]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
24 years agoremove cc1obj from the original SUBDIR list since it's made optional
marcel [Sun, 28 Nov 1999 12:30:20 +0000 (12:30 +0000)]
remove cc1obj from the original SUBDIR list since it's made optional
in the previous commit resulting.

Forgotten by: obrien

24 years agoDefault to not -v.
obrien [Sun, 28 Nov 1999 12:24:16 +0000 (12:24 +0000)]
Default to not -v.
Fix usage() style bug spotted by BDE.

24 years agoScheduler fixes equivalent to the ones logged in the following NetBSD
bde [Sun, 28 Nov 1999 12:12:14 +0000 (12:12 +0000)]
Scheduler fixes equivalent to the ones logged in the following NetBSD
commit to kern_synch.c:

----------------------------
revision 1.55
date: 1999/02/23 02:56:03;  author: ross;  state: Exp;  lines: +39 -10
Scheduler bug fixes and reorganization
* fix the ancient nice(1) bug, where nice +20 processes incorrectly
  steal 10 - 20% of the CPU, (or even more depending on load average)
* provide a new schedclk() mechanism at a new clock at schedhz, so high
  platform hz values don't cause nice +0 processes to look like they are
  niced
* change the algorithm slightly, and reorganize the code a lot
* fix percent-CPU calculation bugs, and eliminate some no-op code

=== nice bug === Correctly divide the scheduler queues between niced and
compute-bound processes. The current nice weight of two (sort of, see
`algorithm change' below) neatly divides the USRPRI queues in half; this
should have been used to clip p_estcpu, instead of UCHAR_MAX.  Besides
being the wrong amount, clipping an unsigned char to UCHAR_MAX is a no-op,
and it was done after decay_cpu() which can only _reduce_ the value.  It
has to be kept <= NICE_WEIGHT * PRIO_MAX - PPQ or processes can
scheduler-penalize themselves onto the same queue as nice +20 processes.
(Or even a higher one.)

=== New schedclk() mechansism === Some platforms should be cutting down
stathz before hitting the scheduler, since the scheduler algorithm only
works right in the vicinity of 64 Hz. Rather than prescale hz, then scale
back and forth by 4 every time p_estcpu is touched (each occurance an
abstraction violation), use p_estcpu without scaling and require schedhz
to be generated directly at the right frequency. Use a default stathz (well,
actually, profhz) / 4, so nothing changes unless a platform defines schedhz
and a new clock.  Define these for alpha, where hz==1024, and nice was
totally broke.

=== Algorithm change === The nice value used to be added to the
exponentially-decayed scheduler history value p_estcpu, in _addition_ to
be incorporated directly (with greater wieght) into the priority calculation.
At first glance, it appears to be a pointless increase of 1/8 the nice
effect (pri = p_estcpu/4 + nice*2), but it's actually at least 3x that
because it will ramp up linearly but be decayed only exponentially, thus
converging to an additional .75 nice for a loadaverage of one. I killed
this, it makes the behavior hard to control, almost impossible to analyze,
and the effect (~~nothing at for the first second, then somewhat increased
niceness after three seconds or more, depending on load average) pointless.

=== Other bugs === hz -> profhz in the p_pctcpu = f(p_cpticks) calcuation.
Collect scheduler functionality. Try to put each abstraction in just one
place.
----------------------------

The details are a little different in FreeBSD:

=== nice bug ===   Fixing this is the main point of this commit.  We use
essentially the same clipping rule as NetBSD (our limit on p_estcpu
differs by a scale factor).  However, clipping at all is fundamentally
bad.  It gives free CPU the hoggiest hogs once they reach the limit, and
reaching the limit is normal for long-running hogs.  This will be fixed
later.

=== New schedclk() mechanism ===  We don't use the NetBSD schedclk()
(now schedclock()) mechanism.  We require (real)stathz to be about 128
and scale by an extra factor of 2 compared with NetBSD's statclock().
We scale p_estcpu instead of scaling the clock.  This is more accurate
and flexible.

=== Algorithm change ===  Same change.

=== Other bugs ===  The p_pctcpu bug was fixed long ago.  We don't try as
hard to abstract functionality yet.

Related changes: the new limit on p_estcpu must be exported to kern_exit.c
for clipping in wait1().

Agreed with by: dufault

24 years agoScheduler fixes equivalent to the ones logged in the following NetBSD
bde [Sun, 28 Nov 1999 12:12:13 +0000 (12:12 +0000)]
Scheduler fixes equivalent to the ones logged in the following NetBSD
commit to kern_synch.c:

  ----------------------------
  revision 1.55
  date: 1999/02/23 02:56:03;  author: ross;  state: Exp;  lines: +39 -10
  Scheduler bug fixes and reorganization
  * fix the ancient nice(1) bug, where nice +20 processes incorrectly
    steal 10 - 20% of the CPU, (or even more depending on load average)
  * provide a new schedclk() mechanism at a new clock at schedhz, so high
    platform hz values don't cause nice +0 processes to look like they are
    niced
  * change the algorithm slightly, and reorganize the code a lot
  * fix percent-CPU calculation bugs, and eliminate some no-op code

  === nice bug === Correctly divide the scheduler queues between niced and
  compute-bound processes. The current nice weight of two (sort of, see
  `algorithm change' below) neatly divides the USRPRI queues in half; this
  should have been used to clip p_estcpu, instead of UCHAR_MAX.  Besides
  being the wrong amount, clipping an unsigned char to UCHAR_MAX is a no-op,
  and it was done after decay_cpu() which can only _reduce_ the value.  It
  has to be kept <= NICE_WEIGHT * PRIO_MAX - PPQ or processes can
  scheduler-penalize themselves onto the same queue as nice +20 processes.
  (Or even a higher one.)

  === New schedclk() mechansism === Some platforms should be cutting down
  stathz before hitting the scheduler, since the scheduler algorithm only
  works right in the vicinity of 64 Hz. Rather than prescale hz, then scale
  back and forth by 4 every time p_estcpu is touched (each occurance an
  abstraction violation), use p_estcpu without scaling and require schedhz
  to be generated directly at the right frequency. Use a default stathz (well,
  actually, profhz) / 4, so nothing changes unless a platform defines schedhz
  and a new clock.  Define these for alpha, where hz==1024, and nice was
  totally broke.

  === Algorithm change === The nice value used to be added to the
  exponentially-decayed scheduler history value p_estcpu, in _addition_ to
  be incorporated directly (with greater wieght) into the priority calculation.
  At first glance, it appears to be a pointless increase of 1/8 the nice
  effect (pri = p_estcpu/4 + nice*2), but it's actually at least 3x that
  because it will ramp up linearly but be decayed only exponentially, thus
  converging to an additional .75 nice for a loadaverage of one. I killed
  this, it makes the behavior hard to control, almost impossible to analyze,
  and the effect (~~nothing at for the first second, then somewhat increased
  niceness after three seconds or more, depending on load average) pointless.

  === Other bugs === hz -> profhz in the p_pctcpu = f(p_cpticks) calcuation.
  Collect scheduler functionality. Try to put each abstraction in just one
  place.
  ----------------------------

The details are a little different in FreeBSD:

=== nice bug ===   Fixing this is the main point of this commit.  We use
essentially the same clipping rule as NetBSD (our limit on p_estcpu
differs by a scale factor).  However, clipping at all is fundamentally
bad.  It gives free CPU the hoggiest hogs once they reach the limit, and
reaching the limit is normal for long-running hogs.  This will be fixed
later.

=== New schedclk() mechanism ===  We don't use the NetBSD schedclk()
(now schedclock()) mechanism.  We require (real)stathz to be about 128
and scale by an extra factor of 2 compared with NetBSD's statclock().
We scale p_estcpu instead of scaling the clock.  This is more accurate
and flexible.

=== Algorithm change ===  Same change.

=== Other bugs ===  The p_pctcpu bug was fixed long ago.  We don't try as
hard to abstract functionality yet.

Related changes: the new limit on p_estcpu must be exported to kern_exit.c
for clipping in wait1().

Agreed with by: dufault

24 years agoTake a shot at implementing the fix for PR 15014 for the a.out kernel
peter [Sun, 28 Nov 1999 12:06:29 +0000 (12:06 +0000)]
Take a shot at implementing the fix for PR 15014 for the a.out kernel
linker as well.

PR: 15014
Submitted by: Vladimir N. Silyaev <vns@delta.odessa.ua>

24 years agoFix an embarresing mistake in the kld symbol lookup for DDB. It should
peter [Sun, 28 Nov 1999 11:59:18 +0000 (11:59 +0000)]
Fix an embarresing mistake in the kld symbol lookup for DDB.  It should
now correctly do a traceback when crashing inside a KLD module.

PR: 15014
Submitted by: Vladimir N. Silyaev <vns@delta.odessa.ua>

24 years agobrucify
mharo [Sun, 28 Nov 1999 09:34:21 +0000 (09:34 +0000)]
brucify

Reviewed by: obrien

24 years agoSync with sys/dev/syscons/syscons.c revision 1.327.
nyan [Sun, 28 Nov 1999 08:57:22 +0000 (08:57 +0000)]
Sync with sys/dev/syscons/syscons.c revision 1.327.

24 years agoSync with sys/i386/conf/GENERIC revision 1.205.
nyan [Sun, 28 Nov 1999 08:55:01 +0000 (08:55 +0000)]
Sync with sys/i386/conf/GENERIC revision 1.205.

24 years ago- Fail VT switching when the controlling program, such as the
yokota [Sun, 28 Nov 1999 07:44:11 +0000 (07:44 +0000)]
- Fail VT switching when the controlling program, such as the
  X server, is not responding to the VT switching protocol. (This part
  of the code has been somewhat wrong in -CURRENT, but -STABLE has the
  correct code...)

24 years ago- Added Estonian keymaps: estonian.iso.kbd, estonian.iso15.kbd,
yokota [Sun, 28 Nov 1999 07:38:19 +0000 (07:38 +0000)]
- Added Estonian keymaps: estonian.iso.kbd, estonian.iso15.kbd,
  and estonian.cp850.kbd.

PR: conf/14658
Submitted by: priit@mig.ee

24 years agocommit (re)generated files last.
imp [Sun, 28 Nov 1999 05:50:45 +0000 (05:50 +0000)]
commit (re)generated files last.

24 years agoUpdate pccard code to latest NetBSD code. This is the last merge
imp [Sun, 28 Nov 1999 05:49:27 +0000 (05:49 +0000)]
Update pccard code to latest NetBSD code.  This is the last merge
before newbusification hits full steam ahead.

All:
Adjust NetBSD labels to reflect new base versions.
dev/pcic/i82365.c:
1.24 Interface change for kernel threads
1.25 Massive unification for cardbus
dev/pcic/i82365var.h
1.8 Massive unification for cardbus
dev/pcic/i82365_isasubr.c
1.3 Massive unification for cardbus
dev/pccard/pccard_cis.c
1.11 Massive unification for cardbus
(better device printing, better memspace calcs)
dev/pccard/pccard_cis_quirks.c
1.4,1.5 Lotsa 3com devices
dev/pccard/pccardchip.h
1.4 Massive unification for cardbus
dev/pccard/pccarddevs
1.33..1.59 Lots of devices

24 years agoadd pthread_cancel, obtained from OpenBSD.
alfred [Sun, 28 Nov 1999 05:38:13 +0000 (05:38 +0000)]
add pthread_cancel, obtained from OpenBSD.

eischen (Daniel Eischen) added wrappers to protect against cancled
threads orphaning internal resources.

the cancelability code is still a bit fuzzy but works for test
programs of my own, OpenBSD's and some examples from ORA's books.

add readdir_r to both libc and libc_r

add some 'const' attributes to function parameters

Reviewed by: eischen, jasone

24 years agoThis commit was generated by cvs2svn to compensate for changes in r53809,
obrien [Sun, 28 Nov 1999 01:35:29 +0000 (01:35 +0000)]
This commit was generated by cvs2svn to compensate for changes in r53809,
which included commits to RCS files with non-trunk default branches.

24 years agoUpdate to the 0.12.0-19991127 patch + my header path & doc changes.
obrien [Sun, 28 Nov 1999 01:35:29 +0000 (01:35 +0000)]
Update to the 0.12.0-19991127 patch + my header path & doc changes.

24 years agoArange the f* floating point instructions to match binutils 2.9.1.
obrien [Sun, 28 Nov 1999 01:11:35 +0000 (01:11 +0000)]
Arange the f* floating point instructions to match binutils 2.9.1.

24 years agoAdd "fild" instruction. GCC 2.95.2 likes to generate this one.
obrien [Sun, 28 Nov 1999 01:05:35 +0000 (01:05 +0000)]
Add "fild" instruction.  GCC 2.95.2 likes to generate this one.

Obtained from: contrib/binutils/include/opcode/i386.h

24 years agoSort PCI SCSI controlers.
obrien [Sun, 28 Nov 1999 00:59:03 +0000 (00:59 +0000)]
Sort PCI SCSI controlers.

24 years ago/sys adjustments to add the `sym' controler driver.
obrien [Sun, 28 Nov 1999 00:48:15 +0000 (00:48 +0000)]
/sys adjustments to add the `sym' controler driver.

This is commented out in GENERIC as you cannot mix `sym' with `ncr' right now.
Note that LINT is no more broken by this commit.

24 years agoAdjust `sym' include file path to match where `sym' landed in our tree.
obrien [Sun, 28 Nov 1999 00:45:27 +0000 (00:45 +0000)]
Adjust `sym' include file path to match where `sym' landed in our tree.

Done as vendor import as I hope the author will accept this patch, and
I'm not ready to take this thing off the vendor branch yet.

24 years agoThis commit was generated by cvs2svn to compensate for changes in r53801,
obrien [Sun, 28 Nov 1999 00:45:27 +0000 (00:45 +0000)]
This commit was generated by cvs2svn to compensate for changes in r53801,
which included commits to RCS files with non-trunk default branches.

24 years agoAdjust `sym' include file path to match where `sym' landed in our tree.
obrien [Sun, 28 Nov 1999 00:39:09 +0000 (00:39 +0000)]
Adjust `sym' include file path to match where `sym' landed in our tree.

Done as vendor import as I hope the author will accept this patch, and
I'm not ready to take this thing off the vendor branch yet.

24 years agoThis commit was generated by cvs2svn to compensate for changes in r53799,
obrien [Sun, 28 Nov 1999 00:39:09 +0000 (00:39 +0000)]
This commit was generated by cvs2svn to compensate for changes in r53799,
which included commits to RCS files with non-trunk default branches.

24 years agoUpdate to the SYM-0.11.0-19991120 patch.
obrien [Sat, 27 Nov 1999 23:37:15 +0000 (23:37 +0000)]
Update to the SYM-0.11.0-19991120 patch.

Submitted by: Gerard Roudier <groudier@club-internet.fr>

24 years agoThis commit was generated by cvs2svn to compensate for changes in r53796,
obrien [Sat, 27 Nov 1999 23:37:15 +0000 (23:37 +0000)]
This commit was generated by cvs2svn to compensate for changes in r53796,
which included commits to RCS files with non-trunk default branches.

24 years agoUpdate to the SYM-0.10.0-19991111 patch.
obrien [Sat, 27 Nov 1999 23:35:46 +0000 (23:35 +0000)]
Update to the SYM-0.10.0-19991111 patch.

Submitted by: Gerard Roudier <groudier@club-internet.fr>

24 years agoThis commit was generated by cvs2svn to compensate for changes in r53793,
obrien [Sat, 27 Nov 1999 23:35:46 +0000 (23:35 +0000)]
This commit was generated by cvs2svn to compensate for changes in r53793,
which included commits to RCS files with non-trunk default branches.

24 years agoNew `sym' device driver optimized for the Symbios/LSI 53C896/53C895A/53C1010
obrien [Sat, 27 Nov 1999 23:32:35 +0000 (23:32 +0000)]
New `sym' device driver optimized for the Symbios/LSI 53C896/53C895A/53C1010
PCI SCSI controllers.  This driver also supports the following Symbios/LSI
PCI SCSI chips: 53C810A53C825A, 53C860, 53C875, 53C876, 53C885, 53C895.

However, it does NOT support earlier chips as the following ones: 53C810,
53C815, 53C825.

See README.sym for more details.

Submitted-by: Gerard Roudier <groudier@club-internet.fr>

24 years agoThis commit was generated by cvs2svn to compensate for changes in r53790,
obrien [Sat, 27 Nov 1999 23:32:35 +0000 (23:32 +0000)]
This commit was generated by cvs2svn to compensate for changes in r53790,
which included commits to RCS files with non-trunk default branches.

24 years agoAdd a commented out 'ATA' driver config block to help assist -CURRENT
obrien [Sat, 27 Nov 1999 23:25:17 +0000 (23:25 +0000)]
Add a commented out 'ATA' driver config block to help assist -CURRENT
people to migrate to this driver since it will be the default IDE/ATA/ATAPI
driver in 4.0-R.

24 years agooptions should be formatted as "#options ^IFOO".
obrien [Sat, 27 Nov 1999 22:46:51 +0000 (22:46 +0000)]
options should be formatted as "#options ^IFOO".

Spammed by: sos, mjacob, and phk

24 years agoMake ObjC bits optional. NO_OBJC ==> don't need `em.
obrien [Sat, 27 Nov 1999 21:52:55 +0000 (21:52 +0000)]
Make ObjC bits optional.  NO_OBJC ==> don't need `em.

24 years agoMake the prompt look like on OpenBoot.
dcs [Sat, 27 Nov 1999 21:44:47 +0000 (21:44 +0000)]
Make the prompt look like on OpenBoot.

24 years agoAdd support for the AMI MegaRAID and Mylex drivers to sysinstall.
msmith [Sat, 27 Nov 1999 21:33:04 +0000 (21:33 +0000)]
Add support for the AMI MegaRAID and Mylex drivers to sysinstall.
There are reports that installs to these controllers still don't
work, but this is at least one step closer.

24 years agoRemove 'sd' support. SCSI disks are known as 'da' these days.
msmith [Sat, 27 Nov 1999 21:20:57 +0000 (21:20 +0000)]
Remove 'sd' support.  SCSI disks are known as 'da' these days.

24 years agoUpdate compiler entries.
obrien [Sat, 27 Nov 1999 21:18:19 +0000 (21:18 +0000)]
Update compiler entries.
Note the 100% total death of /dev/*sd*.

24 years agoAs promised long ago, remove the last evidence of the 'sd' driver.
msmith [Sat, 27 Nov 1999 21:17:02 +0000 (21:17 +0000)]
As promised long ago, remove the last evidence of the 'sd' driver.
SCSI disks are only known as 'da' now.

24 years agoMake fsck even more char/blk dev tolerant.
phk [Sat, 27 Nov 1999 20:02:27 +0000 (20:02 +0000)]
Make fsck even more char/blk dev tolerant.

24 years agoAdd "-v".
obrien [Sat, 27 Nov 1999 19:25:08 +0000 (19:25 +0000)]
Add "-v".

24 years agoMake it possible to make release on archaic kernels.
phk [Sat, 27 Nov 1999 18:48:10 +0000 (18:48 +0000)]
Make it possible to make release on archaic kernels.

24 years agoZeroes structure before using it.
dcs [Sat, 27 Nov 1999 18:31:57 +0000 (18:31 +0000)]
Zeroes structure before using it.

24 years agoRemap the error EEXISTS => 0 *before* using error to determine if we should
eivind [Sat, 27 Nov 1999 18:14:41 +0000 (18:14 +0000)]
Remap the error EEXISTS => 0 *before* using error to determine if we should
return a vp.

24 years agoPass the pppoe packets into do_ppp_print correctly.
brian [Sat, 27 Nov 1999 18:04:57 +0000 (18:04 +0000)]
Pass the pppoe packets into do_ppp_print correctly.

24 years agoRemove abuse of struct nameidata.
bp [Sat, 27 Nov 1999 17:46:04 +0000 (17:46 +0000)]
Remove abuse of struct nameidata.

Pointed by: Eivind Eklund

24 years agoRemove incorrect section name. Terminate .Bl with .El.
charnier [Sat, 27 Nov 1999 17:15:51 +0000 (17:15 +0000)]
Remove incorrect section name. Terminate .Bl with .El.

24 years agoRemove incorrect section name.
charnier [Sat, 27 Nov 1999 17:13:08 +0000 (17:13 +0000)]
Remove incorrect section name.

24 years agoName of program and trailing \n will be added by syslog(3)
charnier [Sat, 27 Nov 1999 17:11:55 +0000 (17:11 +0000)]
Name of program and trailing \n will be added by syslog(3)

24 years agoUse DIAGNOSTICS instead of incorrect section name
charnier [Sat, 27 Nov 1999 17:10:35 +0000 (17:10 +0000)]
Use DIAGNOSTICS instead of incorrect section name

24 years agoRemove incorrect section name
charnier [Sat, 27 Nov 1999 17:09:14 +0000 (17:09 +0000)]
Remove incorrect section name

24 years agoAdd section name to .Xr references.
charnier [Sat, 27 Nov 1999 17:07:59 +0000 (17:07 +0000)]
Add section name to .Xr references.

24 years agoRemove incorrect section name. Incomplete -Wall cleaning.
charnier [Sat, 27 Nov 1999 17:06:40 +0000 (17:06 +0000)]
Remove incorrect section name. Incomplete -Wall cleaning.

24 years agocall ata_reinit in addump trying to make sure the dump will succed,
sos [Sat, 27 Nov 1999 17:06:10 +0000 (17:06 +0000)]
call ata_reinit in addump trying to make sure the dump will succed,
also check for errors underways, and make it possible to abort the
dump by hitting a key.

24 years agoDo not repeat the name of the flag. Change alloc to allocate for better
charnier [Sat, 27 Nov 1999 17:05:08 +0000 (17:05 +0000)]
Do not repeat the name of the flag. Change alloc to allocate for better
spelling of printed messages.

24 years agoDisable on x86 since it conflicts with PNPBIOS.
dfr [Sat, 27 Nov 1999 17:03:08 +0000 (17:03 +0000)]
Disable on x86 since it conflicts with PNPBIOS.

24 years agoMerge fprintf and exit into errx.
charnier [Sat, 27 Nov 1999 17:03:07 +0000 (17:03 +0000)]
Merge fprintf and exit into errx.
Use .Ev for environment variable.

24 years ago.Fn -> .Xr
charnier [Sat, 27 Nov 1999 16:59:06 +0000 (16:59 +0000)]
.Fn -> .Xr
Do not start warn() message with uppercase.

24 years ago.Nm corrections
charnier [Sat, 27 Nov 1999 16:57:27 +0000 (16:57 +0000)]
.Nm corrections

24 years agoTerminate .Bl with .El
charnier [Sat, 27 Nov 1999 16:55:50 +0000 (16:55 +0000)]
Terminate .Bl with .El

24 years agoImplement linux_ustat.
marcel [Sat, 27 Nov 1999 16:55:14 +0000 (16:55 +0000)]
Implement linux_ustat.

Reviewed by: bde

24 years agoRemove incorrect section name. Use .Pa or .Ar instead of .Em.
charnier [Sat, 27 Nov 1999 16:54:36 +0000 (16:54 +0000)]
Remove incorrect section name. Use .Pa or .Ar instead of .Em.

24 years agoPut file names under .Pa.
charnier [Sat, 27 Nov 1999 16:52:41 +0000 (16:52 +0000)]
Put file names under .Pa.

24 years agoChange prototype of linux_ustat.
marcel [Sat, 27 Nov 1999 16:50:54 +0000 (16:50 +0000)]
Change prototype of linux_ustat.

24 years agoAllow root-reloading also for chr devices.
phk [Sat, 27 Nov 1999 16:10:31 +0000 (16:10 +0000)]
Allow root-reloading also for chr devices.

24 years agoAdd ftp.nz.freebsd.org to the list of available FTP sites.
roberto [Sat, 27 Nov 1999 16:02:08 +0000 (16:02 +0000)]
Add ftp.nz.freebsd.org to the list of available FTP sites.

Submitted by: "Dan Langille" <dan@freebsddiary.org>

24 years agoUpdated comments for the move in the previous commit.
bde [Sat, 27 Nov 1999 15:27:11 +0000 (15:27 +0000)]
Updated comments for the move in the previous commit.

24 years agoFixed some comments in statclock(). The previous commit made it clearer
bde [Sat, 27 Nov 1999 14:37:34 +0000 (14:37 +0000)]
Fixed some comments in statclock().  The previous commit made it clearer
that one comment was attached to null code.

24 years agoRemove bad144 program.
phk [Sat, 27 Nov 1999 14:35:22 +0000 (14:35 +0000)]
Remove bad144 program.

24 years agoRemove BAD144 support
phk [Sat, 27 Nov 1999 14:33:30 +0000 (14:33 +0000)]
Remove BAD144 support

24 years agoRemove BAD144 support.
phk [Sat, 27 Nov 1999 14:33:07 +0000 (14:33 +0000)]
Remove BAD144 support.

24 years agoRemove bad144 from floppy images.
phk [Sat, 27 Nov 1999 14:32:58 +0000 (14:32 +0000)]
Remove bad144 from floppy images.

24 years agoCreate a symbolic link in the ftp directory, so that the user can
phk [Sat, 27 Nov 1999 14:32:27 +0000 (14:32 +0000)]
Create a symbolic link in the ftp directory, so that the user can
enter the URL directly to the FTP directory without worrying about
the BUILD_NAME.

Remove some extra quoting in a kernel config file.

24 years agoMoved scheduling-related code to kern_synch.c so that it is easier to fix
bde [Sat, 27 Nov 1999 12:32:27 +0000 (12:32 +0000)]
Moved scheduling-related code to kern_synch.c so that it is easier to fix
and extend.  The new function containing the code is named schedclock()
as in NetBSD, but it has slightly different semantics (it already handles
incrementation of p->p_cpticks, and it should handle any calling frequency).

Agreed with in principle by: dufault

24 years agoSync with sys/i386/conf/GENERIC revision 1.202.
nyan [Sat, 27 Nov 1999 11:41:46 +0000 (11:41 +0000)]
Sync with sys/i386/conf/GENERIC revision 1.202.

24 years agoReplace the block major table with a notice why it isn't needed any more.
phk [Sat, 27 Nov 1999 10:45:31 +0000 (10:45 +0000)]
Replace the block major table with a notice why it isn't needed any more.

Mark /dev/drum as historic.

24 years agoMinor typo nit. Also, remove the redundant entry from fortunes2.
wes [Sat, 27 Nov 1999 07:18:33 +0000 (07:18 +0000)]
Minor typo nit.  Also, remove the redundant entry from fortunes2.

PR: bin/14239
Submitted by: Doug Barton <Doug@gorean.org>

24 years agoAdd another vendor ID(CTL0048) for SB32 PnP.
tanimura [Sat, 27 Nov 1999 06:33:27 +0000 (06:33 +0000)]
Add another vendor ID(CTL0048) for SB32 PnP.

Submitted by: Charles Anderson <caa@columbus.rr.com>

24 years agoAllow extended pap success messages by believing in the PAP headers
brian [Fri, 26 Nov 1999 22:44:45 +0000 (22:44 +0000)]
Allow extended pap success messages by believing in the PAP headers
length field rather than the one byte message length field embedded
in the packet.  This steps slightly outside of the protocol boundaries,
but should not cause any problems.

Limitation noted by: Simon Winwood <simon@winwood.org>

24 years agoChange ``set cd'' so that its default value is device specific. The
brian [Fri, 26 Nov 1999 22:44:33 +0000 (22:44 +0000)]
Change ``set cd'' so that its default value is device specific.  The
default is still 1 second for ttys, but is now 6 seconds for i4b (ISDN)
devices and 5 seconds for ethernet (PPPoE) devices.

24 years agoA bullet in the head to pkg_manage references.
billf [Fri, 26 Nov 1999 22:12:12 +0000 (22:12 +0000)]
A bullet in the head to pkg_manage references.

24 years agoThis fell out of a previous commit: create the opt_md.h and opt_mfs.h
phk [Fri, 26 Nov 1999 21:01:03 +0000 (21:01 +0000)]
This fell out of a previous commit: create the opt_md.h and opt_mfs.h
files for the md module.

24 years agoDo not create bdev nodes anymore. Instead create cdev nodes with the
phk [Fri, 26 Nov 1999 20:45:26 +0000 (20:45 +0000)]
Do not create bdev nodes anymore.  Instead create cdev nodes with the
same names:

        rover# ls -l /dev/*wd0a
        crw-r-----  1 root  operator    3,   0 Nov 26 20:20 /dev/rwd0a
        crw-r-----  1 root  operator    3,   0 Nov 26 20:20 /dev/wd0a

Notice: Over time, no earlier than FreeBSD 5.0, the "r*" names may
be discontinued.  A fair number of programs and scripts need to
(un)learn some tricks before then.

This will take no effect until you either run MAKEDEV by hand or
reinstall your system.

WARNING:  Kernels older than approx November 22 will not be happy
about a /dev created with MAKEDEV after this commit.  Please update
your /kernel.good etc.

24 years agoAdd poul's md information, in abbreviated form, to the UPDATING file.
imp [Fri, 26 Nov 1999 20:38:36 +0000 (20:38 +0000)]
Add poul's md information, in abbreviated form, to the UPDATING file.

24 years agoThe author of this is REALLY: Bruce A. Mah <bmah@CS.Sandia.GOV>
billf [Fri, 26 Nov 1999 20:38:18 +0000 (20:38 +0000)]
The author of this is REALLY: Bruce A. Mah <bmah@CS.Sandia.GOV>

24 years agoAdd the author as co-maintainer, a la mergemaster.
billf [Fri, 26 Nov 1999 20:35:55 +0000 (20:35 +0000)]
Add the author as co-maintainer, a la mergemaster.

24 years agoWelcome pkg_version into the FreeBSD source tree.
billf [Fri, 26 Nov 1999 20:31:58 +0000 (20:31 +0000)]
Welcome pkg_version into the FreeBSD source tree.

pkg_version (as you may well know) matches the existing packages/ports
installed on your system with the ports INDEX and reports which
ports differ from the current INDEX.

Submitted by: Bruce A. Mah <CA.Sandia.GOV>
Reviewed by: ports

24 years agoRetire MFS_ROOT and MFS_ROOT_SIZE options from the MFS implementation.
phk [Fri, 26 Nov 1999 20:08:44 +0000 (20:08 +0000)]
Retire MFS_ROOT and MFS_ROOT_SIZE options from the MFS implementation.

Add MD_ROOT and MD_ROOT_SIZE options to the md driver.

Make the md driver handle MFS_ROOT and MFS_ROOT_SIZE options for compatibility.

Add md driver to GENERIC, PCCARD and LINT.

This is a cleanup which removes the need for some of the worse hacks in
MFS:  We really want to have a rootvnode but MFS on a preloaded image
doesn't really have one.  md is a true device, so it is less trouble.

This has been tested with make release, and if people remember to add
the "md" pseudo-device to their kernels, PicoBSD should be just fine
as well.  If people have no other use for MFS, it can be removed from
the kernel.

24 years agoOops, committed the wrong file..
sos [Fri, 26 Nov 1999 20:05:11 +0000 (20:05 +0000)]
Oops, committed the wrong file..

24 years agoFix atapi_wait_ready, it returned prematurely.
sos [Fri, 26 Nov 1999 19:26:04 +0000 (19:26 +0000)]
Fix atapi_wait_ready, it returned prematurely.

Less verbosity on non-fatal errors.

24 years agoTidy up the VIA support a bit, make it more generic.
sos [Fri, 26 Nov 1999 19:24:26 +0000 (19:24 +0000)]
Tidy up the VIA support a bit, make it more generic.

24 years agoAdd id for ESS 1869 OEM to Compaq on Presario 1621.
dcs [Fri, 26 Nov 1999 18:58:03 +0000 (18:58 +0000)]
Add id for ESS 1869 OEM to Compaq on Presario 1621.

24 years agoRTFREE the correct route entry in dummynet_io(). The previous
luigi [Fri, 26 Nov 1999 13:37:09 +0000 (13:37 +0000)]
RTFREE the correct route entry in dummynet_io(). The previous
code failed in handling things like "forward" actions.

Reported-and-tested-by: Jean-Hugues ROYER jhroyer@joher.com
24 years agos/dir.h/dir.c/
phk [Fri, 26 Nov 1999 12:29:48 +0000 (12:29 +0000)]
s/dir.h/dir.c/

24 years agoSync with sys/i386/conf/files.i386 revision 1.281.
nyan [Fri, 26 Nov 1999 11:47:01 +0000 (11:47 +0000)]
Sync with sys/i386/conf/files.i386 revision 1.281.

24 years agoImplement fdatasync in terms of fsync. The regeneration of proto.h,
marcel [Fri, 26 Nov 1999 10:20:10 +0000 (10:20 +0000)]
Implement fdatasync in terms of fsync. The regeneration of proto.h,
syscall.h and sysent.h was probably forgotten after the last change
syscalls.master.

24 years agoTo avoid confusion, zap libtermcap. ncurses provides both curses, termcap
peter [Fri, 26 Nov 1999 09:33:02 +0000 (09:33 +0000)]
To avoid confusion, zap libtermcap.  ncurses provides both curses, termcap
and termlib (terminfo) support.

24 years agoTo avoid confusion, zap libcurses. ncurses provides both curses, termcap
peter [Fri, 26 Nov 1999 09:28:33 +0000 (09:28 +0000)]
To avoid confusion, zap libcurses.  ncurses provides both curses, termcap
and termlib (terminfo) support.

24 years agoAdd a sysctl to control if argv is disclosed to the world:
phk [Fri, 26 Nov 1999 08:27:16 +0000 (08:27 +0000)]
Add a sysctl to control if argv is disclosed to the world:
kern.ps_argsopen
It defaults to 1 which means that all users can see all argvs in ps(1).

Reviewed by: Warner

24 years agoBelatedly add splash_pcx_load to the documented variables. Reword
dcs [Fri, 26 Nov 1999 08:09:04 +0000 (08:09 +0000)]
Belatedly add splash_pcx_load to the documented variables. Reword
splash_bmp_load.

24 years agoAdd note about how hardwiring ep0 will not likely work.
imp [Fri, 26 Nov 1999 06:36:13 +0000 (06:36 +0000)]
Add note about how hardwiring ep0 will not likely work.

Also correct a small spelling nit from Bill Fumerola.