]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoUpdate to PicoBSD 0.44 from RELENG_3
roger [Mon, 14 Jun 1999 15:45:56 +0000 (15:45 +0000)]
Update to PicoBSD 0.44 from RELENG_3

25 years agoUpdate to PicoBSD 0.44 from RELENG_
roger [Mon, 14 Jun 1999 15:45:41 +0000 (15:45 +0000)]
Update to PicoBSD 0.44 from RELENG_

25 years agoUpdated the BUGS section. BREAK is now implemented.
bde [Mon, 14 Jun 1999 15:10:35 +0000 (15:10 +0000)]
Updated the BUGS section.  BREAK is now implemented.

Don't document non-bugs in the BUGS section, or anywhere else.  It
is not a bug to drop data when overloaded.  The compile-time tuning
options turned out to be not very useful, and aren't supported
offically.

Documented the not so new option CY_PCI_FASTINTR.

25 years agoMan pages for top refer to only 2.x, but not to 3.x and later
n_hibma [Mon, 14 Jun 1999 12:06:11 +0000 (12:06 +0000)]
Man pages for top refer to only 2.x, but not to 3.x and later

25 years agoFix typo confusing `fa' and 'fb' registers.
simokawa [Mon, 14 Jun 1999 11:42:30 +0000 (11:42 +0000)]
Fix typo confusing `fa' and 'fb' registers.
Now, MATLAB works fine under OSF/1 emulation.

25 years agoCome up with something useful when someone searches for
brian [Mon, 14 Jun 1999 10:12:41 +0000 (10:12 +0000)]
Come up with something useful when someone searches for
``dynamic IP''.

25 years ago1) remove the soft reset after a command failed.
n_hibma [Sun, 13 Jun 1999 22:09:14 +0000 (22:09 +0000)]
1) remove the soft reset after a command failed.
2) remove printing of errno and errstr

25 years agoEliminate the bogus procfs private almost struct dirent structure.
phk [Sun, 13 Jun 1999 20:53:16 +0000 (20:53 +0000)]
Eliminate the bogus procfs private almost struct dirent structure.

Spotted by: Lars Hamren
Reviewed by: bde

25 years ago1) Add non-blocking I/O on read
n_hibma [Sun, 13 Jun 1999 20:49:12 +0000 (20:49 +0000)]
1) Add non-blocking I/O on read
2) Add checks to make sure we do not dereference the softc when not found
because the device has been removed.

25 years ago1) Add the USB controllers (will now print pretty strings when no usb
n_hibma [Sun, 13 Jun 1999 20:48:29 +0000 (20:48 +0000)]
1) Add the USB controllers (will now print pretty strings when no usb
   support is compiled in)
2) Add probing for generic USB host controllers as well so we get them all
3) make the returned strings look alike in the whole file

25 years agoShorten the strings a bit (remove 'Host')
n_hibma [Sun, 13 Jun 1999 20:46:10 +0000 (20:46 +0000)]
Shorten the strings a bit (remove 'Host')

25 years agoFix placement of signal trampoline.
dt [Sun, 13 Jun 1999 19:41:50 +0000 (19:41 +0000)]
Fix placement of signal trampoline.

25 years agoUse pmap_kenter instead of pmap_enter to map the message buffer.
alc [Sun, 13 Jun 1999 19:20:25 +0000 (19:20 +0000)]
Use pmap_kenter instead of pmap_enter to map the message buffer.

25 years agoRemoved checking for unit>NBKTR.
roger [Sun, 13 Jun 1999 16:05:00 +0000 (16:05 +0000)]
Removed checking for unit>NBKTR.
No longer need when using device_get_softc.
(also fixed bug where loadable bktr module could only access 1st device)

25 years agoAdd bktr_load for the loadable Bt848/Bt878 driver.
roger [Sun, 13 Jun 1999 15:45:49 +0000 (15:45 +0000)]
Add bktr_load  for the loadable Bt848/Bt878 driver.

25 years agovm_map_protect:
alc [Sat, 12 Jun 1999 23:10:38 +0000 (23:10 +0000)]
vm_map_protect:
The wrong vm_map_entry is used to determine if writes must not be
allowed due to COW.

25 years agoAdd bktr (Bt848/Bt878 driver loadable module) to the Makefile i386 only section
roger [Sat, 12 Jun 1999 15:11:33 +0000 (15:11 +0000)]
Add bktr (Bt848/Bt878 driver loadable module) to the Makefile i386 only section

25 years agoChange perl5 to perl (same change as made in the fxp module)
roger [Sat, 12 Jun 1999 15:08:50 +0000 (15:08 +0000)]
Change perl5 to perl (same change as made in the fxp module)

25 years agoMakefile to build a loadable bktr Bt848/Bt878 device driver
roger [Sat, 12 Jun 1999 15:00:19 +0000 (15:00 +0000)]
Makefile to build a loadable bktr Bt848/Bt878 device driver
This allows the bt848 grabber to be loaded in the boot loader or
after bootup without the need for a custom kernel

25 years agoUpdated to use the new 4.x newbus API
roger [Sat, 12 Jun 1999 14:54:56 +0000 (14:54 +0000)]
Updated to use the new 4.x newbus API
Also removed the BSDI support (for now)
This allows the driver to be loaded/unloaded as a KLD
and loaded in the boot loader phase whithout making a custom kernel.

25 years agoFix a minor style nit in the NAME section.
mpp [Fri, 11 Jun 1999 16:26:29 +0000 (16:26 +0000)]
Fix a minor style nit in the NAME section.

25 years agoMove the description of the strncmp() function from the RETURN VALUES
mpp [Fri, 11 Jun 1999 16:16:32 +0000 (16:16 +0000)]
Move the description of the strncmp() function from the RETURN VALUES
section to the DESCRIPTION section.

PR: 12133
Submitted by: Sheldon Hearn <sheldonh@uunet.co.za>

25 years agoShow the correct error if we fail to open a device.
brian [Fri, 11 Jun 1999 13:28:29 +0000 (13:28 +0000)]
Show the correct error if we fail to open a device.

25 years agoDon't accept divert/tee/pipe rules without corresponding option.
ru [Fri, 11 Jun 1999 11:27:35 +0000 (11:27 +0000)]
Don't accept divert/tee/pipe rules without corresponding option.

PR: 10324
Reviewed by: luigi

25 years agoDescribe why a ``magic'' enddisc is not always a good idea.
brian [Fri, 11 Jun 1999 11:27:18 +0000 (11:27 +0000)]
Describe why a ``magic'' enddisc is not always a good idea.

25 years agoLarger, space separated fields in the output of repquota.
jkoshy [Fri, 11 Jun 1999 10:31:24 +0000 (10:31 +0000)]
Larger, space separated fields in the output of repquota.

PR: 8730
Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>

25 years agoWorkaround the problem that the first (and only first) port name
ru [Fri, 11 Jun 1999 09:43:53 +0000 (09:43 +0000)]
Workaround the problem that the first (and only first) port name
can't have a dash character (it is treated as a ``range'' operator).
One could now use such a name by escaping the ``-'' characters.
For example:

# ipfw add 1 count tcp from any to any "ms\-sql\-s"
# ipfw add 2 count tcp from any ftp\\-data-ftp to any

PR: 7101

25 years agoCompile when NO_REGEX is defined.
hoek [Fri, 11 Jun 1999 04:36:16 +0000 (04:36 +0000)]
Compile when NO_REGEX is defined.

Submitted by: Geordy Kitchen <gkitchen@cs.ubc.ca> (gnu/12117)

25 years agoImprovement in the description of what this does.
markm [Thu, 10 Jun 1999 22:48:27 +0000 (22:48 +0000)]
Improvement in the description of what this does.

Prompted by (and thanks to): W Richard Stevens

25 years agoReplace my previous fix of saving the FP state with a much simpler one: when
dt [Thu, 10 Jun 1999 20:40:59 +0000 (20:40 +0000)]
Replace my previous fix of saving the FP state with a much simpler one: when
we swap out fpcurproc, save its FP state.

Suggested by: bde

25 years agoFor vendor-specific commands sent using the 'camcontrol cmd' facility,
ken [Thu, 10 Jun 1999 16:14:18 +0000 (16:14 +0000)]
For vendor-specific commands sent using the 'camcontrol cmd' facility,
use the passed in CDB length, not 1.

Submitted by: Jean-Marc Zucconi <jmz@FreeBSD.ORG>

25 years agoRemove 'SET_ASM_OP' macro, which is unused now and rather harmful
simokawa [Thu, 10 Jun 1999 12:05:48 +0000 (12:05 +0000)]
Remove 'SET_ASM_OP' macro, which is unused now and rather harmful
for recent egcs.

25 years agoAllow reserved substitution strings to be escaped by preceeding them
brian [Thu, 10 Jun 1999 09:34:57 +0000 (09:34 +0000)]
Allow reserved substitution strings to be escaped by preceeding them
with a backslash.

25 years agoNever bring a link back up after receiving a terminating signal,
brian [Thu, 10 Jun 1999 09:06:30 +0000 (09:06 +0000)]
Never bring a link back up after receiving a terminating signal,
don't rely on already being in PHASE_DEAD.

25 years agoChange variable used for calculating ending address of physical memory
jlemon [Thu, 10 Jun 1999 02:48:51 +0000 (02:48 +0000)]
Change variable used for calculating ending address of physical memory
from 'int' to 'vm_offset_t'.

Spotted by:  Richard Cownie <tich@ma.ikos.com>

25 years agoDon't delete quota files when clean-tmps is enabled.
brian [Thu, 10 Jun 1999 00:34:15 +0000 (00:34 +0000)]
Don't delete quota files when clean-tmps is enabled.

25 years agoAllow a remote IP and port range specification in the
brian [Thu, 10 Jun 1999 00:17:27 +0000 (00:17 +0000)]
Allow a remote IP and port range specification in the
``alias port'' command.

25 years agoInitialise `mp'.
brian [Wed, 9 Jun 1999 20:27:26 +0000 (20:27 +0000)]
Initialise `mp'.

25 years agoAllow our endpoint discriminator to be enabled, disabled, accepted
brian [Wed, 9 Jun 1999 16:54:04 +0000 (16:54 +0000)]
Allow our endpoint discriminator to be enabled, disabled, accepted
and denied.  This is necessary for some MP implementations that
get confused if you accept their endpoint discriminator but reject
their MRRU.

25 years agoAdd a rather complicated but impressive example of how
brian [Wed, 9 Jun 1999 12:32:55 +0000 (12:32 +0000)]
Add a rather complicated but impressive example of how
to implement multi-link ppp over more than one ISP with
the ability to lose ISPs without loss of connectivity.

It *requires* that you either have administrative access
to a machine that's already connected to the 'net or at
least know a really nice person that does.

25 years agoAdd detection code for NVidia Riva 128,TNT,TNT2
ache [Wed, 9 Jun 1999 11:46:43 +0000 (11:46 +0000)]
Add detection code for NVidia Riva 128,TNT,TNT2
(TNT2 from me)

PR: 12094
Submitted by: Ilya Naumov camel@avias.com

25 years agowhere_arge is allocated on stack and return address is overwritten
jkh [Wed, 9 Jun 1999 09:29:53 +0000 (09:29 +0000)]
where_arge is allocated on stack and return address is overwritten
by buffer overflow.

Submitted by: Hidetoshi Shimokawa <simokawa@sat.t.u-tokyo.ac.jp>

25 years agoRemove apple.com from time server list.
jkh [Wed, 9 Jun 1999 09:18:33 +0000 (09:18 +0000)]
Remove apple.com from time server list.

Submitted by: Mike Haro

25 years agoVarious edits to shrink fixit and enable dhcp-client
jkh [Wed, 9 Jun 1999 09:08:22 +0000 (09:08 +0000)]
Various edits to shrink fixit and enable dhcp-client

25 years agoUse the correct pid when substituting PROCESSID.
brian [Wed, 9 Jun 1999 08:47:36 +0000 (08:47 +0000)]
Use the correct pid when substituting PROCESSID.
Problem reported by: Amedeo Beck Peccoz <gea@gressoney.it>

25 years agolittle optimization of locale sensing code
ache [Wed, 9 Jun 1999 00:07:10 +0000 (00:07 +0000)]
little optimization of locale sensing code

25 years agoDon't use static variables if we don't have to.
brian [Tue, 8 Jun 1999 20:12:06 +0000 (20:12 +0000)]
Don't use static variables if we don't have to.

25 years agoDon't drop the last character from lines in ppp.secret unless it's '\n'.
brian [Tue, 8 Jun 1999 20:11:53 +0000 (20:11 +0000)]
Don't drop the last character from lines in ppp.secret unless it's '\n'.

25 years agoUse kmem_alloc_nofault() rather than kmem_alloc_pageable() to allocate
dt [Tue, 8 Jun 1999 17:14:22 +0000 (17:14 +0000)]
Use kmem_alloc_nofault() rather than kmem_alloc_pageable() to allocate
kernel virtual address space for UPAGES.

25 years agoAdd a function kmem_alloc_nofault() - same as kmem_alloc_pageable(), but
dt [Tue, 8 Jun 1999 17:03:28 +0000 (17:03 +0000)]
Add a function kmem_alloc_nofault() - same as kmem_alloc_pageable(), but
create a nofault entry. It will be used to allocate kmem for upages.

(I am not too happy with all this, but it's better than nothing).

25 years agoKeep fpcurproc locked in memory, so that we always can save the FP state
dt [Tue, 8 Jun 1999 16:42:19 +0000 (16:42 +0000)]
Keep fpcurproc locked in memory, so that we always can save the FP state
correctly.

This should fix the "pmap_changebit didn't" panic that some people see.

Reviewed by: dfr

25 years agoJust ignore malformed locale variable unless -d specified
ache [Tue, 8 Jun 1999 14:59:12 +0000 (14:59 +0000)]
Just ignore malformed locale variable unless -d specified

25 years agoIf amd_flags is empty, don't add -p as it makes amd abend.
brian [Tue, 8 Jun 1999 13:00:30 +0000 (13:00 +0000)]
If amd_flags is empty, don't add -p as it makes amd abend.

25 years agoDon't IPCP TLD if we're already doing it. This prevents
brian [Tue, 8 Jun 1999 11:58:27 +0000 (11:58 +0000)]
Don't IPCP TLD if we're already doing it.  This prevents
recursion by doing something like ``down'' or ``quit all''
in ppp.linkdown.

25 years agoMake the ``load'', ``dial'' and ``open'' commands a big clearer.
brian [Tue, 8 Jun 1999 11:57:59 +0000 (11:57 +0000)]
Make the ``load'', ``dial'' and ``open'' commands a big clearer.
Requested by: Michael Heitmeier <MICHAEL_HEITMEIER@HP-Germany-om12.om.hp.com>

25 years agoChange locale name of Traditional Chinese : zh_TW.BIG5 -> zh_TW.Big5.
foxfair [Tue, 8 Jun 1999 02:25:19 +0000 (02:25 +0000)]
Change locale name of Traditional Chinese : zh_TW.BIG5 -> zh_TW.Big5.
It makes compatible with IANA charset defination, and let existent I18N
app happier.

Ref: <http://www.isi.edu/in-notes/iana/assignments/character-sets>

25 years agoFrom the submitter:
msmith [Mon, 7 Jun 1999 20:37:29 +0000 (20:37 +0000)]
From the submitter:

  - this causes POSIX locking to use the thread group leader
   (p->p_leader) as the locking thread for all advisory locks.
   In non-kernel-threaded code p->p_leader == p, so this will have
   no effect.

   This results in (more) correct POSIX threaded flock-ing semantics.

   It also prevents the leader from exiting before any of the children.
   (so that p->p_leader will never be stale) in exit1().

   We have been running this patch for over a month now in our lab
   under load and at customer sites.

Submitted by: John Plevyak <jplevyak@inktomi.com>

25 years agoFix -T doubling
ache [Mon, 7 Jun 1999 19:28:47 +0000 (19:28 +0000)]
Fix -T doubling

25 years agosolve name conflict
andreas [Mon, 7 Jun 1999 19:25:23 +0000 (19:25 +0000)]
solve name conflict
the old STREAM protocol has been obsoleted by "ST2" (RFC 1819)

Detected by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
Suggested by: Matt Crawford <crawdad@fnal.gov>

25 years agoLocalize NEQN part too
ache [Mon, 7 Jun 1999 19:23:04 +0000 (19:23 +0000)]
Localize NEQN part too

25 years agoAdd ability to read/write localized manpages
ache [Mon, 7 Jun 1999 18:36:00 +0000 (18:36 +0000)]
Add ability to read/write localized manpages

25 years agoksprintn() may be called with base=2, so redefine MAXNBUF accordingly.
archie [Mon, 7 Jun 1999 18:26:26 +0000 (18:26 +0000)]
ksprintn() may be called with base=2, so redefine MAXNBUF accordingly.
Other brucification tweaks.

Obtained from: bde@freebsd.org

25 years agoFixed previous fix (dad -> rda via rad).
bde [Mon, 7 Jun 1999 16:36:41 +0000 (16:36 +0000)]
Fixed previous fix (dad -> rda via rad).

25 years agoSeparate the pftp and FTP_PASSIVE_MODE tests so gate mode works again.
des [Mon, 7 Jun 1999 16:35:15 +0000 (16:35 +0000)]
Separate the pftp and FTP_PASSIVE_MODE tests so gate mode works again.

PR: bin/12070
Submitted by: Dominic Mitchell <Dom.Mitchell@palmerharvey.co.uk>

Specifically check that FTP_PASSIVE_MODE is set to YES, rather than
just checking if it is defined.

Discussed on: freebsd-current

25 years agoEnsure that host_addr (which is returned in struct hostent::h_addr_list[0])
dt [Mon, 7 Jun 1999 16:30:27 +0000 (16:30 +0000)]
Ensure that host_addr (which is returned in struct hostent::h_addr_list[0])
is aligned so that pointer to it can be safely casted to struct in_addr *.

25 years agoChanges from OpenBSD:
kris [Mon, 7 Jun 1999 13:53:57 +0000 (13:53 +0000)]
Changes from OpenBSD:
* Better usage() - correct syntax, display available commands
  instead of examples
* Accept command abbreviations
* sprintf -> snprintf (for paranoia)
* manpage capitalisation tweak

Obtained from: OpenBSD

25 years agoCorrect typos.
jkoshy [Mon, 7 Jun 1999 05:18:24 +0000 (05:18 +0000)]
Correct typos.

PR: docs/12068
Submitted by: Steve Coltrin <spcoltri@io.com>

25 years agoSpell the .St macro correctly.
steve [Mon, 7 Jun 1999 03:59:56 +0000 (03:59 +0000)]
Spell the .St macro correctly.

PR: 11704

25 years agoSpell isinf(3) correctly.
steve [Mon, 7 Jun 1999 02:42:31 +0000 (02:42 +0000)]
Spell isinf(3) correctly.

PR: 11974

25 years agoAdd a clarification in the man page suggesting that some tapes
mjacob [Sun, 6 Jun 1999 23:23:50 +0000 (23:23 +0000)]
Add a clarification in the man page suggesting that some tapes
(QIC) written under 2.X may not be easily read under the current
driver without explicitly setting to variable mode or to the blocksize
these tapes were written under 2.X with.

PR: 6681

25 years agoThe FD options are now in /sys/conf/options so need to duplicate them
steve [Sun, 6 Jun 1999 22:45:04 +0000 (22:45 +0000)]
The FD options are now in /sys/conf/options so need to duplicate them
here and keep people from building new kernels. :)

25 years agofix (non destructive) braino to do with quoting in makefiles.
julian [Sun, 6 Jun 1999 20:21:27 +0000 (20:21 +0000)]
fix (non destructive) braino to do with quoting in makefiles.

25 years agoSmall optimization: use wi_seek() in wi_read_record() and wi_write_record()
wpaul [Sun, 6 Jun 1999 16:44:04 +0000 (16:44 +0000)]
Small optimization: use wi_seek() in wi_read_record() and wi_write_record()
instead of frobbing the buffer access path registers directly. Saves a
few lines of duplicated code.

25 years agoCall do_ctors() and do_dtors() using indirect calls through function
jdp [Sun, 6 Jun 1999 15:59:08 +0000 (15:59 +0000)]
Call do_ctors() and do_dtors() using indirect calls through function
pointers.  The calls are in different sections from the functions
being called, and they can potentially be far away.  On a very large
program, the 21-bit displacement field of the BSR instruction
overflowed at link time.

25 years agoCorrect description of -c and -w options
kris [Sun, 6 Jun 1999 14:16:51 +0000 (14:16 +0000)]
Correct description of -c and -w options

Inspired by: PR docs/11774
Submitted by: Chad R. Larson <chad@freeway.dcfinc.com>

25 years agoUpdate of intro(2)/errno(2) in the ongoing effort of the Programmer's
jseger [Sun, 6 Jun 1999 14:15:14 +0000 (14:15 +0000)]
Update of intro(2)/errno(2) in the ongoing effort of the Programmer's
Documentation Project <http://home.wxs.nl/~asmodai/pdp.html>

Submitted by: Jeroen Ruigrok van der Werven <asmodai@wxs.nl>
PR: docs/12030

25 years agoSince FreeBSD can be used as sniffer, update the /etc/protocols file
andreas [Sun, 6 Jun 1999 14:00:10 +0000 (14:00 +0000)]
Since FreeBSD can be used as sniffer, update the /etc/protocols file
after demand of two people and no objections:
- included all registered protocols, so it's now a useful reference as well
- renamed title to reflect new content
  (deleted word "IP", since all registered internet protocols are included now)
- corrected URL (file isn't hosted by iana.org anymore)

25 years agoSync with sys/i386/conf/majors.i386 revision 1.76.
kato [Sun, 6 Jun 1999 12:50:46 +0000 (12:50 +0000)]
Sync with sys/i386/conf/majors.i386 revision 1.76.

25 years agoSync with sys/boot/i386/libi386/bootinfo.c revision 1.19.
kato [Sun, 6 Jun 1999 12:46:40 +0000 (12:46 +0000)]
Sync with sys/boot/i386/libi386/bootinfo.c revision 1.19.

25 years agoReserve cdev major 126 for sync support.
phk [Sun, 6 Jun 1999 10:41:35 +0000 (10:41 +0000)]
Reserve cdev major 126 for sync support.

25 years agotypo in previous commit
phk [Sun, 6 Jun 1999 09:28:01 +0000 (09:28 +0000)]
typo in previous commit

25 years agoIntroduce IFF_SMART bit.
phk [Sun, 6 Jun 1999 09:17:51 +0000 (09:17 +0000)]
Introduce IFF_SMART bit.

This means that the driver will add/delete routes when it knows it is
up/down, rather than have the generic code belive it is up if configured.

This is probably most useful for serial lines, although many PHY chips
could probably tell us if we're connected to the cable/hub as well.

25 years agoThe -perm option takes argument [-]mode, not [-mode]
kris [Sun, 6 Jun 1999 07:37:18 +0000 (07:37 +0000)]
The -perm option takes argument [-]mode, not [-mode]

Obtained from: NetBSD, by way of OpenBSD

25 years agosprintf -> snprintf
kris [Sun, 6 Jun 1999 07:18:49 +0000 (07:18 +0000)]
sprintf -> snprintf

Obtained from: OpenBSD

25 years agoThe function ksprintn(), which is used to convert numbers to ASCII, is not
archie [Sun, 6 Jun 1999 02:41:55 +0000 (02:41 +0000)]
The function ksprintn(), which is used to convert numbers to ASCII, is not
reentrant because it returns a static buffer. This results in a race condition
when/if an interrupt handler calls log(), printf() etc. Fix this.

25 years agoCorrect the way ppp transfers links on the server side in MP
brian [Sat, 5 Jun 1999 21:36:00 +0000 (21:36 +0000)]
Correct the way ppp transfers links on the server side in MP
mode by padding out the ``struct device'' to the maximum
device size.
Bump the ppp version number to indicate the transfer format
change.

This should make MP over tty and udp devices functional again.

25 years agoStore the start time for later so it can be reprinted at the end.
julian [Sat, 5 Jun 1999 21:06:59 +0000 (21:06 +0000)]
Store the start time for later so it can be reprinted at the end.
Reviewed by: terry

25 years ago- When trying to decide if $EDITOR == "vi", match against only the first
hoek [Sat, 5 Jun 1999 20:27:08 +0000 (20:27 +0000)]
- When trying to decide if $EDITOR == "vi", match against only the first
  two characters of $EDITOR.  This allows things like "vim" and "vi -G"
  (although nvi would fail...oh well).

- Avoid certain cases where the editor is passed an invalid line number.

25 years agovm_mmap:
alc [Sat, 5 Jun 1999 18:21:53 +0000 (18:21 +0000)]
vm_mmap:
Insure that device mappings get MAP_PREFAULT(_PARTIAL) set,
so that 4M page mappings are used when possible.

Reviewed by: Luoqi Chen <luoqi@watermarkgroup.com>

25 years agoFix an accounting problem when prefaulting 4M pages.
luoqi [Sat, 5 Jun 1999 16:16:37 +0000 (16:16 +0000)]
Fix an accounting problem when prefaulting 4M pages.

PR: kern/11948

25 years agoremove getopt & friend declarations, now unistd.h included
ache [Sat, 5 Jun 1999 16:05:40 +0000 (16:05 +0000)]
remove getopt & friend declarations, now unistd.h included

25 years agofix typo preventing to include unistd.h
ache [Sat, 5 Jun 1999 15:59:51 +0000 (15:59 +0000)]
fix typo preventing to include unistd.h

25 years agoFix a width.
ru [Sat, 5 Jun 1999 15:03:47 +0000 (15:03 +0000)]
Fix a width.

25 years agoFloppy driver options moved to conf/options.
dfr [Sat, 5 Jun 1999 13:57:10 +0000 (13:57 +0000)]
Floppy driver options moved to conf/options.

25 years ago* Implement ISA dma (only for CIA now but more chipsets to follow).
dfr [Sat, 5 Jun 1999 13:30:13 +0000 (13:30 +0000)]
* Implement ISA dma (only for CIA now but more chipsets to follow).
* Port the fd driver to alpha.

25 years agoDon't discard error output from sysctl(8).
bde [Sat, 5 Jun 1999 12:06:19 +0000 (12:06 +0000)]
Don't discard error output from sysctl(8).

Do discard standard output from the sysctl for approxy_all, and echo
what this sysctl is doing in the usual way.  This fix is probably
backwards.  We should probably just use the standard sysctl output
in all cases (it needs to have a newline filtered out).

Echo what the sysctls for nfs_reserved_port_only and nfs_access_cache
are doing.

25 years agoWhen incrementing through a SIOCGIFCONF list, enforce a lower limit of
archie [Sat, 5 Jun 1999 05:55:07 +0000 (05:55 +0000)]
When incrementing through a SIOCGIFCONF list, enforce a lower limit of
sizeof(ifr->ifr_addr) for the variable length field ifr->ifr_addr.sa_len.
Otherwise the increment will be wrong in certain cases.

Obtained from:  Whistle source tree
For the record: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> suggests
                SIOCGIFCONF should be dropped in favor of a sysctl mechanism.

25 years agoWhen incrementing through a SIOCGIFCONF list, enforce a lower limit of
archie [Sat, 5 Jun 1999 05:52:52 +0000 (05:52 +0000)]
When incrementing through a SIOCGIFCONF list, enforce a lower limit of
sizeof(ifr->ifr_addr) for the variable length field ifr->ifr_addr.sa_len.
Otherwise the increment will be wrong in certain cases.

Obtained from: Whistle source tree
For the record: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> suggests
SIOCGIFCONF should be dropped in favor of a sysctl mechanism.

25 years agoAdd handle to control global TCP keepalives and turn them on as
phk [Sat, 5 Jun 1999 05:45:57 +0000 (05:45 +0000)]
Add handle to control global TCP keepalives and turn them on as
default.

Despite their name it doesn't keep TCP sessions alive, it kills
them if the other end has gone AWOL.  This happens a lot with
clients which use NAT, dynamic IP assignment or which has a 2^32
* 10^-3 seconds upper bound on their uptime.

There is no detectable increase in network trafic because of this:
two minimal TCP packets every two hours for a live TCP connection.

Many servers already enable keepalives themselves.

The host requirements RFC is 10 years old, and doesn't know about
the loosing clients of todays InterNet.

25 years agoVarious changes lifted from the OpenBSD cvs tree:
peter [Sat, 5 Jun 1999 05:35:03 +0000 (05:35 +0000)]
Various changes lifted from the OpenBSD cvs tree:

txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge.

nfs_serv.c: don't call nfsm_adj() with negative values, windows clients
could crash servers when doing a readdir of a large directory.

nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin
loop trying to bind().  Don't clobber a mbuf pointer or we get panics
on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf.

nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB.

Obtained from:  OpenBSD