unix-history/.git
30 years agoDelete some references to sleep() that somehow crept in.
Garrett Wollman [Sun, 20 Mar 1994 00:30:04 +0000 (00:30 +0000)]
Delete some references to sleep() that somehow crept in.

30 years agoAdded comments on Buslogic controllers. Cleaned up some syntax.
Jordan K. Hubbard [Sun, 20 Mar 1994 00:22:08 +0000 (00:22 +0000)]
Added comments on Buslogic controllers.  Cleaned up some syntax.

30 years agoThese are the installation and profile files from the 1.0 cdins*.flp
Rod Grimes [Sun, 20 Mar 1994 00:12:38 +0000 (00:12 +0000)]
These are the installation and profile files from the 1.0 cdins*.flp
floppy images.  Put in as they were in 1.0, about to change them to
work with 1.1.

30 years agoDeclare cpu_model, machine.
Garrett Wollman [Sun, 20 Mar 1994 00:03:33 +0000 (00:03 +0000)]
Declare cpu_model, machine.

30 years agoAdded cpu_model and machine variables.
Garrett Wollman [Sat, 19 Mar 1994 23:58:58 +0000 (23:58 +0000)]
Added cpu_model and machine variables.

30 years agoDeclare ostype, osrelease, etc.
Garrett Wollman [Sat, 19 Mar 1994 23:41:17 +0000 (23:41 +0000)]
Declare ostype, osrelease, etc.

30 years agoDefine new osrelease, ostype, osbuild, osconfig variables with the
Garrett Wollman [Sat, 19 Mar 1994 23:40:24 +0000 (23:40 +0000)]
Define new osrelease, ostype, osbuild, osconfig variables with the
pre-broken-down information from version.

30 years agoAs per Rod's wishes, man uses uid/gid 9 now.
Jordan K. Hubbard [Sat, 19 Mar 1994 23:31:39 +0000 (23:31 +0000)]
As per Rod's wishes, man uses uid/gid 9 now.

30 years agoRemove e-mail headers.
Garrett Wollman [Sat, 19 Mar 1994 22:55:43 +0000 (22:55 +0000)]
Remove e-mail headers.

30 years agoRemove man group - no longer necessary (that was quick! :). I'll let Rod
Jordan K. Hubbard [Sat, 19 Mar 1994 22:45:04 +0000 (22:45 +0000)]
Remove man group - no longer necessary (that was quick! :).  I'll let Rod
pick the uid for the `man' user, since he staked a claim on that, but he'd
better not forget or the make install will break badly! :)

30 years agoBack out previous change to vm_page.h.
David Greenman [Sat, 19 Mar 1994 22:31:09 +0000 (22:31 +0000)]
Back out previous change to vm_page.h.

30 years agoNo longer necessary to make catpages group write.
Jordan K. Hubbard [Sat, 19 Mar 1994 22:27:30 +0000 (22:27 +0000)]
No longer necessary to make catpages group write.

30 years agoMake man the default owner, don't bother with sgid.
Jordan K. Hubbard [Sat, 19 Mar 1994 22:25:17 +0000 (22:25 +0000)]
Make man the default owner, don't bother with sgid.

30 years agoCorrect pre-installation target.
Jordan K. Hubbard [Sat, 19 Mar 1994 22:24:48 +0000 (22:24 +0000)]
Correct pre-installation target.

30 years agoCheck for error from VOP_BMAP(). Not doing so leads to very strange
David Greenman [Sat, 19 Mar 1994 22:24:40 +0000 (22:24 +0000)]
Check for error from VOP_BMAP(). Not doing so leads to very strange
results (panics) when the filesystem doesn't support bmap().

30 years agoAdded vfs_update_interval variable and UPDATE_INTERVAL configuration
Garrett Wollman [Sat, 19 Mar 1994 22:19:11 +0000 (22:19 +0000)]
Added vfs_update_interval variable and UPDATE_INTERVAL configuration
parameter.  (Normally this would go in param.c, but since I'll be
making this run-time configurable sometime soon, I'd just as soon not
move it.)

30 years agoGo to userid man, groupid bin for catpages (and uid/gid bin for man dirs)
Jordan K. Hubbard [Sat, 19 Mar 1994 22:18:22 +0000 (22:18 +0000)]
Go to userid man, groupid bin for catpages (and uid/gid bin for man dirs)

30 years agoMore symtab space.
Garrett Wollman [Sat, 19 Mar 1994 22:17:56 +0000 (22:17 +0000)]
More symtab space.

30 years agovm_page_hold isn't quite up to handling the case where shared pages
David Greenman [Sat, 19 Mar 1994 22:13:37 +0000 (22:13 +0000)]
vm_page_hold isn't quite up to handling the case where shared pages
are involved, so do the old thing until this is fixed.

30 years agoMistake: Man files should not be group owned by man (only catfiles).
Jordan K. Hubbard [Sat, 19 Mar 1994 22:02:35 +0000 (22:02 +0000)]
Mistake:  Man files should not be group owned by man (only catfiles).
Engage brain, reverse commit.

30 years agoBrain fart - wrong location for ansi.h.
Garrett Wollman [Sat, 19 Mar 1994 21:43:18 +0000 (21:43 +0000)]
Brain fart - wrong location for ansi.h.

30 years agoUpgrade to sendmail 8.6.8
Andrew Moore [Sat, 19 Mar 1994 07:37:22 +0000 (07:37 +0000)]
Upgrade to sendmail 8.6.8
Included Karl Strickland's CHOWN_RESTRICTED patch

30 years agoRemove ttyfree from sioclose
Andrey Chernov [Fri, 18 Mar 1994 23:35:37 +0000 (23:35 +0000)]
Remove ttyfree from sioclose
Example:
Application use port cua01
Getty open ttyd1 (allocates rawq,outq,etc) and waits while application
done
Application quits, sioclose issued, ttyfree issued (getty calls revoke)
Getty awakes and goes to panic into initrb (NULL rawq)

30 years agoOnly format the last patch better with correct tabsettings.
Andreas Schulz [Fri, 18 Mar 1994 17:42:20 +0000 (17:42 +0000)]
Only format the last patch better with correct tabsettings.

30 years agoDisabled the free of the tty structure. Else we have a very reproducable
Andreas Schulz [Fri, 18 Mar 1994 17:18:12 +0000 (17:18 +0000)]
Disabled the free of the tty structure. Else we have a very reproducable
panic, if you reboot the machine remote. The machine doesn't reboot but
hang until you type a key on the console. Then a panic happens in ttyinput.
The console interrupt is never switched off and the console driver
tries to write the character into the freed raw queue :-).
Maybe the same bug is lurking in syscons ?

30 years agoFrom: Keith Bostic <bostic@vangogh.CS.Berkeley.edu>
Jordan K. Hubbard [Fri, 18 Mar 1994 14:08:57 +0000 (14:08 +0000)]
From: Keith Bostic <bostic@vangogh.CS.Berkeley.edu>

Here are the patches that make stty support vmin/vtime set
and display.

I'm not sure that reset should also handle this problem.  (I
wouldn't mind if reset were to fix this, but I'd argue that it
should do a full "stty sane" then, not just add vmin/vtime to
what it already does.)

--keith

30 years agoSome changes from Julian Stacey installed
Andrey Chernov [Fri, 18 Mar 1994 13:40:01 +0000 (13:40 +0000)]
Some changes from Julian Stacey installed

30 years agoFix memory leak.
Jordan K. Hubbard [Fri, 18 Mar 1994 13:38:56 +0000 (13:38 +0000)]
Fix memory leak.

30 years agoChange default MANGRP to `man'.
Jordan K. Hubbard [Fri, 18 Mar 1994 11:58:47 +0000 (11:58 +0000)]
Change default MANGRP to `man'.

30 years agoChange group id to man
Jordan K. Hubbard [Fri, 18 Mar 1994 11:55:53 +0000 (11:55 +0000)]
Change group id to man

30 years agoMake man files group owned by man. Make cat dirs group write.
Jordan K. Hubbard [Fri, 18 Mar 1994 11:46:12 +0000 (11:46 +0000)]
Make man files group owned by man.  Make cat dirs group write.

30 years agoAdded a man group ID.
Jordan K. Hubbard [Fri, 18 Mar 1994 11:45:49 +0000 (11:45 +0000)]
Added a man group ID.

30 years ago>From Chris Christoph P. U. Kukulies kuku@acds.physik.rwth-aachen.de
Rod Grimes [Fri, 18 Mar 1994 03:09:51 +0000 (03:09 +0000)]
>From Chris  Christoph P. U. Kukulies    kuku@acds.physik.rwth-aachen.de
Change references to we0 to be ed0 since the we driver is long gone.

30 years ago#ifdef KERNEL the include of systm.h, as vm_page.h is sometimes included
David Greenman [Thu, 17 Mar 1994 23:07:19 +0000 (23:07 +0000)]
#ifdef KERNEL the include of systm.h, as vm_page.h is sometimes included
externally.

30 years agoIncreased maximum protection for data segment to VM_PROT_ALL because
David Greenman [Thu, 17 Mar 1994 22:21:02 +0000 (22:21 +0000)]
Increased maximum protection for data segment to VM_PROT_ALL because
some (lisp) programs try to execute code out of it.

30 years agoReturn EPERM if hard linking to a directory as required by POSIX. Pointed
David Greenman [Wed, 16 Mar 1994 23:32:58 +0000 (23:32 +0000)]
Return EPERM if hard linking to a directory as required by POSIX. Pointed
out by Sean Eric Fagan.

30 years agoAdded recent changes by me and David.
Garrett Wollman [Wed, 16 Mar 1994 20:55:05 +0000 (20:55 +0000)]
Added recent changes by me and David.

30 years agoPollution control per bde.
Garrett Wollman [Wed, 16 Mar 1994 20:29:53 +0000 (20:29 +0000)]
Pollution control per bde.

30 years agoDefined _POSIX_* to be one, rather than just empty.
Garrett Wollman [Wed, 16 Mar 1994 19:18:34 +0000 (19:18 +0000)]
Defined _POSIX_* to be one, rather than just empty.

30 years agoForgot to correct prototype of execve(). Now done.
Garrett Wollman [Wed, 16 Mar 1994 19:10:39 +0000 (19:10 +0000)]
Forgot to correct prototype of execve().  Now done.

30 years agoDocumented execve()'s action wrt SUGID, and corrected reference to and value of
Garrett Wollman [Wed, 16 Mar 1994 19:06:28 +0000 (19:06 +0000)]
Documented execve()'s action wrt SUGID, and corrected reference to and value of
NCARGS to ARG_MAX.

30 years agoDocumented new logic of set*id calls. Moved setuid.3 to setuid.2, since
Garrett Wollman [Wed, 16 Mar 1994 19:01:10 +0000 (19:01 +0000)]
Documented new logic of set*id calls.  Moved setuid.3 to setuid.2, since
everything there describes system calls.  Moved setr[ug]id.c to compat-43;
it mgiht be better to remove them, since they don't actually do anything,
and the POSIX security model (such as it is) forbids changing the real ID
without simultaneously changing the effective and saved IDs.

30 years ago Defined _POSIX_SAVED_IDS per bde and P1003.1.
Garrett Wollman [Wed, 16 Mar 1994 17:38:10 +0000 (17:38 +0000)]
Defined _POSIX_SAVED_IDS per bde and P1003.1.

30 years agoMake man sgid bin, not sgid AND suid. There is no reason to make man suid
Jordan K. Hubbard [Wed, 16 Mar 1994 13:05:31 +0000 (13:05 +0000)]
Make man sgid bin, not sgid AND suid.  There is no reason to make man suid
bin since all you need to do is make the cat dirs group owned and writable
by bin - much less dangerous.

30 years agoAdded call to setlogin().
David Greenman [Wed, 16 Mar 1994 07:28:35 +0000 (07:28 +0000)]
Added call to setlogin().

30 years agoIncreased VM_PAGEOUT_PAGE_COUNT from 4 to 6 pages. ...better all-around
David Greenman [Wed, 16 Mar 1994 05:54:08 +0000 (05:54 +0000)]
Increased VM_PAGEOUT_PAGE_COUNT from 4 to 6 pages. ...better all-around
number.

30 years agoCatch the divide by zero bug earlier. If the drive give back
Andreas Schulz [Tue, 15 Mar 1994 20:49:09 +0000 (20:49 +0000)]
Catch the divide by zero bug earlier. If the drive give back
a zero cylinder and/or head count, set them and the sector count
to some fictious values. The same already happens if the request
sense is unsuccessfull.

30 years agoRemove x-bit from regular msdos files to reflect real situation,
Andrey Chernov [Tue, 15 Mar 1994 20:28:54 +0000 (20:28 +0000)]
Remove x-bit from regular msdos files to reflect real situation,
now x-bit mapped to hidden attribute.
So, 99% of your msdos files looks now rw-rw-rw-,
hidden msdos files and unix executable files into msdos filesys
looks now rwxrwxrwx.
When you boot DOS, unix executable files becomes hidden :-)

30 years agoDeleted references to SPAGE and added SUGID. When the functionality of
Garrett Wollman [Tue, 15 Mar 1994 18:51:23 +0000 (18:51 +0000)]
Deleted references to SPAGE and added SUGID.  When the functionality of
SPAGE is re-implemented, that code should be re-activated.

30 years agoFix yet one security hole with overflowing bufer (and zeroing userid)
Andrey Chernov [Tue, 15 Mar 1994 11:22:10 +0000 (11:22 +0000)]
Fix yet one security hole with overflowing bufer (and zeroing userid)

30 years agoDon't ever allow hard links to directories.
David Greenman [Tue, 15 Mar 1994 05:55:08 +0000 (05:55 +0000)]
Don't ever allow hard links to directories.

30 years agoAdded system call stubs for ntp_gettime and ntp_adjtime. These will
Garrett Wollman [Tue, 15 Mar 1994 04:02:32 +0000 (04:02 +0000)]
Added system call stubs for ntp_gettime and ntp_adjtime.  These will
some day be documented.

30 years agoForgot to set SUGID in set-id exec.
Garrett Wollman [Tue, 15 Mar 1994 03:55:35 +0000 (03:55 +0000)]
Forgot to set SUGID in set-id exec.

30 years agoImplement 4.4's SUGID flag, which keeps track of processes which
Garrett Wollman [Tue, 15 Mar 1994 02:48:51 +0000 (02:48 +0000)]
Implement 4.4's SUGID flag, which keeps track of processes which
call set*id() (reset on exec()).  Also, implement 4.4's logic for
setre[ug]id(), which is much simpler and closer to being the correct
behavior (given that we're stuck with setre[ug]id() for a variety of reasons).

30 years agoAdded David Mills' kernel NTP PLL code. The current version of NTP does
Garrett Wollman [Tue, 15 Mar 1994 01:58:40 +0000 (01:58 +0000)]
Added David Mills' kernel NTP PLL code.  The current version of NTP does
not yet speak to this code.

30 years agoFix server return code
Andrey Chernov [Tue, 15 Mar 1994 00:30:06 +0000 (00:30 +0000)]
Fix server return code

30 years agoFix security hole:
Andrey Chernov [Tue, 15 Mar 1994 00:01:34 +0000 (00:01 +0000)]
Fix security hole:
change creat to open to correct security flaw involving symlinks

30 years agoFix two security holes:
Andrey Chernov [Mon, 14 Mar 1994 23:54:34 +0000 (23:54 +0000)]
Fix two security holes:
1) IFS-related, don't bother us, because 'ash' don't use it
2) chown-related (instead of fchown).

30 years agoPut a safety check in for the secsiz divide for the MB printf. Else
Andreas Schulz [Mon, 14 Mar 1994 23:09:34 +0000 (23:09 +0000)]
Put a safety check in for the secsiz divide for the MB printf. Else
we get a panic integer divide fault in the kernel if the drive returns
a bogus zero secsize.

30 years agoPerformance improvements from John Dyson.
David Greenman [Mon, 14 Mar 1994 21:54:34 +0000 (21:54 +0000)]
Performance improvements from John Dyson.

1) A new mechanism has been added to prevent pages from being paged
out called "vm_page_hold". Similar to vm_page_wire, but
much lower overhead.
2) Scheduling algorithm has been changed to improve interactive
performance.
3) Paging algorithm improved.
4) Some vnode and swap pager bugs fixed.

30 years agoAdd Russian site ftp.kiae.su
Rod Grimes [Mon, 14 Mar 1994 19:20:45 +0000 (19:20 +0000)]
Add Russian site ftp.kiae.su

30 years agoChecking l_close in unneeded (one of my previous fixes),
Andrey Chernov [Mon, 14 Mar 1994 18:52:14 +0000 (18:52 +0000)]
Checking l_close in unneeded (one of my previous fixes),
so back it out.

30 years agoFix from Bruce Evans; dup() would turn itself into dup2() bogusly
Garrett Wollman [Sat, 12 Mar 1994 00:59:33 +0000 (00:59 +0000)]
Fix from Bruce Evans; dup() would turn itself into dup2() bogusly
for fd values >= 64.

30 years agoAdded explanation of EXCLUDE_<driver> options for sound drivers.
Steven Wallace [Fri, 11 Mar 1994 22:18:19 +0000 (22:18 +0000)]
Added explanation of EXCLUDE_<driver> options for sound drivers.

30 years agoRemoved EXCLUDE_<driver> for sound drivers so that all sound drivers
Steven Wallace [Fri, 11 Mar 1994 22:17:24 +0000 (22:17 +0000)]
Removed EXCLUDE_<driver> for sound drivers so that all sound drivers
are compiled into kernel.  Users are referred to sound doc for info.

30 years agoUpdated documentation for the sound drivers.
Steven Wallace [Fri, 11 Mar 1994 10:32:25 +0000 (10:32 +0000)]
Updated documentation for the sound drivers.

30 years agoAdded appropriate entries into files.i386 for snd drivers in /sys/i386/isa/sound
Steven Wallace [Fri, 11 Mar 1994 10:31:05 +0000 (10:31 +0000)]
Added appropriate entries into files.i386 for snd drivers in /sys/i386/isa/sound
Added new snd drivers and EXCLUDE_<driver> options to LINT.

30 years agoUpdated include files for sound drivers.
Steven Wallace [Fri, 11 Mar 1994 10:28:36 +0000 (10:28 +0000)]
Updated include files for sound drivers.

30 years agoIntegrated Hannu Savolainen's new VoxWare sound drivers, version 2.4.
Steven Wallace [Fri, 11 Mar 1994 10:27:25 +0000 (10:27 +0000)]
Integrated Hannu Savolainen's new VoxWare sound drivers, version 2.4.
These drivers now have full SoundBlaster 16 support.

30 years agoDue to the deletion of the gcc support from libc we need again the
Andreas Schulz [Thu, 10 Mar 1994 23:19:54 +0000 (23:19 +0000)]
Due to the deletion of the gcc support from libc we need again the
-lgcc_pic library. rtld uses the udivd3 routine from it.
Repeat the bug by simply compiling ld on current.

30 years agoDon't call ttyclose twice into sioclose and move ttyclose
Andrey Chernov [Thu, 10 Mar 1994 10:06:27 +0000 (10:06 +0000)]
Don't call ttyclose twice into sioclose and move ttyclose
before comhardclose

30 years agowarned not set in __initdb
Andrey Chernov [Wed, 9 Mar 1994 20:59:03 +0000 (20:59 +0000)]
warned not set in __initdb

30 years agoFrom Jordan via. Paul K.
Nate Willams [Wed, 9 Mar 1994 17:12:59 +0000 (17:12 +0000)]
From Jordan via. Paul K.

This fixes the problems Warner's having with ctors not being called
again with the latest round of ld changes and updates the file-names to what
Paul is using now.

The name change will not affect anything as we are not (yet) using it.

30 years agoDocument existance of -Z
David Greenman [Wed, 9 Mar 1994 14:28:02 +0000 (14:28 +0000)]
Document existance of -Z

30 years agoFirst open:
Andrey Chernov [Tue, 8 Mar 1994 22:11:10 +0000 (22:11 +0000)]
First open:
now HUPCL set only in bidir case for callin lines
(this prevents set HUPCL on mouse)
comhardclose:
in addition to HUPCL case now DTR dropped for bidir case
if line was active in and no carrier present now.
(this prevents DTR sleep on mouse)

30 years agoReduced pending IO max to make things a little less drastic when the
David Greenman [Tue, 8 Mar 1994 21:44:38 +0000 (21:44 +0000)]
Reduced pending IO max to make things a little less drastic when the
machine must page out lots of stuff. Reduced in half, but this still
might not be enough.

30 years agoFrom: Jim Babb <babb@sedhps01.mdc.com>
Nate Willams [Tue, 8 Mar 1994 16:25:29 +0000 (16:25 +0000)]
From: Jim Babb <babb@sedhps01.mdc.com>
Subject: Re: Bugs with floppy drives
Date: Tue, 8 Mar 94 9:11:54 CST

The transfer speed was only set in the retry after error, not  when
switching drives.

30 years agoWe've got libmalloc, let's build it.
Nate Willams [Tue, 8 Mar 1994 16:20:53 +0000 (16:20 +0000)]
We've got libmalloc, let's build it.

30 years agoFixed a bug, where a rdump doesn't close the rmt connection. You
Andreas Schulz [Tue, 8 Mar 1994 16:20:06 +0000 (16:20 +0000)]
Fixed a bug, where a rdump doesn't close the rmt connection. You
can repeat these by using "rdump -u" and make the /etc/dumpdates file
not accessable. It writes out correctly, that the dumpdates file cannot be
rewritten and then forgots to close the connection to the remote
server.

30 years agoModified pccnprobe to not set cn_tp = CONSOLE_TTY if __FreeBSD__ is
David Greenman [Tue, 8 Mar 1994 15:17:41 +0000 (15:17 +0000)]
Modified pccnprobe to not set cn_tp = CONSOLE_TTY if __FreeBSD__ is
defined. This code should probably be yanked out.

30 years agoForget half of the static declarations. Now the xxx_rcv, xx_op etc
Andreas Schulz [Tue, 8 Mar 1994 14:08:14 +0000 (14:08 +0000)]
Forget half of the static declarations. Now the xxx_rcv, xx_op etc
routines are all declared static now in this source files.

30 years agoMake the shmat,shmctl,shmdt,shmget routines static.
Andreas Schulz [Tue, 8 Mar 1994 13:01:44 +0000 (13:01 +0000)]
Make the shmat,shmctl,shmdt,shmget routines static.

30 years agoAdd mcd1 at 0x340 irq 11.
Rod Grimes [Tue, 8 Mar 1994 05:56:47 +0000 (05:56 +0000)]
Add mcd1 at 0x340 irq 11.

30 years agoFix a divide by zero trap in fsck. If the super-block is really hosed
Andreas Schulz [Mon, 7 Mar 1994 22:27:23 +0000 (22:27 +0000)]
Fix a divide by zero trap in fsck. If the super-block is really hosed
(filled with zeroes), you can get this.
Thanks for Gerd Truschinski for finding that out in his tests for the
PCI SCSI controller :-).

30 years agoChanges: added descriptions of latest stuff from David and John.
Garrett Wollman [Mon, 7 Mar 1994 17:41:42 +0000 (17:41 +0000)]
Changes: added descriptions of latest stuff from David and John.
options: document now reflects the system between 1.1 and 1.2, so say
so.

30 years agoFixed a wrong name in semconfig.c . The else part faulty declared semconfig
Andreas Schulz [Mon, 7 Mar 1994 16:58:43 +0000 (16:58 +0000)]
Fixed a wrong name in semconfig.c . The else part faulty declared semconfig
as semctl. Should have no affect on the code. Not needed for 1.1 .

30 years agoDescription:
Andreas Schulz [Mon, 7 Mar 1994 16:37:08 +0000 (16:37 +0000)]
Description:
'dump W' or 'dump w' expects to find the file /etc/dumpdates and dumps
core if it does not exist.
Bug fixed, so that it now exists cleanly.
Bug found by Adam David.

30 years agoModified mbuf macros to reduce calls to malloc. From John Dyson.
David Greenman [Mon, 7 Mar 1994 11:48:12 +0000 (11:48 +0000)]
Modified mbuf macros to reduce calls to malloc. From John Dyson.

30 years ago1) enhanced in_cksum from Bruce Evans.
David Greenman [Mon, 7 Mar 1994 11:47:32 +0000 (11:47 +0000)]
1) enhanced in_cksum from Bruce Evans.
2) minor comment change in machdep.c
3) enhanced bzero from John Dyson (twice as fast on a 486DX/33)

30 years ago 1) "Pre-faulting" in of pages into process address space
David Greenman [Mon, 7 Mar 1994 11:39:18 +0000 (11:39 +0000)]
1) "Pre-faulting" in of pages into process address space
Eliminates vm_fault overhead on process startup and
mmap referenced data for in-memory pages.

(process startup time using in-memory segments *much* faster)

2) Even more efficient pmap code.  Code partially cleaned up.
More comments yet to follow.

(generally more efficient pte management)

3) Pageout clustering ( in addition to the FreeBSD V1.1 pagein
clustering.)

(much faster paging performance on non-write behind disk
subsystems, slightly faster performance on other systems.)

4) Slightly changed vm_pageout code for more efficiency and
better statistics.  Also, resist swapout a little more.

(less likely to pageout a recently used page)

5) Slight improvement to the page table page trap efficiency.

(generally faster system VM fault performance)

6) Defer creation of unnamed anonymous regions pager until needed.

(speeds up shared memory bss creation)

7) Remove possible deadlock from swap_pager initialization.

8) Enhanced procfs to provide "vminfo" about vm objects and user
pmaps.

9) Increased MCLSHIFT/MCLBYTES from 2K to 4K to improve net &
socket performance and to prepare for things to come.

John Dyson
dyson@implode.root.com
David Greenman
davidg@root.com

30 years agoChanged ipcaccess from static to extern as it is no longer in this module.
David Greenman [Mon, 7 Mar 1994 11:03:02 +0000 (11:03 +0000)]
Changed ipcaccess from static to extern as it is no longer in this module.

30 years agoAdd mitsumi cdrom /dev entries.
Rod Grimes [Mon, 7 Mar 1994 00:57:38 +0000 (00:57 +0000)]
Add mitsumi cdrom /dev entries.

30 years ago>From BETA_1_1 branch
Rod Grimes [Mon, 7 Mar 1994 00:43:11 +0000 (00:43 +0000)]
>From BETA_1_1 branch
Add missing ${cpioverbose} option.

30 years ago>From BETA_1_1 branch
Rod Grimes [Mon, 7 Mar 1994 00:42:56 +0000 (00:42 +0000)]
>From BETA_1_1 branch
Correction for wrong permissions on /root, /usr/lib, /usr/libexec,
/usr/share and /usr/share/misc caused by filesyst.flp cpio file.

30 years agoNew freely redistributable version of malloc and friends which are much
Nate Willams [Sun, 6 Mar 1994 23:02:21 +0000 (23:02 +0000)]
New freely redistributable version of malloc and friends which are much
easier on memory yet still perform very well.

His version of malloc.h is installed in /usr/include.  This can change
if it's still something folks are not happy with, though it is suggested
in the installation notes.

I hope to replace the standard malloc in our system with this one and
remove gnumalloc once people shake this out and verify that it works for
them.

More notes can be found in the libmalloc directory.

Performance info. from the author:

                          Yet another malloc()
                          --------------------
                              Mark Moraes
                        <moraes@csri.toronto.edu>

Performance

The 4.3 malloc, (variants of which are distributed with perl, tcsh,
GNU and so on) is slightly faster than this malloc but it wastes more
space because it allocates blocks in powers of 2. It does not coalesce
free space, which can lead to it wasting lots of space. (There are some
pathological allocation sequences where it will ask for more space
from the system even though it has enough free space)

The 4.1 malloc is much slower than this malloc, and wastes about the
same space.
~

30 years agoFixing a local not initialized server.sin_addr.s_addr variable.
Andreas Schulz [Sun, 6 Mar 1994 21:38:56 +0000 (21:38 +0000)]
Fixing a local not initialized server.sin_addr.s_addr variable.
Bug found by Bob Willcox.

30 years agoRemoved a #warning that I left here.
Guido van Rooij [Sun, 6 Mar 1994 20:56:26 +0000 (20:56 +0000)]
Removed a #warning that I left here.

30 years agoUsed definable status codes (some may be sharable, esp the door open codes, but
Jordan K. Hubbard [Sun, 6 Mar 1994 14:14:49 +0000 (14:14 +0000)]
Used definable status codes (some may be sharable, esp the door open codes, but
it still looks a little suspicious that so many of the status codes are missing
so I'm not going to adopt all of the existing ones yet.  Try to be more
descriptive in the use of hex constants.

30 years agoStop count getty spacing problem, if we issue kill -1 1
Andrey Chernov [Sun, 6 Mar 1994 08:55:02 +0000 (08:55 +0000)]
Stop count getty spacing problem, if we issue kill -1 1

30 years agoFix typing error in bootstrappwd target
Andrey Chernov [Sun, 6 Mar 1994 08:40:31 +0000 (08:40 +0000)]
Fix typing error in bootstrappwd target