]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
26 years agoThis commit was generated by cvs2svn to compensate for changes in r30549,
peter [Sat, 18 Oct 1997 13:23:35 +0000 (13:23 +0000)]
This commit was generated by cvs2svn to compensate for changes in r30549,
which included commits to RCS files with non-trunk default branches.

26 years agoImport Matt's current de driver... He appears to be maintaining it in
peter [Sat, 18 Oct 1997 13:23:35 +0000 (13:23 +0000)]
Import Matt's current de driver...  He appears to be maintaining it in
the NetBSD source tree now.  This is slightly newer than the updated
.tar.gz on the 3am-software web site.

26 years agoImport Matt's current de driver... He appears to be maintaining it in
peter [Sat, 18 Oct 1997 13:23:35 +0000 (13:23 +0000)]
Import Matt's current de driver...  He appears to be maintaining it in
the NetBSD source tree now.  This is slightly newer than the updated
.tar.gz on the 3am-software web site.

26 years agoFix a #include "regex.h" that slipped through the net.
jraynard [Sat, 18 Oct 1997 13:12:39 +0000 (13:12 +0000)]
Fix a #include "regex.h" that slipped through the net.

26 years agoMF2.2: rev 1.17.2.36, and parts of 1.17.2.35 and 1.17.2.37
joerg [Sat, 18 Oct 1997 11:20:02 +0000 (11:20 +0000)]
MF2.2: rev 1.17.2.36, and parts of 1.17.2.35 and 1.17.2.37

26 years agoMake all the documented (in pcvt(4)) options supported options. While
joerg [Sat, 18 Oct 1997 10:59:47 +0000 (10:59 +0000)]
Make all the documented (in pcvt(4)) options supported options.  While
i was at it, do no longer insist on `PCVT_FREEBSD' being declared in
the config file, but default it to a reasonable value.

More cleanup to follow, but this part is safe for RELENG_2_2, too.

26 years agoSomething that should have been done 2.5 years ago: install pcvt's
joerg [Sat, 18 Oct 1997 10:40:02 +0000 (10:40 +0000)]
Something that should have been done 2.5 years ago: install pcvt's
accompanying doc files into /usr/share/pcvt/.  (This goes to the `doc'
distribution, not `bin'.)

26 years agoMF22: Minor cosmetic tweak.
jkh [Sat, 18 Oct 1997 10:11:04 +0000 (10:11 +0000)]
MF22: Minor cosmetic tweak.

26 years agoDon't doc PPP_BSDCOMP and PPP_DEFLATE twice. slight reorder so that the
peter [Sat, 18 Oct 1997 10:10:36 +0000 (10:10 +0000)]
Don't doc PPP_BSDCOMP and PPP_DEFLATE twice.  slight reorder so that the
options are not in the middle of the pseudo-device list.

Prompted by: bde

26 years agoBraino on my part.. a #define isn't a reference to a structure, so the
peter [Sat, 18 Oct 1997 09:02:39 +0000 (09:02 +0000)]
Braino on my part..  a #define isn't a reference to a structure, so the
struct only needs to be defined if the macro is used.

Pointed out by: bde

26 years ago - do string-magic on the right variable when trying
jkh [Sat, 18 Oct 1997 05:54:17 +0000 (05:54 +0000)]
 - do string-magic on the right variable when trying
   to find dependent packages on the local disk (i.e. package given as
   filename, not URL)
 - document $PKG_PATH
 - use -v on pkg_add for dependent packages

Submitted by: hubertf@NetBSD.ORG
Obtained from: NetBSD

26 years agoDeal with PLIST_UNEXEC properly.
jkh [Sat, 18 Oct 1997 05:42:05 +0000 (05:42 +0000)]
Deal with PLIST_UNEXEC properly.
PR: 4784
Submitted by: Stefan `Sec` Zehl <sec@quit.42.org>

26 years agoUpdate Makefile to know about PPP_FILTER and fix the PPP_BSDCOMP and
peter [Sat, 18 Oct 1997 01:37:52 +0000 (01:37 +0000)]
Update Makefile to know about PPP_FILTER and fix the PPP_BSDCOMP and
PPP_DEFLATE options... The code uses #if defined(PPP_DEFLATE) etc, so the
original method of "#define PPP_DEFLATE 0" did not actually disable
anything.  It was not possible to configure out bsdcomp or zlib
compression.  Also, join the settings for PPP_FILTER and NBPFILTER together
since they are related here (can't have PPP_FILTER without NBPFILTER).

26 years agoDon't say that the filtering options are only available under NetBSD, and
peter [Sat, 18 Oct 1997 01:29:18 +0000 (01:29 +0000)]
Don't say that the filtering options are only available under NetBSD, and
don't claim that _any_ serial speed is settable, mention that it has to
be supported by the device driver.  Some have very flexible baud rate
selection (ie: based on divisors etc, such as the com ports), while others
have explicit baud rate choices.

26 years agoDoc PPP_* options and add PPP_FILTER
peter [Sat, 18 Oct 1997 01:24:20 +0000 (01:24 +0000)]
Doc PPP_* options and add PPP_FILTER

26 years agoConvert PPP_FILTER to an option, like PPP_BSDCOMP and PPP_DEFLATE.
peter [Sat, 18 Oct 1997 01:20:28 +0000 (01:20 +0000)]
Convert PPP_FILTER to an option, like PPP_BSDCOMP and PPP_DEFLATE.
It requires bpf, I'll note this in LINT.

26 years agoConvert PPP_FILTER to an option
peter [Sat, 18 Oct 1997 01:15:32 +0000 (01:15 +0000)]
Convert PPP_FILTER to an option

26 years agoBetter fix for the bpf dependency that doesn't have such a large impact
peter [Sat, 18 Oct 1997 00:56:23 +0000 (00:56 +0000)]
Better fix for the bpf dependency that doesn't have such a large impact
on the code and pppd in userland.  PPP_FILTER is meant to be an option (or
negatable option).

26 years agoBack out the `PPP_FILTER => NBPFILTER' changes.
peter [Sat, 18 Oct 1997 00:49:59 +0000 (00:49 +0000)]
Back out the `PPP_FILTER => NBPFILTER' changes.

26 years agoBack out the `PPP_FILTER => NBPFILTER' changes
peter [Sat, 18 Oct 1997 00:46:10 +0000 (00:46 +0000)]
Back out the `PPP_FILTER => NBPFILTER' changes

26 years agoKilled non-sensical call to splimp/splx in crfree().
dg [Fri, 17 Oct 1997 23:52:56 +0000 (23:52 +0000)]
Killed non-sensical call to splimp/splx in crfree().

26 years agoA better fix for both kernel and LKM compilation.
roberto [Fri, 17 Oct 1997 21:50:32 +0000 (21:50 +0000)]
A better fix for both kernel and LKM compilation.

26 years agoChange PPP_FILTER into NBPFILTER to fix kernel compilation.
roberto [Fri, 17 Oct 1997 21:30:30 +0000 (21:30 +0000)]
Change PPP_FILTER into NBPFILTER to fix kernel compilation.
It should probably be changed in ppp_tty.c for consistency but I'll let
Brian deal with this.

Forgotten by: brian

26 years agoUpdate expected results for random number generator.
jraynard [Fri, 17 Oct 1997 19:24:33 +0000 (19:24 +0000)]
Update expected results for random number generator.

26 years agoRemove GNU regex.h (it's already in /usr/include).
jraynard [Fri, 17 Oct 1997 19:16:02 +0000 (19:16 +0000)]
Remove GNU regex.h (it's already in /usr/include).

26 years agoLEarn about the TI 1131.
wollman [Fri, 17 Oct 1997 16:34:08 +0000 (16:34 +0000)]
LEarn about the TI 1131.

26 years agoOops... back out the change to recognize the TI 1131; there's a better place.
wollman [Fri, 17 Oct 1997 16:26:14 +0000 (16:26 +0000)]
Oops... back out the change to recognize the TI 1131; there's a better place.

26 years agoTeach the PCI code about the TI 1131 and NeoMagic NM2160 in my laptop.
wollman [Fri, 17 Oct 1997 16:15:43 +0000 (16:15 +0000)]
Teach the PCI code about the TI 1131 and NeoMagic NM2160 in my laptop.

26 years agoMake a set of VOP standard lock, unlock & islocked VOP operators, which
phk [Fri, 17 Oct 1997 12:36:19 +0000 (12:36 +0000)]
Make a set of VOP standard lock, unlock & islocked VOP operators, which
depend on the lock being located at vp->v_data.  Saves 3x3 identical
vop procs, more as the other filesystems becomes lock aware.

26 years agoAdd $Id
ache [Fri, 17 Oct 1997 09:40:08 +0000 (09:40 +0000)]
Add $Id

26 years agoFix LONG_MAX overflowing
ache [Fri, 17 Oct 1997 09:35:50 +0000 (09:35 +0000)]
Fix LONG_MAX overflowing
Return seconds if errno other than EINTR
Add $Id
Submitted by: bde with minor optimization by me

26 years agoMention the problem with the Intel AL440LX BIOS; hopefully we will
msmith [Fri, 17 Oct 1997 06:48:35 +0000 (06:48 +0000)]
Mention the problem with the Intel AL440LX BIOS; hopefully we will
have more details from Intel about this shortly.

Submitted by: David Filo <filo@yahoo.com>

26 years agoFixed a bug where input packets were counted twice - messing up the
dg [Fri, 17 Oct 1997 06:27:44 +0000 (06:27 +0000)]
Fixed a bug where input packets were counted twice - messing up the
stats.

26 years agoFix two bugs which caused various RPC programs (mountd, nfsd, ...)
jdp [Fri, 17 Oct 1997 04:59:56 +0000 (04:59 +0000)]
Fix two bugs which caused various RPC programs (mountd, nfsd, ...)
to fail under certain circumstances.

1. In one spot, the ifr_flags member was being examined in the
wrong structure, thus it contained garbage.  On a machine in which
only the loopback interface was up, this caused everything that
wanted to talk to the portmapper to fail -- a particular problem
with laptops, where the pccard ethernet interface is likely to come
up long after the attempt to start mountd, nfsd, amd, etc.

2. Compounding the above problem, get_myaddress() returned a
successful status even though it failed to find an address that it
considered good enough.

26 years agoUpdate XFree86 reference.
jkh [Fri, 17 Oct 1997 02:27:24 +0000 (02:27 +0000)]
Update XFree86 reference.
Submitted by: luigi

26 years agoAsk for VJ slot id compression by default.
brian [Fri, 17 Oct 1997 00:20:30 +0000 (00:20 +0000)]
Ask for VJ slot id compression by default.

26 years agoPPP_FILTER => NBPFILTER
brian [Fri, 17 Oct 1997 00:12:38 +0000 (00:12 +0000)]
PPP_FILTER => NBPFILTER

26 years agocosmetic:
brian [Thu, 16 Oct 1997 23:55:19 +0000 (23:55 +0000)]
cosmetic:
o  Log client connection commands when "set log +command"
o  Don't display PAP password in the log unless we're
   "set log +debug".

26 years agoVFS clean up "hekto commit"
phk [Thu, 16 Oct 1997 22:01:05 +0000 (22:01 +0000)]
VFS clean up "hekto commit"

1.  Add defaults for more VOPs
        VOP_LOCK        vop_nolock
        VOP_ISLOCKED    vop_noislocked
        VOP_UNLOCK      vop_nounlock
    and remove direct reference in filesystems.

2.  Rename the nfsv2 vnop tables to improve sorting order.

26 years agoUpdate FREEBSD-upgrade file after removing random.*
jraynard [Thu, 16 Oct 1997 21:45:46 +0000 (21:45 +0000)]
Update FREEBSD-upgrade file after removing random.*

26 years agoDelete redundant files (thanks, Andrei!)
jraynard [Thu, 16 Oct 1997 21:39:00 +0000 (21:39 +0000)]
Delete redundant files (thanks, Andrei!)

26 years agoCopy time_to_sleep to time_remaining since it can be left
ache [Thu, 16 Oct 1997 21:31:43 +0000 (21:31 +0000)]
Copy time_to_sleep to time_remaining since it can be left
uninitialized if nanosleep returns early with agr error

26 years agoAnother VFS cleanup "kilo commit"
phk [Thu, 16 Oct 1997 20:32:40 +0000 (20:32 +0000)]
Another VFS cleanup "kilo commit"

1.  Remove VOP_UPDATE, it is (also) an UFS/{FFS,LFS,EXT2FS,MFS}
    intereface function, and now lives in the ufsmount structure.

2.  Remove VOP_SEEK, it was unused.

3.  Add mode default vops:

    VOP_ADVLOCK          vop_einval
    VOP_CLOSE            vop_null
    VOP_FSYNC            vop_null
    VOP_IOCTL            vop_enotty
    VOP_MMAP             vop_einval
    VOP_OPEN             vop_null
    VOP_PATHCONF         vop_einval
    VOP_READLINK         vop_einval
    VOP_REALLOCBLKS      vop_eopnotsupp

    And remove identical functionality from filesystems

4.   Add vop_stdpathconf, which returns the canonical stuff.  Use
     it in the filesystems.  (XXX: It's probably wrong that specfs
     and fifofs sets this vop, shouldn't it come from the "host"
     filesystem, for instance ufs or cd9660 ?)

5.   Try to make system wide VOP functions have vop_* names.

6.   Initialize the um_* vectors in LFS.

(Recompile your LKMS!!!)

26 years agoOops. forgot the blasted cvs add.
phk [Thu, 16 Oct 1997 17:48:22 +0000 (17:48 +0000)]
Oops. forgot the blasted cvs add.

Pointed hat sent from: Karl Denninger  <karl@Mcs.Net>

26 years agoAlphabetize.
jkh [Thu, 16 Oct 1997 17:47:00 +0000 (17:47 +0000)]
Alphabetize.

26 years agoAdd /usr/share/examples/libvgl so we don't fall over. :(
jkh [Thu, 16 Oct 1997 17:38:55 +0000 (17:38 +0000)]
Add /usr/share/examples/libvgl so we don't fall over. :(
Forgotten by: sos

26 years agoHandle machine-dependent (stdlib) sources more automatically.
bde [Thu, 16 Oct 1997 14:58:30 +0000 (14:58 +0000)]
Handle machine-dependent (stdlib) sources more automatically.

This fixes bugs in the manual handling.  abs.[cS] was handled too
specially and the wrong (.c) variant for each of div.[cS], labs.[cS]
and ldiv.[cS] was added to SRCS.  This caused the .c variant to be
used if `depend' was made and the .S version to be used otherwise.

26 years agoRemoved bogus .PATH statement.
bde [Thu, 16 Oct 1997 14:41:25 +0000 (14:41 +0000)]
Removed bogus .PATH statement.

26 years agoRemoved the subdirectory paths from the definitions of MAN[1-9]. They
bde [Thu, 16 Oct 1997 14:26:13 +0000 (14:26 +0000)]
Removed the subdirectory paths from the definitions of MAN[1-9].  They
were a workaround for limitations in bsd.man.mk that were fixed about
2 years ago.

26 years agoHandle machine-dependent (m-d) (string) sources more automatically.
bde [Thu, 16 Oct 1997 13:46:50 +0000 (13:46 +0000)]
Handle machine-dependent (m-d) (string) sources more automatically.
The names of m-d variants are now added (manually) to MDSRCS instead
of to SRCS, and the names of all machine-independent (m-i) variants
that can reasonably be replaced by an m-d variant are now added
(manually) to MISRCS instead of to SRCS, so that a simple substitution
can be used to discard the unused m-i variants.  MISRCS is potentially
all m-i sources, but the substitution is too simple to be fast, so
MISRCS should be kept reasonably small.

libc/Makefile.inc:
Do the substitution.

libc/i386/string/Makefile.inc:
Add to MDSRCS instead of to SRCS.  Add the names of all sources in this
directory, but no others.

libc/string/Makefile.inc
Add to MISRCS instead of to SRCS.  Add the names of all sources in this
directory.  Don't use (broken) explicit rules for special cases.

26 years agoReflect current sleep/usleep implementations state
ache [Thu, 16 Oct 1997 13:42:03 +0000 (13:42 +0000)]
Reflect current sleep/usleep implementations state

26 years agoCleanup #includes
ache [Thu, 16 Oct 1997 13:35:25 +0000 (13:35 +0000)]
Cleanup #includes

26 years agoMakes html table with all VOPs in it.
phk [Thu, 16 Oct 1997 12:40:58 +0000 (12:40 +0000)]
Makes html table with all VOPs in it.

26 years agoStaticize the ufs vnops member functions.
phk [Thu, 16 Oct 1997 11:59:09 +0000 (11:59 +0000)]
Staticize the ufs vnops member functions.

26 years agoRemove an overlapping variable I created in last round.
phk [Thu, 16 Oct 1997 11:58:30 +0000 (11:58 +0000)]
Remove an overlapping variable I created in last round.

Don't do pointer subtraction on void *

Use VOP_STRATEGY instead of homegrown stuff.

Add an XXX warning for LFS freaks to ponder.

26 years agoVFS mega cleanup commit (x/N)
phk [Thu, 16 Oct 1997 10:50:27 +0000 (10:50 +0000)]
VFS mega cleanup commit (x/N)

1.  Add new file "sys/kern/vfs_default.c" where default actions for
    VOPs go. Implement proper defaults for ABORTOP, BWRITE, LEASE,
    POLL, REVOKE and STRATEGY.  Various stuff spread over the entire
    tree belongs here.

2.  Change VOP_BLKATOFF to a normal function in cd9660.

3.  Kill VOP_BLKATOFF, VOP_TRUNCATE, VOP_VFREE, VOP_VALLOC.  These
    are private interface functions between UFS and the underlying
    storage manager layer (FFS/LFS/MFS/EXT2FS).  The functions now
    live in struct ufsmount instead.

4.  Remove a kludge of VOP_ functions in all filesystems, that did
    nothing but obscure the simplicity and break the expandability.
    If a filesystem doesn't implement VOP_FOO, it shouldn't have an
    entry for it in its vnops table.  The system will try to DTRT
    if it is not implemented.  There are still some cruft left, but
    the bulk of it is done.

5.  Fix another VCALL in vfs_cache.c (thanks Bruce!)

26 years agoFix a couble of typos.
sos [Thu, 16 Oct 1997 09:12:31 +0000 (09:12 +0000)]
Fix a couble of typos.

Submitted by: "Vanill I. Shu" <vanilla@MinJe.com.TW>

26 years agoremove forgotten debug printf()
julian [Thu, 16 Oct 1997 08:18:38 +0000 (08:18 +0000)]
remove forgotten debug printf()

26 years agoTwo more places where root filesystems were mounted, put them at the head of
julian [Thu, 16 Oct 1997 08:16:34 +0000 (08:16 +0000)]
Two more places where root filesystems were mounted, put them at the head of
the mount list  in case there is already  DEVFS present.

26 years agoWe are mounting the root.
julian [Thu, 16 Oct 1997 07:32:14 +0000 (07:32 +0000)]
We are mounting the root.
mount it at the HEAD of the queue, DEVFS might already be there..

26 years agoAdd a little example on libvgl usage.
sos [Thu, 16 Oct 1997 07:29:26 +0000 (07:29 +0000)]
Add a little example on libvgl usage.

26 years ago1/ by default make all versions of the same device get the same vnode.
julian [Thu, 16 Oct 1997 07:28:50 +0000 (07:28 +0000)]
1/ by default make all versions of the same device get the same vnode.
2/ Show the dummy mount in the mount list. it cannot be reached (that I know of)
but puting it there, means that disks mounted from devfs will have their things such as the superblock and the bitmaps, synced to disk :)

26 years agosome cleanups of init code, and changes needed to support disk layering.
julian [Thu, 16 Oct 1997 06:29:27 +0000 (06:29 +0000)]
some cleanups of init code, and changes needed to support disk layering.

26 years agoMake the supported media info output the default.
peter [Thu, 16 Oct 1997 03:55:29 +0000 (03:55 +0000)]
Make the supported media info output the default.
The -m switch is now accepted for backwards compatability.

Suggested by: davidg

26 years agoProper spacing in the Synopsis.
max [Thu, 16 Oct 1997 01:19:15 +0000 (01:19 +0000)]
Proper spacing in the Synopsis.

26 years agosrandom(time()) is too weak even for joke, use srandom(time() ^ getpid())
ache [Wed, 15 Oct 1997 22:49:19 +0000 (22:49 +0000)]
srandom(time()) is too weak even for joke, use srandom(time() ^ getpid())
at least. Note that srandomdev can't be used here since gawk preserve old
seed value.

26 years agoOn execing a sgid program, do not set P_SUGID when cr_gid and cr)_uid
guido [Wed, 15 Oct 1997 18:28:34 +0000 (18:28 +0000)]
On execing a sgid program, do not set P_SUGID when  cr_gid and cr)_uid
do not change.
PR: 4755
Reviewed by: Bruce Evans

26 years agoRemove Xref to format(8) which does not exist.
max [Wed, 15 Oct 1997 17:33:47 +0000 (17:33 +0000)]
Remove Xref to format(8) which does not exist.

26 years agoInclude the machine-dependent Makefile.inc for sys in the correct place.
bde [Wed, 15 Oct 1997 16:29:14 +0000 (16:29 +0000)]
Include the machine-dependent Makefile.inc for sys in the correct place.

26 years agoRemoved the subdirectory paths from the definitions of MAN[1-9]. They
bde [Wed, 15 Oct 1997 16:16:41 +0000 (16:16 +0000)]
Removed the subdirectory paths from the definitions of MAN[1-9].  They
were a workaround for limitations in bsd.man.mk that were fixed about
2 years ago.

26 years agoUnspam l10n ranges check
ache [Wed, 15 Oct 1997 15:37:46 +0000 (15:37 +0000)]
Unspam l10n ranges check

26 years agoAdded some 2-line source files to get a direct correspondence
bde [Wed, 15 Oct 1997 15:27:19 +0000 (15:27 +0000)]
Added some 2-line source files to get a direct correspondence
between sources and objects.  This will be used to avoid messy
special cases in Makefile.inc.

26 years agoRemove old SIGALRM absorbing back-compat code. It wasn't working at all
peter [Wed, 15 Oct 1997 14:11:08 +0000 (14:11 +0000)]
Remove old SIGALRM absorbing back-compat code.  It wasn't working at all
for the entire time that it was there, so obviously nothing needs it
anymore.

Note, unix98/single-unix spec v2 says that usleep() returns an int rather
than a void, to indicate whether the entire time period elapsed (0) or an
error (eg: signal handler) interrupted it (returns -1, errno = EINTR)
It is probably useful to make this change but I'll test it locally first
to see if this will break userland programs [much]...

Reviewed by: ache, bde

26 years agoGive up on the "try and compensate for broken programs" cruft and revert
peter [Wed, 15 Oct 1997 14:06:15 +0000 (14:06 +0000)]
Give up on the "try and compensate for broken programs" cruft and revert
back to the original single nanosleep() implementation.  This is POSIX and
Unix98 (aka single-unix spec v2) compliant behavior.  If a program sets
alarm(2) or an interval timer (setitimer(2)) without a SIGALRM handler
being active, sleep(3) will no longer absorb it, and the program will get
what it asked for..... :-]

The original reason for this in the first place (apache) doesn't seem to
need it anymore, according to Andrey.

Reviewed by: ache, bde

26 years agoDo not use gawk random, we have better one in libc
ache [Wed, 15 Oct 1997 14:01:31 +0000 (14:01 +0000)]
Do not use gawk random, we have better one in libc

26 years agoSigh. Signal handlers are executed on leaving the system call, not
peter [Wed, 15 Oct 1997 13:58:52 +0000 (13:58 +0000)]
Sigh.  Signal handlers are executed on leaving the system call, not
at moment of delivery.  Restoring the signal mask after the tsleep()
is next to useless since the signal is still queued.. This was interacting
with usleep(3) on receipt of a SIGALRM causing it to near busy loop.

Now, we set the new signal mask "permanently" for signanosleep().

Problem noted by:  bde

26 years agovnops megacommit
phk [Wed, 15 Oct 1997 13:24:07 +0000 (13:24 +0000)]
vnops megacommit

1.  Use the default function to access all the specfs operations.
2.  Use the default function to access all the fifofs operations.
3.  Use the default function to access all the ufs operations.
4.  Fix VCALL usage in vfs_cache.c
5.  Use VOCALL to access specfs functions in devfs_vnops.c
6.  Staticize most of the spec and fifofs vnops functions.
7.  Make UFS panic if it lacks bits of the underlying storage handling.

26 years agoOne could be surprised how much bugs can still be found here...
joerg [Wed, 15 Oct 1997 12:16:37 +0000 (12:16 +0000)]
One could be surprised how much bugs can still be found here...

Properlay clean the global RM variable if cgetstr() failed for it.
Otherwise, a connection attempt to a remote machine was made (and a
bogus result code printed) if a local printer followed a remote one in
printcap, and you did a `lpq -a', since checkremote() falsely assumed
the printer to be a remote one.

While i was at it, removed a gratuituous newline printed in front of
the remote machine's name, thus making the output more consistent (and
better machine-parseable) now.

26 years agoHide a silly ``unknown board type'' message behind bootverbose. The
joerg [Wed, 15 Oct 1997 10:09:24 +0000 (10:09 +0000)]
Hide a silly ``unknown board type'' message behind bootverbose.  The
``ie0: not found'' message is sufficient as a default.

26 years agoHmm, realign the vnops into two columns.
phk [Wed, 15 Oct 1997 10:05:29 +0000 (10:05 +0000)]
Hmm, realign the vnops into two columns.

26 years agoStylistic overhaul of vnops tables.
phk [Wed, 15 Oct 1997 09:22:02 +0000 (09:22 +0000)]
Stylistic overhaul of vnops tables.
        1. Remove comment stating the blatantly obvious.
        2. Align in two columns.
        3. Sort all but the default element alphabetically.
        4. Remove XXX comments pointing out entries not needed.

26 years agoWhen the default vnops funtion is vn_default_error(), there is no reason to
phk [Wed, 15 Oct 1997 07:42:34 +0000 (07:42 +0000)]
When the default vnops funtion is vn_default_error(), there is no reason to
implement small functions that just return EOPNOTSUPP for things we don't do.

The removed functions only apply to UFS based filesystems anyway.

26 years agoMention that the Zip driver (vpo) requires SCSI disk support, and works
msmith [Wed, 15 Oct 1997 07:35:48 +0000 (07:35 +0000)]
Mention that the Zip driver (vpo) requires SCSI disk support, and works
best with EPP 1.9 mode selected.
Submitted by: Gianmarco Giovannelli <gmarco@giovannelli.it>

26 years agoIN_HASHED goes in the in-core flags ip->i_flag, not in the on-disk flags
bde [Wed, 15 Oct 1997 07:32:45 +0000 (07:32 +0000)]
IN_HASHED goes in the in-core flags ip->i_flag, not in the on-disk flags
ip->i_flags.

Rev.1.18 completely broke ufs.  My root directory went away about 10
seconds after booting.  I think file system damage was null, since
IN_HASHED = 0x80 is not used in the disk flags (it would probably
be UF_SOMETHING if it were used).

26 years agoUse err(3). Add prototypes.
charnier [Wed, 15 Oct 1997 06:43:54 +0000 (06:43 +0000)]
Use err(3). Add prototypes.

26 years agoUse err(3). Rewrote man page in mdoc format.
charnier [Wed, 15 Oct 1997 06:42:28 +0000 (06:42 +0000)]
Use err(3). Rewrote man page in mdoc format.

26 years agoAdd usage().
charnier [Wed, 15 Oct 1997 06:41:19 +0000 (06:41 +0000)]
Add usage().

26 years agoFix various bogons reported in this PR.
jkh [Wed, 15 Oct 1997 04:37:26 +0000 (04:37 +0000)]
Fix various bogons reported in this PR.
PR: 4765
Submitted by: fdiv

26 years agoInitialise LCP, IPCP and CCP as soon as LCP is down.
brian [Wed, 15 Oct 1997 01:01:17 +0000 (01:01 +0000)]
Initialise LCP, IPCP and CCP as soon as LCP is down.
Without this, in -auto mode, we stay in ST_STOPPING
and never check our dial filters to see if it's time
to bring the line up again.

This may make "set stopped" redundant.

26 years agoCorrect a bug in the 'allow arbitrary number of socket descriptors' changes
wpaul [Tue, 14 Oct 1997 21:50:17 +0000 (21:50 +0000)]
Correct a bug in the 'allow arbitrary number of socket descriptors' changes
made to the RPC code some months ago. The value of __svc_fdsetsize is being
calculated incorrectly.

Logically, one would assume that __svc_fdsetsize is being used as a
substitute for FD_SETSIZE, with the difference being that __svc_fdsetsize
can be expanded on the fly to accomodate more descriptors if need be.
There are two problems: first, __svc_fdsetsize is not initialized to 0.
Second, __svc_fdsetsize is being calculated in svc.c:xprt_registere() as:

                __svc_fdsetsize = howmany(sock+1, NFDBITS);

This is wrong. If we are adding a socket with index value 4 to the
descriptor set, then __svc_fdsetsize will be 1 (since fds_bits is
an unsigned long, it can support any descriptor from 0 to 31, so we
only need one of them). In order for this to make sense with the
rest of the code though, it should be:

                __svc_fdsetsize = howmany(sock+1, NFDBITS) * NFDBITS;

Now if sock == 4, __svc_fdsetsize will be 32.

This bug causes 2 errors to occur. First, in xprt_register(), it
causes the __svc_fdset descriptor array to be freed and reallocated
unnecessarily. The code checks if it needs to expand the array using
the test: if (sock + 1 > __svc_fdsetsize). The very first time through,
__svc_fdsetsize is 0, which is fine: an array has to be allocated the
first time out. However __svc_fdsetsize is incorrectly set to 1, so
on the second time through, the test (sock + 1 > __svc_fdsetsize)
will still succeed, and the __svc_fdset array will be destroyed and
reallocated for no reason.

Second, the code in svc_run.c:svc_run() can become hopelessly confused.
The svc_run() routine malloc()s its own fd_set array using the value
of __svc_fdsetsize to decide how much memory to allocate. Once the
xprt_register() function expands the __svc_fdset array the first time,
the value for __svc_fdsetsize becomes 2, which is too small: the resulting
calculation causes the code to allocate an array that's only 32 bits wide
when it actually needs 64 bits. It also uses the valuse of __svc_fdsetsize
when copying the contents of the __svc_fdset array into the new array.
The end result is that all but the first 32 file descriptors get lost.

Note: from what I can tell, this bug originated in OpenBSD and was
brought over to us when the code was merged. The bug is still there
in the OpenBSD source.

Total nervous breakdown averted by: Electric Fence 2.0.5

26 years agoReset the flag right away, could catch a bogon someday.
phk [Tue, 14 Oct 1997 18:51:07 +0000 (18:51 +0000)]
Reset the flag right away, could catch a bogon someday.

26 years agoI think my previous change may have opened a race conditio.
phk [Tue, 14 Oct 1997 18:46:48 +0000 (18:46 +0000)]
I think my previous change may have opened a race conditio.
This patch does the same thing, with no change in semantics.

26 years agoMakefile for contributed version of awk.
jraynard [Tue, 14 Oct 1997 18:32:38 +0000 (18:32 +0000)]
Makefile for contributed version of awk.

26 years agoRemove old version of awk.
jraynard [Tue, 14 Oct 1997 18:29:32 +0000 (18:29 +0000)]
Remove old version of awk.

26 years agoAdd a new startup variables menu for tweaking rc.conf variables in more
jkh [Tue, 14 Oct 1997 18:17:35 +0000 (18:17 +0000)]
Add a new startup variables menu for tweaking rc.conf variables in more
detail (also added more of them to the networking services menu).  Add
new dmenuISetVariable() function for interactive setting of non-boolean
flags.

26 years agoThis commit was generated by cvs2svn to compensate for changes in r30411,
jraynard [Tue, 14 Oct 1997 18:17:11 +0000 (18:17 +0000)]
This commit was generated by cvs2svn to compensate for changes in r30411,
which included commits to RCS files with non-trunk default branches.

26 years agoVirgin import of GNU awk 3.0.3
jraynard [Tue, 14 Oct 1997 18:17:11 +0000 (18:17 +0000)]
Virgin import of GNU awk 3.0.3

26 years agoSheesh! Who's put a non-comment comment into this file?
joerg [Tue, 14 Oct 1997 16:08:11 +0000 (16:08 +0000)]
Sheesh!  Who's put a non-comment comment into this file?

26 years agoImprove my hack from rev 1.6 of displayq.c, and make the TCP
joerg [Tue, 14 Oct 1997 16:01:02 +0000 (16:01 +0000)]
Improve my hack from rev 1.6 of displayq.c, and make the TCP
connection timeout controllable by a new printcap(5) capability named
`ct' (connectiom timeout), defaulting to 120 seconds (which is the
default TCP connection timeout).

Would anybody see a problem with merging all this into RELENG_2_2?

26 years agoDeleted obsolete non-Intel CPU options.
kato [Tue, 14 Oct 1997 14:58:18 +0000 (14:58 +0000)]
Deleted obsolete non-Intel CPU options.