]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
27 years agoThe defines INTR_FAST and INTR_EXCL are part of the public interface. The
Doug Rabson [Mon, 2 Jun 1997 10:46:28 +0000 (10:46 +0000)]
The defines INTR_FAST and INTR_EXCL are part of the public interface.  The
previous commit made them private which broke things.

27 years agoChange isa_device.h to intr_machdep.h
Doug Rabson [Mon, 2 Jun 1997 10:44:08 +0000 (10:44 +0000)]
Change isa_device.h to intr_machdep.h

27 years agoMake this thing actually compile.
Doug Rabson [Mon, 2 Jun 1997 10:43:41 +0000 (10:43 +0000)]
Make this thing actually compile.

27 years agoBack out /var/run/*.pid cleaning since some programms (like innd)
Andrey A. Chernov [Mon, 2 Jun 1997 10:21:12 +0000 (10:21 +0000)]
Back out /var/run/*.pid cleaning since some programms (like innd)
use .pid to detect crash

27 years agoSynchronize with sys/i386/conf/files.i386 revision up to 1.164.
KATO Takenori [Mon, 2 Jun 1997 09:22:55 +0000 (09:22 +0000)]
Synchronize with sys/i386/conf/files.i386 revision up to 1.164.

27 years agoSynchronize with sys/i386/conf/Makefile.i386 revision 1.99.
KATO Takenori [Mon, 2 Jun 1997 09:20:46 +0000 (09:20 +0000)]
Synchronize with sys/i386/conf/Makefile.i386 revision 1.99.

27 years agoMove interrupt handling code from isa.c to a new file. This should make
Doug Rabson [Mon, 2 Jun 1997 08:19:06 +0000 (08:19 +0000)]
Move interrupt handling code from isa.c to a new file.  This should make
isa.c (slightly) more portable and will make my life developing the really
portable version much easier.

Reviewed by: peter, fsmp

27 years agoIf the localid is set, don't forget to expand it...
Peter Wemm [Mon, 2 Jun 1997 07:06:35 +0000 (07:06 +0000)]
If the localid is set, don't forget to expand it...

27 years agoGrr. Remove not-ready LDT modification fluff that also crept in with
Mike Smith [Mon, 2 Jun 1997 06:45:17 +0000 (06:45 +0000)]
Grr.  Remove not-ready LDT modification fluff that also crept in with
the last commit.

27 years agoCorrect the way IRQs are passed to rndcontrol. If there are more than
Mark Murray [Mon, 2 Jun 1997 06:43:52 +0000 (06:43 +0000)]
Correct the way IRQs are passed to rndcontrol. If there are more than
one IRQ being used, the -s options must be specified for each of them.
In this case rc.conf would only allow 1, as the -s was coded into
rc.i386.

27 years agoTypo fix.
Philippe Charnier [Mon, 2 Jun 1997 06:40:08 +0000 (06:40 +0000)]
Typo fix.

27 years agoOops, remove some bogus debugging code that crept in with the last commit.
Mike Smith [Mon, 2 Jun 1997 06:33:22 +0000 (06:33 +0000)]
Oops, remove some bogus debugging code that crept in with the last commit.

27 years agoTypo (the the -> the).
Philippe Charnier [Mon, 2 Jun 1997 06:32:26 +0000 (06:32 +0000)]
Typo (the the -> the).

27 years agoAdd support for the SIOCGIFHWADDR ioctl, commonly used by
Mike Smith [Mon, 2 Jun 1997 06:31:49 +0000 (06:31 +0000)]
Add support for the SIOCGIFHWADDR ioctl, commonly used by
license managers to obtain the host's ethernet address as
a key.

Note that this implementation takes the first hardware address for
the first ethernet interface found, and disregards the interface name
that may be passed in, as linux ethernet devices are all "ethX".

27 years agoOff by 1 adjustment. Properly NUL terminate after strncpy.
Philippe Charnier [Mon, 2 Jun 1997 06:30:06 +0000 (06:30 +0000)]
Off by 1 adjustment. Properly NUL terminate after strncpy.
Obtained from: {Net|Open}BSD

27 years agoOff by 1 adjustment.
Philippe Charnier [Mon, 2 Jun 1997 06:27:00 +0000 (06:27 +0000)]
Off by 1 adjustment.

27 years agoRemove unused variables. Suppress -Wall warnings. Check for potential buffer
Philippe Charnier [Mon, 2 Jun 1997 06:25:19 +0000 (06:25 +0000)]
Remove unused variables. Suppress -Wall warnings. Check for potential buffer
overflow. Display available command in usage string.
Obtained from: {Net|Open}BSD

27 years agoSubmitted by: Whistle Communications (archie Cobbs)
Julian Elischer [Mon, 2 Jun 1997 06:24:52 +0000 (06:24 +0000)]
Submitted by: Whistle Communications (archie Cobbs)

These changes add the ability to specify that a UFS file/directory
cannot be unlinked. This is basically a scaled back version
of the IMMUTABLE flag. The reason is to allow an administrator
to create a directory hierarchy that a group of users
can arbitrarily add/delete files from, but that the hierarchy
itself is safe from removal by them.
If the NOUNLINK definition is set to 0
then this results in no change to what happens normally.
(and results in identical binary (in the kernel)).
It can be proven that if this bit is never set by the admin,
no new behaviour is introduced..
Several "good idea" comments from reviewers plus one grumble
about creeping featurism.

This code is in production in 2.2 based systems

27 years agoSubmitted by: Whistle Communications (archie Cobbs)
Julian Elischer [Mon, 2 Jun 1997 05:02:37 +0000 (05:02 +0000)]
Submitted by: Whistle Communications (archie Cobbs)

these are quite extensive additions to the ipfw code.
they include a change to the API because the old method was
broken, but the user view is kept the same.

The new code allows a particular match to skip forward to a particular
line number, so that blocks of rules can be
used without checking all the intervening rules.
There are also many more ways of rejecting
connections especially TCP related, and
many many more ...

see the man page for a complete description.

27 years agotiny spelling fix in comment
Julian Elischer [Mon, 2 Jun 1997 04:56:38 +0000 (04:56 +0000)]
tiny spelling fix in comment

27 years agoWas superseded by ns16550.h 4-5 years ago.
Bruce Evans [Mon, 2 Jun 1997 03:10:10 +0000 (03:10 +0000)]
Was superseded by ns16550.h 4-5 years ago.

27 years agotimed_enabled -> timed_enable
Jordan K. Hubbard [Mon, 2 Jun 1997 02:58:08 +0000 (02:58 +0000)]
timed_enabled -> timed_enable
Submitted by: "James D. Stewart" <jds@c4systm.com>

27 years agoBack out last change (which in fact breaks rfc1661)
Brian Somers [Mon, 2 Jun 1997 00:04:40 +0000 (00:04 +0000)]
Back out last change (which in fact breaks rfc1661)
and modify the behaviour so that a timer is started
and the restart counter is zeroed (as per the State
Machine Diagram).

Pointed-out-by: joerg

27 years agoAdd rm -f /var/run/*.pid to prevent random process killing due to
Andrey A. Chernov [Sun, 1 Jun 1997 23:33:00 +0000 (23:33 +0000)]
Add rm -f /var/run/*.pid to prevent random process killing due to
picking old .pid file

27 years agoI've given up on the idea of sizing FIFO's. You can set it with a flag
Poul-Henning Kamp [Sun, 1 Jun 1997 20:42:01 +0000 (20:42 +0000)]
I've given up on the idea of sizing FIFO's.  You can set it with a flag
value now, if you know that you have a fifo deeper than 16.

27 years agoDon't use -fomit-frame-pointer for ipl_funcs.c if ${PROF} is nonempty,
Bruce Evans [Sun, 1 Jun 1997 20:25:55 +0000 (20:25 +0000)]
Don't use -fomit-frame-pointer for ipl_funcs.c if ${PROF} is nonempty,
is incompatible with -pg.  (We use a different version of mcount for
profiling frame-pointer-less assembler functions, but gcc doesn't know
about this.)

Added a missing dependency.

Cleaned up trailing backslashes.

Added comment about config's limitations/bugs handling dependencies and
backslashe/newlines.

Finished removing support for isdn drivers.

27 years agoMove "typedef struct intrec {} intrec" from sys/interrupt.h to kern_intr.c
Peter Wemm [Sun, 1 Jun 1997 16:05:14 +0000 (16:05 +0000)]
Move "typedef struct intrec {} intrec" from sys/interrupt.h to kern_intr.c
since that's the only place that it's used.

Submitted by: se  (apparently on suggestion from dfr)

27 years agoon second thoughts, it doesn't even need <machine/ipl.h> either.
Peter Wemm [Sun, 1 Jun 1997 16:03:13 +0000 (16:03 +0000)]
on second thoughts, it doesn't even need <machine/ipl.h> either.

27 years agoPCI_CLASS_MASS_STORAGE (under PCI_COMPAT) was used in a driver in LINT
Peter Wemm [Sun, 1 Jun 1997 16:00:43 +0000 (16:00 +0000)]
PCI_CLASS_MASS_STORAGE (under PCI_COMPAT) was used in a driver in LINT
still (stallion.c).

27 years ago<machine/spl.h> -> <machine/ipl.h>
Peter Wemm [Sun, 1 Jun 1997 15:59:11 +0000 (15:59 +0000)]
<machine/spl.h> -> <machine/ipl.h>

27 years agotypo fix, s/imp/inp'; move lookup call inside splnet since there were
Peter Wemm [Sun, 1 Jun 1997 15:58:44 +0000 (15:58 +0000)]
typo fix, s/imp/inp';  move lookup call inside splnet since there were
comments on it being outside.

27 years agoGo directly to ST_STOPPED when a TerminateReq is
Brian Somers [Sun, 1 Jun 1997 14:37:19 +0000 (14:37 +0000)]
Go directly to ST_STOPPED when a TerminateReq is
received and after the TerminateAck is sent (as
per rfc1661) rather than to ST_STOPPING.  Going
to ST_STOPPING will leave us in a state where
we're waiting for the other side to do something -
not a good idea, especially as the client side sends
a TerminateReq then exits on idle timeout.

27 years agoReduce MAX_M[RT]U back to 2048. Even on a null-modem
Brian Somers [Sun, 1 Jun 1997 11:35:04 +0000 (11:35 +0000)]
Reduce MAX_M[RT]U back to 2048.  Even on a null-modem
ppp link, we gain nothing with a larger mtu.

Suggested by: joerg

Fix pred1 buffer size (must be == MAX_MTU)

27 years agoAdd xref to signanosleep(2)
Peter Wemm [Sun, 1 Jun 1997 10:34:39 +0000 (10:34 +0000)]
Add xref to signanosleep(2)

27 years agoDoc signanosleep, add to links
Peter Wemm [Sun, 1 Jun 1997 09:47:06 +0000 (09:47 +0000)]
Doc signanosleep, add to links

27 years agoUpdate the sleep(3)/usleep(3) code to use signanosleep(2) if compiled with
Peter Wemm [Sun, 1 Jun 1997 09:27:03 +0000 (09:27 +0000)]
Update the sleep(3)/usleep(3) code to use signanosleep(2) if compiled with
-DUSE_NANOSLEEP.  Also, seperate the code for _THREAD_SAFE so that it uses
the simpler threaded nanosleep() call in libc_r..  We don't go to the same
extremes for emulating traditional sleep semantics (ie: eating any SIGALRM
that might happen) which things like apache seem to depend on.

27 years agoGenerate signanosleep(2) syscall wrapper
Peter Wemm [Sun, 1 Jun 1997 09:20:30 +0000 (09:20 +0000)]
Generate signanosleep(2) syscall wrapper

27 years agooops, fix a braino that I noticed during the commit.. Don't verify the
Peter Wemm [Sun, 1 Jun 1997 09:05:19 +0000 (09:05 +0000)]
oops, fix a braino that I noticed during the commit..  Don't verify the
remaining time pointer if it's NULL, since we don't write back in that
case! (*blush*!)

27 years ago- implement signanosleep(2) by moving common code from nanosleep() into a
Peter Wemm [Sun, 1 Jun 1997 09:01:07 +0000 (09:01 +0000)]
- implement signanosleep(2) by moving common code from nanosleep() into a
shared function.
- use p->p_sleepend to try and get more accurate "time remaining" results
when the time has been adjusted.
- verify writeability of return address so that we can fail before sleeping
if the address for the result is bogus.

27 years agoRegenerate
Peter Wemm [Sun, 1 Jun 1997 08:56:12 +0000 (08:56 +0000)]
Regenerate

27 years agoNew syscall, signanosleep(), which is a hybrid of sigsuspend(2) and
Peter Wemm [Sun, 1 Jun 1997 08:52:38 +0000 (08:52 +0000)]
New syscall, signanosleep(), which is a hybrid of sigsuspend(2) and
nanosleep(2).  It sleeps until either the time expires, or a signal
permitted by the supplied mask arrives (eg: SIGALRM if appropriate)

27 years agoNew field: p_sleepend; so that settime() can adjust the expected wakeup
Peter Wemm [Sun, 1 Jun 1997 08:49:49 +0000 (08:49 +0000)]
New field: p_sleepend; so that settime() can adjust the expected wakeup
time for things like nanosleep.  These sleep in terms of "ticks" and
calculate the elapsed time relative to the expected wakeup time and do
not return good results when the system time is adjusted.

27 years agoCreate the release name of SNAPs automagically.
Joerg Wunsch [Sun, 1 Jun 1997 08:06:48 +0000 (08:06 +0000)]
Create the release name of SNAPs automagically.

27 years agoMake this puppy compile again - add two includes.
Jordan K. Hubbard [Sun, 1 Jun 1997 05:04:39 +0000 (05:04 +0000)]
Make this puppy compile again - add two includes.

27 years agoRemove initial startup printf's in -direct mode.
Brian Somers [Sun, 1 Jun 1997 03:43:26 +0000 (03:43 +0000)]
Remove initial startup printf's in -direct mode.
Make "same magic" error send a NAK (as per rfc).

27 years agoImplement "set mtu" command to allow the client to
Brian Somers [Sun, 1 Jun 1997 01:13:04 +0000 (01:13 +0000)]
Implement "set mtu" command to allow the client to
reduce the interface mtu.
Allow max mru spec of 16k.
Add "show m[rt]u" ability.

27 years agoFix a typo.
Tor Egge [Sat, 31 May 1997 18:29:58 +0000 (18:29 +0000)]
Fix a typo.
Submitted by: Wolfgang Helbig <helbig@MX.BA-Stuttgart.De>

27 years agospecify compile-with option to get -fomit-frame-pointer on ipl_funcs.c
Peter Wemm [Sat, 31 May 1997 18:01:38 +0000 (18:01 +0000)]
specify compile-with option to get -fomit-frame-pointer on ipl_funcs.c

27 years agoRevert previous change, use "compile-with" in files.i386 instead.
Peter Wemm [Sat, 31 May 1997 17:59:56 +0000 (17:59 +0000)]
Revert previous change, use "compile-with" in files.i386 instead.

27 years agoRemove "set mtu ..." ability. Currently, this is an
Brian Somers [Sat, 31 May 1997 16:37:19 +0000 (16:37 +0000)]
Remove "set mtu ..." ability.  Currently, this is an
"alias" for "set mru ...", but there's no such thing
as setting your mtu in the ppp protocol (rfc1661).

27 years agoNow understand password in ftp:// URLs
Andrey A. Chernov [Sat, 31 May 1997 14:45:41 +0000 (14:45 +0000)]
Now understand password in ftp:// URLs

27 years agoInclude file update: <machine/spl.h> --> <machine/ipl.h>.
KATO Takenori [Sat, 31 May 1997 12:45:37 +0000 (12:45 +0000)]
Include file update: <machine/spl.h> --> <machine/ipl.h>.

27 years agoSynchronize with following changes:
KATO Takenori [Sat, 31 May 1997 12:43:17 +0000 (12:43 +0000)]
Synchronize with following changes:

>  Revision  Changes    Path
>   1.97      +2 -1      src/sys/i386/i386/trap.c
>   1.86      +2 -1      src/sys/i386/isa/clock.c
>   1.88      +2 -1      src/sys/i386/isa/isa.c
>   1.44      +3 -2      src/sys/i386/isa/npx.c

27 years agoSynchronize with sys/i386/conf/files.i386 revision 1.162.
KATO Takenori [Sat, 31 May 1997 12:39:57 +0000 (12:39 +0000)]
Synchronize with sys/i386/conf/files.i386 revision 1.162.

27 years agoSynchronize with sys/i386/conf/Makefile.i386 revision 1.98.
KATO Takenori [Sat, 31 May 1997 12:39:08 +0000 (12:39 +0000)]
Synchronize with sys/i386/conf/Makefile.i386 revision 1.98.

27 years agoBruce mentioned to me that Paul Traina had noticed that the ppp_tty
Peter Wemm [Sat, 31 May 1997 10:13:46 +0000 (10:13 +0000)]
Bruce mentioned to me that Paul Traina had noticed that the ppp_tty
interrupt mask hackery wasn't happening when being modloaded via the
if_ppp lkm.  It seems that the lkm system doesn't particularly like having
two sets of load/unload/etc routines. :-]  This really should be fixed
by having a seperate if_ppp and ppp_tty lkm, but that requires that ppp_tty
is loaded after if_ppp, and needs to be able to link with symbols in
if_ppp.  This gets messy, it is a better task for the in-kernel linker.
(if_ppp is generic, ppp_tty is a tty-specific bottom end for if_ppp, it's
 not _too_ hard to have another "provider" (such as a hdlc sync card)
 connected to if_ppp)

27 years agodon't refer to SWI_*_MASK, it's not SMP/UP kernel portable for the lkm.
Peter Wemm [Sat, 31 May 1997 09:49:35 +0000 (09:49 +0000)]
don't refer to SWI_*_MASK, it's not SMP/UP kernel portable for the lkm.

27 years agoAdd prototypes for the spl* funcs and add externs for *_imask. Leaving
Peter Wemm [Sat, 31 May 1997 09:43:22 +0000 (09:43 +0000)]
Add prototypes for the spl* funcs and add externs for *_imask.  Leaving
the *_imask down in the isa machine dependent layers requires code changes
to all pci drivers, but the interrupt registration mechanism is in flux
at the moment. These can go away when the interface is cleaned and settled.

27 years agos/intrmask/intrmask_t/g
Peter Wemm [Sat, 31 May 1997 09:33:29 +0000 (09:33 +0000)]
s/intrmask/intrmask_t/g
remove warning about ICU_LEN > 32, it's not likely to happen like this,
and besides, ICU_LEN is not the ideal indicator of "number of interrupts".

Reviewed by: bde, se

27 years ago<machine/spl.h> -> <machine/ipl.h>
Peter Wemm [Sat, 31 May 1997 09:30:39 +0000 (09:30 +0000)]
<machine/spl.h> -> <machine/ipl.h>
s/intrmask/intrmask_t/g

Reviewed by: bde, se

27 years agoInclude file updates.. <machine/spl.h> -> <machine/ipl.h>, add
Peter Wemm [Sat, 31 May 1997 09:27:31 +0000 (09:27 +0000)]
Include file updates..  <machine/spl.h> -> <machine/ipl.h>, add
<machine/ipl.h> to those files that were depending on getting SWI_*
implicitly via <machine/cpufunc.h>

27 years agoadd ipl_funcs.c (Hmm.. should probably use a "compile-with" arg rather
Peter Wemm [Sat, 31 May 1997 09:23:44 +0000 (09:23 +0000)]
add ipl_funcs.c  (Hmm.. should probably use a "compile-with" arg rather
than a Makefile.i386 hook)

27 years agocompile ipl_funcs.c with -fomit-frame-pointer, as suggested by Bruce. This
Peter Wemm [Sat, 31 May 1997 09:19:19 +0000 (09:19 +0000)]
compile ipl_funcs.c with -fomit-frame-pointer, as suggested by Bruce.  This
cuts the cost of a function call instead of an inline.

27 years agomove spl.h to ipl.h, minus the inline spl macros. The inline spl
Peter Wemm [Sat, 31 May 1997 09:16:36 +0000 (09:16 +0000)]
move spl.h to ipl.h, minus the inline spl macros.  The inline spl
implementation is now in isa/ipl_funcs.c (at least for the time being),
leaving the definitions of the SWI_* and cpl/idelayed/ipending.

27 years agoremove #include of <machine/spl.h> - they are externed now
Peter Wemm [Sat, 31 May 1997 09:13:03 +0000 (09:13 +0000)]
remove #include of <machine/spl.h> - they are externed now

Reviewed by: bde

27 years agomove intrhand2_t from isa_device.h to machine/types.h - it's used far more
Peter Wemm [Sat, 31 May 1997 09:07:36 +0000 (09:07 +0000)]
move intrhand2_t from isa_device.h to machine/types.h - it's used far more
than in the isa dependent code.
create intrmask_t in machine/types.h  (replaces "intrmask" typedef from
sys/interrupt.h)

Reviewed by: bde

27 years agoStore the macro values for SWI_TTY_MASK and SWI_NET_MASK in variables to
Peter Wemm [Sat, 31 May 1997 09:03:52 +0000 (09:03 +0000)]
Store the macro values for SWI_TTY_MASK and SWI_NET_MASK in variables to
that lkm's can use them for fiddling the masks without being dependent on
which mode the kernel is compiled in (SMP or UP).  This is particularly
for ppp_tty.c which has some domain crossing between the net and tty
subsystems.  The values are not used in the spl code, they are for
reference only (ie: the compiled code uses immediate values rather than
an indirect 32 bit address and 32 bit data fetch).

27 years agoThe SWI_NET_MASK and SWI_TTY_MASK handlers are now back adjacent to the
Peter Wemm [Sat, 31 May 1997 08:59:51 +0000 (08:59 +0000)]
The SWI_NET_MASK and SWI_TTY_MASK handlers are now back adjacent to the
top of the hardware interrupt handlers.  Apparently this is slightly
faster with the bit scanning instruction that looks these up - this set of
changes reverts the original change.

Reviewed by: bde

27 years agoBruce's original implementation of the splxxx() routines, but as C code
Peter Wemm [Sat, 31 May 1997 08:57:05 +0000 (08:57 +0000)]
Bruce's original implementation of the splxxx() routines, but as C code
rather than inlines.  These are compiled with -fomit-frame-pointer and
work out pretty close to the original routines, but it might be a fraction
slower.  The reason for doing this is to prevent the SWI_* and HWI_* values
from being compiled into drivers and lkms etc which is one of the things
that prevents the same lkm from being used on both SMP and UP kernels.

This gives us a lot more scope for experimenting with the splxxx
implementaton for SMP parallelism etc.

Reviewed by: bde

27 years agoBruce's original definitions for the spl/ipl code when running on the old
Peter Wemm [Sat, 31 May 1997 08:50:34 +0000 (08:50 +0000)]
Bruce's original definitions for the spl/ipl code when running on the old
PIC's.

Reviewed by: bde

27 years agoDefinitions for the spl/ipl code when running on an IO APIC.
Peter Wemm [Sat, 31 May 1997 08:49:08 +0000 (08:49 +0000)]
Definitions for the spl/ipl code when running on an IO APIC.

Obtained from: fsmp  (part of what Steve added for IO APIC support)
Reviewed by:   bde

27 years ago- Use `6x86MX' instead of `M2'. Cyrix officially use `6x86MX' for the
KATO Takenori [Sat, 31 May 1997 08:45:24 +0000 (08:45 +0000)]
- Use `6x86MX' instead of `M2'.  Cyrix officially use `6x86MX' for the
  CPU code-named `M2'.

- Use the result of cpuid instruction instead of DIR to identify
  6x86MX cpu.  DIR0 and DIR1 are not documented in the data sheet, and
  cpuid instruction is enabled at reset time.

- Add a function, init_6x86MX() to initialize 6x86MX cpu.  It supports
  CPU_SUSP_HLT and CPU_IORT options.  It always sets NC1 (640K - 1M is
  not cached.), and enables L1 cache in write-back mode.

- Fix typo in the comment in identblue().

27 years agoFix some old typos and misspellings.
John Polstra [Sat, 31 May 1997 04:28:53 +0000 (04:28 +0000)]
Fix some old typos and misspellings.

27 years agoModified code in direction of supporting MULTIPLE_IOAPICS.
Steve Passe [Sat, 31 May 1997 03:29:57 +0000 (03:29 +0000)]
Modified code in direction of supporting MULTIPLE_IOAPICS.

 - moved read_io_apic_maskc24() from isa/mpapic.h, disabled it, currently unused.
 - cleaned up various panic() calls.

27 years agoModified code in direction of supporting MULTIPLE_IOAPICS.
Steve Passe [Sat, 31 May 1997 03:29:06 +0000 (03:29 +0000)]
Modified code in direction of supporting MULTIPLE_IOAPICS.

 - removed set_io_apic_mask() & clr_io_apic(_mask)
 - moved read_io_apic_maskc24() to i386/mpapic.c, disabled it, currently unused.

27 years agoSubmitted by: Rich Murphey (ages ago) and Gene Stark
Gene Stark [Sat, 31 May 1997 02:39:32 +0000 (02:39 +0000)]
Submitted by:   Rich Murphey (ages ago) and Gene Stark
Hopefully I've done the proper magic to merge changes between 1.17 and
1.17.2.1 into the main trunk.  Description of those changes follows:

Brought in changes sent to me in late 1995 by Rich Murphey.
I cleaned up a few things and am currently running these under
2.2-970205-GAMMA.

The changes deal with software debouncing apparently necessary on
todays faster hardware, and also some problems with the use of the -Select
line for the TW-523 sync.  This driver allows use of +PaperEnd as an
alternative.

27 years agoThe previous fix didn't work properly for small block size filesystems,
Doug Rabson [Fri, 30 May 1997 22:25:35 +0000 (22:25 +0000)]
The previous fix didn't work properly for small block size filesystems,
which caused very slow file access for cd9660 and some ext2fs filesystems.

Reviewed by: bde

27 years agoAdd code to correctly probe all buses on the Intel XXPRESS motherboard.
Stefan Eßer [Fri, 30 May 1997 21:01:47 +0000 (21:01 +0000)]
Add code to correctly probe all buses on the Intel XXPRESS motherboard.
Add a few Intel PCI chip-set names (VX) and fix Orion entries.

27 years agoRemove use of intrec*, use void* instead.
Stefan Eßer [Fri, 30 May 1997 20:58:04 +0000 (20:58 +0000)]
Remove use of intrec*, use void* instead.
Disable test entries for wired PCI devices on bus 1.

27 years agosysctlbyname allows acces to sysctl variables by name.
Poul-Henning Kamp [Fri, 30 May 1997 20:53:13 +0000 (20:53 +0000)]
sysctlbyname allows acces to sysctl variables by name.

The manpage has been sent to linquistic decontamination and will arrive
when released from the quarantine

Reviewed by: peter

27 years agoMalloc flag X makes malloc behave like the canonical xmalloc() wrapper.
Poul-Henning Kamp [Fri, 30 May 1997 20:39:32 +0000 (20:39 +0000)]
Malloc flag X makes malloc behave like the canonical xmalloc() wrapper.

Untested support for Solaris from John-Mark Gurney

Reviewed by: phk
Submitted by: (partially) John-Mark Gurney <gurney_j@resnet.uoregon.edu>

27 years agoTypo fix.
Masafumi Max NAKANE [Fri, 30 May 1997 15:43:36 +0000 (15:43 +0000)]
Typo fix.

27 years agoSynchronize with sys/i386/isa/isa.c revision up to 1.87.
KATO Takenori [Fri, 30 May 1997 10:05:56 +0000 (10:05 +0000)]
Synchronize with sys/i386/isa/isa.c revision up to 1.87.

27 years agoSynchronize with sys/i386/i386/microtime.s revision 1.23.
KATO Takenori [Fri, 30 May 1997 10:00:58 +0000 (10:00 +0000)]
Synchronize with sys/i386/i386/microtime.s revision 1.23.

27 years agoSynchronize with sys/i386/i386/trap.c revision 1.96.
KATO Takenori [Fri, 30 May 1997 09:59:11 +0000 (09:59 +0000)]
Synchronize with sys/i386/i386/trap.c revision 1.96.

27 years agoSynchronize with sys/i386/isa/clock.c and sio.c revisions 1.85 and
KATO Takenori [Fri, 30 May 1997 09:57:09 +0000 (09:57 +0000)]
Synchronize with sys/i386/isa/clock.c and sio.c revisions 1.85 and
1.168, respectively.

27 years agoFix typos (of the printed output kind).
Jordan K. Hubbard [Fri, 30 May 1997 01:03:09 +0000 (01:03 +0000)]
Fix typos (of the printed output kind).

27 years agoDocument new -Bforcedynamic flag and -assert pure-text.
Doug Rabson [Thu, 29 May 1997 15:34:34 +0000 (15:34 +0000)]
Document new -Bforcedynamic flag and -assert pure-text.

Reviewed by: jdp

27 years agoGrr... curse my fumble-fingered hide: logmane -> logname
Bill Paul [Thu, 29 May 1997 15:03:32 +0000 (15:03 +0000)]
Grr... curse my fumble-fingered hide: logmane -> logname

Inscribed onto my chest with razor blades and then doused with
lemon juice by: Jason Young <doogie@forbidden-donut.anet-stl.com>

27 years agoAnother (harmless, luckily) mind-o: somehow I typed 'rpc.ypupdated.'
Bill Paul [Thu, 29 May 1997 14:21:39 +0000 (14:21 +0000)]
Another (harmless, luckily) mind-o: somehow I typed 'rpc.ypupdated.'
instead of 'rpc.ypupdated' when I added rpc.ypupdated to the SUBDIRS
list.

Carved into my forehead with an rusty railroad spike by: Mark Murray

27 years agoMind-o: MAN8= yes --> NOMAN= yes
Bill Paul [Thu, 29 May 1997 13:59:22 +0000 (13:59 +0000)]
Mind-o: MAN8= yes --> NOMAN= yes

Etched onto my brain with acid by: Mark Murray.

27 years agoDon't remove the controlling tty from the session if the vnode is being
Tor Egge [Thu, 29 May 1997 13:29:13 +0000 (13:29 +0000)]
Don't remove the controlling tty from the session if the vnode is being
cleaned. This should help for PR kern/3581.

27 years agoCode such as apic_base[APIC_ID] converted to lapic__id
Steve Passe [Thu, 29 May 1997 05:58:41 +0000 (05:58 +0000)]
Code such as apic_base[APIC_ID] converted to lapic__id

Changes to pmap.c for lapic_t lapic && ioapic_t ioapic pointers,
currently equal to apic_base && io_apic_base, will stand alone with the
private page mapping.

27 years agoapic.h now has structure definitions for both the local APIC and io APIC.
Steve Passe [Thu, 29 May 1997 05:57:43 +0000 (05:57 +0000)]
apic.h now has structure definitions for both the local APIC and io APIC.

apic.h has defines like:
#define lapic__id       lapic->id

Once private pages and "known virtual addr" mapping of the APICs is
ready all 'lapic__XXX' will be changed to 'lapic.XXX', and the defines
will be removed.

Changes to smp.h for lapic_t lapic && ioapic_t ioapic pointers,
currently equal to apic_base && io_apic_base, will stand alone with the
private page mapping.

27 years agoAdded code to manage the local and io APICs as structures.
Steve Passe [Thu, 29 May 1997 05:56:12 +0000 (05:56 +0000)]
Added code to manage the local and io APICs as structures.

27 years agoremove no longer needed opt_smp.h includes
Peter Wemm [Thu, 29 May 1997 05:11:11 +0000 (05:11 +0000)]
remove no longer needed opt_smp.h includes

27 years agominor style police (recent divergence from KNF code)
Peter Wemm [Thu, 29 May 1997 05:07:10 +0000 (05:07 +0000)]
minor style police (recent divergence from KNF code)

27 years agoremove opt_smp.h and fix the reason it was needed.
Peter Wemm [Thu, 29 May 1997 05:04:30 +0000 (05:04 +0000)]
remove opt_smp.h and fix the reason it was needed.

27 years agoNo longer need opt_smp.h here
Peter Wemm [Thu, 29 May 1997 05:00:35 +0000 (05:00 +0000)]
No longer need opt_smp.h here

27 years agoremove opt_smp.h from this well-included file, minor style police
Peter Wemm [Thu, 29 May 1997 04:58:04 +0000 (04:58 +0000)]
remove opt_smp.h from this well-included file, minor style police