]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoAfter a successfull poll of the cloning functions, match on the
Poul-Henning Kamp [Mon, 14 May 2001 08:20:46 +0000 (08:20 +0000)]
After a successfull poll of the cloning functions, match on the
returned dev_t rather than the original name.

This allows cloning from one name to another which is useful for
/dev/tty and later for the pty's.

23 years agoMinor style(9) changes:
Warner Losh [Mon, 14 May 2001 06:15:24 +0000 (06:15 +0000)]
Minor style(9) changes:
return (VALUE);

23 years agoFix a possible segfault introduced in my previous commit.
Maxim Sobolev [Mon, 14 May 2001 06:15:07 +0000 (06:15 +0000)]
Fix a possible segfault introduced in my previous commit.

23 years agoTake a stab at making this less dependent on having pcic as a parent.
Warner Losh [Mon, 14 May 2001 06:04:29 +0000 (06:04 +0000)]
Take a stab at making this less dependent on having pcic as a parent.
For memory for the pccard attribute/common memory mapping allocate on
the pccard.  For other allocations, use whatever is the parent of this
device.  There's no doubt other issues lurking, but this should make
things closer to being independent.

23 years agoChange machdep.pccard.pcic_mem_{start,end} to machdep.pccard.mem_{start,end}
Warner Losh [Mon, 14 May 2001 05:56:12 +0000 (05:56 +0000)]
Change machdep.pccard.pcic_mem_{start,end} to machdep.pccard.mem_{start,end}
since they have nothing to do with pcic and it makes grepping for pcic
infection in pccard harder.

MFC after: 3 days

23 years ago- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9).
Seigo Tanimura [Mon, 14 May 2001 05:26:48 +0000 (05:26 +0000)]
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9).

- Since polling should not involve sleeping, keep holding a
  process lock upon scanning file descriptors.

- Hold a reference to every file descriptor prior to entering
  polling loop in order to avoid lock order reversal between
  lockmgr and p_mtx upon calling fdrop() in fo_poll().
  (NOTE: this work has not been done for netncp and netsmb
  yet because a socket itself has no reference counts.)

Reviewed by: jhb

23 years agoDEVFS is opt in not opt out. Warn about July 1.
Warner Losh [Mon, 14 May 2001 05:11:31 +0000 (05:11 +0000)]
DEVFS is opt in not opt out.  Warn about July 1.

23 years agoAdd some clarification to the release note entry introducing the
Bruce A. Mah [Mon, 14 May 2001 05:08:09 +0000 (05:08 +0000)]
Add some clarification to the release note entry introducing the
sis(4) driver.

23 years agoGeneral corrections and fixings of the hardware list:
Bruce A. Mah [Mon, 14 May 2001 05:05:13 +0000 (05:05 +0000)]
General corrections and fixings of the hardware list:

1.  Everywhere I could figure out what driver supported a device or
class of device, there is now a cross-reference via a &man entity.
For cases where a driver has no manpage (and hence no &man entity),
we now at least give the name of the driver.  For the most part,
this was done by examining driver manpages.

2.  A number of devices which are i386-only are now marked as such,
determined by noting manpages or kernel source files in
architecture-specific directories.

3.  Added hardware supported by the vpo(4), wl(4), awi(4), and bktr(4)
drivers, based on a read of the manpages.

The manpages and source files in question were taken from 4-STABLE,
(which is what was running on my off-net laptop at the time)
but at this level of detail, I don't expect there to be any appreciable
differences between 4-STABLE and 5-CURRENT.

23 years agoWhen activating or deactivating a resource, only attempt to deal with
Warner Losh [Mon, 14 May 2001 04:53:02 +0000 (04:53 +0000)]
When activating or deactivating a resource, only attempt to deal with
the resource activation if we're dealing with our grandchild.
Otherwise, we run into two problems.  One, if the pccard layer wanted
to allocate and activate something, we'd wind up trying to do the
wrong thing twice: the ivars are wrong and we don't want the bridge to
map the resource to the slot.  If we're more than a grandchild, then
who knows what kind of ivar is present.  In either of these cases, we
just pass it up the food chain.

23 years agoMinor SGML-ification...convert ASCII characters to <quote></quote> and
Bruce A. Mah [Mon, 14 May 2001 04:49:14 +0000 (04:49 +0000)]
Minor SGML-ification...convert ASCII characters to <quote></quote> and
<emphasis></emphasis>.  No content changes.

23 years agoIf the effective link count is zero when an NFS file handle request
Kirk McKusick [Sun, 13 May 2001 23:30:45 +0000 (23:30 +0000)]
If the effective link count is zero when an NFS file handle request
comes in for it, the file is really gone, so return ESTALE.

The problem arises when the last reference to an FFS file is
released because soft-updates may delay the actual freeing of the
inode for some time. Since there are no filesystem links or open
file descriptors referencing the inode, from the point of view of
the system, the file is inaccessible. However, if the filesystem
is NFS exported, then the remote client can still access the inode
via ufs_fhtovp() until the inode really goes away. To prevent this
anomoly, it is necessary to begin returning ESTALE at the same time
that the file ceases to be accessible to the local filesystem.

Obtained from: Ian Dowse <iedowse@maths.tcd.ie>

23 years agoinstall libroken and roken.h
Assar Westerlund [Sun, 13 May 2001 20:57:48 +0000 (20:57 +0000)]
install libroken and roken.h

23 years agoConvert DEVFS from an "opt-in" to an "opt-out" option.
Poul-Henning Kamp [Sun, 13 May 2001 20:52:40 +0000 (20:52 +0000)]
Convert DEVFS from an "opt-in" to an "opt-out" option.

If for some reason DEVFS is undesired, the "NODEVFS" option is
needed now.

Pending any significant issues, DEVFS will be made mandatory in
-current on july 1st so that we can start reaping the full
benefits of having it.

23 years agoBack out rev 1.260. The 'streams' module is a registered dependency of
Peter Wemm [Sun, 13 May 2001 20:43:30 +0000 (20:43 +0000)]
Back out rev 1.260.  The 'streams' module is a registered dependency of
the 'svr4' module already.  This was a 4.x-only problem and should not
have been MFS'ed.

23 years ago+ add u_long sector_size to struct disk (documented in libdisk.3)
Jordan K. Hubbard [Sun, 13 May 2001 20:08:54 +0000 (20:08 +0000)]
+ add u_long sector_size to struct disk (documented in libdisk.3)
+ make Open_Disk sense the sector size by trying 512, 1024 and 2048
  in this order. This makes the kernel note that
  dscheck(cd1): bio_bcount 512 is not on a sector boundary (ssize 2048)
  dscheck(cd1): bio_bcount 1024 is not on a sector boundary (ssize 2048)
  if 2048 is the sector size. If this worries anyone: the message is from
  /usr/src/sys/kern/subr_diskslice.c and shutups are to be placed there.
+ Have read_block and write_block use an additional parameter, the
  sector size.
+ replace all barfout calls with return NULL, 0, __LINE__, etc.
  Note that this does NOT emit diagnostics. More often than not,
  you don't want library functions to scribble on stderr -- it may
  not even be available. The right thing is to propagate the error
  condition to upper management. The app should take care of errors.
+ use d1->sector_size instead of 512 in various places. I've left many
  places untouched, especially those writing MBRs. I simply added
  another arg hardcoded as 512. This is because I would not know what
  I'm doing... I felt this approach would be reasonably backward
  compatible and not introduce any new bugs in critical software.
  Famous last words. Messing with MBRs might soon put me in the same
  screwup meister category as, uh, never mind.  :-)
+ bump the max no of disks from 20 to 32 (due to PR 24503).

PR: 8434 / 8436 / 24503
Submitted by: Jens Schweikhardt <schweikh@schweikhardt.net>

23 years agoAdd sk_SK
Andrey A. Chernov [Sun, 13 May 2001 16:32:45 +0000 (16:32 +0000)]
Add sk_SK

PR:
Submitted by:
Reviewed by:
Approved by:
Obtained from:

23 years agoAdd sk_SK
Andrey A. Chernov [Sun, 13 May 2001 16:28:38 +0000 (16:28 +0000)]
Add sk_SK

PR: 27290
Submitted by: Juraj Bednar <juraj@bednar.sk>
Reviewed by:
Approved by:
Obtained from:

23 years agoAdd sk_SK
Andrey A. Chernov [Sun, 13 May 2001 16:18:56 +0000 (16:18 +0000)]
Add sk_SK

PR:
Submitted by:
Reviewed by:
Approved by:
Obtained from:

23 years agoAdd sk_SK
Andrey A. Chernov [Sun, 13 May 2001 16:13:49 +0000 (16:13 +0000)]
Add sk_SK

PR:
Submitted by:
Reviewed by:
Approved by:
Obtained from:

23 years agoUse a critical region to protect pushing of the parent's npx state to the
Bruce Evans [Sun, 13 May 2001 07:44:14 +0000 (07:44 +0000)]
Use a critical region to protect pushing of the parent's npx state to the
pcb for fork().  It was possible for the state to be saved twice when an
interrupt handler saved it concurrently.  This corrupted (reset) the state
because fnsave has the (in)convenient side effect of doing an implicit
fninit.  Mundane null pointer bugs were not possible, because we save to
an "arbitrary" process's pcb and not to the "right" place (npxproc).

Push the parent's %gs to the pcb for fork().  Changes to %gs before
fork() were not preserved in the child unless an accidental context
switch did the pushing.  Updated the list of pcb contents which is
supposed to inhibit bugs like this.  pcb_dr*, pcb_gs and pcb_ext were
missing.  Copying is correct for pcb_dr*, and pcb_ext is already
handled specially (although XXX'ly).

Reducing the savectx() call to an npxsave() call in rev.1.80 was a
mistake.  The above bugs are duplicated in many places, including in
savectx() itself.

The arbitraryness of the parent process pointer for the fork()
subroutines, the pcb pointer for savectx(), and the save87 pointer
for npxsave(), is illusory.  These functions don't work "right" unless
the pointers are precisely curproc, curpcb, and the address of npxproc's
save87 area, respectively, although the special context in which they
are called allows savectx(&dumppcb) to sort of work and npxsave(&dummy)
to work.  cpu_fork() just doesn't work unless the parent process
pointer is curproc, or the caller has pushed %gs to the pcb, or %gs
happens to already be in the pcb.

23 years agoRemove safety belt that checks for miibus in the config file. This
Jonathan Lemon [Sun, 13 May 2001 05:38:59 +0000 (05:38 +0000)]
Remove safety belt that checks for miibus in the config file.  This
was only intended for -stable, not -current.

23 years agoReturn errors for unsupported operations on pcic_get_res_flags rather
Warner Losh [Sun, 13 May 2001 04:44:45 +0000 (04:44 +0000)]
Return errors for unsupported operations on pcic_get_res_flags rather
than 0.

23 years agoChange #ifdef PC98 to #ifdef MECIA_SUPPORT and define MECIA_SUPPORT
Warner Losh [Sun, 13 May 2001 04:16:09 +0000 (04:16 +0000)]
Change #ifdef PC98 to #ifdef MECIA_SUPPORT and define MECIA_SUPPORT
when PC98 is defined.  This is in perparation for a mecia driver
separate from pcic, assuming that all goes well with that effort.
MECIA_SUPPORT won't be removed until after that support is working.

23 years agoI'll be making some rather substantial changes to the pci attachment
Warner Losh [Sun, 13 May 2001 01:52:55 +0000 (01:52 +0000)]
I'll be making some rather substantial changes to the pci attachment
of the pcic class of devices.  Go ahead and move it to the "usual"
place.  I say "usual" in quotes since it isn't exactly right (not in
dev/blah), but it is closer than before.

23 years agoo Get rid of static array of slots in pccard layer. Move this to the
Warner Losh [Sun, 13 May 2001 01:44:27 +0000 (01:44 +0000)]
o Get rid of static array of slots in pccard layer.  Move this to the
  softc.
o Store pointers to softc in dev_t in si_drv1.
o Change 'kludge version' to 'classic version' since things are getting less
  kludgy.
o Minor code shuffling so that we probe and attach the pccard slots.
o Minor style(9) changes.

23 years ago- Split out the {family,socktype}->netid conversion into a separate
Ian Dowse [Sun, 13 May 2001 00:07:03 +0000 (00:07 +0000)]
- Split out the {family,socktype}->netid conversion into a separate
  function; we now handle unknown protocols more gracefully.

- Cache the return from getnetconfigent() so that we don't have to
  remember to call freenetconfigent() each time. This fixes a memory
  leak that would cause retrying background mount_nfs processes to
  slowly increase their memory usage.

23 years agoUpdate makefile to reflect vlan support.
Jonathan Lemon [Sun, 13 May 2001 00:04:02 +0000 (00:04 +0000)]
Update makefile to reflect vlan support.

23 years agoAdd few cosmetic style fixes, and some debug information for SCB timeouts.
Jonathan Lemon [Sun, 13 May 2001 00:03:39 +0000 (00:03 +0000)]
Add few cosmetic style fixes, and some debug information for SCB timeouts.
Add VLAN support, obtained from Pedro J. Lobo (through Mike Tancsa).

23 years agoAdd a few more register definitions.
Jonathan Lemon [Sat, 12 May 2001 23:59:48 +0000 (23:59 +0000)]
Add a few more register definitions.

23 years agoRevert part of last commit. Instead of using %fs for KSD/TSD, we'll
Daniel Eischen [Sat, 12 May 2001 22:54:53 +0000 (22:54 +0000)]
Revert part of last commit.  Instead of using %fs for KSD/TSD, we'll
follow Linux' convention and use %gs.  This adds back the setting of
%fs to a sane value in sendsig().  The value of %gs remains preserved
to whatever it was in user context.

23 years agoAdd some gravitational values.
Eric Melville [Sat, 12 May 2001 21:21:38 +0000 (21:21 +0000)]
Add some gravitational values.

Approved by: murray

23 years agoExtract the path from an AF_LOCAL sockaddr_un in a way that correctly
Ian Dowse [Sat, 12 May 2001 20:05:26 +0000 (20:05 +0000)]
Extract the path from an AF_LOCAL sockaddr_un in a way that correctly
terminates the string in all cases, based on code from netstat(1).
The path in a sockaddr_un is terminated either by a '\0', or by
the end of the sockaddr as defined by sun_len.

Previously, the code could write the "safety" '\0' beyond the end
of the sockaddr (sockaddr_un's need only be large enough to store
sun_len bytes), and writing into the the supplied sockaddr is bad
anyway.

23 years agoUnbreak release. *sigh*
Bill Paul [Sat, 12 May 2001 19:51:40 +0000 (19:51 +0000)]
Unbreak release. *sigh*

23 years agoCheck for both hint files in /var/run (elf and aout) and eventually create
Jimmy Olgeni [Sat, 12 May 2001 19:39:13 +0000 (19:39 +0000)]
Check for both hint files in /var/run (elf and aout) and eventually create
them.

Sysinstall used to check /var/run/ld.so.hints (aout related) and create
hints with the ldconfig command, but the ldconfig command alone will
generate elf hints only. The correct behavior is:

* If /var/run/ld-elf.so.hints does not exist, generate elf hints

* If /var/run/ld.so.hints does not exist, generate aout hints
  (using ldconfig with the -aout option)

This will help ports that check for aout libraries using ldconfig in their
pkg-req scripts.

Approved by: jkh
MFC after: 1 weeks

23 years agoclarify comment about MAKE_KERBEROS5. noticed by Peter Pentchev
Assar Westerlund [Sat, 12 May 2001 14:38:00 +0000 (14:38 +0000)]
clarify comment about MAKE_KERBEROS5.  noticed by Peter Pentchev
<roam@orbitel.bg>

23 years agoFixed a missing library in DPADD.
Bruce Evans [Sat, 12 May 2001 14:37:53 +0000 (14:37 +0000)]
Fixed a missing library in DPADD.

Fixed some style bugs.

23 years agoFixed missing DPADD.
Bruce Evans [Sat, 12 May 2001 14:34:46 +0000 (14:34 +0000)]
Fixed missing DPADD.

Don't clobber CFLAGS by assigning to it using "=".

Fixed some style bugs.

23 years agoFixed some bitrot:
Bruce Evans [Sat, 12 May 2001 14:21:14 +0000 (14:21 +0000)]
Fixed some bitrot:
- the changes that renamed libf2c to libg2c had not reached here
- there were no definitions for LIBDEVINFO, LIBMENU, LIBPANEL, LIBTINFO,
  LIBUSB or LIBVGL.  LIBUSB was used without it being defined, and
  LIBDEVINFO and LIBVGL should have been used.
- the definitions of LIBDESCRYPT, LIBGCC_PIC, LIBGPLUSPLUS, LIBKZHEAD,
  LIBKZTAIL, LIBSCRYPT and LIBSCSI were garbage.

Fixed some old bugs:
- LIBC_PIC and LIBCOM_ERR were assigned to using "=" instead of "?=".
- the definition of LIBC_R was disordered.
- LIBFORM was misspelled LIBFORMS (but not actually used).

23 years agoFix typo introduced in a comment in the previous commit
Kris Kennaway [Sat, 12 May 2001 10:28:47 +0000 (10:28 +0000)]
Fix typo introduced in a comment in the previous commit

23 years agoClean up some of the evil string handling in this code, replace
Kris Kennaway [Sat, 12 May 2001 09:44:32 +0000 (09:44 +0000)]
Clean up some of the evil string handling in this code, replace
warnx()+exit() with errx() and replace a big if..then..else construct
to determine the package download directory with a lookup table.

Reviewed by: jkh
MFC after: 2 weeks

23 years agoTake keyboard map files from ${.CURDIR}/../../share/syscons/keymaps, not from
Maxim Sobolev [Sat, 12 May 2001 09:19:36 +0000 (09:19 +0000)]
Take keyboard map files from ${.CURDIR}/../../share/syscons/keymaps, not from
/usr/share/syscons/keymaps. This should prevent word breakage when new keymaps
have been added.

Prompted by: Matthew D. Fuller <fullermd@over-yonder.net>

23 years agoAllow path where keyboard maps are looked for to be altered using environment
Maxim Sobolev [Sat, 12 May 2001 09:16:09 +0000 (09:16 +0000)]
Allow path where keyboard maps are looked for to be altered using environment
variable.

23 years agoTh -> The in a comment
Warner Losh [Sat, 12 May 2001 06:04:02 +0000 (06:04 +0000)]
Th -> The in a comment

23 years agoChoose the right LIB_SPEC for threads based on "__FreeBSD_version".
David E. O'Brien [Sat, 12 May 2001 04:57:49 +0000 (04:57 +0000)]
Choose the right LIB_SPEC for threads based on "__FreeBSD_version".
This allows someone with an older -current to update their compiler [only].

Inspired by: Loren James Rittle <rittle@rsch.comm.mot.com>

23 years agoNew release notes: ELF rtld(1) performance improvements, catch up
Bruce A. Mah [Sat, 12 May 2001 04:19:26 +0000 (04:19 +0000)]
New release notes:  ELF rtld(1) performance improvements, catch up
RELNOTESng with RELNOTES.TXT for nge(4) entries (committed originally
by wpaul).

MFCs noted:  ifconfig(8) slash/CIDR support

23 years agoPreserve the state of the %gs register when setting up the signal
Daniel Eischen [Sat, 12 May 2001 03:23:10 +0000 (03:23 +0000)]
Preserve the state of the %gs register when setting up the signal
handler in Linux emulation.  According to bde, this is what Linux
does.

Recent versions of linuxthreads use %gs for thread-specific data,
while FreeBSD uses %fs (mostly because WINE uses %gs).

Tested by: drew

23 years agoFix a spelling error.
Dima Dorfman [Sat, 12 May 2001 02:26:58 +0000 (02:26 +0000)]
Fix a spelling error.

23 years agoDocument the -h option.
Dima Dorfman [Fri, 11 May 2001 23:53:46 +0000 (23:53 +0000)]
Document the -h option.

23 years agoSimplify the vm fault trap handling code a bit by using if-else instead of
John Baldwin [Fri, 11 May 2001 23:50:08 +0000 (23:50 +0000)]
Simplify the vm fault trap handling code a bit by using if-else instead of
duplicating code in the then case and then using a goto to jump around
the else case.

23 years agoBring the interface IFF_UP from ipcp_InterfaceUp() (called from IPCP's
Brian Somers [Fri, 11 May 2001 23:42:03 +0000 (23:42 +0000)]
Bring the interface IFF_UP from ipcp_InterfaceUp() (called from IPCP's
TLU event handler).

This used to be done as a side effect of SIOCAIFADDR'ing the interface,
but now that duplicate SIOCAIFADDRs are optimised out, we can't depend
on that behaviour.

23 years agoGC prototype for procfs_bmap() missed during a previous commit.
John Baldwin [Fri, 11 May 2001 23:37:37 +0000 (23:37 +0000)]
GC prototype for procfs_bmap() missed during a previous commit.

23 years agoSimply the vm fault trap handling code a bit by using if-else instead of
John Baldwin [Fri, 11 May 2001 23:17:54 +0000 (23:17 +0000)]
Simply the vm fault trap handling code a bit by using if-else instead of
duplicating code in the then case and then using a goto to jump around
the else case.

23 years agoAllow sendsig() to hand the proc lock off to osendsig() rather than having
John Baldwin [Fri, 11 May 2001 23:17:05 +0000 (23:17 +0000)]
Allow sendsig() to hand the proc lock off to osendsig() rather than having
sendsig() release the lock just so it can call osendsig() and have
osendsig() acquire the lock.

23 years agoAdd a new macro to test if a process' proc lock is held by the current
John Baldwin [Fri, 11 May 2001 21:28:37 +0000 (21:28 +0000)]
Add a new macro to test if a process' proc lock is held by the current
thread: PROC_LOCKED().

23 years agoIt's vlan.h, not opt_vlan.h.
Bill Paul [Fri, 11 May 2001 20:55:31 +0000 (20:55 +0000)]
It's vlan.h, not opt_vlan.h.

23 years agoIn vrele() and vput(), avoid triggering the confusing "missed vn_close"
Ian Dowse [Fri, 11 May 2001 20:42:41 +0000 (20:42 +0000)]
In vrele() and vput(), avoid triggering the confusing "missed vn_close"
KASSERT when vp->v_usecount is zero or negative. In this case, the
"v*: negative ref cnt" panic that follows is much more appropriate.

Reviewed by: mckusick

23 years agoRegenerate.
Jonathan Lemon [Fri, 11 May 2001 20:41:20 +0000 (20:41 +0000)]
Regenerate.

23 years agoCorrectly recognize the i82562{EM} PHYs.
Jonathan Lemon [Fri, 11 May 2001 20:34:38 +0000 (20:34 +0000)]
Correctly recognize the i82562{EM} PHYs.

Obtained from: OpenBSD

23 years agoRegenerate
Bill Paul [Fri, 11 May 2001 20:27:39 +0000 (20:27 +0000)]
Regenerate

23 years agoCheck witness_dead in more functions to avoid panic'ing when assertions
John Baldwin [Fri, 11 May 2001 20:25:29 +0000 (20:25 +0000)]
Check witness_dead in more functions to avoid panic'ing when assertions
fail due to witness exhausting its internal resources and shutting down.

Reported by: Szilveszter Adam <sziszi@petra.hos.u-szeged.hu>
Tested by: David Wolfskill <david@catwhisker.org>

23 years agoDisable the 'stats counter overflow' interrupts since they can happen
Bill Paul [Fri, 11 May 2001 20:15:41 +0000 (20:15 +0000)]
Disable the 'stats counter overflow' interrupts since they can happen
much more often that expected and negatively impact performance when
running at 100mbps. I need to figure out if there's a better way to
handle this, but for now this shouldn't hurt anything.

23 years agoAdd support for gigabit ethernet cards based on the NatSemi DP83820
Bill Paul [Fri, 11 May 2001 19:56:39 +0000 (19:56 +0000)]
Add support for gigabit ethernet cards based on the NatSemi DP83820
and DP83821 gigabit ethernet MAC chips and the NatSemi DP83861 10/100/1000
copper PHY. There are a whole bunch of very low cost cards available with
this chipset selling for $150USD or less. This includes the SMC9462TX,
D-Link DGE-500T, Asante GigaNIX 1000TA and 1000TPC, and a couple cards
from Addtron.

This chip supports TCP/IP checksum offload, VLAN tagging/insertion.
2048-bit multicast filter, jumbograms and has 8K TX and 32K RX FIFOs.
I have not done serious performance testing with this driver. I know
it works, and I want it under CVS control so I can keep tabs on it.
Note that there's no serious mutex stuff in here yet either: I need
to talk more with jhb to figure out the right way to do this. That
said, I don't think there will be any problems.

This driver should also work on the alpha. It's not turned on in
GENERIC.

23 years agoAS800 uses EDO DIMMs, not SDRAM
Wilko Bulte [Fri, 11 May 2001 18:02:37 +0000 (18:02 +0000)]
AS800 uses EDO DIMMs, not SDRAM

Submitted by: Brodie Hynes <brodie@alum.mit.edu>

23 years agoAS800 uses EDO DIMMs, not SDRAM DIMMs.
Wilko Bulte [Fri, 11 May 2001 18:01:27 +0000 (18:01 +0000)]
AS800 uses EDO DIMMs, not SDRAM DIMMs.

Submitted by: Brodie Hynes <brodie@alum.mit.edu>

23 years agoRegenerate.
Tor Egge [Fri, 11 May 2001 17:05:47 +0000 (17:05 +0000)]
Regenerate.

23 years agogettimeofday() is MP safe on both -current and -stable.
Tor Egge [Fri, 11 May 2001 17:05:12 +0000 (17:05 +0000)]
gettimeofday() is MP safe on both -current and -stable.

23 years agode-constify to make it compatible with (krb4) and make-print-version
Assar Westerlund [Fri, 11 May 2001 16:56:36 +0000 (16:56 +0000)]
de-constify to make it compatible with (krb4) and make-print-version

Submitted by: Peter Pentchev <roam@orbitel.bg> (just inverted)

23 years agoIn in_ifadown(), differentiate between whether the interface goes
Ruslan Ermilov [Fri, 11 May 2001 14:37:34 +0000 (14:37 +0000)]
In in_ifadown(), differentiate between whether the interface goes
down or interface address is deleted.  Only delete static routes
in the latter case.

Reported by: Alexander Leidinger <Alexander@leidinger.net>

23 years agoAdd utility PAMs for finer userland control
Mark Murray [Fri, 11 May 2001 10:12:55 +0000 (10:12 +0000)]
Add utility PAMs for finer userland control

23 years agoFix PAMized telnet in exactly the same way as BDE did it for the other
Mark Murray [Fri, 11 May 2001 09:39:11 +0000 (09:39 +0000)]
Fix PAMized telnet in exactly the same way as BDE did it for the other
telnet instances.

23 years agoFix make world in the kerberosIV case.
Mark Murray [Fri, 11 May 2001 09:36:17 +0000 (09:36 +0000)]
Fix make world in the kerberosIV case.

23 years agoAdd 470.status-named.
Dirk Froemberg [Fri, 11 May 2001 09:32:48 +0000 (09:32 +0000)]
Add 470.status-named.

Reminded by: gshapiro

23 years agoAdd exact names for suggested screenmaps
Andrey A. Chernov [Fri, 11 May 2001 08:45:34 +0000 (08:45 +0000)]
Add exact names for suggested screenmaps

23 years agoUpdate the fsck command in /etc/rc to use the new background
Kirk McKusick [Fri, 11 May 2001 07:40:39 +0000 (07:40 +0000)]
Update the fsck command in /etc/rc to use the new background
fsck checking. Applying these changes (typically via mergemaster)
will cause your system to start running background checks on all
your soft update enabled filesystems (provided that you have
a kernel with the required functionality, e.g., one built since
the end of April). Please report any and all problems to
mckusick@mckusick.com (not mckusick@freebsd.org which I read
infrequently). See the comment above the fsck command in /etc/rc
for instructions on how to disable background checking should it
cause you too much trouble.

Several FAQs:

1) Can I reboot before the background checks are done?

Ans) Yes, when the system restarts the checks will pick up
     where they left off.

2) Can a crash during checking corrupt my filesystem?

Ans) No, recovered resources are returned to the system using soft
     updates which ensure that the freeing is done in a safe order.

3) How will I know if any background checks are being done?

Ans) Filesystems that are to be checked in background will be listed
     as `DEFER FOR BACKGROUND CHECKING' at the usual fsck check time
     during system startup.

4) What happens to the output of the background checks?

Ans) It is sent to syslog `daemon' facility log level `notice'.

5) When will this feature be available in the 4.X kernel?

Ans) Never. It is much too radical and extensive a change to be
     MFC'ed. Besides, it needs many months of experience and
     tuning before it is ready for widespread use.

6) What happens if a background fsck fails (i.e., fsck finds
   errors that would normally require a manual fsck)?

Ans) The filesystem will be marked as needing a manual fsck.
     At the next system reboot, the check will be done in
     foreground and the usual actions taken (usually a failure
     to go multi-user until fsck has been run by hand on the
     affected filesystem).

23 years agoSync picobsd script with the one in -stable, and also
Luigi Rizzo [Fri, 11 May 2001 07:39:40 +0000 (07:39 +0000)]
Sync picobsd script with the one in -stable, and also
patch boot2 so it goes to load /kernel right away

Perl code from Josef Karthauser

23 years agoRemove yet another deadlock case.
Kirk McKusick [Fri, 11 May 2001 07:12:03 +0000 (07:12 +0000)]
Remove yet another deadlock case.

23 years agoFix world-breaking typo in previous commit.
Greg Lehey [Fri, 11 May 2001 07:06:06 +0000 (07:06 +0000)]
Fix world-breaking typo in previous commit.

23 years agoUn-swap irq/link byte values so that printf works.
Mike Smith [Fri, 11 May 2001 04:52:29 +0000 (04:52 +0000)]
Un-swap irq/link byte values so that printf works.

23 years agoAllow mail(1) to be able to read Eudora mailboxes by transforming
Mike Heffner [Fri, 11 May 2001 03:07:11 +0000 (03:07 +0000)]
Allow mail(1) to be able to read Eudora mailboxes by transforming
lines that end in <CR><LF> to just <LF>.

Reviewed by: imp
Obtained from: OpenBSD

23 years agomerge imported changes into HEAD
Assar Westerlund [Fri, 11 May 2001 00:14:02 +0000 (00:14 +0000)]
merge imported changes into HEAD

23 years agoNew release notes: statfs(2) and df(1) track blocks committed to being
Bruce A. Mah [Thu, 10 May 2001 23:38:32 +0000 (23:38 +0000)]
New release notes:  statfs(2) and df(1) track blocks committed to being
freed by softupdates, ifconfig(8) accepts CIDR notation, rc(8) clean-out
of /var/run and /var/spool/lock, c89(1) is now a binary, pax(1)
enhancements and cpio(1)/tar(1) compatability, Ukranian language console
support.

Other:  Update/make (more) consistent the list of WaveLAN devices
supported.

MFCs noted:  ln(1) -h/-n, find(1) timestamp flags.

23 years agoAdd information about the 2400A controllers. Fix erroneous reference to
Scott Long [Thu, 10 May 2001 22:34:45 +0000 (22:34 +0000)]
Add information about the 2400A controllers.  Fix erroneous reference to
BSD/OS.

23 years agoSync KOI8-U with latest KOI8-R changes. Also restore correct numbering
Maxim Sobolev [Thu, 10 May 2001 21:04:36 +0000 (21:04 +0000)]
Sync KOI8-U with latest KOI8-R changes. Also restore correct numbering
in the font selection dialog broken by the removal of KOI8-R.

Prompted by: ache
MFC after: 2 weeks

23 years agoBring the doc up to date with regard to the /etc bootstrapping code
Brian S. Dean [Thu, 10 May 2001 20:50:54 +0000 (20:50 +0000)]
Bring the doc up to date with regard to the /etc bootstrapping code
and the role of /conf/default/etc.

23 years agoMFS: bring Luigi's updated diskless man page into -current.
Brian S. Dean [Thu, 10 May 2001 20:45:46 +0000 (20:45 +0000)]
MFS: bring Luigi's updated diskless man page into -current.

23 years agoAdd Romania to ftp mirrors.
Jordan K. Hubbard [Thu, 10 May 2001 18:38:25 +0000 (18:38 +0000)]
Add Romania to ftp mirrors.

23 years agoTrim lots of stuff that is now in MI code along with MD alpha code.
John Baldwin [Thu, 10 May 2001 17:58:35 +0000 (17:58 +0000)]
Trim lots of stuff that is now in MI code along with MD alpha code.

23 years ago- Split out the support for per-CPU data from the SMP code. UP kernels
John Baldwin [Thu, 10 May 2001 17:45:49 +0000 (17:45 +0000)]
- Split out the support for per-CPU data from the SMP code.  UP kernels
  have per-CPU data and gdb on the i386 at least needs access to it.
- Clean up includes in kern_idle.c and subr_smp.c.

Reviewed by: jake

23 years agoAdd mousechar_start=3 for Russian font
Andrey A. Chernov [Thu, 10 May 2001 17:32:31 +0000 (17:32 +0000)]
Add mousechar_start=3 for Russian font
(-current only)

23 years agoTry to read the station address twice during the probe. I've seen
Bill Paul [Thu, 10 May 2001 17:17:24 +0000 (17:17 +0000)]
Try to read the station address twice during the probe. I've seen
a LinkSys card here in the office where reading the station address
fails the first time, but works find afterwards. Without this, the
probe fails. I don't think this will negatively impact any existing
cards, but I want to confirm this before MFC'ing.

23 years agoHandle sbsize as a size, not a number.
Sheldon Hearn [Thu, 10 May 2001 17:15:06 +0000 (17:15 +0000)]
Handle sbsize as a size, not a number.

Submitted by: Paul Herman <pherman@frenchfries.net>

23 years agoMove cons25w entry last, index is off by one otherwise if RunningAsInit
Andrey A. Chernov [Thu, 10 May 2001 17:14:29 +0000 (17:14 +0000)]
Move cons25w entry last, index is off by one otherwise if RunningAsInit

23 years agoRemove CP866 keyboard and KOI8-R font to not confuse users.
Andrey A. Chernov [Thu, 10 May 2001 17:03:14 +0000 (17:03 +0000)]
Remove CP866 keyboard and KOI8-R font to not confuse users.
Explain more about CP866 font usage.

23 years ago- Use sched_lock and critical regions to ensure that LDT updates are thread
John Baldwin [Thu, 10 May 2001 17:03:03 +0000 (17:03 +0000)]
- Use sched_lock and critical regions to ensure that LDT updates are thread
  safe from preemption and concurrent access to the LDT.
- Move the prototype for i386_extend_pcb() to <machine/pcb_ext.h>.

Reviewed by: silence on -hackers

23 years agoAdd support for Ukrainian console.
Maxim Sobolev [Thu, 10 May 2001 15:57:16 +0000 (15:57 +0000)]
Add support for Ukrainian console.

Submitted by: Olexander Kunytsa <kunia@istc.kiev.ua>
MFC after: 2 weeks

23 years agoAdd the ability to specify alternate PID file for syslogd.
David Malone [Thu, 10 May 2001 15:48:33 +0000 (15:48 +0000)]
Add the ability to specify alternate PID file for syslogd.

PR: 25784
Submitted by: Jon Villarreal <jonv@ivmg.net>
Reviewed by: iedowse

23 years agoDelete everything that's not a directory in /var/run and /var/spool/lock
Brian Somers [Thu, 10 May 2001 09:25:10 +0000 (09:25 +0000)]
Delete everything that's not a directory in /var/run and /var/spool/lock
at boot time.

MFC after: 3 weeks

23 years agoTell the user what group they must be in to su to root.
David E. O'Brien [Thu, 10 May 2001 07:37:25 +0000 (07:37 +0000)]
Tell the user what group they must be in to su to root.

Submitted by: Seth Kingsley <sethk@osd.bsdi.com>

23 years agomdoc(7) police: fix markup.
Ruslan Ermilov [Thu, 10 May 2001 07:29:35 +0000 (07:29 +0000)]
mdoc(7) police: fix markup.