]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
26 years agoImport the (Fast) Etherlink XL driver. I'm reasonally confident in its
wpaul [Sun, 16 Aug 1998 17:14:59 +0000 (17:14 +0000)]
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its
stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC
and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate
support for the XL adapters in the vortex driver. LAstly, add a man
page.

(Also added an MLINKS entry for the ThunderLAN man page which I forgot
previously.)

26 years agoEnable kernel dumps on SLICE systems.
des [Sun, 16 Aug 1998 11:27:19 +0000 (11:27 +0000)]
Enable kernel dumps on SLICE systems.

26 years agoFix typo in previous commit.
des [Sun, 16 Aug 1998 10:38:02 +0000 (10:38 +0000)]
Fix typo in previous commit.

PR: 7621
Submitted by: Mark Huizer

26 years agoAdd new zh_TW.BIG5 locale
jkh [Sun, 16 Aug 1998 10:32:15 +0000 (10:32 +0000)]
Add new zh_TW.BIG5 locale

26 years agoMake ELF kernels build again.
jdp [Sun, 16 Aug 1998 04:19:03 +0000 (04:19 +0000)]
Make ELF kernels build again.

26 years agoRevamp the ELF include files to better support architecture-independent
jdp [Sun, 16 Aug 1998 03:03:38 +0000 (03:03 +0000)]
Revamp the ELF include files to better support architecture-independent
applications.  Here's how it works.

The kernel should include <machine/elf.h> to get the definitions
for the native architecture.  It can reference the various ELF
structures with generic names like Elf_Sym, Elf_Shdr, etc.  A define
__ELF_WORD_SIZE is also available with the value 32 or 64 as
appropriate for the native architecture.

Generic applications should include <elf.h>, which is just a wrapper
for <machine/elf.h>.

Applications such as object file dumpers that need to deal with
foreign ELF files can include <sys/elf32.h> and/or <sys/elf64.h>.
Both can be included from the same source file if desired.  The
structure names must be referenced using wordsize-specific names
like Elf32_Sym, Elf64_Shdr, etc.

I haven't change the alpha stuff, but I haven't broken it either.

26 years agoFixed a style bug (a long line) in the previous commit.
bde [Sun, 16 Aug 1998 01:47:19 +0000 (01:47 +0000)]
Fixed a style bug (a long line) in the previous commit.

26 years agoMake provisions for a pre-build script that one can use to frob the
jkh [Sun, 16 Aug 1998 01:24:40 +0000 (01:24 +0000)]
Make provisions for a pre-build script that one can use to frob the
chroot tree right after everything has been checked out and we're ready
to go.

26 years agoUse [u]intptr_t instead of [u_]long for casts between pointers and
bde [Sun, 16 Aug 1998 01:21:52 +0000 (01:21 +0000)]
Use [u]intptr_t instead of [u_]long for casts between pointers and
integers.  Don't forget to cast to (void *) as well.

26 years agoCast an int to (intptr_t) before casting it to (void *).
bde [Sun, 16 Aug 1998 01:04:48 +0000 (01:04 +0000)]
Cast an int to (intptr_t) before casting it to (void *).

Don't cast a pointer to a long just to print it.

26 years agoAdd BINFORMAT awareness.
jkh [Sun, 16 Aug 1998 00:57:08 +0000 (00:57 +0000)]
Add BINFORMAT awareness.

26 years agoFixed yet more ioctl breakage due to the type of the `cmd' arg chaninging
bde [Sun, 16 Aug 1998 00:57:07 +0000 (00:57 +0000)]
Fixed yet more ioctl breakage due to the type of the `cmd' arg chaninging
from int to u_long but not changing here.

26 years agoTrim more out of the boot floppy so that it fits in 1.44MB again.
jkh [Sun, 16 Aug 1998 00:44:29 +0000 (00:44 +0000)]
Trim more out of the boot floppy so that it fits in 1.44MB again.

26 years agopmap.c:
bde [Sun, 16 Aug 1998 00:41:40 +0000 (00:41 +0000)]
pmap.c:
Cast pointers to (vm_offset_t) instead of to (u_long) (as before) or to
(uintptr_t)(void *) (as would be more correct).  Don't cast vm_offset_t's
to (u_long) just to do arithmetic on them.

mp_machdep.c:
Cast pointers to (uintptr_t) instead of to (u_long).  Don't forget
to cast pointers to (void *) first or to recover from integral
possible integral promotions, although this is too much work for
machine-dependent code.

vm code generally avoids warnings for pointer vs long size mismatches
by using vm_offset_t to represent pointers; pmap.c often uses plain
`unsigned int' instead of vm_offset_t and didn't use u_long elsewhere,
but this style was messed up by code apparently imported from mp_machdep.c.

26 years agoUse an array of uintptr_t's instead of an array of u_longs to hold
bde [Sun, 16 Aug 1998 00:05:05 +0000 (00:05 +0000)]
Use an array of uintptr_t's instead of an array of u_longs to hold
address constants.  This fixes some warnings for conversions from
64-bit integers to 32-bit pointers on i386's with 64-bit longs.
vm86 still uses too many u_longs.

26 years agoCast to `char *' instead of to u_long to help add byte offsets to
bde [Sat, 15 Aug 1998 23:06:38 +0000 (23:06 +0000)]
Cast to `char *' instead of to u_long to help add byte offsets to
pointers.  Neither is portable, but "correct" casts to integral
types are much uglier - they lead to expressions like

        ptr = (struct struct_with_too_long_a_name *)(void *)(uintptr_t)
    ((uintptr_t)(void *)ptr + offset);

Here the cast to the struct pointer is to match the surrounding
style of this file (and not depend on C's feature of properly
converting `void *' on assignment or cast), the `void *' casts are
because uintptr_t is only specified to work on `void *' pointers
(I missed this in about 100 lines of previous changes from [u]long
to [u]intptr_t), the outer cast to a uintptr_t is in case the
addition promoted the type, and the inner cast to a uintptr_t
corresponds to the one cast to an integer in the original code.

Don't depend on gcc's feature of casting lvalues.

26 years agoOops, the printf format error fixes confused curp->area with a pointer.
bde [Sat, 15 Aug 1998 22:42:20 +0000 (22:42 +0000)]
Oops, the printf format error fixes confused curp->area with a pointer.

26 years agoOops, the previous fix confused Linux's sigset_t with a pointer type.
bde [Sat, 15 Aug 1998 22:29:43 +0000 (22:29 +0000)]
Oops, the previous fix confused Linux's sigset_t with a pointer type.
It can be integral or a struct in POSIX, so it is difficult to print,
but it is actually declared as unsigned long.  Assume that it is
unsigned integral.

26 years agoFixed yet more ioctl breakage due to the type of the `cmd' arg changing
bde [Sat, 15 Aug 1998 21:58:09 +0000 (21:58 +0000)]
Fixed yet more ioctl breakage due to the type of the `cmd' arg changing
from int to u_long but not changing here.

26 years agoMade some disgusting ifdefs even more disgusting to enable the support
bde [Sat, 15 Aug 1998 21:51:53 +0000 (21:51 +0000)]
Made some disgusting ifdefs even more disgusting to enable the support
for `u_long cmd' ioctl args if __FreeBSD_version >= 300003.  Some ioctls
were broken on machines with 32-bit ints and 64-bit longs.

26 years agoUse offsetof() to avoid some casts from pointers to integers (of a
bde [Sat, 15 Aug 1998 21:36:16 +0000 (21:36 +0000)]
Use offsetof() to avoid some casts from pointers to integers (of a
possibly different size).

26 years agoAdd missing #include of <sys.types.h>
phk [Sat, 15 Aug 1998 20:53:34 +0000 (20:53 +0000)]
Add missing #include of <sys.types.h>

26 years agoBump version number for latest perl
markm [Sat, 15 Aug 1998 17:34:00 +0000 (17:34 +0000)]
Bump version number for latest perl

26 years agoClean up the kerberos entries, and add example CVS entries
markm [Sat, 15 Aug 1998 17:32:27 +0000 (17:32 +0000)]
Clean up the kerberos entries, and add example CVS entries

26 years agoAdd kerberised CVS to the dependant programs.
markm [Sat, 15 Aug 1998 17:28:06 +0000 (17:28 +0000)]
Add kerberised CVS to the dependant programs.

26 years agoFix LIBDIR so kerberised CVS will work.
markm [Sat, 15 Aug 1998 17:26:36 +0000 (17:26 +0000)]
Fix LIBDIR so kerberised CVS will work.

26 years agoAll but two if these build on alpha now, but most are untested.
jb [Sat, 15 Aug 1998 13:21:41 +0000 (13:21 +0000)]
All but two if these build on alpha now, but most are untested.
ldconfig isn't required.

26 years agoI have added the support for BIG5 encoding into libc/libxpg4/mklocale.
phk [Sat, 15 Aug 1998 12:51:49 +0000 (12:51 +0000)]
I have added the support for BIG5 encoding into libc/libxpg4/mklocale.
the diff is attached below. This is done on the 3.0 source-tree.
I have test this on 2.2-stable before, but I don't have a 3.0 machine
right now.

This patch is mainly to make libc support BIG5 encoding, thus add
zh_TW.BIG5 locale to 3.0.

Submitted by: Chen Hsiung Chan <frankch@waru.life.nthu.edu.tw>

26 years agoRemove half the file, somehow it got doubled...
phk [Sat, 15 Aug 1998 12:36:50 +0000 (12:36 +0000)]
Remove half the file, somehow it got doubled...

26 years agoBuild libkvm on alpha too.
jb [Sat, 15 Aug 1998 12:36:42 +0000 (12:36 +0000)]
Build libkvm on alpha too.

26 years agoNow that alpha uses FreeBSD syscalls, all these work.
jb [Sat, 15 Aug 1998 12:27:23 +0000 (12:27 +0000)]
Now that alpha uses FreeBSD syscalls, all these work.

26 years agoAdd an alpha machdep for kvm. The vatop functions are stubbed out for
dfr [Sat, 15 Aug 1998 12:12:22 +0000 (12:12 +0000)]
Add an alpha machdep for kvm.  The vatop functions are stubbed out for
now (mainly because I haven't ported them from the NetBSD crash dump
environment).

Obtained from: NetBSD

26 years agoUse explicitly sized types when laying out the cylinder groups. This
dfr [Sat, 15 Aug 1998 10:07:33 +0000 (10:07 +0000)]
Use explicitly sized types when laying out the cylinder groups.  This
bug was the cause of the 'freeing free frag' panics that people have been
seeing with FreeBSD/alpha. I have a similar patch to newfs but I've not
finished testing it.

26 years ago- change at2pc98() into global from static.
kato [Sat, 15 Aug 1998 09:08:26 +0000 (09:08 +0000)]
- change at2pc98() into global from static.
- Oops, I forgot to merge several lines from sys/i386/isa/syscons.c.

26 years agoSync with sys/i386/isa/syscons.c revision 1.274.
kato [Sat, 15 Aug 1998 09:04:03 +0000 (09:04 +0000)]
Sync with sys/i386/isa/syscons.c revision 1.274.

26 years agoBuild the makedefs program static to avoid trying to use the shared
jb [Sat, 15 Aug 1998 07:56:07 +0000 (07:56 +0000)]
Build the makedefs program static to avoid trying to use the shared
loader before it has been installed in an aout to elf transition build.

26 years agoBuild the setup program static so that we don't try to use the shared
jb [Sat, 15 Aug 1998 07:46:35 +0000 (07:46 +0000)]
Build the setup program static so that we don't try to use the shared
loader before it has been installed in a transition build from aout to elf.

26 years agoMake the locally built and executed build tools in this directory
jb [Sat, 15 Aug 1998 07:15:51 +0000 (07:15 +0000)]
Make the locally built and executed build tools in this directory
static to avoid them trying to use shared libraries before we're
ready. During the initial elf buildworld, the shared loader only
exists in the obj tree, so it can't be used.

26 years agoRemoved explict PATH which tended to cause the *wrong* path to be used
jb [Sat, 15 Aug 1998 07:10:21 +0000 (07:10 +0000)]
Removed explict PATH which tended to cause the *wrong* path to be used
more often than not.

Also included Doug Rabson's changes to make this script better handle the
two varieties of nm output we now have.

26 years agoOnly write a filemark on close when data has actually been written.
sos [Fri, 14 Aug 1998 15:54:57 +0000 (15:54 +0000)]
Only write a filemark on close when data has actually been written.

26 years agoPR: 7613
thepish [Fri, 14 Aug 1998 13:49:57 +0000 (13:49 +0000)]
PR: 7613
Submitted by: Sheldon Hearn <axl@iafrica.com>
set crt='' so that mail(1) will page according to stty(1) setting

26 years agoPR: 7613
thepish [Fri, 14 Aug 1998 12:26:43 +0000 (12:26 +0000)]
PR: 7613
Submitted by: axl@iafrica.com
change default setting of the crt env var from 24 to blank (mail(1) then uses
stty(1) setting

26 years agoChange cmd argument to dsioctl from int to u_long.
dfr [Fri, 14 Aug 1998 08:04:39 +0000 (08:04 +0000)]
Change cmd argument to dsioctl from int to u_long.

26 years agoIn /etc/rc.network, near line 242, setting up Kerberos,
phk [Fri, 14 Aug 1998 06:55:17 +0000 (06:55 +0000)]
In /etc/rc.network, near line 242, setting up Kerberos,
variable "stash_flag" is set.  A few lines later, it is evaluated
as "stash_flags" with a trailing "s", and then a bit later the
singular version is unset.

PR: 7609
Reviewed by: phk
Submitted by: Walt Howard <howard@ee.utah.edu>

26 years agoAdd "Mounting NFS file systems" progress indicator
phk [Fri, 14 Aug 1998 06:49:05 +0000 (06:49 +0000)]
Add "Mounting NFS file systems" progress indicator

PR: 7605
Reviewed by: phk
Submitted by Nick Hibma <nick.hibma@jrc.it>:

26 years ago/usr/src/usr.bin/lex/Makefile puts FlexLexer.h into /usr/include/g++,
phk [Fri, 14 Aug 1998 06:35:37 +0000 (06:35 +0000)]
/usr/src/usr.bin/lex/Makefile puts FlexLexer.h into /usr/include/g++,
but flex still generates "#include <FlexLexer.h".  As a result,
C++ sources flex generates failed to be compiled.

PR: 7575
Reviewed by: phk
Submitted by: MOROHOSHI Akihiko <moro@remus.dti.ne.jp>

26 years agoChanged \n to \r in the mouse cutbuffer, this makes pine & Emacs
sos [Fri, 14 Aug 1998 06:32:03 +0000 (06:32 +0000)]
Changed \n to \r in the mouse cutbuffer, this makes pine & Emacs
behave better when using the cut&paste functionality.

Submitted by: Barry Bierbauch <pivrnec@vszbr.cz>

26 years agoWrapper script for PC Card Ethernet Adapters only supports ISC DHCP;
phk [Fri, 14 Aug 1998 06:31:58 +0000 (06:31 +0000)]
Wrapper script for PC Card Ethernet Adapters only supports ISC DHCP;
I prefer wide-dhcp, and since both are packages, it should get
equal billing.

PR: 7568
Reviewed by: phk
Submitted by: Drew Derbyshire <software@kew.com>

26 years ago/usr/sbin/spkrtest and /usr/sbin/vidfont (aka kbdmap) use very
phk [Fri, 14 Aug 1998 06:26:58 +0000 (06:26 +0000)]
/usr/sbin/spkrtest and /usr/sbin/vidfont (aka kbdmap) use very
predictable /tmp files (static prefix + process number) which are
overwritten blindly, and follow links.

These workarounds change it from a process number to a very random
(9999 with tons of decimal places) number.

PR: 7565
Submitted by: Thomas Stromberg <tstrombe@rtci.com>

26 years agoA long time ago, in a galaxy far, far away, someone made the (bonehead)
phk [Fri, 14 Aug 1998 06:21:34 +0000 (06:21 +0000)]
A long time ago, in a galaxy far, far away, someone made the (bonehead)
decision to limit the sound driver to one device.  With Luigi's pcm
driver, the unit number for the PnP detected cards is nearly always 1,
and so you can't create the symlinks that will make your sound work.

PR: 7564
Submitted by: Doug White <dwhite@resnet.uoregon.edu>

26 years agoSubmitted by: Peter Hawkins <thepish@freebsd.org>
thepish [Fri, 14 Aug 1998 03:47:38 +0000 (03:47 +0000)]
Submitted by: Peter Hawkins <thepish@freebsd.org>
Document last mod (extended gcos)

26 years agoSubmitted by: Peter Hawkins <thepish@FreeBSD.org>
thepish [Fri, 14 Aug 1998 03:18:32 +0000 (03:18 +0000)]
Submitted by: Peter Hawkins <thepish@FreeBSD.org>
Add functionality for support for more than 4 subfields within gcos. chsh,
chpass etc did not parse beyond the 4th field previously and so truncated
gcos on updating the database.

26 years agoFix BNC/AUI autodetection.
msmith [Thu, 13 Aug 1998 20:31:17 +0000 (20:31 +0000)]
Fix BNC/AUI autodetection.
Submitted by: MIHIRA "Sanpei" Yoshiro <sanpei@sanpei.org>
Submitted by: Oleg Sharoiko <os@rsu.ru>

26 years agoUse "baseclass" instead of "class" for storing the contents of PCI register
gibbs [Thu, 13 Aug 1998 19:12:20 +0000 (19:12 +0000)]
Use "baseclass" instead of "class" for storing the contents of PCI register
0xB so that C++ programs can use the PCI conf interface.

26 years agoCleaned up previous commit, mainly by moving repetitive calculations
bde [Thu, 13 Aug 1998 19:03:22 +0000 (19:03 +0000)]
Cleaned up previous commit, mainly by moving repetitive calculations
of invariants to cyattach().

Fixed minor bugs:
- cyparam() returned without restoring the ipl in the error cases.  This
  was harmless because cyparam() is always called at spltty().
- one check for "rev. J or higher" actually checked for precisely rev. J.

26 years agoUpdated for not-so-new version of Cyclom-Y boards (with 60MHz clock and
bde [Thu, 13 Aug 1998 13:54:10 +0000 (13:54 +0000)]
Updated for not-so-new version of Cyclom-Y boards (with 60MHz clock and
swapped RTS/DTR).  Merge the vendor's modification of the 2.2.6-release
version into -current for reference.  Will be cleaned up in next commit.

Obtained from: ftp://ftp.cyclades.com/pub/cyclades/cyclom-y/freebsd/2.2.6/cyy226.tar.gz

26 years agoUpdate URL of BSD Hypertext Man Pages.
wosch [Thu, 13 Aug 1998 10:10:10 +0000 (10:10 +0000)]
Update URL of BSD Hypertext Man Pages.

26 years agoI managed to break sparse configuration accesses with my last commit.
dfr [Thu, 13 Aug 1998 08:11:27 +0000 (08:11 +0000)]
I managed to break sparse configuration accesses with my last commit.

Pointed out by: Andrew Gallatin <gallatin@cs.duke.edu>

26 years agoProtect all modifications to v_numoutput with splbio().
dfr [Thu, 13 Aug 1998 08:09:08 +0000 (08:09 +0000)]
Protect all modifications to v_numoutput with splbio().

26 years agoProtect all modifications to paging_in_progress with splvm().
dfr [Thu, 13 Aug 1998 08:05:13 +0000 (08:05 +0000)]
Protect all modifications to paging_in_progress with splvm().

26 years agoSync with sys/i386/isa/sio.c revision 1.210.
kato [Thu, 13 Aug 1998 07:36:40 +0000 (07:36 +0000)]
Sync with sys/i386/isa/sio.c revision 1.210.

26 years agoRemove some #if 0'd and commented-out code.
brian [Wed, 12 Aug 1998 23:44:22 +0000 (23:44 +0000)]
Remove some #if 0'd and commented-out code.
Fix a COMPAT_43 type.

26 years agoOne-liner: add a call to the underlying device driver's SIOCDELMULTI
wpaul [Wed, 12 Aug 1998 22:51:59 +0000 (22:51 +0000)]
One-liner: add a call to the underlying device driver's SIOCDELMULTI
ioctl() routine at the end of if_delmulti() so that interfaces with
hardware multicast filtering can update their filters in a timely
manner.

If the interface doesn't support hardware multicast filtering, then
reception of multicast frames is done using 'promiscious mode' or
'capture all multicast frames' mode and software filtering in the
kernel. In this case, it doesn't matter if if_delmulti() ever does
an SCIODELMULTI on the interface or not: if MULTICAST support is
enabled, then we join the 'all hosts' group when the interface is
configured, and remain in it until the interface is brought down.
Without hardware filtering, joining one group means joining all
groups, so it makes no difference if we call the SIOCDELMULTI
routine.

If the interface does support hardware multicast filtering, then
by not reprogramming the hardware filter in if_delmulti(), we have
to wait until somebody calls if_setmulti(), during which time the
interface is receiving frames for multicast groups in which we are
no longer interested.

26 years ago"The releaseing of the reference and lock is not temporary and belongs
msmith [Wed, 12 Aug 1998 21:42:54 +0000 (21:42 +0000)]
"The releaseing of the reference and lock is not temporary and belongs
 where it is.  The reference and lock(s) are acquired just above the
 code in VREF() and relookup()."

Submitted by: Michael Hancock <michaelh@cet.co.jp>

26 years agoHandle the case of moving a directory onto the top of a sibling's
julian [Wed, 12 Aug 1998 20:46:47 +0000 (20:46 +0000)]
Handle the case of moving a directory onto the top of a sibling's
child of the same name.

Submitted by: Kirk Mckusick with fixes from luoqi Chen
Obtained from:   Whistle test tree.

26 years agoAdd missing initialization of the length variable passed to
jdp [Wed, 12 Aug 1998 20:29:29 +0000 (20:29 +0000)]
Add missing initialization of the length variable passed to
getsockopt().  Use a separate variable for clarity, instead of
overloading an unrelated variable.

26 years agoDon't configure compatibility code for pre-Lite2 mount() calls by
bde [Wed, 12 Aug 1998 20:17:42 +0000 (20:17 +0000)]
Don't configure compatibility code for pre-Lite2 mount() calls by
default.  This code should go away soon.

26 years agoRemoved support for FreeBSD-1.x.
bde [Wed, 12 Aug 1998 18:32:42 +0000 (18:32 +0000)]
Removed support for FreeBSD-1.x.

26 years agoFixed printf format errors.
bde [Wed, 12 Aug 1998 18:16:38 +0000 (18:16 +0000)]
Fixed printf format errors.

Removed support for FreeBSD-1.x.

26 years agoRemoved unused includes.
bde [Wed, 12 Aug 1998 18:02:48 +0000 (18:02 +0000)]
Removed unused includes.

26 years agoSynced with dgb.c. Important changes from 1.29 (ioctl handling
bde [Wed, 12 Aug 1998 17:38:09 +0000 (17:38 +0000)]
Synced with dgb.c.  Important changes from 1.29 (ioctl handling
was broken), 1.30 (COMPAT_43 option header was missing), 1.31 (DEVFS
option header was missing), 1.33 (garbage pointers were followed
in debugging code).  Cosmetic changes from 1.27, 1.32, 1.36, 1.37.

Of course, the DEVFS code didn't even compile.  Fixed.  Not tested.

Forgotten by: brian

This file should not exist.  It is the same as dgb.c except for lots of
renamed variables, about 250 lines removed, and only about 100 lines of
real differences.

26 years agoFixed printf format errors.
bde [Wed, 12 Aug 1998 16:16:10 +0000 (16:16 +0000)]
Fixed printf format errors.

This file is disgusting.

26 years agoBump this one more time before going on a witch hunt for space.
jkh [Wed, 12 Aug 1998 11:49:22 +0000 (11:49 +0000)]
Bump this one more time before going on a witch hunt for space.

26 years agoBring proflibs distro back from the dead - taught about BINFORMAT.
jkh [Wed, 12 Aug 1998 11:43:24 +0000 (11:43 +0000)]
Bring proflibs distro back from the dead - taught about BINFORMAT.

26 years agoAdd missing .TP in "INTERACTIVE MODE" section.
des [Wed, 12 Aug 1998 10:38:37 +0000 (10:38 +0000)]
Add missing .TP in "INTERACTIVE MODE" section.

26 years agoMerge from stable: support for the o, order, command
wosch [Wed, 12 Aug 1998 09:58:15 +0000 (09:58 +0000)]
Merge from stable: support for the o, order, command

26 years agoMerge 1.227.2.45 (perl version substitution in PLIST).
asami [Wed, 12 Aug 1998 09:47:47 +0000 (09:47 +0000)]
Merge 1.227.2.45 (perl version substitution in PLIST).

26 years agoModify the internal interfaces to the kernel linker to make it possible
dfr [Wed, 12 Aug 1998 08:44:21 +0000 (08:44 +0000)]
Modify the internal interfaces to the kernel linker to make it possible
for DDB to use its symbol tables.

26 years agoForgot to remove a ';' in my previous commit.
charnier [Wed, 12 Aug 1998 06:07:43 +0000 (06:07 +0000)]
Forgot to remove a ';' in my previous commit.

26 years agoAdd "-aout" option when calling the linker, so that this will work
jdp [Wed, 12 Aug 1998 02:39:23 +0000 (02:39 +0000)]
Add "-aout" option when calling the linker, so that this will work
in an ELF world.  This will have to be revisited when the kernel
moves to ELF.

26 years agoMerge 1.227.2.44.
asami [Wed, 12 Aug 1998 01:47:47 +0000 (01:47 +0000)]
Merge 1.227.2.44.

26 years agoRemoved a bogus ifdef that broke using SWI_* in LKMs.
bde [Tue, 11 Aug 1998 19:21:17 +0000 (19:21 +0000)]
Removed a bogus ifdef that broke using SWI_* in LKMs.

Fixed a comment.

26 years agoFixed printf format errors (ntohl() returns in_addr_t = u_int32_t != long
bde [Tue, 11 Aug 1998 19:08:42 +0000 (19:08 +0000)]
Fixed printf format errors (ntohl() returns in_addr_t = u_int32_t != long
on some 64-bit systems).  print_ip() should use inet_ntoa() instead of
bloated inline code with 4 ntohl()s.

26 years agoBe careful about using ``.Bd -literal'' as groff sometimes keeps
brian [Tue, 11 Aug 1998 18:59:36 +0000 (18:59 +0000)]
Be careful about using ``.Bd -literal'' as groff sometimes keeps
the `-literal' after the closing .Ed.
Where this happens, use ``.Bd -unfilled'' with ``.It Li'' to dodge
the problem - it looks better too.
Problem reported by: Dom Mitchell <dom@phmit.demon.co.uk>

26 years agoDon't depend on "implicit int".
bde [Tue, 11 Aug 1998 17:22:42 +0000 (17:22 +0000)]
Don't depend on "implicit int".

26 years agoRegister tty software interrupt handlers at run time using register_swi()
bde [Tue, 11 Aug 1998 17:01:32 +0000 (17:01 +0000)]
Register tty software interrupt handlers at run time using register_swi()
instead of at compile time using ifdefs.

Use _swi_null instead of dummycamisr.  CAM and dpt should call
register_swi() instead of hacking on ihandlers[] directly.

26 years agoFixed printf format errors.
bde [Tue, 11 Aug 1998 16:06:10 +0000 (16:06 +0000)]
Fixed printf format errors.

26 years agoImplemented dynamic registration of software interrupt handlers. Not
bde [Tue, 11 Aug 1998 15:08:13 +0000 (15:08 +0000)]
Implemented dynamic registration of software interrupt handlers.  Not
used yet.

Use dummy SWI handlers to avoid some checks for null pointers.

26 years agoAdd code from NetBSD to print CIA capabilities and try to detect Pyxis bugs.
dfr [Tue, 11 Aug 1998 08:51:09 +0000 (08:51 +0000)]
Add code from NetBSD to print CIA capabilities and try to detect Pyxis bugs.

26 years agoMake /etc/security bitch about passwordless accounts.
des [Tue, 11 Aug 1998 08:48:54 +0000 (08:48 +0000)]
Make /etc/security bitch about passwordless accounts.
Use awk -F: rather than 'BEGIN {FS=":"}'

26 years agoSync with i386/isa/diskslice_machdep.c 1.31.
dfr [Tue, 11 Aug 1998 07:17:36 +0000 (07:17 +0000)]
Sync with i386/isa/diskslice_machdep.c 1.31.

26 years agoBump BOOTINODE value a bit more.
jkh [Tue, 11 Aug 1998 03:59:12 +0000 (03:59 +0000)]
Bump BOOTINODE value a bit more.

26 years agoFixed null setting of multiple wakeup prevention variable.
bde [Mon, 10 Aug 1998 20:47:22 +0000 (20:47 +0000)]
Fixed null setting of multiple wakeup prevention variable.

Fixed printf format errors.

26 years agoAdd Id keyword.
brian [Mon, 10 Aug 1998 19:54:12 +0000 (19:54 +0000)]
Add Id keyword.

26 years agoDon't blat /var/run and /var/spool/lock *after* network_pass1()
brian [Mon, 10 Aug 1998 19:53:50 +0000 (19:53 +0000)]
Don't blat /var/run and /var/spool/lock *after* network_pass1()
unless /var is nfs-mounted.
This was removing ppps device lock (when ppp's started in
/etc/start_if.tun0 in -ddial or -dedicated mode) and allowing
other programs (such as mgetty) to confuse ppp by playing with
the modem while ppp's using it.

People that nfs-mount /var had better not start ppp in this
way....

26 years agoCosmetic MF2.2: Add a missing (commented) parenthesis.
brian [Mon, 10 Aug 1998 19:53:22 +0000 (19:53 +0000)]
Cosmetic MF2.2: Add a missing (commented) parenthesis.

26 years agoDNS lookups for partially qualified hosts didn't work right. I was trying
wpaul [Mon, 10 Aug 1998 19:43:33 +0000 (19:43 +0000)]
DNS lookups for partially qualified hosts didn't work right. I was trying
to be clever by avoiding the 'check all domains in the search list'
cycle in certain cases, but this would lose if handed a name like
"foo.ctr" which refers to an FQDN of "foo.ctr.columbia.edu". If
"columbia.edu" is in the search list in /etc/resolv.conf then the
DNS lookup code should resolve it, but it didn't.

26 years agoFixed restoring of cpl after trap handling. The wrong cpl (SWI_AST_MASK
bde [Mon, 10 Aug 1998 19:41:07 +0000 (19:41 +0000)]
Fixed restoring of cpl after trap handling.  The wrong cpl (SWI_AST_MASK
instead of 0) was "restored" after handling a trap that occurred while
returning to user mode.  This bug was most noticeable for VM86 and is
still detected and fixed up (on return from the next exception) in doreti
if VM86 is configured.

26 years agoUse [u]intptr_t instead of [unsigned] long to convert and/or represent
bde [Mon, 10 Aug 1998 17:21:49 +0000 (17:21 +0000)]
Use [u]intptr_t instead of [unsigned] long to convert and/or represent
pointers.

This finishes fixing conversions between pointers and integers of
possibly different sizes in GENERIC.

26 years agoHide compiler warnings for casting from a long to a pointer of a
bde [Mon, 10 Aug 1998 17:06:28 +0000 (17:06 +0000)]
Hide compiler warnings for casting from a long to a pointer of a
different size (on i386's with 64 bit longs).  Cosmetic.  Non-cosmetic
unportabilities were already hidden by using ntohl() to convert a
32-bit user DMA address to a long.