]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
26 years agoMFMP: Make ppp multilink capable.
Brian Somers [Thu, 21 May 1998 21:49:08 +0000 (21:49 +0000)]
MFMP: Make ppp multilink capable.
      See the file README.changes, and re-read the man page.

26 years agoMFMP: Make ppp multilink capable.
Brian Somers [Thu, 21 May 1998 21:45:49 +0000 (21:45 +0000)]
MFMP: Make ppp multilink capable.

26 years agoMove __FreeBSD_version from src/sys/conf/newvers.sh to
Eivind Eklund [Thu, 21 May 1998 19:21:46 +0000 (19:21 +0000)]
Move __FreeBSD_version from src/sys/conf/newvers.sh to
src/sys/sys/param.h, to facilitate access from the kernel.  This make
it possible to do outside kernel development and have it actually work
properly.

26 years agoYeargh! After all that, I forgot to remove the #include.
Jordan K. Hubbard [Thu, 21 May 1998 17:05:32 +0000 (17:05 +0000)]
Yeargh!  After all that, I forgot to remove the #include.

26 years agoDon't use __FreeBSD_version explicitly - none of the other
Jordan K. Hubbard [Thu, 21 May 1998 16:24:05 +0000 (16:24 +0000)]
Don't use __FreeBSD_version explicitly - none of the other
drivers here do and it also blows up in building GENERIC during
a release build if you try and include <osreldate.h> (which shot
my SNAP dead - argh!).  Use __FreeBSD__ instead.

26 years agoThe incorrect select() timeout calculation that I fixed in svc_tcp.c
Bill Paul [Thu, 21 May 1998 15:22:39 +0000 (15:22 +0000)]
The incorrect select() timeout calculation that I fixed in svc_tcp.c
also exists here (the timeout can expire much sooner than it's supposed
to).

26 years agoSync with sys/i386/conf/GENERIC revision 1.109.
KATO Takenori [Thu, 21 May 1998 09:22:03 +0000 (09:22 +0000)]
Sync with sys/i386/conf/GENERIC revision 1.109.

26 years agoAdded contributed entry for Linux consoles.
Joseph Koshy [Thu, 21 May 1998 09:10:55 +0000 (09:10 +0000)]
Added contributed entry for Linux consoles.

PR: 5977
Submitted by: Christian Weisberger <naddy@bigeye.rhein-neckar.de>

26 years agoCosmetic fix. Compute header field widths from UT_{HOST,LINE,NAME}SIZE
Joseph Koshy [Thu, 21 May 1998 08:46:48 +0000 (08:46 +0000)]
Cosmetic fix.  Compute header field widths from UT_{HOST,LINE,NAME}SIZE
instead of hardcoding header line.

PR: 6325

26 years agoMake flushing dirty pages work correctly on filesystems that
John Dyson [Thu, 21 May 1998 07:47:58 +0000 (07:47 +0000)]
Make flushing dirty pages work correctly on filesystems that
unexpectedly do not complete writes even with sync I/O requests.
This should help the behavior of mmaped files when using
softupdates (and perhaps in other circumstances also.)

26 years agoNote presence of the ThunderLAN driver
Jordan K. Hubbard [Thu, 21 May 1998 04:12:03 +0000 (04:12 +0000)]
Note presence of the ThunderLAN driver

26 years agoAnd entries for ThunderLAN driver.
Bill Paul [Thu, 21 May 1998 03:22:20 +0000 (03:22 +0000)]
And entries for ThunderLAN driver.

26 years agoAdd entry for ThunderLAN driver to files.
Bill Paul [Thu, 21 May 1998 03:21:34 +0000 (03:21 +0000)]
Add entry for ThunderLAN driver to files.

26 years agoAdd man page for ThunderLAN driver.
Bill Paul [Thu, 21 May 1998 03:20:39 +0000 (03:20 +0000)]
Add man page for ThunderLAN driver.

26 years agoAdd Texas Instruments TNET100 'ThunderLAN' PCI NIC driver to the tree.
Bill Paul [Thu, 21 May 1998 03:19:56 +0000 (03:19 +0000)]
Add Texas Instruments TNET100 'ThunderLAN' PCI NIC driver to the tree.
This driver supports the following cards/integrated ethernet controllers:

Compaq Netelligent 10, Compaq Netelligent 10/100, Compaq Netelligent 10/100,
Compaq Netelligent 10/100 Proliant, Compaq Netelligent 10/100 Dual Port,
Compaq NetFlex-3/P Integrated, Compaq NetFlex-3/P Integrated,
Compaq NetFlex 3/P w/ BNC, Compaq Deskpro 4000 5233MMX.

It should also support Texas Instruments NICs that use the ThunderLAN
chip, though I don't have any to test. If you've got a card that uses
the ThunderLAN chip but isn't listed in the PCI vendor/product list in
if_tl.c, try adding it and see what happens.

The driver supports any MII compliant PHY at 10 or 100Mbps speeds in
full or half duplex. (Those I've personally tested are the National
Semiconductor DP83840A (Prosignia server), the Level 1 LXT970 (Deskpro
desktop), and the ThunderLAN's internal 10baseT PHY.) Autonegotiation,
hardware multicast filtering, BPF and ifmedia support are included.

This chip is pretty fast; Prosignia servers with NCR SCSI, ThunderLAN
ethernet and FreeBSD make for a nice combination.

26 years agoDon't accept IPCP packets before NETWORK phase.
Brian Somers [Thu, 21 May 1998 01:26:10 +0000 (01:26 +0000)]
Don't accept IPCP packets before NETWORK phase.
Pointed out by: Ralf Rutherford <ralf@philips.oz.au>

26 years agoo Add `set autoload'. You can now set the minimum and maximum
Brian Somers [Thu, 21 May 1998 01:13:32 +0000 (01:13 +0000)]
o Add `set autoload'.  You can now set the minimum and maximum
  thresholds (in terms of queued packets for a period of time)
  where -auto links will be brought up and down.  By default,
  all auto links come up when we reach NETWORK phase and never
  go down.
o Display current autoload state in `show bundle'.
o Disable the idle timer as soon as it's called.
o Disable the idle and autoload timers when exiting (in case
  we're abending).

26 years agoo Give a dialback example.
Brian Somers [Thu, 21 May 1998 01:12:38 +0000 (01:12 +0000)]
o Give a dialback example.
  Submitted by: Kenneth Furge <kenneth.furge@us.endress.com>
o Set MRRU in multilink server mode.

26 years agoo Set utmp::ut_host to getenv("CONNECT"). Mgetty sets
Brian Somers [Thu, 21 May 1998 01:12:20 +0000 (01:12 +0000)]
o Set utmp::ut_host to getenv("CONNECT").  Mgetty sets
  this to the connection speed.
  Submitted by: forgotten (sorry)
o Use all UT_LINESIZE characters rather than leaving the
  last one NUL.

26 years agoBacked out last fix and fixed my typo:
David Greenman [Thu, 21 May 1998 00:33:16 +0000 (00:33 +0000)]
Backed out last fix and fixed my typo:
ipflow(fastforward -> ipflow_fastforward

26 years agoComplete back-out of the bsd.locale.mk change.
Jordan K. Hubbard [Wed, 20 May 1998 22:27:18 +0000 (22:27 +0000)]
Complete back-out of the bsd.locale.mk change.
OK'd by: asami

26 years agoBack out locale change until it's better understood by all just
Jordan K. Hubbard [Wed, 20 May 1998 22:24:48 +0000 (22:24 +0000)]
Back out locale change until it's better understood by all just
what it is we're trying to do here. :)
OK'd by: asami

26 years agoFix typo in prompt.
James Raynard [Wed, 20 May 1998 22:23:16 +0000 (22:23 +0000)]
Fix typo in prompt.

26 years agoBack out previous commit - it breaks when called from 'buildworld' :-(
Eivind Eklund [Wed, 20 May 1998 20:54:10 +0000 (20:54 +0000)]
Back out previous commit - it breaks when called from 'buildworld' :-(
(Damn, I wanted that in the -snap).

26 years agoMove __FreeBSD_version into <sys/param.h> (with automated duplication
Eivind Eklund [Wed, 20 May 1998 20:35:42 +0000 (20:35 +0000)]
Move __FreeBSD_version into <sys/param.h> (with automated duplication
to <osreldate.h>).  This allow kernel drivers access to it.

Approved by: -current

26 years agoChange the sanity test here. It's not correct to assume that the record
Bill Paul [Wed, 20 May 1998 15:56:11 +0000 (15:56 +0000)]
Change the sanity test here. It's not correct to assume that the record
size we receive here should fit into the receive buffer. Unfortunately,
there's no 100% foolproof way to distinguish a ridiculously large record
size that a client actually meant to send us from a ridiculously large
record size that was sent as a spoof attempt.

The one value that we can positively identify as bogus is zero. A
zero-sized record makes absolutely no sense, and sending an endless
supply of zeroes will cause the server to loop forever trying to
fill its receive buffer.

Note that the changes made to readtcp() make it okay to revert this
sanity test since the deadlock case where a client can keep the server
occupied forever in the readtcp() select() loop can't happen anymore.
This solution is not ideal, but is relatively easy to implement. The
ideal solution would be to re-arrange the way dispatching is handled
so that the select() loop in readtcp() can be eliminated, but this is
difficult to implement. I do plan to implement the complete solution
eventually but in the meantime I don't want to leave the RPC library
totally vulnerable.

That you very much Sun, may I have another.

26 years agoAdd missing close paren
Peter Dufault [Wed, 20 May 1998 14:08:43 +0000 (14:08 +0000)]
Add missing close paren

26 years agoSync with sys/i386/isa/sio.c revision 1.202.
KATO Takenori [Wed, 20 May 1998 13:39:06 +0000 (13:39 +0000)]
Sync with sys/i386/isa/sio.c revision 1.202.

26 years agoSync with sys/i386/isa/clock.c revision 1.120.
KATO Takenori [Wed, 20 May 1998 13:38:42 +0000 (13:38 +0000)]
Sync with sys/i386/isa/clock.c revision 1.120.

26 years agoAdd mid scripts. Mid is a tool which create a Message-ID database
Wolfram Schneider [Wed, 20 May 1998 09:20:02 +0000 (09:20 +0000)]
Add mid scripts. Mid is a tool which create a Message-ID database
for mailing lists.

26 years agoOnly ignore "owner" permissions selectively rather than always. In some
Peter Wemm [Wed, 20 May 1998 09:05:48 +0000 (09:05 +0000)]
Only ignore "owner" permissions selectively rather than always.  In some
cases we ignore it (eg: read/write) to maintain chmod-after-open semantics
but in other cases we do care, eg: creating files, access() etc.  Never
ignore errors from VOP_ACCESS() on immutable files.

This apparently comes from BSDI (from Keith Bostic) via NetBSD.

PR: 5148
Submitted by: Yoshiro MIHIRA <sanpei@yy.cs.keio.ac.jp>

26 years agos/flags/flag/
Peter Wemm [Wed, 20 May 1998 08:05:45 +0000 (08:05 +0000)]
s/flags/flag/

26 years agoA cleaner fix for PR#5102, clear nonsense flags at mount time rather than
Peter Wemm [Wed, 20 May 1998 08:02:24 +0000 (08:02 +0000)]
A cleaner fix for PR#5102, clear nonsense flags at mount time rather than
in the core of nfs_bio.c at the 11th hour.

PR: 5102

26 years agoDon't change argp->flags after it's been copied.
Peter Wemm [Wed, 20 May 1998 07:59:21 +0000 (07:59 +0000)]
Don't change argp->flags after it's been copied.

26 years agoLoadSoftModem() routine at sio.c does not trap general serial I/Os.
Poul-Henning Kamp [Wed, 20 May 1998 06:46:58 +0000 (06:46 +0000)]
LoadSoftModem() routine at sio.c does not trap general serial I/Os.
It fauls to probe eather DSI Modem or others.

PR: 4657
Reviewed by: phk
Submitted by: Kenji Saito <marukun@mx2.nisiq.net>

26 years agoAdd rcsid.
Matthew Hunt [Wed, 20 May 1998 05:43:01 +0000 (05:43 +0000)]
Add rcsid.

26 years agoIf LOCALE is not defined, default to USA. Otherwise, bsd.locale.mk
Matthew Hunt [Wed, 20 May 1998 05:41:20 +0000 (05:41 +0000)]
If LOCALE is not defined, default to USA.  Otherwise, bsd.locale.mk
blows up with errors.

26 years agoAdd bsd.locale.mk to the list of files. Whether bsd.locale.mk is
Matthew Hunt [Wed, 20 May 1998 04:57:16 +0000 (04:57 +0000)]
Add bsd.locale.mk to the list of files.  Whether bsd.locale.mk is
the right solution or not, bsd.port.mk is broken unless bsd.locale.mk
is installed.

Note that if LOCALE is not defined, port-building explodes:

"/usr/share/mk/bsd.locale.mk", line 135: if-less elif
"/usr/share/mk/bsd.locale.mk", line 135: Need an operator
(For each .if testing LOCALE)

26 years agoAlpha MAKEDEV as directly imported from NetBSD-current. This will
Jordan K. Hubbard [Wed, 20 May 1998 03:05:05 +0000 (03:05 +0000)]
Alpha MAKEDEV as directly imported from NetBSD-current.  This will
almost certainly have to change as the FreeBSD/Alpha kernel comes
along, but for the transitionary period this is the most correct.
Obtained from: NetBSD

26 years agoAdd some place-holder files here to keep the distribution target
Jordan K. Hubbard [Wed, 20 May 1998 03:01:42 +0000 (03:01 +0000)]
Add some place-holder files here to keep the distribution target
from blowing up.  The disktab file isn't necessarily invalid either
for a PCI Alpha.

26 years agoDon't pack distfiles left over from docs builds into ports.tgz.
Jordan K. Hubbard [Wed, 20 May 1998 01:06:32 +0000 (01:06 +0000)]
Don't pack distfiles left over from docs builds into ports.tgz.

26 years agoChange the description of where to get the Soft Updates files.
Julian Elischer [Tue, 19 May 1998 23:23:09 +0000 (23:23 +0000)]
Change the description of where to get the Soft Updates files.

26 years agoEnsure that there is enough information here, so that people can use
Julian Elischer [Tue, 19 May 1998 23:18:37 +0000 (23:18 +0000)]
Ensure that there is enough information here, so that people can use
soft updates should they desire.

26 years agoBring up-to-date with Whistle's current version
Julian Elischer [Tue, 19 May 1998 23:07:25 +0000 (23:07 +0000)]
Bring up-to-date with Whistle's current version
Includes some debugging code.

26 years agoDon't bring the interface down with the last link when we're
Brian Somers [Tue, 19 May 1998 23:05:10 +0000 (23:05 +0000)]
Don't bring the interface down with the last link when we're
in -auto mode (broken a few commits ago).

26 years agoMerge with Kirk's version as of Feb 20
Julian Elischer [Tue, 19 May 1998 22:54:53 +0000 (22:54 +0000)]
Merge with Kirk's version as of Feb 20

His version 9.23 == our version 1.5 of ffs_softdep.c
His version 9.5 ==  our version 1.4 of softdep.c

26 years agoClose some file descriptors.
Brian Somers [Tue, 19 May 1998 21:51:24 +0000 (21:51 +0000)]
Close some file descriptors.
Pointed at by: Ruslan Ermilov <ru@ucb.crimea.ua>

26 years agoClose forgotten descriptor.
Brian Somers [Tue, 19 May 1998 21:49:54 +0000 (21:49 +0000)]
Close forgotten descriptor.
PR: 6690
Submitted by: Ruslan Ermilov <ru@ucb.crimea.ua>

26 years agoMerge in Kirk's changes to stop softupdates from hogging all of memory.
Julian Elischer [Tue, 19 May 1998 21:45:53 +0000 (21:45 +0000)]
Merge in Kirk's changes to stop softupdates from hogging all of memory.

26 years ago1. Add new defs for mins and maxs for the POSIX flavor priorities. They
Peter Dufault [Tue, 19 May 1998 21:11:53 +0000 (21:11 +0000)]
1. Add new defs for mins and maxs for the POSIX flavor priorities.  They
end up being the same, but it doesn't look like you're comparing
apples and oranges.

2. Use need_resched instead of reset_priority.  This isn't right
either, since for example you'll round-robin against equal priority FIFO
processes when lowering the priority of another process,
but this works better and a real fix needs to be in kern_synch and
not out here.

3. This is not a device driver: copyin/copyout the structure.

26 years agoRemove option for SCHED_FIFO. With this optional, SCHED_FIFO
Peter Dufault [Tue, 19 May 1998 20:59:07 +0000 (20:59 +0000)]
Remove option for SCHED_FIFO.  With this optional,  SCHED_FIFO
is the same as RTPRIO_IDLE when it falls through to the default.

26 years agoTreat RTP_PRIO_FIFO the same as RTP_PRIO_REALTIME for status display.
Peter Dufault [Tue, 19 May 1998 20:52:31 +0000 (20:52 +0000)]
Treat RTP_PRIO_FIFO the same as RTP_PRIO_REALTIME for status display.

26 years agoChange to stop a silly panic. This should be understood better.
Julian Elischer [Tue, 19 May 1998 20:50:41 +0000 (20:50 +0000)]
Change to stop a silly panic. This should be understood better.
Change a buffer swizzle trick to a bcopy. It would be nice if the efficient
trick could be used in the future.

26 years agoSupport integer constant unsigned-suffix and long-long-suffix.
Robert Nordier [Tue, 19 May 1998 20:41:20 +0000 (20:41 +0000)]
Support integer constant unsigned-suffix and long-long-suffix.

26 years agoFirst published FreeBSD version of soft updates Feb 5.
Julian Elischer [Tue, 19 May 1998 20:18:42 +0000 (20:18 +0000)]
First published FreeBSD version of soft updates Feb 5.

26 years agoThis commit was generated by cvs2svn to compensate for changes in r36206,
Julian Elischer [Tue, 19 May 1998 20:03:29 +0000 (20:03 +0000)]
This commit was generated by cvs2svn to compensate for changes in r36206,
which included commits to RCS files with non-trunk default branches.

26 years agoImport the next version received from kirk after some
Julian Elischer [Tue, 19 May 1998 20:03:29 +0000 (20:03 +0000)]
Import the next version received from kirk after some
FreeBSD feedback.

26 years agoImport the next version received from kirk after some
Julian Elischer [Tue, 19 May 1998 20:03:29 +0000 (20:03 +0000)]
Import the next version received from kirk after some
FreeBSD feedback.

26 years agoImport the next version received from kirk after some
Julian Elischer [Tue, 19 May 1998 20:03:29 +0000 (20:03 +0000)]
Import the next version received from kirk after some
FreeBSD feedback.

26 years agoo Indicate which commands require context and which have optional
Brian Somers [Tue, 19 May 1998 19:58:21 +0000 (19:58 +0000)]
o Indicate which commands require context and which have optional
  context in the `help' text.
o Remove some redundant code.
o Fix some comments.

26 years agoThis commit was generated by cvs2svn to compensate for changes in r36201,
Julian Elischer [Tue, 19 May 1998 19:47:22 +0000 (19:47 +0000)]
This commit was generated by cvs2svn to compensate for changes in r36201,
which included commits to RCS files with non-trunk default branches.

26 years agoImport the earliest version of the soft update code that I have.
Julian Elischer [Tue, 19 May 1998 19:47:22 +0000 (19:47 +0000)]
Import the earliest version of the soft update code that I have.

26 years agoImport the earliest version of the soft update code that I have.
Julian Elischer [Tue, 19 May 1998 19:47:22 +0000 (19:47 +0000)]
Import the earliest version of the soft update code that I have.

26 years agoImport the earliest version of the soft update code that I have.
Julian Elischer [Tue, 19 May 1998 19:47:22 +0000 (19:47 +0000)]
Import the earliest version of the soft update code that I have.

26 years agoMissing parens caused cpu features not to be printed for cyrix >= M2/MX.
Peter Wemm [Tue, 19 May 1998 19:40:45 +0000 (19:40 +0000)]
Missing parens caused cpu features not to be printed for cyrix >= M2/MX.
Althought the comments say the datasheet doesn't list the device ID
registers on the M2/MX, they seem to be there and quite alive.
(It's interesting to note that the M2/MX calls itself a 686 class cpu but
 is missing a heck of a lot of features, including VME, PGE, PSE, etc)

26 years agoChange a data type internal to the timecounters, and remove the "delta"
Poul-Henning Kamp [Tue, 19 May 1998 18:55:02 +0000 (18:55 +0000)]
Change a data type internal to the timecounters, and remove the "delta"
function.

Reviewed, but not entirely approved by: bde

26 years agoChange a data type internal to the timecounters, and remove the "delta"
Poul-Henning Kamp [Tue, 19 May 1998 18:48:30 +0000 (18:48 +0000)]
Change a data type internal to the timecounters, and remove the "delta"
function.

Reviewed, but not entirely approved by: bde

26 years agoRedo the previous commit in a more Bruce-friendly fashion.
Poul-Henning Kamp [Tue, 19 May 1998 18:42:09 +0000 (18:42 +0000)]
Redo the previous commit in a more Bruce-friendly fashion.

Urged by: bde

26 years agoFix a typo-bug in ipflow_reap that could cause a NULL pointer
John Polstra [Tue, 19 May 1998 17:47:11 +0000 (17:47 +0000)]
Fix a typo-bug in ipflow_reap that could cause a NULL pointer
dereference.  I have also sent this fix to Matt Thomas.

26 years agoMake "netstat -s" display stats on fast-forwarded packets.
Pierre Beyssac [Tue, 19 May 1998 16:00:55 +0000 (16:00 +0000)]
Make "netstat -s" display stats on fast-forwarded packets.
Obtained from: NetBSD code

26 years agoMove (private) struct ipflow out of ip_var.h, to reduce dependencies
Pierre Beyssac [Tue, 19 May 1998 15:53:50 +0000 (15:53 +0000)]
Move (private) struct ipflow out of ip_var.h, to reduce dependencies
(for ipfw for example) on internal implementation details.
Add $Id$ where missing.

26 years agoMoved #define of IPFLOW_HASHBITS to ip_flow.c where I think it belongs.
David Greenman [Tue, 19 May 1998 14:19:03 +0000 (14:19 +0000)]
Moved #define of IPFLOW_HASHBITS to ip_flow.c where I think it belongs.

26 years agoAdded fast IP forwarding code by Matt Thomas <matt@3am-software.com> via
David Greenman [Tue, 19 May 1998 14:04:36 +0000 (14:04 +0000)]
Added fast IP forwarding code by Matt Thomas <matt@3am-software.com> via
NetBSD, ported to FreeBSD by Pierre Beyssac <pb@fasterix.freenix.org> and
minorly tweaked by me.
This is a standard part of FreeBSD, but must be enabled with:
"sysctl -w net.inet.ip.fastforwarding=1" ...and of course forwarding must
also be enabled. This should probably be modified to use the zone
allocator for speed and space efficiency. The current algorithm also
appears to lose if the number of active paths exceeds IPFLOW_MAX (256),
in which case it wastes lots of time trying to figure out which cache
entry to drop.

26 years agoPR: misc/6031
Peter Hawkins [Tue, 19 May 1998 13:14:11 +0000 (13:14 +0000)]
PR: misc/6031
Patch applied as per PR - enables preferential fetch from local mirrors

26 years agoPR: misc/6031
Peter Hawkins [Tue, 19 May 1998 13:02:04 +0000 (13:02 +0000)]
PR: misc/6031

Patch applied as per PR - enables preferential fetch from local mirrors

26 years agoSync with sys/i386/i386/machdep.c revision 1.296.
KATO Takenori [Tue, 19 May 1998 12:58:05 +0000 (12:58 +0000)]
Sync with sys/i386/i386/machdep.c revision 1.296.

26 years agoReminded by: Alex Nash
Daniel O'Callaghan [Tue, 19 May 1998 12:50:35 +0000 (12:50 +0000)]
Reminded by: Alex Nash
Bring man page up to date with -q flag behaviour.

26 years agoRemove a bogus dependency on the size of the inkernel msgbuf.
Poul-Henning Kamp [Tue, 19 May 1998 12:02:41 +0000 (12:02 +0000)]
Remove a bogus dependency on the size of the inkernel msgbuf.
This change is likely to introduce a few linebreaks in the boot
messages, but that is not easy to solve without breaking syslogd
semantics.  Maybe the right  fix is to return an integral number
of lines from the kernel driver.

Noticed by: dg

26 years agoAdd new NetBSD FFS with FDISK partition ID 0xa9
David E. O'Brien [Tue, 19 May 1998 11:15:44 +0000 (11:15 +0000)]
Add new NetBSD FFS with FDISK partition ID 0xa9

Obtained from: Frank van der Linden <frank@wins.uva.nl>

26 years agoAdd "show msgbuf" command
Poul-Henning Kamp [Tue, 19 May 1998 11:02:24 +0000 (11:02 +0000)]
Add "show msgbuf" command

26 years agoMake the size of the msgbuf (dmesg) a "normal" option.
Poul-Henning Kamp [Tue, 19 May 1998 08:58:53 +0000 (08:58 +0000)]
Make the size of the msgbuf (dmesg) a "normal" option.

26 years agoSupport changing the attribute cache limits per-mount. We don't have
Peter Wemm [Tue, 19 May 1998 07:18:28 +0000 (07:18 +0000)]
Support changing the attribute cache limits per-mount.  We don't have
many option letters left, I used long names only (like the previous
port= option)

26 years agoMake the previous commit compile..
Peter Wemm [Tue, 19 May 1998 07:13:21 +0000 (07:13 +0000)]
Make the previous commit compile..

26 years agoAllow control of the attribute cache timeouts at mount time.
Peter Wemm [Tue, 19 May 1998 07:11:27 +0000 (07:11 +0000)]
Allow control of the attribute cache timeouts at mount time.

We had run out of bits in the nfs mount flags, I have moved the internal
state flags into a seperate variable.  These are no longer visible via
statfs(), but I don't know of anything that looks at them.

26 years agoImprove description of absolute modes.
Joseph Koshy [Tue, 19 May 1998 06:24:50 +0000 (06:24 +0000)]
Improve description of absolute modes.

Submitted by: Josh Gilliam <josh@quick.net>
PR: 6634

26 years agocosmetic: clean up startup messages and rearrange some options
Jordan K. Hubbard [Tue, 19 May 1998 04:36:31 +0000 (04:36 +0000)]
cosmetic: clean up startup messages and rearrange some options
to go in a more proper order.

26 years agoUse the correct name of the login.conf(5) capability (`passwordperiod' ->
Joseph Koshy [Tue, 19 May 1998 03:48:07 +0000 (03:48 +0000)]
Use the correct name of the login.conf(5) capability (`passwordperiod' ->
`passwordtime').

26 years agoReviewed by: ahasty@freebsd.org
Amancio Hasty [Tue, 19 May 1998 03:38:31 +0000 (03:38 +0000)]
Reviewed by: ahasty@freebsd.org
Submitted by: Roger Hardiman <roger@cs.strath.ac.uk>
options BROOKTREE_SYSTEM_DEFAULT=BROOKTREE_PAL
in the kernel config file makes the driver's video_open() function
select PAL rather than NTSC. This fixed all the hangs on my
Dual Crystal card when using a PAL video signal.

As a result, you can loose the tsleep (of 2 seconds - now 0.25!!)
which I previously added. (Unless someone else wanted the 0.25
second tsleep).

26 years agoRemoved reference to obsolete file "README.iBCS2".
Joseph Koshy [Tue, 19 May 1998 03:31:34 +0000 (03:31 +0000)]
Removed reference to obsolete file "README.iBCS2".

Submitted by: Max Euston <max@jmrodgers.com>
PR: 6679

26 years agoTypo fix.
Masafumi Max NAKANE [Tue, 19 May 1998 03:10:14 +0000 (03:10 +0000)]
Typo fix.

26 years agoBack out part of revision 1.198 commit (clearing kernel stack pages).
Tor Egge [Tue, 19 May 1998 00:06:35 +0000 (00:06 +0000)]
Back out part of revision 1.198 commit (clearing kernel stack pages).
By request from David Greenman <dg@root.com>

26 years agoDisallow reading the current kernel stack. Only the user structure and
Tor Egge [Tue, 19 May 1998 00:00:14 +0000 (00:00 +0000)]
Disallow reading the current kernel stack. Only the user structure and
the current registers should be accessible.
Reviewed by: David Greenman <dg@root.com>

26 years agoTalk about multilink ppp
Brian Somers [Mon, 18 May 1998 23:24:24 +0000 (23:24 +0000)]
Talk about multilink ppp

26 years agoToss publickey.c in the attic and remove it from the Makefile.
Bill Paul [Mon, 18 May 1998 21:59:53 +0000 (21:59 +0000)]
Toss publickey.c in the attic and remove it from the Makefile.
If we have it in libc, we don't need it here.

26 years agoReplace the getpublickey() stub with the real thing.
Bill Paul [Mon, 18 May 1998 21:59:15 +0000 (21:59 +0000)]
Replace the getpublickey() stub with the real thing.

26 years agoPlug hole reported on Bugtraq: do not allow mmap with WRITE privs for
Guido van Rooij [Mon, 18 May 1998 18:26:27 +0000 (18:26 +0000)]
Plug hole reported on Bugtraq: do not allow mmap with WRITE  privs for
append-only and immutable files.

Obtained from: OpenBSD (partly)

26 years agoGrumble...It seems I'm suffering from some mental disease. Do it correct now.
Guido van Rooij [Mon, 18 May 1998 17:11:24 +0000 (17:11 +0000)]
Grumble...It seems I'm suffering from some mental disease. Do it correct now.

26 years agoAdd some parenthesis for clarity and fix a bug
Guido van Rooij [Mon, 18 May 1998 17:07:58 +0000 (17:07 +0000)]
Add some parenthesis for clarity and fix a bug
Pointed out by: Garrett Wollmand

26 years agoAdd rules to generate PDF files.
Wolfram Schneider [Mon, 18 May 1998 16:57:10 +0000 (16:57 +0000)]
Add rules to generate PDF files.

26 years agoImprove DoS avoidance in RPC stream oriented transports. The TCP transport
Bill Paul [Mon, 18 May 1998 16:12:13 +0000 (16:12 +0000)]
Improve DoS avoidance in RPC stream oriented transports. The TCP transport
uses readtcp() to gather data from the network; readtcp() uses select(),
with a timeout of 35 seconds. The problem with this is that if you
connect to a TCP server, send two bytes of data, then just pause, the
server will remain blocked in readtcp() for up to 35 seconds, which is
sort of a long time. If you keep doing this every 35 seconds, you can
keep the server occupied indefinitely.

To fix this, I modified readtcp() (and its cousin, readunix() in svc_unix.c)
to monitor all service transport handles instead of just the current socket.
This allows the server to keep handling new connections that arrive while
readtcp() is running. This prevents one client from potentially monopolizing
a server.

Also, while I was here, I fixed a bug in the timeout calculations. Someone
attempted to adjust the timeout so that if select() returned EINTR and the
loop was restarted, the timeout would be reduced so that rather than waiting
for another 35 seconds, you could never wait for more than 35 seconds total.
Unfortunately, the calculation was wrong, and the timeout could expire much
sooner than 35 seconds.