

22 years agoGenerally improve documentation of kern_prot.c:
rwatson [Mon, 27 Aug 2001 16:01:52 +0000 (16:01 +0000)]
Generally improve documentation of kern_prot.c:

o Add comments for:
  - kern.security.suser_permitted
  - p_cansee()
  - p_cansignal()
  - p_cansched()
  - kern.security.unprivileged_procdebug_permitted
  - p_candebug()

Update copyright.

Obtained from: TrustedBSD

22 years agoFix typo in my last commit
imp [Mon, 27 Aug 2001 15:18:26 +0000 (15:18 +0000)]
Fix typo in my last commit

22 years agoo Hook up vaccess.9 to the build.
rwatson [Mon, 27 Aug 2001 13:57:49 +0000 (13:57 +0000)]
o Hook up vaccess.9 to the build.

Submitted by: ru

22 years agoo Improve conformance to mdoc requirements, provided by the amazing
rwatson [Mon, 27 Aug 2001 13:56:16 +0000 (13:56 +0000)]
o Improve conformance to mdoc requirements, provided by the amazing
  mdocguard!

Submitted by: ru

22 years agoNew release note: config(8) ENV. I put this under the "kernel
dd [Mon, 27 Aug 2001 13:41:15 +0000 (13:41 +0000)]
New release note: config(8) ENV.  I put this under the "kernel
changes" section, but I'm not sure it's appropriate there.  config(8)
is a userland utility, but this is actually a "kernel" feature.  There
are release notes about config(8) in both sections, but "kernel" seems
more appropriate for this one.

22 years ago- On module unload try to detach all configured disks and let unload proceed
sobomax [Mon, 27 Aug 2001 13:25:47 +0000 (13:25 +0000)]
- On module unload try to detach all configured disks and let unload proceed
  if all disks were detached sucessfully;
- use consistent style for return statements and fix several others style
  inconsistencies.

Reviewed by: ru
Approved by: phk

22 years agostyle fixes (by the way of bde)
assar [Mon, 27 Aug 2001 13:25:43 +0000 (13:25 +0000)]
style fixes (by the way of bde)

22 years agoThis is not jail(2), or anything else suitable to be referenced with .Fn.
dd [Mon, 27 Aug 2001 12:15:44 +0000 (12:15 +0000)]
This is not jail(2), or anything else suitable to be referenced with .Fn.

22 years agoChange the name of KRNLDEFDIR to KERNCONFDIR.
paul [Mon, 27 Aug 2001 12:01:46 +0000 (12:01 +0000)]
Change the name of KRNLDEFDIR to KERNCONFDIR.

Suggested by Bruce, since the latter is more acceptable for a variable
that is externally visible.

Fix a style nit with a long line.

22 years agoMake pccard_product_lookup non-static again...
jon [Mon, 27 Aug 2001 11:28:00 +0000 (11:28 +0000)]
Make pccard_product_lookup non-static again...

When something is exported, do not mark it as static for no apparent reason...

Broken by: me

22 years agoFix the hack that allocates memory when the bios haven't already done so.
jon [Mon, 27 Aug 2001 11:23:05 +0000 (11:23 +0000)]
Fix the hack that allocates memory when the bios haven't already done so.

We can't rely on rman to give us a useable memory window if we tell it to
do auto-allocation.  This should probably be fixed on the pci bus/rman side.

22 years agoDon't make it sound like -A is turning on the default behavior.
dd [Mon, 27 Aug 2001 11:04:09 +0000 (11:04 +0000)]
Don't make it sound like -A is turning on the default behavior.

22 years agoSend a reset request for every packet received when our encryption
brian [Mon, 27 Aug 2001 10:42:21 +0000 (10:42 +0000)]
Send a reset request for every packet received when our encryption
dictionaries are out of sync.

This avoids the complications that happen when our original reset
request gets lost in transit (quite likely in hind sight, given a
lossy link) when we end up ignoring the peer for the next (up to)
256 packets.

Submitted by: Nick Sayer <nsayer@quack.kfu.com>

22 years agoBackout last change. I didnt follow the thread and made a mistake
ps [Mon, 27 Aug 2001 10:37:50 +0000 (10:37 +0000)]
Backout last change.  I didnt follow the thread and made a mistake
with this.  localisations is a valid spelling.  Oops

22 years agoCorrectly spell localizations
ps [Mon, 27 Aug 2001 10:20:02 +0000 (10:20 +0000)]
Correctly spell localizations

22 years agoThere shuldn't be whitespace before a question mark.
dd [Mon, 27 Aug 2001 09:34:39 +0000 (09:34 +0000)]
There shuldn't be whitespace before a question mark.

22 years agosort SUBDIR
assar [Mon, 27 Aug 2001 09:09:10 +0000 (09:09 +0000)]
sort SUBDIR

Requested by: bde

22 years agomdoc(7) police: removed whitespace at EOL.
ru [Mon, 27 Aug 2001 08:37:35 +0000 (08:37 +0000)]
mdoc(7) police: removed whitespace at EOL.

22 years agomdoc(7) police: markup and spelling fixes.
ru [Mon, 27 Aug 2001 08:26:57 +0000 (08:26 +0000)]
mdoc(7) police: markup and spelling fixes.

22 years agomdoc(7) police: removed whitespace at EOL, sorted SEE ALSO xrefs.
ru [Mon, 27 Aug 2001 08:12:37 +0000 (08:12 +0000)]
mdoc(7) police: removed whitespace at EOL, sorted SEE ALSO xrefs.

22 years ago- Add missed closing `"';
sobomax [Mon, 27 Aug 2001 08:05:52 +0000 (08:05 +0000)]
- Add missed closing `"';
- mention koi8-u and cp866u fonts;
- describe 1024-byte limit on amount of data that is possible to paste from
  the syscons cut buffer.

MFC after: 1 day

22 years agomdoc(7) police: markup nits.
ru [Mon, 27 Aug 2001 08:01:01 +0000 (08:01 +0000)]
mdoc(7) police: markup nits.

22 years agoRemove grammatical bogon.
sobomax [Mon, 27 Aug 2001 08:00:15 +0000 (08:00 +0000)]
Remove grammatical bogon.

22 years agoCorrectly #ifdef COMPAT_43 around osendsig() prototype.
peter [Mon, 27 Aug 2001 06:19:18 +0000 (06:19 +0000)]
Correctly #ifdef COMPAT_43 around osendsig() prototype.

22 years agoIf a file has been completely unlinked, stop automatically syncing the
peter [Mon, 27 Aug 2001 06:09:56 +0000 (06:09 +0000)]
If a file has been completely unlinked, stop automatically syncing the
file.  ffs will discard any pending dirty pages when it is closed,
so we may as well not waste time trying to clean them.  This doesn't
stop other things from writing it out, eg: pageout, fsync(2) etc.

22 years agoThere is nothing more embarresing than having three goes at correcting
peter [Mon, 27 Aug 2001 05:18:12 +0000 (05:18 +0000)]
There is nothing more embarresing than having three goes at correcting
typos in the same paragraph.  s/in in/in/

Submitted by: iedowse

22 years agoEnable hardwiring of things like tunables from embedded enironments
peter [Mon, 27 Aug 2001 05:11:53 +0000 (05:11 +0000)]
Enable hardwiring of things like tunables from embedded enironments
that do not start from loader(8).

22 years agoTry to make some sense out of the mess that used to be INSTALL.TXT.
bmah [Mon, 27 Aug 2001 03:37:56 +0000 (03:37 +0000)]
Try to make some sense out of the mess that used to be INSTALL.TXT.

1.  Bring floppies.sgml section in-line into install.sgml, where it
makes more sense.

2.  Slightly reorganize some sections of Installation section and
do some wordsmithing.

3.  Update distribution layout to reflect RELNOTESng and new compat
distributions.

4.  Update upgrade file list from 4-STABLE.

5.  It's been a long time since 2.2.X; get rid of instructions dealing
with "new" handling of compatability slices and fix up other references
in the text.

6.  Hunt down and kill emoticons with extreme prejudice.  Try to
tone down the use of exclamation points.

7.  Cross-reference new and improved Installation chapter in
Handbook.

8.  Add a proper abstract for this document.

22 years agoCL-PD6729 and CLPD-6730 chips (the only ones with I/O bars in the pci
imp [Mon, 27 Aug 2001 01:59:57 +0000 (01:59 +0000)]
CL-PD6729 and CLPD-6730 chips (the only ones with I/O bars in the pci
config space that I'm aware of) work.  I'm committing this from such
a machine.

Remove warning about I/O based bridges.  Warn users that the PCI routing
of interrupts still doesn't work for these cards.

22 years agoNote difficulties we had in OLDCARD using NEWCARD's methods of debouncing card event...
imp [Mon, 27 Aug 2001 01:30:46 +0000 (01:30 +0000)]
Note difficulties we had in OLDCARD using NEWCARD's methods of debouncing card event interrupts.

22 years agoAdd interfaces for SCSI LOG SELECT and LOG SENSE commands.
kbyanc [Mon, 27 Aug 2001 01:29:30 +0000 (01:29 +0000)]
Add interfaces for SCSI LOG SELECT and LOG SENSE commands.

Reviewed by: ken

22 years agoMore notes to the reader about issues in pccard code here.
imp [Mon, 27 Aug 2001 01:24:33 +0000 (01:24 +0000)]
More notes to the reader about issues in pccard code here.

22 years agoMerge notes, but not code, from my latest tree on pccard_function_init and pccard_fun...
imp [Mon, 27 Aug 2001 01:18:21 +0000 (01:18 +0000)]
Merge notes, but not code, from my latest tree on pccard_function_init and pccard_function_free

22 years agonow we have the rate feeder, we don't need to constrain the default channel
cg [Mon, 27 Aug 2001 01:02:13 +0000 (01:02 +0000)]
now we have the rate feeder, we don't need to constrain the default channel
speed.  however, continue to do so for record channels until the feederchain
builder is fixed.

22 years agoA number of fixes for the TI-1130 and ISA interrupt routing cases:
imp [Mon, 27 Aug 2001 01:00:27 +0000 (01:00 +0000)]
A number of fixes for the TI-1130 and ISA interrupt routing cases:

o For TI PCI-1130, you need to set bit 5 of register 91 if you want
  ANY pci interrupts.  Then set bits 3 and/or 4 as appropriate.  This
  will fix those people with 1030, 1130 and 1131 in their machines
  trying to do PCI interrupts.
o Fix case where we were trying to automatically fail back to ISA
  interrupt routing.  We were dereferencing a NULL pointer.  This
  was true of ANY chipset.
o The bus_setup_intr method needs to be pcic_setup_intr so that "FAST"
  interrupts fail on PCI case (modems act flakey if we don't force
  them to fall back to normal interrupts).  Also needed so that the
  proper ISA IRQ can be set in the ExCA register.  This fixes the
  people whose ISA routing was failing[*].
o When we find a generic yenta/pccard bridge, go ahead and print its
  vendor ID in boot verbose.

Machine with theses symptoms and a serial console by: jedgar

[*] Looks like my pc98 machine has some interrupt source on IRQ 15
that gave about 30 interrupts per second, which masked this problem on
my PC-9821Nr15.

22 years agoMFS rev 1.8.2.8:
cg [Mon, 27 Aug 2001 00:54:26 +0000 (00:54 +0000)]
MFS rev 1.8.2.8:
release resources on detach

Submitted by: "Christopher N . Harrell" <cnh@netvmg.com>

22 years agoPart two of this NEWCARD update:
jon [Mon, 27 Aug 2001 00:09:42 +0000 (00:09 +0000)]
Part two of this NEWCARD update:

Briefly, the significant changes include:
 * Way better resource management in pccbb, pccard and cardbus.
 * pccard hot-removal now appears to work.
 * support pre-fetchable memory in cardbus.
 * update cardbus to support new pci bus interface functions.
 * Fix CIS reading to no longer use rman_get_virtual().

What's not there, but in the works:
 * pccard needs to do interrupt properly and not read the ISR on single
   function cards.
 * real resource management for pccard
 * a complete implementation of CIS parsing
 * need to look into how to correctly use mutex in pccbb

22 years agoadd krb5-config
assar [Mon, 27 Aug 2001 00:08:53 +0000 (00:08 +0000)]
add krb5-config

22 years agoNon-functional changes to NEWCARD stuff.
jon [Sun, 26 Aug 2001 23:56:49 +0000 (23:56 +0000)]
Non-functional changes to NEWCARD stuff.

This is the first part of a two-part update to NEWCARD.  Changes in this
commit are non-functional, and includes the following:
 * indentation and other changes to meet style(9).
 * other minor style consistancy changes
 * addition of comments
 * renaming of device_t variables to be consistant across all of NEWCARD.

(note that not all style violations are fixed in this commit -- those that
 aren't will be clobbered by the next commit.)

22 years agoNon-functional changes to NEWCARD stuff.
jon [Sun, 26 Aug 2001 23:55:34 +0000 (23:55 +0000)]
Non-functional changes to NEWCARD stuff.
This is the first part of a two-part update to NEWCARD.  Changes in this
commit are non-functional, and includes the following:
 * indentation and other changes to meet style(9).
 * other minor style consistancy changes
 * addition of comments
 * renaming of device_t variables to be consistant across all of NEWCARD.

(note that not all style violations are fixed in this commit -- those that
 aren't will be clobbered by the next commit.)

22 years agoUpdate to match new debugging subsystem/level codes. Also mention
msmith [Sun, 26 Aug 2001 22:51:52 +0000 (22:51 +0000)]
Update to match new debugging subsystem/level codes.  Also mention
upcoming load-time module autoloading feature.

22 years agoUpdates to match the ACPI CA 20010816 import:
msmith [Sun, 26 Aug 2001 22:50:15 +0000 (22:50 +0000)]
Updates to match the ACPI CA 20010816 import:

 - New debug macro (ACPI_DEBUG_PRINT), reducing debug-case code size.
 - New debug level/subsystem codes.

22 years agoNuke the Windows platform defines; we don't need them.
msmith [Sun, 26 Aug 2001 22:46:28 +0000 (22:46 +0000)]
Nuke the Windows platform defines; we don't need them.

22 years agoMerge local changes.
msmith [Sun, 26 Aug 2001 22:45:42 +0000 (22:45 +0000)]
Merge local changes.

22 years agoThis commit was generated by cvs2svn to compensate for changes in r82367,
msmith [Sun, 26 Aug 2001 22:28:18 +0000 (22:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r82367,
which included commits to RCS files with non-trunk default branches.

22 years agoVendor import of the Intel ACPI CA 20010816 update.
msmith [Sun, 26 Aug 2001 22:28:18 +0000 (22:28 +0000)]
Vendor import of the Intel ACPI CA 20010816 update.

22 years agoI missed a typo in the last commit: s/whach/which/
peter [Sun, 26 Aug 2001 21:55:54 +0000 (21:55 +0000)]
I missed a typo in the last commit: s/whach/which/

Submitted by: bde

22 years agoThe EXTENDED_GEOMETRY flag is really bit 7. This affects >2GB drives.
peter [Sun, 26 Aug 2001 21:51:35 +0000 (21:51 +0000)]
The EXTENDED_GEOMETRY flag is really bit 7.  This affects >2GB drives.

PR: 29454
Submitted by: Rob Manchester <rmanches@vmware.com>

22 years agoStop using dirhash when a directory is removed, and ensure that we
iedowse [Sun, 26 Aug 2001 20:47:19 +0000 (20:47 +0000)]
Stop using dirhash when a directory is removed, and ensure that we
never attempt to hash directories once they are deleted. This fixes
a problem where operations on a deleted directory could trigger
dirhash sanity panics.

22 years agoAdd support for the SiS 7018
greid [Sun, 26 Aug 2001 19:15:28 +0000 (19:15 +0000)]
Add support for the SiS 7018

PR: 30100
Submitted by: Ada Lim <adal@cse.unsw.edu.au>

22 years agoAdd a variable KRNLDEFDIR that specifies where to find the kernel
paul [Sun, 26 Aug 2001 18:39:07 +0000 (18:39 +0000)]
Add a variable KRNLDEFDIR that specifies where to find the kernel
config files.

It defaults to KRNLCONFDIR.

22 years agoTidy, reorder and adjust to more correctly reflect FreeBSD default
markm [Sun, 26 Aug 2001 18:15:32 +0000 (18:15 +0000)]
Tidy, reorder and adjust to more correctly reflect FreeBSD default
policy.

22 years agoIntroduce a "noroot_ok" option to make this module ignore authentications
markm [Sun, 26 Aug 2001 18:09:00 +0000 (18:09 +0000)]
Introduce a "noroot_ok" option to make this module ignore authentications
to a non-superuser if required.

22 years agoIntroduce better logging, error reporting and use of login_cap data.
markm [Sun, 26 Aug 2001 18:05:35 +0000 (18:05 +0000)]
Introduce better logging, error reporting and use of login_cap data.

22 years agoForced commit.
markm [Sun, 26 Aug 2001 18:03:29 +0000 (18:03 +0000)]
Forced commit.

Previous commit improved use of login_cap database. This can probably
be further improved.

22 years agoAdd extra logging detail. This needs a more general solution.
markm [Sun, 26 Aug 2001 17:57:44 +0000 (17:57 +0000)]
Add extra logging detail. This needs a more general solution.

22 years agoAdjust dependancies and build order. PAM needs RPC.
markm [Sun, 26 Aug 2001 17:49:51 +0000 (17:49 +0000)]
Adjust dependancies and build order. PAM needs RPC.

22 years agoMore libss removal.
markm [Sun, 26 Aug 2001 17:47:21 +0000 (17:47 +0000)]
More libss removal.

22 years agoAdjust dependancies; now that a PAM module (pam_unix) can change
markm [Sun, 26 Aug 2001 17:45:56 +0000 (17:45 +0000)]
Adjust dependancies; now that a PAM module (pam_unix) can change
NIS passwords, libpam needs rpc dependancies.

22 years agoBig module makeover; improve logging, standardise variable names,
markm [Sun, 26 Aug 2001 17:41:13 +0000 (17:41 +0000)]
Big module makeover; improve logging, standardise variable names,
introduce ability to change passwords for both "usual" Unix methods
and NIS.

22 years agoUse &os.numports; to track how many ports we have. Don't need to give
bmah [Sun, 26 Aug 2001 17:31:03 +0000 (17:31 +0000)]
Use &os.numports; to track how many ports we have.  Don't need to give
an example of a specific release in CDROM/DVD ordering.

22 years agoCosmetique fixes from bde
ache [Sun, 26 Aug 2001 10:38:29 +0000 (10:38 +0000)]
Cosmetique fixes from bde

22 years agoCosmetique & style fixes from bde
ache [Sun, 26 Aug 2001 10:28:58 +0000 (10:28 +0000)]
Cosmetique & style fixes from bde

22 years agoCosmetique & style fixes from bde
ache [Sun, 26 Aug 2001 10:23:49 +0000 (10:23 +0000)]
Cosmetique & style fixes from bde

22 years agothe IP_FW_GET code in ip_fw_ctl() sizes a buffer to hold information
billf [Sun, 26 Aug 2001 10:09:47 +0000 (10:09 +0000)]
the IP_FW_GET code in ip_fw_ctl() sizes a buffer to hold information
about rules and dynamic rules. it later fills this buffer with these
rules.

it also takes the opporunity to compare the expiration of the dynamic
rules with the current time and either marks them for deletion or simply
charges the countdown.

unfortunatly it does this all (the sizing, the buffer copying, and the
expiration GC) with no spl protection whatsoever. it was possible for
the dynamic rule(s) to be ripped out from under the request before it
had completed, resulting in corrupt memory dereferencing.

Reviewed by: ps
MFC before: 4.4-RELEASE, hopefully.

22 years agoAlpha uses openjade i.s.o. jade for docs generation. So go out
wilko [Sun, 26 Aug 2001 07:52:56 +0000 (07:52 +0000)]
Alpha uses openjade i.s.o. jade for docs generation. So go out
and build the right port for make release.

Submitted by: bmah
Approved by: bmah
MFC after: 2 days

22 years agoPass NULL instead of MAXPHYS to the DMA allocation method. Be a bit more
benno [Sun, 26 Aug 2001 07:07:47 +0000 (07:07 +0000)]
Pass NULL instead of MAXPHYS to the DMA allocation method.  Be a bit more
verbose if we fail to allocate the DMA buffer.

22 years agoImprove the documentation for the various tunables. Document how to
imp [Sun, 26 Aug 2001 04:43:16 +0000 (04:43 +0000)]
Improve the documentation for the various tunables.  Document how to
force ISA routing of interrupts.  Warn the user that with ISA routing
of interrupts for PCI devices with more than one slot, polling mode is
in order.  Minor markup fixes as well and some white space/sentence
break changes (I did these as one commit since this file has been in
the tree only 4 days and I doubt that translation has begun).

22 years agoStop dereferencing 'r' unconditionally. Maybe it is NULL when ISA
imp [Sun, 26 Aug 2001 04:05:07 +0000 (04:05 +0000)]
Stop dereferencing 'r' unconditionally.  Maybe it is NULL when ISA
mode and using polling mode.

22 years agoEnd sentence with a period.
dd [Sun, 26 Aug 2001 02:37:22 +0000 (02:37 +0000)]
End sentence with a period.

22 years agoWhen compacting directories, ufs_direnter() always trusted DIRSIZ()
iedowse [Sun, 26 Aug 2001 01:25:12 +0000 (01:25 +0000)]
When compacting directories, ufs_direnter() always trusted DIRSIZ()
to supply the number of bytes to be bcopy()'d to move an entry. If
d_ino == 0 however, DIRSIZ() is not guaranteed to return a sensible
length, so ufs_direnter could end up corrupting a directory during
compaction. In practice I believe this can only happen after fsck_ffs
has fixed a previously-corrupted directory.

We now deal with any mid-block unused entries specially to avoid
using DIRSIZ() or bcopy() on such entries. We also ensure that the
variables 'dsize' and 'spacefree' contain meaningful values at all
times. Add a few comments to describe better this intricate piece
of code.

The special handling of mid-block unused entries makes the dirhash-
specific bugfix in the previous revision (1.53) now uncecessary,
so this change removes it.

Reviewed by: mckusick

22 years agoPut a parenthesis in the right place (DUH!).
brian [Sat, 25 Aug 2001 23:41:37 +0000 (23:41 +0000)]
Put a parenthesis in the right place (DUH!).
This fixes the apparent immediate client timeout problem.

22 years agoFix last second typo
imp [Sat, 25 Aug 2001 22:53:47 +0000 (22:53 +0000)]
Fix last second typo