]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
28 years agoFix a bogon. cpu_fork & savectx ecpected cpu_switch to restore %eax,
Poul-Henning Kamp [Thu, 18 Apr 1996 21:34:53 +0000 (21:34 +0000)]
Fix a bogon.  cpu_fork & savectx ecpected cpu_switch to restore %eax,
they shouldn't.

28 years ago- Save/restore the FS data segment.
Nate Williams [Thu, 18 Apr 1996 19:22:59 +0000 (19:22 +0000)]
- Save/restore the FS data segment.

Reviewed by: bde

28 years ago- Addition of my name to the APM Copyright
Nate Williams [Thu, 18 Apr 1996 19:21:47 +0000 (19:21 +0000)]
- Addition of my name to the APM Copyright
- More code cleanups
- #ifdef DEBUG debugging code
- More consistant printfs
- Better handling of the apm_int() assembly code (mostly from Bruce Evans)

Reviewed by: bde

28 years agoUpdate an offset field for i386 a.out core files, so that the program
John Polstra [Thu, 18 Apr 1996 19:05:58 +0000 (19:05 +0000)]
Update an offset field for i386 a.out core files, so that the program
name comes out right again.

Note: Don't bring this change directly into the -stable branch.  The
appropriate offset for -stable is different.

28 years agoFix a typo in the syntax for indirect offsets.
John Polstra [Thu, 18 Apr 1996 18:17:14 +0000 (18:17 +0000)]
Fix a typo in the syntax for indirect offsets.

28 years agoThree speed-ups in the output path (two small, one substantial):
Garrett Wollman [Thu, 18 Apr 1996 15:49:06 +0000 (15:49 +0000)]
Three speed-ups in the output path (two small, one substantial):

1) Require all callers to pass a valid route pointer to ip_output()
   so that we don't have to check and allocate one off the stack
   as was done before.  This eliminates one test and some stack
    bloat from the common (UDP and TCP) case.

2) Perform the IP header checksum in-line if it's of the usual length.
    This results in about a 5% speed-up in my packet-generation test.

3) Use ip_vhl field rather than ip_v and ip_hl bitfields.

28 years agoDefine a few macros useful in the _IP_VHL case.
Garrett Wollman [Thu, 18 Apr 1996 15:42:50 +0000 (15:42 +0000)]
Define a few macros useful in the _IP_VHL case.

28 years agoFix a warning by not referencing ip_output() as a pr_output() member.
Garrett Wollman [Thu, 18 Apr 1996 15:41:51 +0000 (15:41 +0000)]
Fix a warning by not referencing ip_output() as a pr_output() member.

28 years agoAlways call ip_output() with a valid route pointer. For igmp, also get the
Garrett Wollman [Thu, 18 Apr 1996 15:41:11 +0000 (15:41 +0000)]
Always call ip_output() with a valid route pointer.  For igmp, also get the
multicast option structure off the stack rather than malloc.

28 years agoAdd an inline function to compute the common-case IP header checksum.
Garrett Wollman [Thu, 18 Apr 1996 15:39:27 +0000 (15:39 +0000)]
Add an inline function to compute the common-case IP header checksum.

28 years agoRemove a long-standing window leak I just noticed.
Jordan K. Hubbard [Thu, 18 Apr 1996 13:36:39 +0000 (13:36 +0000)]
Remove a long-standing window leak I just noticed.

28 years agoRemove the special-case behavior for fire actions that return
Jordan K. Hubbard [Thu, 18 Apr 1996 13:21:26 +0000 (13:21 +0000)]
Remove the special-case behavior for fire actions that return
DITEM_FAILURE - formerly they would simply act as an implicit "continue",
but this is wrong.  If you want this behavior, you should now return
with the DITEM_CONTINUE flag set.

Also make the semantics of DITEM_RESTORE quite a bit different - rather
than restore the screen back to pre-menu state, we restore the menu
itself.  This is more correct for a variety of reasons when dealing with
nested menus (whoops!).

28 years agoUse -X to be xargs-friendly
Andrey A. Chernov [Thu, 18 Apr 1996 10:34:07 +0000 (10:34 +0000)]
Use -X to be xargs-friendly
Check devices too, follow original BSD intention
Find only executable files with s-bits, close PR bin/1022
Reset locale to C to have equal results in any case

28 years agoFix sgetrune/sputrune arg type: was unsigned int instead of size_t
Andrey A. Chernov [Thu, 18 Apr 1996 07:13:42 +0000 (07:13 +0000)]
Fix sgetrune/sputrune arg type: was unsigned int instead of size_t

28 years agoFix error in wcstombs: byte count not counted
Andrey A. Chernov [Thu, 18 Apr 1996 07:01:46 +0000 (07:01 +0000)]
Fix error in wcstombs: byte count not counted
Remove unneded casts in sgetrune/sputrune
Submitted by: wcstombs fix by Mihoko Tanaka <m_tonaka@pa.yokogawa.co.jp>

28 years agoFix usage: message to print out corrrectly.
Nate Williams [Thu, 18 Apr 1996 05:23:47 +0000 (05:23 +0000)]
Fix usage: message to print out corrrectly.

Submitted by: Tatsumi Hosokawa <hosokawa@mt.cs.keio.ac.jp>

28 years agoAdded RCS Id and BSD-style copyrights to individual files.
Nate Williams [Thu, 18 Apr 1996 04:25:17 +0000 (04:25 +0000)]
Added RCS Id and BSD-style copyrights to individual files.

28 years agoAdded a disabled psm0 (PS/2) mouse device, using the new 'disable'
Nate Williams [Thu, 18 Apr 1996 04:02:30 +0000 (04:02 +0000)]
Added a disabled psm0 (PS/2) mouse device, using the new 'disable'
keyword.

28 years agoMention swapinfo as .Nm, so it will be referenced by mkwhatis.
Joerg Wunsch [Wed, 17 Apr 1996 20:50:44 +0000 (20:50 +0000)]
Mention swapinfo as .Nm, so it will be referenced by mkwhatis.

28 years agoDocument the possible EPERM return.
Joerg Wunsch [Wed, 17 Apr 1996 20:49:29 +0000 (20:49 +0000)]
Document the possible EPERM return.

Submitted by: imp@village.org (Warner Losh)

28 years agoImplement the RFNOWAIT flag for rfork(). If set this flag will cause the
Sujal Patel [Wed, 17 Apr 1996 17:05:08 +0000 (17:05 +0000)]
Implement the RFNOWAIT flag for rfork().  If set this flag will cause the
forked child to be dissociated from the parent).

Cleanup fork1(), implement vfork() and fork() in terms of rfork() flags.

Remove RFENVG, RFNOTEG, RFCNAMEG, RFCENVG which are Plan9 specific and cannot
possibly be implemented in FreeBSD.

Renumbered the flags to make up for the removal of the above flags.

Reviewed by: peter, smpatel
Submitted by: Mike Grupenhoff <kashmir@umiacs.umd.edu>

28 years agoSupport FreeBSD v2.1
Paul Traina [Wed, 17 Apr 1996 09:37:17 +0000 (09:37 +0000)]
Support FreeBSD v2.1

28 years agoAdd in linux support for the quickcam driver bottom half and
Paul Traina [Wed, 17 Apr 1996 09:00:53 +0000 (09:00 +0000)]
Add in linux support for the quickcam driver bottom half and
add a new "invasive camera scan" option for folks who have weird
cameras or weird parallel ports.

28 years agoFix a few bugs in argument processing.
Paul Traina [Wed, 17 Apr 1996 08:57:50 +0000 (08:57 +0000)]
Fix a few bugs in argument processing.

Submitted by: Timo Kokkonen <tjko@jyu.fi>

28 years agoMore changes to attempt to make this whole new dialog scheme more
Jordan K. Hubbard [Tue, 16 Apr 1996 12:17:27 +0000 (12:17 +0000)]
More changes to attempt to make this whole new dialog scheme more
robust.  The new "fire" actions, while affording signficantly more
interactivity to libdialog, come at a cost - if the fire action trashes
the screen then you're not going to be in Kansas anymore when you
come back to the menu and there had better be considerable extra
smarts in place for coping with such a situation.  These changes are my
attempt to do just that.

28 years agoBack out my MOTIF_STATIC change, it is unnecessary
Andrey A. Chernov [Tue, 16 Apr 1996 11:47:29 +0000 (11:47 +0000)]
Back out my MOTIF_STATIC change, it is unnecessary

28 years agoRemove old cruft.
Mark Murray [Tue, 16 Apr 1996 06:38:56 +0000 (06:38 +0000)]
Remove old cruft.

28 years agoFix oder of subdirs so this will rebuild if .depends are blown away.
Mark Murray [Tue, 16 Apr 1996 06:06:53 +0000 (06:06 +0000)]
Fix oder of subdirs so this will rebuild if .depends are blown away.

28 years agoFix for PR #1146: the "next" pointer must be cached before calling soabort
David Greenman [Tue, 16 Apr 1996 03:50:08 +0000 (03:50 +0000)]
Fix for PR #1146: the "next" pointer must be cached before calling soabort
since the struct containing it may be freed.

28 years agoNIS cleanups and fixes, the next generation, continued.
Bill Paul [Tue, 16 Apr 1996 00:23:59 +0000 (00:23 +0000)]
NIS cleanups and fixes, the next generation, continued.

pwd_mkdb.c:

- Don't save the PLUSCNT and MINUSCNT tokens: we don't need them anymore.

- Count the + and - entires for NIS together instead of counting + and -
  entries seperately. Index all special NIS entries using new _PW_KEYYPBYNUM
  token.

pwd.h:

- Remove the PLUSBYNUM, MINUSBYNUM, PLUSCNT and MINUSCNT tokens and replace
  then with a single _PW_KEYYPBYNUM token.

28 years agoNIS cleanups and fixes, the next generation.
Bill Paul [Tue, 16 Apr 1996 00:22:41 +0000 (00:22 +0000)]
NIS cleanups and fixes, the next generation.

getnetgrent.c:

- Catch one bogon that snuck by: in _listmatch(), check for '\0'
  rather than '\n'; strings returned from yp_match() are terminated
  with a nul, not a newline.

getpwent.c:

- Rip out all of the +inclusion/-exclusion stuff from before and
  replace it with something a little less grotty. The main problem
  with the old mechanism was that it wasted many cycles processing
  NIS entries even after it already knew they were to be exlcuded
  (or not included, depending on your pointof view). The highlights
  of these changes include:

  o Uses an in-memory hash database table to keep track of all the
    -@netgroup, -user, and -@group exclusions.

  o Tries harder to duplicate the behavior normally obtained when using
    NIS inclusions/exclusions on a flat /etc/passwd file (meaning things
    come out in much the same order).

  o Uses seperate methods for handling getpwent() and getpwnam()/getpwuid()
    operations instead of trying to do everything with one general
    function, which didn't work as well as I thought it would.

  o Uses both getnetgrent() and innetgr() to try to save time where
    possible.

  o Use only one special token in the local password database
    (_PW_KEYYPBYNUM) instead of seperate tokens to mark + and -
    entries (and stop using the counter tokens too). If this new
    token doesn't exist, the code will make due with the standard
    _PW_KEYBYNUM token in order to support older databases that
    won't have the new token in them.

  All this is an attempt to make this stuff work better in environments
  with large NIS passwd databases.

28 years agoAdd man pages for the copy*, fetch*, and store* families of
Mike Pritchard [Mon, 15 Apr 1996 21:37:07 +0000 (21:37 +0000)]
Add man pages for the copy*, fetch*, and store* families of
kernel functions.

Also added timeout(9) to the makefile.

Reviewed by: joerg
Obtained from: NetBSD

28 years agoFix a few NIS-related bogons:
Bill Paul [Mon, 15 Apr 1996 16:17:04 +0000 (16:17 +0000)]
Fix a few NIS-related bogons:

- Clear the _yp_innetgr flag immediately after calling setnetgrent() from
  innetgr(). We only need the flag set to temporarily alter setnetgrent()'s
  behavior. Previously, it was being cleared too late.

- When in NIS-only mode, innetgr() was wasting time doing unecessary
  extra processing after it had already found a match.

- Remember to free memory allocated by the NIS functions during innetgr()
  searches.

28 years agoTwo fixes from Rich Stevens:
David Greenman [Mon, 15 Apr 1996 03:46:33 +0000 (03:46 +0000)]
Two fixes from Rich Stevens:

1) Set the persist timer to help time-out connections in the CLOSING state.
2) Honor the keep-alive timer in the CLOSING state.

   This fixes problems with connections getting "stuck" due to incompletion
of the final connection shutdown which can be a BIG problem on busy WWW
servers.

28 years agoUpdate PCI bus code from my current sources:
Stefan Eßer [Sun, 14 Apr 1996 20:14:36 +0000 (20:14 +0000)]
Update PCI bus code from my current sources:

- always use pci_conf_read() and pci_conf_write(). (This is required to
  simulate non-existant devices in my system for PCI bridge code tests.)

- reorder some functions (put the main functions at the end).

- correct off by one bug in the code dealing with unitialized PCI to PCI
  bridge chips. (Bug found by ASAMI Satoshi.)

- print function number for multi-function devices.

28 years agoadd keymap entries for
Wolfram Schneider [Sun, 14 Apr 1996 17:22:43 +0000 (17:22 +0000)]
add keymap entries for
Italian ISO, Japanese 106/106x, Spanish ISO  [en,de]
add US dvorak [en,de,fr]

remove old test examples

28 years agoNow that I have started to use netboot, I see what is missing...
Poul-Henning Kamp [Sun, 14 Apr 1996 09:54:14 +0000 (09:54 +0000)]
Now that I have started to use netboot, I see what is missing...
Load symbols and set boot args more nicely.

28 years agoDocument that ipx is a valid protocol for the -f option.
John Hay [Sat, 13 Apr 1996 21:23:57 +0000 (21:23 +0000)]
Document that ipx is a valid protocol for the -f option.
Comment out ns and iso. They are not supported anymore.

28 years agoImplemented a new keyword `disable'. This should be useful for controlling
Bruce Evans [Sat, 13 Apr 1996 19:57:38 +0000 (19:57 +0000)]
Implemented a new keyword `disable'.  This should be useful for controlling
dangerous drivers in GENERIC.

Removed non-comments on #endifs in config.y.

Improved output formatting in mkioconf.c.

28 years agoFixed handling of device flags. The real flags were never used.
Bruce Evans [Sat, 13 Apr 1996 18:33:04 +0000 (18:33 +0000)]
Fixed handling of device flags.  The real flags were never used.

Submitted by: Michael Smith <msmith@atrad.adelaide.edu.au>

28 years agoOops, include opt_pcvt_hdr.h before the key value in it is used.
Bruce Evans [Sat, 13 Apr 1996 16:59:29 +0000 (16:59 +0000)]
Oops, include opt_pcvt_hdr.h before the key value in it is used.

28 years agoDescribe timeout() and untimeout().
Joerg Wunsch [Sat, 13 Apr 1996 16:59:07 +0000 (16:59 +0000)]
Describe timeout() and untimeout().

28 years agoFixed `PCVT_FREEBSD=210'.
Bruce Evans [Sat, 13 Apr 1996 16:50:23 +0000 (16:50 +0000)]
Fixed `PCVT_FREEBSD=210'.

28 years agoFixed a couple of format strings to match the change of pid_t from long
Bruce Evans [Sat, 13 Apr 1996 15:47:43 +0000 (15:47 +0000)]
Fixed a couple of format strings to match the change of pid_t from long
to int32_t.  I only fixed the ones that I noticed the warnings for.
Perhaps most of the format strings are correct now because they were
wrong before.  Except of course if int32_t isn't compatible with `int'.

28 years agoThis commit was generated by cvs2svn to compensate for changes in r15249,
Joerg Wunsch [Sat, 13 Apr 1996 15:33:16 +0000 (15:33 +0000)]
This commit was generated by cvs2svn to compensate for changes in r15249,
which included commits to RCS files with non-trunk default branches.

28 years agoImport NetBSD's termios'ed getty into a vendor branch.
Joerg Wunsch [Sat, 13 Apr 1996 15:33:16 +0000 (15:33 +0000)]
Import NetBSD's termios'ed getty into a vendor branch.

Obtained from: NetBSD 1.1R

28 years agoAccept and use the content of packets received that is bigger than the
John Hay [Sat, 13 Apr 1996 15:13:30 +0000 (15:13 +0000)]
Accept and use the content of packets received that is bigger than the
Novell spec, but still only transmit according to the spec.
Add a feature to dump the RIP and SAP tables when a SIGINFO signal is
received.

28 years agoAnother try: fixed bogus change of the fifo settings for the non-speed of 0.
Bruce Evans [Sat, 13 Apr 1996 14:55:18 +0000 (14:55 +0000)]
Another try: fixed bogus change of the fifo settings for the non-speed of 0.
rev.1.30 incorrectly changed the behaviour from always disabling the fifo
to always enabling it.

28 years agoDon't use a newfangled auto initializer. Initialize everything by
John Hay [Sat, 13 Apr 1996 14:37:22 +0000 (14:37 +0000)]
Don't use a newfangled auto initializer.  Initialize everything by
assignment to avoid one bug and several pessimizations.

In the old version, gcc-2.6.3 (i386 version) generates 16 bytes
of static data and copies it using 4 4-byte load-stores.  gcc-2.7.2
generates 2 1-byte stores and calls memset() to zero 14 bytes.
Linking fails because memset() doesn't exist in the kernel.

In both versions, the 2 bytes stored directly are all that is
actually used unless the null padding at the end is used, since
the 3 4-byte words in the middle are initialized again by struct
assignment.  These words are misaligned.  gcc generates misaligned
load-stores for (small) misaligned struct copies.

Submitted by: Bruce Evans

28 years agoChanged _BSD_PID_T_ from long to int. Lite2 changed pid_t from long to
Bruce Evans [Sat, 13 Apr 1996 14:25:06 +0000 (14:25 +0000)]
Changed _BSD_PID_T_ from long to int.  Lite2 changed pid_t from long to
int32_t in <sys/types.h> but this change was missed when the Lite2 types
were merged.

28 years agoChanged `noreturn' to `__noreturn__' so that all headers don't break if
Bruce Evans [Sat, 13 Apr 1996 14:23:29 +0000 (14:23 +0000)]
Changed `noreturn' to `__noreturn__' so that all headers don't break if
an application #defines `noreturn'.

Changed one instance of `const' similarly.  This is less like to be a
problem since applications shouldn't #define `const', and the common
hack of #defining `const' as nothing gives harmless (?) null attributes
instead of syntax errors.

Fixed comments on #endifs to match code.

28 years ago1. Update all the copyrights to delete useless clauses 3 and 4.
Jordan K. Hubbard [Sat, 13 Apr 1996 13:32:15 +0000 (13:32 +0000)]
1. Update all the copyrights to delete useless clauses 3 and 4.
2. Change more of the menu code over to new system.
3. Streamline label editor.

28 years agoFix a longstanding bug and a buglet of no significance.
Poul-Henning Kamp [Sat, 13 Apr 1996 13:28:54 +0000 (13:28 +0000)]
Fix a longstanding bug and a buglet of no significance.
Now net.ipx works.

Noticed by: John Hay -- John.Hay@csir.co.za

28 years agoRemove some of the grot I added to try and save/restore screen contents.
Jordan K. Hubbard [Sat, 13 Apr 1996 13:13:32 +0000 (13:13 +0000)]
Remove some of the grot I added to try and save/restore screen contents.
This shouldn't be done here, it should be done in the client as needed.

28 years agoEliminated sloppy common-style declarations. Now there are no duplicated
Bruce Evans [Sat, 13 Apr 1996 12:53:53 +0000 (12:53 +0000)]
Eliminated sloppy common-style declarations.  Now there are no duplicated
common labels for LINT.  There are still some common declarations for the
!KERNEL case in tcp_debug.h and spx_debug.h.  trpt depends on the ones in
tcp_debug.h.

28 years agoEliminated sloppy common-style declarations. Now there are no duplicated
Bruce Evans [Sat, 13 Apr 1996 12:45:57 +0000 (12:45 +0000)]
Eliminated sloppy common-style declarations.  Now there are no duplicated
common labels for LINT.  There are still some common declarations for the
!KERNEL case in tcp_debug.h and spx_debug.h.  trpt depends on the ones in
tcp_debug.h.

28 years agoDeclared `unittab' as static. It was bogusly shared between the asc and
Bruce Evans [Sat, 13 Apr 1996 12:18:45 +0000 (12:18 +0000)]
Declared `unittab' as static.  It was bogusly shared between the asc and
gsc drivers.

28 years agoDocument how to drop into DDB from a serial console.
Jordan K. Hubbard [Sat, 13 Apr 1996 11:57:18 +0000 (11:57 +0000)]
Document how to drop into DDB from a serial console.

28 years agoIntroduce a -b option for sites who are not interested in the old biff
Joerg Wunsch [Sat, 13 Apr 1996 11:44:12 +0000 (11:44 +0000)]
Introduce a -b option for sites who are not interested in the old biff
service.  (Avoid a ``in_vain'' warning...)

28 years agoRemove the legacy -d option from the arguments to mail.local in 4.4bsd.
Joerg Wunsch [Sat, 13 Apr 1996 11:42:46 +0000 (11:42 +0000)]
Remove the legacy -d option from the arguments to mail.local in 4.4bsd.

28 years agoChanged all per-file errors to warnings. Exit with a nonzero status of
Bruce Evans [Sat, 13 Apr 1996 11:35:54 +0000 (11:35 +0000)]
Changed all per-file errors to warnings.  Exit with a nonzero status of
if there was a per-file error.  My test case of `wc /proc/curproc/*'
works reasonably now (much like `size /proc/curproc/*'.

Reviewed by: wosch

28 years agoUse PCB_SAVEFPU_SIZE instead of a too-small size in savectx(). This
Bruce Evans [Sat, 13 Apr 1996 11:24:13 +0000 (11:24 +0000)]
Use PCB_SAVEFPU_SIZE instead of a too-small size in savectx().  This
bug only affected FPU emulators.  It might have caused bogus FPU states
in core dumps and in the child pcb after a fork.  Emulated FPU states
in core dumps don't work for other reasons, and the child FPU state
is reinitialized by exec, so the problem might not have caused any
noticeable affects.

Cleaned up #includes.

28 years agoGenerate #define of PCB_SAVEFPU_SIZE for use in savectx().
Bruce Evans [Sat, 13 Apr 1996 11:22:57 +0000 (11:22 +0000)]
Generate #define of PCB_SAVEFPU_SIZE for use in savectx().

28 years agoA little paper about phkmalloc.
Poul-Henning Kamp [Sat, 13 Apr 1996 08:30:21 +0000 (08:30 +0000)]
A little paper about phkmalloc.

28 years agoSplit libcrypt and libcipher man pages.
Mark Murray [Sat, 13 Apr 1996 08:18:24 +0000 (08:18 +0000)]
Split libcrypt and libcipher man pages.

28 years agoSplit the libcrypt and libcipher man pages.
Mark Murray [Sat, 13 Apr 1996 08:03:24 +0000 (08:03 +0000)]
Split the libcrypt and libcipher man pages.

28 years agoDocument MOTIF_STATIC
Andrey A. Chernov [Sat, 13 Apr 1996 06:40:19 +0000 (06:40 +0000)]
Document MOTIF_STATIC

28 years agodefault to udp if no flags specified...closes PR#bin/759
Marc G. Fournier [Sat, 13 Apr 1996 06:08:24 +0000 (06:08 +0000)]
default to udp if no flags specified...closes PR#bin/759
- should this be documented in the man page?

28 years agoFix a typo.
Mike Pritchard [Sat, 13 Apr 1996 02:09:03 +0000 (02:09 +0000)]
Fix a typo.

Submitted by: David Leonard <d@scry.dstc.edu.au>

28 years agoUpdate section on mounting DOS extended partitions.
Jordan K. Hubbard [Sat, 13 Apr 1996 00:12:41 +0000 (00:12 +0000)]
Update section on mounting DOS extended partitions.
Document the fact that we're now working on BSDI's dos emulator
and point people at the new freebsd-emulation@freebsd.org mailing
list.

28 years agoDocument the freebsd-emulation mailing list.
Jordan K. Hubbard [Sat, 13 Apr 1996 00:05:30 +0000 (00:05 +0000)]
Document the freebsd-emulation mailing list.

28 years agoImprove the handling of receive errors. Fix a nasty bug in the receive
John Hay [Fri, 12 Apr 1996 19:57:44 +0000 (19:57 +0000)]
Improve the handling of receive errors. Fix a nasty bug in the receive
buffer code.

28 years agoUpdate the Entropy-Gatherer to reflect a better setup - do not
Mark Murray [Fri, 12 Apr 1996 14:24:59 +0000 (14:24 +0000)]
Update the Entropy-Gatherer to reflect a better setup - do not
use IRQ1, as the keyboard hook already gets this timing info.

Also some slight improvements to the comments.

28 years agoMake alltraps a .globl so that DDB doesn't make people belive they have
Poul-Henning Kamp [Fri, 12 Apr 1996 12:22:14 +0000 (12:22 +0000)]
Make alltraps a .globl so that DDB doesn't make people belive they have
an ALIGNFLT on their hands all the time.

28 years agoAttempt to break a $Log$ snafu where a *** /--- (minus space)
Mark Murray [Fri, 12 Apr 1996 11:37:32 +0000 (11:37 +0000)]
Attempt to break a $Log$ snafu where a *** /--- (minus space)
was fouling up a comment in the checked-out code.

28 years agoinstall bsd.obj.mk
Wolfram Schneider [Fri, 12 Apr 1996 11:13:29 +0000 (11:13 +0000)]
install bsd.obj.mk
Submitted by: Asami

28 years agoAdd commented-out definitions of HAVE_MOTIF and MOTIFLIB.
Satoshi Asami [Fri, 12 Apr 1996 09:28:14 +0000 (09:28 +0000)]
Add commented-out definitions of HAVE_MOTIF and MOTIFLIB.

28 years agoFix a bogon I introduced with my last change.
Poul-Henning Kamp [Fri, 12 Apr 1996 09:24:22 +0000 (09:24 +0000)]
Fix a bogon I introduced with my last change.

Submitted by: Andreas Klemm <andreas@knobel.gun.de>

28 years agoAdd another good book to the required reading.
Poul-Henning Kamp [Fri, 12 Apr 1996 09:16:42 +0000 (09:16 +0000)]
Add another good book to the required reading.
make a couple of rules more sensible.

Reviewed by: phk
Submitted by: jmb

28 years agoAdd NO_CDROM, RESTRICTED, BROKEN and Motif support.
Satoshi Asami [Fri, 12 Apr 1996 08:08:36 +0000 (08:08 +0000)]
Add NO_CDROM, RESTRICTED, BROKEN and Motif support.

(1) The new NO_CDROM Boolean variable means "don't put the distfile/
    package on the CDROM you're going to sell".  It will basically
    turn off everything if FOR_CDROM is set.

    Many of the NO_PACKAGE ports are actually "don't sell for profit"
    types, which we shouldn't have any problem distributing via ftp.

(2) The new RESTRICTED Boolean variable means don't build this unless
    you know what you are doing.  It doesn't have any effect unless
    NO_RESTRICTED is also set.

(3) BROKEN means this port is broken.  At least it will now show up in
    INDEX and README.html, and give people more incentive to fix (I
    hope).

    RESTRICTED and BROKEN are expected to replace the pseudo-targets
    in parent Makefiles.  (The RESTRICTED and BROKEN list didn't do
    anything before, they were solely for grepping purposes.)

(4) The Motif support brings in four new variables: REQUIRES_MOTIF,
    which the porter sets for ports that require Motif to build;
    HAVE_MOTIF, which the user sets to indicate the system has Motif;
    MOTIF_STATIC, which the user sets to indicate that the static
    libXm, instead af the default dynamic library, is to be used; and
    MOTIFLIB, which is set to "${X11BASE}/lib/libXm.a" or
    "-L${X11BASE}/lib -lXm", depending on whether MOTIF_STATIC is set.

    The porter is expected to replace all occurrences of libXm in the
    {Im,M}akefiles with ${MOTIFLIB}, and this will allow both dynamic
    linkage (for users with Motif) and static linkage (for those who
    build packages to be used by those withot Motif, i.e., me ;)
    automatically.

Original Motif support idea by: graichen

28 years agoremove variables KMODGRP?=, KMODOWN?=, KMODMODE?=
Wolfram Schneider [Thu, 11 Apr 1996 23:03:10 +0000 (23:03 +0000)]
remove variables KMODGRP?=, KMODOWN?=, KMODMODE?=
replace ${BIN*} variables with ${KMOD*} variables

cleanup manpage code

include bsd.obj.mk
remove targets clean, cleandir, obj (included by bsd.obj.mk)

28 years agoMoved AUTO_EOI_[12] and most sio and pcvt options out of the makefile.
Bruce Evans [Thu, 11 Apr 1996 21:18:49 +0000 (21:18 +0000)]
Moved AUTO_EOI_[12] and most sio and pcvt options out of the makefile.

28 years agoFixed a spl hog. The vmdaemon process ran entirely at splhigh. It
Bruce Evans [Thu, 11 Apr 1996 21:05:25 +0000 (21:05 +0000)]
Fixed a spl hog.  The vmdaemon process ran entirely at splhigh.  It
sometimes disabled clock interrupts for 60 msec or more on a P133.
Clock interrupts were lost ...

Reviewed by: dyson

28 years agoSpell cpu_switch() with an i in a comment.
Bruce Evans [Thu, 11 Apr 1996 20:56:29 +0000 (20:56 +0000)]
Spell cpu_switch() with an i in a comment.

28 years agoFix for memory leak: specify NULL as openinfo parameter when calling
Bill Paul [Thu, 11 Apr 1996 20:54:15 +0000 (20:54 +0000)]
Fix for memory leak: specify NULL as openinfo parameter when calling
dbopen() to open an NIS map.

Testing with very large maps (e.g. a sample password database with 31,000+
entries) has shown that ypserv will leak memory (ps shows RSS and VSZ
growing to 4000 pages or more) when performing repeated yp_next()s or
a yp_all(). The problem with yp_all() is not immediately obvious since
the ypproc_all service is handled in a child process which exits once
the transfer is finished, but with repeated yp_next()s (like what you
get when you use getpwent() to scroll through the password database),
the parent ypserv grows to enormous size and never shrinks again.

It seems this is related to the HASHINFO parameters I used in yp_dblookup.c,
which I actually stole from pwd_mkdb. Calling dbopen() with the default
parameters (specifying openinfo as NULL) fixes the problem.

I still need to see how this impacts the other NIS tools. I'm also
considering changing from hash to btree databases: the hash database
method doesn't support R_CURSOR, which means yp_next_record() has to
do a lot of ugly work in order to reach an arbitrary location in the
database.

28 years agoFixed the PF_REMOTE case of ptcwrite(). I broke it in rev.1.22. Output
Bruce Evans [Thu, 11 Apr 1996 18:43:37 +0000 (18:43 +0000)]
Fixed the PF_REMOTE case of ptcwrite().  I broke it in rev.1.22.  Output
was duplicated until the canq filled up, and write() normally returned 0.
This case is apparently rare.  It was reported for Jove's shell buffer in
PR 1130.

28 years agoAdded a note about OPTi 82C929 based cards.
Sujal Patel [Thu, 11 Apr 1996 15:34:22 +0000 (15:34 +0000)]
Added a note about OPTi 82C929 based cards.
Clarified GUS DMA Settings.
Other misc. changes.

This should hold us over until I can finish cleaning up TASD, and finish
reintegrating all of the FreeBSD changes to the sound driver.  At that time
this document will be removed, and it's information moved to the handbook.

28 years agoImplemented a "-D" option that causes ftpd to detach and become a daemon -
David Greenman [Thu, 11 Apr 1996 10:22:16 +0000 (10:22 +0000)]
Implemented a "-D" option that causes ftpd to detach and become a daemon -
accepting connections on the FTP port and forking children processes to
handling them. This is lower overhead than spawning ftpd from inetd and
can be a significant win on busy FTP servers. Be sure to disable ftpd in
inetd.conf if you decide to use this option.
These changes are based on similar changes I made to wu-ftpd and have
been in use on wcarchive for several months.

28 years agoPriorities were broken. If there was an Index: line and ***/--- lines
Mark Murray [Thu, 11 Apr 1996 10:13:40 +0000 (10:13 +0000)]
Priorities were broken. If there was an Index: line and ***/--- lines
with valid names, the ***/---names were taken first.
this broke eg:
Index: foo/Makefile
==========
RCS <blah>
Retrieving <blah>
diff <blah>
*** Makefile <blah>
--- Makefile <blah>

By trying to patch the Makefile in the _curent_ directory, rather than
the one in the foo/ directory.

28 years agoAdd Daniel M. Eischen <deischen@iworks.InterWorks.org> for his xmcd
Satoshi Asami [Thu, 11 Apr 1996 08:49:21 +0000 (08:49 +0000)]
Add Daniel M. Eischen <deischen@iworks.InterWorks.org> for his xmcd
port contribution.

28 years agoFix editing mistake in last commit...sorry.
David Greenman [Thu, 11 Apr 1996 08:24:04 +0000 (08:24 +0000)]
Fix editing mistake in last commit...sorry.

28 years agoBugfix based on a kernel fix:
David Greenman [Thu, 11 Apr 1996 08:14:44 +0000 (08:14 +0000)]
Bugfix based on a kernel fix:

When PPP gets an uncompressed packet, it attempts to save off the TCP/IP
header for use in decompressing subsequant packets. If PPP gets garbage
(such as what happens when there is a port speed mismatch or modem line
noise), it will occasionally mistake the packet as a valid uncompressed
packet. When it tries to save off the header, it doesn't bother to check
for the validity of the header length and will happily clobber not only
the PPP VJC data structure, but parts of other process memory that happens
to follow it...causing, ahem, undesired behavior.

28 years agoremoved xref to icheck(8), ncheck(8) as they don't exist
Marc G. Fournier [Thu, 11 Apr 1996 07:49:34 +0000 (07:49 +0000)]
removed xref to icheck(8), ncheck(8) as they don't exist

28 years agoWhen cslip gets an uncompressed packet, it attempts to save off the TCP/IP
David Greenman [Thu, 11 Apr 1996 06:46:24 +0000 (06:46 +0000)]
When cslip gets an uncompressed packet, it attempts to save off the TCP/IP
header for use in decompressing subsequant packets. If cslip gets garbage
(such as what happens when there is a port speed mismatch or modem line
noise), it will occasionally mistake the packet as a valid uncompressed
packet. When it tries to save off the header, it doesn't bother to check
for the validity of the header length and will happily clobber not only
the cslip data structure, but parts of other kernel memory that happens
to follow it...causing, ahem, undesired behavior.

28 years agoAdded documentation for "options PAS_JOYSTICK_ENABLE"
Marc G. Fournier [Thu, 11 Apr 1996 06:23:18 +0000 (06:23 +0000)]
Added documentation for "options PAS_JOYSTICK_ENABLE"

28 years agomentioned support for enabling gameport on ProAudio Spectrum with
Marc G. Fournier [Thu, 11 Apr 1996 06:19:44 +0000 (06:19 +0000)]
mentioned support for enabling gameport on ProAudio Spectrum with
appropriately commented out "options" line for PAS_JOYSTICK_ENABLE
(PR#i386/960 - partial closer)

28 years agoPrint some warnings if root invokes chpass and sets the
Mike Pritchard [Thu, 11 Apr 1996 05:30:18 +0000 (05:30 +0000)]
Print some warnings if root invokes chpass and sets the
shell to one of the following:

- a non-existent file
- a non-regular file
- a file without any execute bits set

The shell is still set to whatever they entered even if the above
conditions exist (hey, it is the super user doing this after all :-),
but this might give the admin. some warning that they are about to screw
themselves and give them a chance to fix it before it is too late.

Inspired by: some new FreeBSD user on USENET who set his root shell
  to a shell that doesn't exist and now can't gain access to root (don't
  worry, I sent him some mail on how to recover from this).

28 years agoDisable sio3 in GENERIC - it messes with ATI cards.
Jordan K. Hubbard [Wed, 10 Apr 1996 23:03:36 +0000 (23:03 +0000)]
Disable sio3 in GENERIC - it messes with ATI cards.

28 years agoDo not exit if a file is not readable. This is a short hack
Wolfram Schneider [Wed, 10 Apr 1996 22:21:01 +0000 (22:21 +0000)]
Do not exit if a file is not readable. This is a short hack
until someone rewrite wc(1).

28 years agoremoved RS_IBUFSIZE #ifndef - have been informed it was a bad idea
Marc G. Fournier [Wed, 10 Apr 1996 18:09:58 +0000 (18:09 +0000)]
removed RS_IBUFSIZE #ifndef - have been informed it was a bad idea