]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoAdd note about pccard rototill and swapping breakage.
imp [Mon, 21 May 2001 16:40:50 +0000 (16:40 +0000)]
Add note about pccard rototill and swapping breakage.

23 years agoChange pmap_emulate_reference() so that it only touches the vm_page
gallatin [Mon, 21 May 2001 16:09:29 +0000 (16:09 +0000)]
Change pmap_emulate_reference() so that it only touches the vm_page
flags if it is safe to do so, otherwise it will just alter the pmap state
(eg, clear the appropriate PG_FOx bits).

This gets alpha booting in the face of the vm_mtx introduction.

Reviewed by: dfr

23 years agocatch these files up to their i386 neighbors to make alpha boot
gallatin [Mon, 21 May 2001 16:04:24 +0000 (16:04 +0000)]
catch these files up to their i386 neighbors to make alpha boot
prior to the vm_mtx

23 years agoUpdate pc98 memory probe functions.
nyan [Mon, 21 May 2001 12:51:44 +0000 (12:51 +0000)]
Update pc98 memory probe functions.
 - pc98_getmemsize() function returns available memory size under 16MB.
 - getmemsize() function is merged from PC-AT's one.

Submitted by: chi@bd.mbn.or.jp (Chiharu Shibata) and
NOKUBI Hirotaka <nokubi@ff.iij4u.or.jp>

23 years agoGet rid of LSYMSUBDIRS by merely setting up symlinks to LNOHEADERDIRS.
ru [Mon, 21 May 2001 12:37:04 +0000 (12:37 +0000)]
Get rid of LSYMSUBDIRS by merely setting up symlinks to LNOHEADERDIRS.

23 years agoMerged from sys/i386/isa/npx.c revisions 1.99 and 1.100.
nyan [Mon, 21 May 2001 12:20:22 +0000 (12:20 +0000)]
Merged from sys/i386/isa/npx.c revisions 1.99 and 1.100.

23 years agoMerged from sys/i386/i386/machdep.c revisions 1.452 and 1.453.
nyan [Mon, 21 May 2001 11:57:54 +0000 (11:57 +0000)]
Merged from sys/i386/i386/machdep.c revisions 1.452 and 1.453.

23 years agoMerged from sys/boot/i386/btx/btx/btx.s revision 1.24.
nyan [Mon, 21 May 2001 11:49:21 +0000 (11:49 +0000)]
Merged from sys/boot/i386/btx/btx/btx.s revision 1.24.

23 years agoAdded pc98 note menu.
nyan [Mon, 21 May 2001 11:25:58 +0000 (11:25 +0000)]
Added pc98 note menu.

Submitted by: Nobuyuki Koganemaru <n-kogane@syd.odn.ne.jp>

23 years agoCorrect error message of an example.
ume [Mon, 21 May 2001 10:05:36 +0000 (10:05 +0000)]
Correct error message of an example.

Obtained from: KAME

23 years agoSubmitted by: Juha-Matti Liukkonen, Cubical Solutions Ltd (jml@cubical.fi)
hm [Mon, 21 May 2001 09:24:48 +0000 (09:24 +0000)]
Submitted by: Juha-Matti Liukkonen, Cubical Solutions Ltd (jml@cubical.fi)
Reviewed by: hm

Bug in i4btel driver read routine corrected. The conditions in the
while() clause caused the receive queue to be referenced before checking
if a channel is connected, leading to kernel panic (do a 'dd
if=/dev/i4btel0 of=/dev/null' on an unconnected tel device, panic will
follow). Correction was to reorder the while clause conditions to check
for connectedness first.

23 years agoMove allocation of ExCA registers from the base driver into the bus
imp [Mon, 21 May 2001 07:32:46 +0000 (07:32 +0000)]
Move allocation of ExCA registers from the base driver into the bus
attachment code.

23 years agoMove setting of Vcc bit to before the vcc switch statement. The
imp [Mon, 21 May 2001 05:49:15 +0000 (05:49 +0000)]
Move setting of Vcc bit to before the vcc switch statement.  The
datasheets I have seem to indicate that generally this bit is viewed
as a toggle.  Correct comments to match code.

23 years agoNext step on the road to pci: power taming.
imp [Mon, 21 May 2001 04:44:14 +0000 (04:44 +0000)]
Next step on the road to pci: power taming.

Work through the various power commands and convert them from a "is
this a foo controller or a foo' controller or a foo''' controller" to
a cabability based scheme.  We have bits in the softc that tell us
what kind of power control scheme the controller uses, rather than
relying on being able to enumerate them all.  Cardbus bridges are
numerous, but nearly all implement the i82365sl-DF scheme (well, a few
implement cirrus CL-PD67xx, but those were made by Cirrus Logic!).

Add a pointer back to the softc in each pcic_slot so we can access
these flags.

Add comments that talk about the issues here.  Also note in passing
that there are two differ Vpp schemes in use and that we may need to
adjust the code to deal with both of them.  Note why it usually works
now.

We have 5 power management modes right now: KING, AB, DF, PD and VG.
AB is for the i82365 stpes A, B and C.  DF is for step DF.  PD is the
cirrus logic extensions for 3.3V while VG is the VADEM extensions for
3.3V.  KING is for the IBM KING controller found on some old cards.
# I'm looking for one of those old cards or a laptop that has the KING
# bridge in it.

We have to still cheat and treat the AB parts like the DF parts
because pci isn't here yet.  As far as I can tell, this is harmless
for actual old parts and necessary to work with 3.3V cards in some
laptops.

This almost eliminates all tests for controller in the code.  There
are still a few unrelated to power that need taming as well.

23 years agoNext step towards pcic_pci: the ability to allocate mapped memory in attach.
imp [Mon, 21 May 2001 03:22:52 +0000 (03:22 +0000)]
Next step towards pcic_pci: the ability to allocate mapped memory in attach.

o Introduce flags word to the softc.  This will be used to control various
  aspects of the driver.  Right now there are two bits defined, PCIC_IO_MAPPED
  and PCIC_MEM_MAPPED.  One for ISA cards that are I/O mapped, the other is
  for PCI cards that are memory mapped.  Only the ISA side is implemented
  with this commit.
o Introduce a pcic_dealloc which will cleanly dealloc resources used.  Right
  now it is only supported when called from probe/attach.
o Keep track of resources allocated in the pcic_softc.
o move pcictimeout_ch to the softc so we can support multiple devices
  in polling mode.
o In ISA probe, set PCIC_IO_MAPPED.
o Introduce and compute the slot mask.  This will be used later when
  we expand the number of slots on ISA from 2 to 4.  In such a case, we
  appear to have to use polling mode otherwise we get two different cards
  trying to drive the same interrupt line.  I don't have hardware to
  test this configuration, so I'll stop here.

23 years agoAdd description for 82801BA controller.
kuriyama [Mon, 21 May 2001 01:24:14 +0000 (01:24 +0000)]
Add description for 82801BA controller.

MFC after: 1 week

23 years agoTwo comments and one bug fix:
imp [Mon, 21 May 2001 00:55:44 +0000 (00:55 +0000)]
Two comments and one bug fix:
o Add defines for the VS[12]# bits in register 0x16.
o Add comment about what we're doing reading register 0x16 (PCIC_CDGC)
  in the DF case.
o Check bit VS1# rather than a random bit I was checking due to a bogus
  transcrition on my part from nakagawa-san's article.
o Add note about IBM KING and 3.3V operation from information larned from
  wildboard.

23 years agoAdd a test for PR 24345.
jasone [Sun, 20 May 2001 23:12:13 +0000 (23:12 +0000)]
Add a test for PR 24345.

23 years agoUpdate the verify script.
jasone [Sun, 20 May 2001 23:11:54 +0000 (23:11 +0000)]
Update the verify script.

23 years agoDon't define _REENTRANT, since the Makefile does so.
jasone [Sun, 20 May 2001 23:11:09 +0000 (23:11 +0000)]
Don't define _REENTRANT, since the Makefile does so.

23 years agoFix a typo.
jasone [Sun, 20 May 2001 23:10:30 +0000 (23:10 +0000)]
Fix a typo.

23 years agoInstead of using a join queue for each thread, use a single pointer to
jasone [Sun, 20 May 2001 23:08:33 +0000 (23:08 +0000)]
Instead of using a join queue for each thread, use a single pointer to
keep track of a joiner.  POSIX only supports a single joiner, so this
simplification is acceptable.

At the same time, make sure to mark a joined thread as detached so that
its resources can be freed.

Reviewed by: deischen
PR: 24345

23 years agoAquire vm mutex when releasing sysv shm segments.
alfred [Sun, 20 May 2001 20:37:47 +0000 (20:37 +0000)]
Aquire vm mutex when releasing sysv shm segments.

Obtained from: Dima Dorfman <dima@unixfreak.org>

23 years agoPrint commas in the group list in user(). current() already did this.
dd [Sun, 20 May 2001 20:32:13 +0000 (20:32 +0000)]
Print commas in the group list in user().  current() already did this.

Reviewed by: ru, sheldonh
Inspired by: OpenBSD, NetBSD

23 years agoThrow away the complications in npxsave() and their infrastructure.
bde [Sun, 20 May 2001 20:04:40 +0000 (20:04 +0000)]
Throw away the complications in npxsave() and their infrastructure.
npxsave() went to great lengths to excecute fnsave with interrupts
enabled in case executing it froze the CPU.  This case can't happen,
at least for Intel CPU/NPX's.  Spurious IRQ13's don't imply spurious
freezes.  Anyway, the complications were usually no-ops because IRQ13
is not used on i486's and newer CPUs, and because SMPng broke them in
rev.1.84.  Forcible enabling of interrupts was changed to
write_eflags(old_eflags), but since SMPng usually calls npxsave() from
cpu_switch() with interrupts disabled, write_eflags() usually just
kept interrupts disabled.

23 years agoUse a critical region to protect almost everything in npxinit().
bde [Sun, 20 May 2001 18:05:44 +0000 (18:05 +0000)]
Use a critical region to protect almost everything in npxinit().
npxinit() didn't have the usual race because it doesn't save to curpcb,
but it may have had a worse form of it since it uses the npx when it
doesn't "own" it.  I'm not sure if locking prevented this.  npxinit()
is normally caled with the proc lock but not sched_lock.

Use a critical region to protect pushing of curproc's npx state to
curpcb in npxexit().  Not doing so was harmless since it at worst
saved a wrong state to a dieing pcb.

23 years agoUse a critical region to protect pushing of curproc's npx state to
bde [Sun, 20 May 2001 17:01:20 +0000 (17:01 +0000)]
Use a critical region to protect pushing of curproc's npx state to
curpcb in vm86_bioscall().  I don't know if the state is ever in the
npx at that point.

23 years agoUse a critical region to protect saving of the npx state in savectx().
bde [Sun, 20 May 2001 16:51:08 +0000 (16:51 +0000)]
Use a critical region to protect saving of the npx state in savectx().
Not doing this was fairly harmless because savectx() is only called
for panic dumps and the bug could at worse reset the state.

savectx() is still missing saving of (volatile) debug registers, and
still isn't called for core dumps.

23 years agofix vm_mtx related compiler warning
gallatin [Sun, 20 May 2001 16:41:37 +0000 (16:41 +0000)]
fix vm_mtx related compiler warning

23 years agofix alpha-MD compile errors after the vm_mtx commit
gallatin [Sun, 20 May 2001 16:22:46 +0000 (16:22 +0000)]
fix alpha-MD compile errors after the vm_mtx commit

23 years agoUpdate softdep_setup_directory_add prototype to reflect changes in
mckusick [Sun, 20 May 2001 15:59:55 +0000 (15:59 +0000)]
Update softdep_setup_directory_add prototype to reflect changes in
actual function.

Obtained from: Jim Bloom <bloom@jbloom.jbloom.org>

23 years agoPlug memoly leak in overlaps fragment cases.
sumikawa [Sun, 20 May 2001 15:33:46 +0000 (15:33 +0000)]
Plug memoly leak in overlaps fragment cases.

Obtained from: KAME

23 years agoRemoved -I${.CURDIR}/../../sys from CFLAGS.
ru [Sun, 20 May 2001 12:45:53 +0000 (12:45 +0000)]
Removed -I${.CURDIR}/../../sys from CFLAGS.

23 years ago- Sort usage() output;
sobomax [Sun, 20 May 2001 12:17:31 +0000 (12:17 +0000)]
- Sort usage() output;
- sync usage() with reality and manpage;
- display usage() even if terminal is not a syscons.

Submitted by: ru, sobomax

23 years agoFixed `objwarn' so that it doesn't issue spurious warnings,
ru [Sun, 20 May 2001 12:14:17 +0000 (12:14 +0000)]
Fixed `objwarn' so that it doesn't issue spurious warnings,
especially now that ${.OBJDIR} is canonicalized by make(1).

Urged by: bde
Reviewed by: bde

23 years agoSpell FreeBSD.org correctly.
ru [Sun, 20 May 2001 11:39:53 +0000 (11:39 +0000)]
Spell FreeBSD.org correctly.

23 years ago- Fix markup;
sobomax [Sun, 20 May 2001 11:33:12 +0000 (11:33 +0000)]
- Fix markup;
- add nik into Authors section.

Submitted by: ru

23 years agoAdd a flag to "ipfw show" which supresses the display of dynamic
dwmalone [Sun, 20 May 2001 10:01:39 +0000 (10:01 +0000)]
Add a flag to "ipfw show" which supresses the display of dynamic
rules.  Also, don't show dynamic rules if you only asked to see a
certain rule number.

PR: 18550
Submitted by: Lyndon Nerenberg <lyndon@orthanc.ab.ca>
Approved by: luigi
MFC after: 2 weeks

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386
kris [Sun, 20 May 2001 06:21:12 +0000 (06:21 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386

MFC after:      1 week

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386
kris [Sun, 20 May 2001 06:18:37 +0000 (06:18 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386

MFC after: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386
kris [Sun, 20 May 2001 06:16:53 +0000 (06:16 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386

MFC After: 1 week

23 years agoThis passes WARNS=2 on alpha and i386
kris [Sun, 20 May 2001 06:15:09 +0000 (06:15 +0000)]
This passes WARNS=2 on alpha and i386

23 years agoSilence WARNS=2 and BDECFLAGS except for stupid mode_t warnings.
kris [Sun, 20 May 2001 05:49:20 +0000 (05:49 +0000)]
Silence WARNS=2 and BDECFLAGS except for stupid mode_t warnings.

MFC after: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386
kris [Sun, 20 May 2001 05:44:03 +0000 (05:44 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386

MFC after: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386
kris [Sun, 20 May 2001 05:37:46 +0000 (05:37 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386

MFC after: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386
kris [Sun, 20 May 2001 05:33:53 +0000 (05:33 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386

MFC After: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386
kris [Sun, 20 May 2001 05:25:37 +0000 (05:25 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386

MFC after: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386
kris [Sun, 20 May 2001 05:13:21 +0000 (05:13 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386

MFC After: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386
kris [Sun, 20 May 2001 05:10:04 +0000 (05:10 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386

MFC after: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on i386 and alpha.
kris [Sun, 20 May 2001 05:01:48 +0000 (05:01 +0000)]
Silence WARNS=2 and BDECFLAGS on i386 and alpha.

MFC After: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on alpha and i386, except for mode_t
kris [Sun, 20 May 2001 05:00:16 +0000 (05:00 +0000)]
Silence WARNS=2 and BDECFLAGS on alpha and i386, except for mode_t
warnings.

MFC after: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on i386 and alpha
kris [Sun, 20 May 2001 04:54:15 +0000 (04:54 +0000)]
Silence WARNS=2 and BDECFLAGS on i386 and alpha

MFC After: 1 week

23 years agoSilence WARNS=2 and BDECFLAGS on the i386 and alpha
kris [Sun, 20 May 2001 04:52:37 +0000 (04:52 +0000)]
Silence WARNS=2 and BDECFLAGS on the i386 and alpha

MFC After: 1 week

23 years agoSilence warnings on alpha. Unfortunately we can't add WARNS to this
kris [Sun, 20 May 2001 04:47:55 +0000 (04:47 +0000)]
Silence warnings on alpha.  Unfortunately we can't add WARNS to this
because of that stupid mode_t warning bug.

MFC After: 1 week

23 years agoMake this pass WARNS=2 and BDECFLAGS
kris [Sun, 20 May 2001 04:37:28 +0000 (04:37 +0000)]
Make this pass WARNS=2 and BDECFLAGS

23 years agoBDECFLAGS cleanup (modulo long long issues). Add WARNS. Tested on alpha.
kris [Sun, 20 May 2001 04:12:30 +0000 (04:12 +0000)]
BDECFLAGS cleanup (modulo long long issues).  Add WARNS.  Tested on alpha.

Reviewed by: the great man himself (except alpha cleanups)

23 years agoUpdate for OpenSSL 0.9.6a
kris [Sun, 20 May 2001 03:43:14 +0000 (03:43 +0000)]
Update for OpenSSL 0.9.6a

MFC after: 2 weeks

23 years agoMake this compile with WARNS=2
kris [Sun, 20 May 2001 03:24:24 +0000 (03:24 +0000)]
Make this compile with WARNS=2

23 years agoResolve conflicts
kris [Sun, 20 May 2001 03:17:35 +0000 (03:17 +0000)]
Resolve conflicts

23 years agoAdd SmartLink 5634PCV SurfRider
sanpei [Sun, 20 May 2001 03:12:55 +0000 (03:12 +0000)]
Add SmartLink 5634PCV SurfRider

PR: kern/26952
Submitted by: Simon Dick <simond@irrelevant.org>

23 years agoThis commit was generated by cvs2svn to compensate for changes in r76866,
kris [Sun, 20 May 2001 03:07:21 +0000 (03:07 +0000)]
This commit was generated by cvs2svn to compensate for changes in r76866,
which included commits to RCS files with non-trunk default branches.

23 years agoInitial import of OpenSSL 0.9.6a
kris [Sun, 20 May 2001 03:07:21 +0000 (03:07 +0000)]
Initial import of OpenSSL 0.9.6a

23 years agoDocument ECONNREFUSED.
archie [Sun, 20 May 2001 00:12:06 +0000 (00:12 +0000)]
Document ECONNREFUSED.

Submitted by: Richard Hodges <rh@matriplex.com>

23 years agoOnly add -Werror if in fact we have set WARNS or WFORMAT to >0
kris [Sat, 19 May 2001 23:32:19 +0000 (23:32 +0000)]
Only add -Werror if in fact we have set WARNS or WFORMAT to >0

23 years ago(Previous commit was made prematurely before I could finish composing
kris [Sat, 19 May 2001 23:24:17 +0000 (23:24 +0000)]
(Previous commit was made prematurely before I could finish composing
the log message)

Bring in the WARNS directive from NetBSD, localized in a new bsd.sys.mk
to avoid polluting sys.mk.  This directive controls the addition of
compiler warning flags to CFLAGS in a relatively compiler-neutral manner.

The idea is that WARNS can be set in Makefile.inc or in individual
Makefiles as they become clean, to prevent the introduction of new
warnings in the code (-Werror is added by default when WARNS > 0).

This behaviour can be overridden with the following two knobs:
NO_WARNS disables the checking of WARNS completely
NO_WERROR disables the addition of -Werror to the flags.

Additionally, setting WFORMAT to 1 (in makefiles) and/or defining
FORMAT_AUDIT (by the user) will enable increased checking of format
string arguments.

Obtained from: NetBSD
Reviewed by: peter

23 years agoBring in the WARNS directive from NetBSD, localized in a new bsd.sys.mk
kris [Sat, 19 May 2001 23:18:21 +0000 (23:18 +0000)]
Bring in the WARNS directive from NetBSD, localized in a new bsd.sys.mk
to avoid polluting sys.mk.  This directive controls the addition of
compiler warning flags to CFLAGS in a relatively compiler-neutral manner.

The idea is that WARNS can be set in Makefile.inc or in individual
Makefiles as they become clean, to prevent the introduction of new
warnings in the code.  -Werror is added by default

23 years agoFix a range checking bug in ng_int32_parse which affected 64-bit
jdp [Sat, 19 May 2001 19:36:32 +0000 (19:36 +0000)]
Fix a range checking bug in ng_int32_parse which affected 64-bit
machines.  The code formerly read:

    long val;
    if (val < (long)-0x80000000 || ...)
            return EINVAL;

The constant 0x80000000 has type unsigned int.  The unary `-'
operator does not change the type (or the value, in this case).
Therefore the promotion to long is done by 0-extension, giving
0x0000000080000000 instead of the desired 0xffffffff80000000.  I
got rid of the `-' and changed the cast to (int32_t) to give proper
sign-extension on all architectures and to better reflect the fact
that we are range-checking a 32-bit value.

This commit also makes the analogous changes to ng_int{8,16}_parse
for consistency.

MFC after: 3 days

23 years agoMust ensure that all the entries on the pd_pendinghd list have been
mckusick [Sat, 19 May 2001 19:24:26 +0000 (19:24 +0000)]
Must ensure that all the entries on the pd_pendinghd list have been
committed to disk before clearing them. More specifically, when
free_newdirblk is called, we know that the inode claims the new
directory block. However, if the associated pagedep is still linked
onto the directory buffer dependency chain, then some of the entries
on the pd_pendinghd list may not be committed to disk yet. In this
case, we will simply note that the inode claims the block and let
the pd_pendinghd list be processed when the pagedep is next written.
If the pagedep is no longer on the buffer dependency chain, then
all the entries on the pd_pending list are committed to disk and
we can free them in free_newdirblk. This corrects a window of
vulnerability introduced in the code added in version 1.95.

23 years agoMake /compat a relative symlink instead of an absolute symlink.
jkh [Sat, 19 May 2001 19:08:56 +0000 (19:08 +0000)]
Make /compat a relative symlink instead of an absolute symlink.

Requested by: jens

23 years agoEnsure the generated BPF in the example deliver the whole packet.
archie [Sat, 19 May 2001 17:47:51 +0000 (17:47 +0000)]
Ensure the generated BPF in the example deliver the whole packet.

23 years ago#include <digi/*.h> -> #include <dev/digi/*.h>
brian [Sat, 19 May 2001 17:06:48 +0000 (17:06 +0000)]
#include <digi/*.h> -> #include <dev/digi/*.h>

Suggested by: bde

23 years agoUse CONS_GETINFO ioctl to get geometry of the current text mode instead of
sobomax [Sat, 19 May 2001 17:05:52 +0000 (17:05 +0000)]
Use CONS_GETINFO ioctl to get geometry of the current text mode instead of
TIOCGWINSZ. The former is more correct in this context.

23 years agoFairwell digiio.h (moved to src/sys/sys)
brian [Sat, 19 May 2001 09:40:10 +0000 (09:40 +0000)]
Fairwell digiio.h (moved to src/sys/sys)

23 years agoBack out previous commit. digiio.h has moved to /usr/include/sys
brian [Sat, 19 May 2001 09:29:12 +0000 (09:29 +0000)]
Back out previous commit.  digiio.h has moved to /usr/include/sys

23 years agodigiio.h has moved to /usr/include/sys
brian [Sat, 19 May 2001 09:28:59 +0000 (09:28 +0000)]
digiio.h has moved to /usr/include/sys

23 years agoAnti-foot-shooting for pcvt users: ignore isdn_screenflags which is
schweikh [Sat, 19 May 2001 08:17:35 +0000 (08:17 +0000)]
Anti-foot-shooting for pcvt users: ignore isdn_screenflags which is
only for syscons.

Requested by: Alexander Leidinger <Alexander@Leidinger.net>
MFC after: 7 days

23 years agoMove scrshot(1) functionality into vidcontrol(1).
sobomax [Sat, 19 May 2001 06:47:36 +0000 (06:47 +0000)]
Move scrshot(1) functionality into vidcontrol(1).

Suggested by: many
Not objected to by: nik (scrshot co-author)

23 years agoAdd back the plain i82365 to the list of bridges that do special
imp [Sat, 19 May 2001 06:36:09 +0000 (06:36 +0000)]
Add back the plain i82365 to the list of bridges that do special
things to get 3.3V.  It appears that some cardbus chipsets have id
registers that say they are C step parts, but they really support the
DF step 3.3V functionality.

# Need to verify that IBM KING is handled properly since the MISC1
# register is really a cirrus logic only register.

23 years agoNote ncurses import and ata ioctl changes.
imp [Sat, 19 May 2001 06:27:16 +0000 (06:27 +0000)]
Note ncurses import and ata ioctl changes.

23 years agoInitialize cinfo structure at compile time rather than run time since
imp [Sat, 19 May 2001 06:17:37 +0000 (06:17 +0000)]
Initialize cinfo structure at compile time rather than run time since
they are now constant.

23 years agoslots and next haven't been used in a while. GC them.
imp [Sat, 19 May 2001 06:13:12 +0000 (06:13 +0000)]
slots and next haven't been used in a while.  GC them.

23 years agoAdd new 'loadavg' entry, fix overflow with meminfo.
jlemon [Sat, 19 May 2001 05:54:26 +0000 (05:54 +0000)]
Add new 'loadavg' entry, fix overflow with meminfo.

PR: 27253, 27350
Submitted by: Jim Pirzyk

23 years agoRemove these libs that are the same .so version as the ones in
obrien [Sat, 19 May 2001 05:54:11 +0000 (05:54 +0000)]
Remove these libs that are the same .so version as the ones in
5-CURRENT.  These libs will not be used [sitting in /usr/lib/compat] as
the -current ones in /usr/lib will be found first by ld-elf.so.1.

23 years agoRemove these libs that are the samed .so version as the ones in
obrien [Sat, 19 May 2001 05:53:50 +0000 (05:53 +0000)]
Remove these libs that are the samed .so version as the ones in
5-CURRENT.  These libs will not be used [sitting in /usr/lib/compat] as
the -current ones in /usr/lib will be found first by ld-elf.so.1.

23 years agoNow that we've moved the mecia support out of pcic.c to its own
imp [Sat, 19 May 2001 05:50:42 +0000 (05:50 +0000)]
Now that we've moved the mecia support out of pcic.c to its own
driver, we no longer need to go through the cinfo.XXXX indirections.
restore the direct calls that were replaced earlier.

23 years agoUse new kernel_sysctlbyname function. Remove private copy.
jlemon [Sat, 19 May 2001 05:48:07 +0000 (05:48 +0000)]
Use new kernel_sysctlbyname function.  Remove private copy.

23 years agoAdd convenience function kernel_sysctlbyname() for kernel consumers,
jlemon [Sat, 19 May 2001 05:45:55 +0000 (05:45 +0000)]
Add convenience function kernel_sysctlbyname() for kernel consumers,
so they don't have to roll their own sysctlbyname function.

23 years agoMove ISA specific code into pcic_isa. This is the probe routine, the
imp [Sat, 19 May 2001 05:21:23 +0000 (05:21 +0000)]
Move ISA specific code into pcic_isa.  This is the probe routine, the
get/setb1 routines.  Also expose clrb and setb as pcic_{clrb,setb} so
we can use it from the probe.  pcic_probe is no longer needed.

23 years agoIt turns out that Intel's i82365sl-DF step has the same ID as the VLSI
imp [Sat, 19 May 2001 04:53:20 +0000 (04:53 +0000)]
It turns out that Intel's i82365sl-DF step has the same ID as the VLSI
82C146.  The Intel i82365SL-DF supports 3.3V cards.  The Step A/B/C
parts do not appear to support this.  This is hard to know for sure
since it was deduced from "compatible" parts' data sheets and the
article mentioned below.

Rework the VLSI detection to be a little nicer and not depend on
scanning cards twice.  This would allow bad VLSI cards to coexist with
a good intel card, for example.  We now detect i82365SL-DF cards where
before we'd detect a VLSI.  For the most part, this is good, but we
run a small chance of detecting a single slot 82C146 as a i82365SL-DF.
Since I can't find a datasheet for the 82c146, I don't know if this is
a problem or not.

This work is based on an excellent article, in Japanese, by NAKAGAWA,
Yoshihisa-san that appeared in FreeBSD Press Number 4.  He provided a
patch against PAO3 in his article.  Since the pcic.c code has changed
some since then, I've gone ahead and cleaned up his patch somewhat and
changed how the code detects the buggy '146 cards.

I also removed the comment asking if there were other cards that
matched the 82C146 since we found one and additional information isn't
necessary.

23 years agoremove my private assertions from tsleep.
alfred [Sat, 19 May 2001 01:40:48 +0000 (01:40 +0000)]
remove my private assertions from tsleep.
add one assertion to ensure we don't sleep while holding vm.

23 years agoRegen syscalls that were made mpsafe via vm_mtx
alfred [Sat, 19 May 2001 01:37:12 +0000 (01:37 +0000)]
Regen syscalls that were made mpsafe via vm_mtx

obreak, getpagesize, sbrk, sstk, mmap, ovadvise, munmap, mprotect,
madvise, mincore, mmap, mlock, munlock, minherit, msync, mlockall,
munlockall

23 years agoUpdate the AXP compat4x dist to the 4.3-FreeBSD libs.
obrien [Sat, 19 May 2001 01:32:36 +0000 (01:32 +0000)]
Update the AXP compat4x dist to the 4.3-FreeBSD libs.
There were some security issues fixed between 4.2 -> 4.3.

23 years agoIntroduce a global lock for the vm subsystem (vm_mtx).
alfred [Sat, 19 May 2001 01:28:09 +0000 (01:28 +0000)]
Introduce a global lock for the vm subsystem (vm_mtx).

vm_mtx does not recurse and is required for most low level
vm operations.

faults can not be taken without holding Giant.

Memory subsystems can now call the base page allocators safely.

Almost all atomic ops were removed as they are covered under the
vm mutex.

Alpha and ia64 now need to catch up to i386's trap handlers.

FFS and NFS have been tested, other filesystems will need minor
changes (grabbing the vm lock when twiddling page properties).

Reviewed (partially) by: jake, jhb

23 years agoUpdate the i386 compat4x dist to the FreeBSD 4.3 libs.
obrien [Sat, 19 May 2001 01:20:02 +0000 (01:20 +0000)]
Update the i386 compat4x dist to the FreeBSD 4.3 libs.
There were some security issues fixed between 4.2 -> 4.3.

23 years agoMust be a bit less aggressive about freeing pagedep structures.
mckusick [Fri, 18 May 2001 22:16:28 +0000 (22:16 +0000)]
Must be a bit less aggressive about freeing pagedep structures.

Obtained from: Robert Watson <rwatson@FreeBSD.org> and
Matthew Jacob <mjacob@feral.com>

23 years agoThe function clnt_dg_call(), which is used for UDP RPC calls, could
iedowse [Fri, 18 May 2001 19:43:18 +0000 (19:43 +0000)]
The function clnt_dg_call(), which is used for UDP RPC calls, could
accidentally clobber the server address if a stray packet arrived
at the client port. This would result in any further retransmits
going to the wrong address.

For now, fix this by not saving the source address of the reply; this
matches the pre-tirpc behaviour.

23 years agoString operations could silently truncate long strings, leaving the output
gshapiro [Fri, 18 May 2001 18:35:34 +0000 (18:35 +0000)]
String operations could silently truncate long strings, leaving the output
corrupted.  Mark's patch fixes this be removing the MAXTOK limitation on
substring operations and allowing the putback buffer size to be the limiting
factor.  If the putback buffer size if reached, m4 gives an error instead of
silently truncating the string.

PR: bin/26619
Submitted by: Mark Peek <mark-ml@whistle.com>
MFC after: 5 days

23 years agoRestore the RSA host key to /etc/ssh/ssh_host_key.
obrien [Fri, 18 May 2001 18:10:02 +0000 (18:10 +0000)]
Restore the RSA host key to /etc/ssh/ssh_host_key.
Also fix $FreeBSD$ spamage in crypto/openssh/sshd_config rev. 1.16.

23 years agoI appear to be badly out of practice. Remember to flush the output stream
nik [Fri, 18 May 2001 14:07:51 +0000 (14:07 +0000)]
I appear to be badly out of practice.  Remember to flush the output stream
when mixing printf and write.

23 years agoGrab headers from /usr/include.
ru [Fri, 18 May 2001 13:58:08 +0000 (13:58 +0000)]
Grab headers from /usr/include.