]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoAdd support for Dlink DL10022 to the ed driver. This is a mii part
imp [Sat, 3 Mar 2001 08:31:37 +0000 (08:31 +0000)]
Add support for Dlink DL10022 to the ed driver.  This is a mii part
bolted to a ne-2000 chip.  This is necessary for the NetGear FA-410TX
and other cards.

This also requires you add mii to your kernel if you have an ed driver
configured.

This code will result in a couple of timeout messages for ed on the
impacted cards.  Additional work will be needed, but this does work
right now, and many people need these cards.

Submitted by: Ian Dowse <iedowse@maths.tcd.ie>

23 years agoFix typo present since 1997: single used mode -> single user mode.
asmodai [Sat, 3 Mar 2001 08:12:58 +0000 (08:12 +0000)]
Fix typo present since 1997: single used mode -> single user mode.

23 years agoFix INVARIANT_SUPPORT-only builds (without INVARIANTS). The required
bmilekic [Sat, 3 Mar 2001 06:06:00 +0000 (06:06 +0000)]
Fix INVARIANT_SUPPORT-only builds (without INVARIANTS). The required
`infrastructure' built with INVARIANT_SUPPORT for kern_mutex.c essentially
involves _mtx_assert(), which makes use of constants that were defined
under #ifdef INVARIANTS here.

23 years agoCleanups to Macros for sending data between netgraph nodes.
julian [Sat, 3 Mar 2001 05:52:49 +0000 (05:52 +0000)]
Cleanups to Macros for sending data between netgraph nodes.

23 years agoAdd parenthesis to a macro.
julian [Sat, 3 Mar 2001 05:50:47 +0000 (05:50 +0000)]
Add parenthesis to a macro.
This took me 2 whole days to track down. (bleah)

23 years agoSmall optimization: set use_ampm only when needed
ache [Sat, 3 Mar 2001 01:46:58 +0000 (01:46 +0000)]
Small optimization: set use_ampm only when needed

23 years agoDon't use "," as load avg. separator because it can be precision point in some
ache [Sat, 3 Mar 2001 01:32:46 +0000 (01:32 +0000)]
Don't use "," as load avg. separator because it can be precision point in some
locales, so leads to confusion

23 years agoUse AM/PM time only when available in locale
ache [Fri, 2 Mar 2001 23:53:36 +0000 (23:53 +0000)]
Use AM/PM time only when available in locale

23 years agoOops, remove tabs sneaked in wrong place
ache [Fri, 2 Mar 2001 23:49:07 +0000 (23:49 +0000)]
Oops, remove tabs sneaked in wrong place

23 years agoUse AM/PM time only when available in locale
ache [Fri, 2 Mar 2001 23:11:38 +0000 (23:11 +0000)]
Use AM/PM time only when available in locale

23 years agoNew release notes: TCP NewReno, TCP more aggressive timeouts on
bmah [Fri, 2 Mar 2001 23:02:10 +0000 (23:02 +0000)]
New release notes:  TCP NewReno, TCP more aggressive timeouts on
SYN segments, IP verifies destination of received packet with
ingress interface, bc update to 1.06, ipfstat -t.

23 years agoClear am and pm fields, add empty ampm_fmt for locales I know is not AM/PM
ache [Fri, 2 Mar 2001 22:27:27 +0000 (22:27 +0000)]
Clear am and pm fields, add empty ampm_fmt for locales I know is not AM/PM

23 years agoActually implement T_FMT_AMPM
ache [Fri, 2 Mar 2001 22:12:19 +0000 (22:12 +0000)]
Actually implement T_FMT_AMPM

23 years agoImplement ampm_fmt (%r) per POSIX
ache [Fri, 2 Mar 2001 22:10:04 +0000 (22:10 +0000)]
Implement ampm_fmt (%r) per POSIX

23 years agoAdd a new sysctl net.inet.ip.check_interface, which will verify that
jlemon [Fri, 2 Mar 2001 20:54:03 +0000 (20:54 +0000)]
Add a new sysctl net.inet.ip.check_interface, which will verify that
an incoming packet arrivees on an interface that has an address matching
the packet's address.  This is turned on by default.

23 years agoOnly pick up so_error the first time through with EISCONN, as advertised.
jlemon [Fri, 2 Mar 2001 19:29:53 +0000 (19:29 +0000)]
Only pick up so_error the first time through with EISCONN, as advertised.
The sense of the test was reversed, so we were returning EISCONN, then 0.

Pointed out and tested by: Martin Blapp <mb@imp.ch>

23 years agosetlocale(3) has been fixed to match POSIX standard:
ru [Fri, 2 Mar 2001 16:52:14 +0000 (16:52 +0000)]
setlocale(3) has been fixed to match POSIX standard:
LC_ALL takes precedence over other LC_* envariables.

23 years agoPOSIX wrote:
ru [Fri, 2 Mar 2001 16:19:49 +0000 (16:19 +0000)]
POSIX wrote:

: LC_TIME   This variable shall determine the format and
:           contents of date and time strings when the -v
:           option is specified.

Developers took this wrong.  LC_TIME specifies the locale
name, not the ``format'' argument of strftime().

Oops:
pax -w -f /tmp/foo /dev/null
LC_TIME=de_DE.ISO_8859-1 pax -v -f /tmp/foo

23 years agoMismatched MFSNAMELEN and MNAMELEN with fstype / fspath.
adrian [Fri, 2 Mar 2001 14:05:49 +0000 (14:05 +0000)]
Mismatched MFSNAMELEN and MNAMELEN with fstype / fspath.

Submitted by: Naoki Kobayashi <shibata@geo.titech.ac.jp>

23 years agoRemoved duplicate $FreeBSD$.
ru [Fri, 2 Mar 2001 13:00:26 +0000 (13:00 +0000)]
Removed duplicate $FreeBSD$.

23 years agoFix setlocale() to conform to the ISO C and POSIX standards.
ru [Fri, 2 Mar 2001 12:45:52 +0000 (12:45 +0000)]
Fix setlocale() to conform to the ISO C and POSIX standards.
The below text is quoted from the latest POSIX draft:

: The values of locale categories shall be determined by a precedence
: order; the first condition met below determines the value:
:
: 1. If the LC_ALL environment variable is defined and is not null,
:    the value of LC_ALL shall be used.
: 2. If the LC_* environment variable (LC_COLLATE, LC_CTYPE, LC_MESSAGES,
:    LC_MONETARY, LC_NUMERIC, LC_TIME) is defined and is not null, the
:    value of the environment variable shall be used to initialize the
:    category that corresponds to the environment variable.
: 3. If the LANG environment variable is defined and is not null, the
:    value of the LANG environment variable shall be used.
: 4. If the LANG environment variable is not set or is set to the empty
:    string, the implementation-defined default locale shall be used.

The conditions 1 and 2 were interchanged, i.e., LC_* were looked first,
then LC_ALL, then LANG (note that LC_ALL and LANG were essentially the
same, providing the default, with LC_ALL taking precedence over LANG).
Now, LC_ALL and LANG serve the different purposes.  LC_ALL overrides
any LC_*, and LANG provides the default fallback.

Testcase:

/usr/bin/env LC_ALL=C LC_TIME=de_DE.ISO_8859-1 /bin/date

Should return date in the "C" locale format.

Inspired by: date(1) reference page in the Draft

23 years agomdoc(7) police: fix markup.
ru [Fri, 2 Mar 2001 09:59:58 +0000 (09:59 +0000)]
mdoc(7) police: fix markup.

23 years agomdoc(7) police: in -offset clause, spell the ``indent'' correctly.
ru [Fri, 2 Mar 2001 09:38:50 +0000 (09:38 +0000)]
mdoc(7) police: in -offset clause, spell the ``indent'' correctly.

23 years agoChange "NO_PROFILE_DATA" to "NO_PROFILE_COUNTERS" to match the commit by
obrien [Fri, 2 Mar 2001 08:46:15 +0000 (08:46 +0000)]
Change "NO_PROFILE_DATA" to "NO_PROFILE_COUNTERS" to match the commit by
the FSF/GCC people in the stock 2.97 source that is like our custom change
modulo the conditional compilation symbol.

23 years agoMerged from sys/i386/include/bus_at386.h revision 1.13.
kato [Fri, 2 Mar 2001 08:31:38 +0000 (08:31 +0000)]
Merged from sys/i386/include/bus_at386.h revision 1.13.

23 years agoAck! I finally got annoyed enough to actually kill this. There is no
peter [Fri, 2 Mar 2001 08:15:41 +0000 (08:15 +0000)]
Ack! I finally got annoyed enough to actually kill this.  There is no
need to manually force the network_interfaces variable in /etc/rc.conf,
and it only ever gets in the way.  rc.network and rc.network6 DTRT with
the default of 'auto'.  This should have died over a year ago.

23 years agoFix the rest of formatting.
ru [Fri, 2 Mar 2001 08:09:19 +0000 (08:09 +0000)]
Fix the rest of formatting.

23 years agoAdd -g group to usage message.
imp [Fri, 2 Mar 2001 07:30:37 +0000 (07:30 +0000)]
Add -g group to usage message.
Don't treat pointer as a boolean, but instead test it against NULL.
Add warning for groups that don't exist

Submitted by: ru

23 years agoSwitch to using 16 bit handles instead of 32 bit handles.
mjacob [Fri, 2 Mar 2001 06:28:55 +0000 (06:28 +0000)]
Switch to using 16 bit handles instead of 32 bit handles.
This is a pretty invasive change, but there are three good
reasons to do this:

1. We'll never have > 16 bits of handle.
2. We can (eventually) enable the RIO (Reduced Interrupt Operation)
bits which return multiple completing 16 bit handles in mailbox
registers.
3. The !)$*)$*~)@$*~)$* Qlogic target mode for parallel SCSI spec
changed such that at_reserved (which was 32 bits) was split into
two pieces- and one of which was a 16 bit handle id that functions
like the at_rxid for Fibre Channel (a tag for the f/w to correlate
CTIOs with a particular command). Since we had to muck with that
and this changed the whole handler architecture, we might as well...

Propagate new at_handle on through int ct_fwhandle. Follow
implications of changing to 16 bit handles.

These above changes at least get Qlogic 1040 cards working in target
mode again. 1080/12160 cards don't work yet.

In isp.c:
Prepare for doing all loop management in outer layers.

23 years agoOk, the kernel will panic in kmem_malloc() if the kernel map is full, so
jhb [Fri, 2 Mar 2001 06:07:38 +0000 (06:07 +0000)]
Ok, the kernel will panic in kmem_malloc() if the kernel map is full, so
malloc with M_WAITOK can't actually return NULL.  I wish I could get two
people to give me the same answer about this when I ask...

Submitted by: jake

23 years agoBack out a removal that I was far to quick to apply. The root cause
markm [Fri, 2 Mar 2001 05:57:39 +0000 (05:57 +0000)]
Back out a removal that I was far to quick to apply. The root cause
has been fixed.

23 years agoversion 1.7 made some changes to correct problems identifed by compiling
mdodd [Fri, 2 Mar 2001 05:33:53 +0000 (05:33 +0000)]
version 1.7 made some changes to correct problems identifed by compiling
with egcs-1.1.1.  bus_space_write_multi_2() had an extra operation that
should have been removed.

Remove it.

This fixes the panic when bus_space_write_multi_2() is used.

Obtained from: jake

23 years ago- Check to see if malloc() returned NULL even with M_WAITOK.
jhb [Fri, 2 Mar 2001 05:33:03 +0000 (05:33 +0000)]
- Check to see if malloc() returned NULL even with M_WAITOK.
- Add a KASSERT() to ensure an ithread has a backing kernel thread when we
  schedule it.
- Don't attempt to preemptively switch to an ithread if p_stat of curproc
  is not SRUN.

23 years agoFix isp_print_qentry to print all four lines- it's been broken for months.
mjacob [Fri, 2 Mar 2001 04:48:41 +0000 (04:48 +0000)]
Fix isp_print_qentry to print all four lines- it's been broken for months.

23 years agoAdd link for resource_query_unit, now documented in the
imp [Fri, 2 Mar 2001 04:32:50 +0000 (04:32 +0000)]
Add link for resource_query_unit, now documented in the
resource_query_string man page.

23 years agoFlesh out this man page a bit more. Add information about
imp [Fri, 2 Mar 2001 04:30:16 +0000 (04:30 +0000)]
Flesh out this man page a bit more.  Add information about
resource_query_unit and improve the descriptions of the parameters
passed to these functions.

Plus a couple minor formatting/markup changes:
o Quote -1 as \-1.
o .Dq hints to match resource_int_value().

23 years agoInclude `dbxelf.h' via tm.h (the approved FSF/GNU way) vs. in our MI header.
obrien [Fri, 2 Mar 2001 03:00:41 +0000 (03:00 +0000)]
Include `dbxelf.h' via tm.h (the approved FSF/GNU way) vs. in our MI header.

23 years agoDefine "FREEBSD_NATIVE" in the freebsd-native header to reduce the amount
obrien [Fri, 2 Mar 2001 02:56:59 +0000 (02:56 +0000)]
Define "FREEBSD_NATIVE" in the freebsd-native header to reduce the amount
of stuff (and thus length of error output) we put on the invocation command
line.  Also follow the new FSF/GNU style of giving the symbol a value so it
can be used in `if()' statements in addition to `#if' so seldomly compiled
in code (on some platforms) gets compiled always, to help reduce bit-rot.

23 years agoMove the obvious bits of mixed ELF and a.out support down into this MD
obrien [Fri, 2 Mar 2001 02:46:05 +0000 (02:46 +0000)]
Move the obvious bits of mixed ELF and a.out support down into this MD
header to reduce the difference of our sources to the stock GNU/FSF ones.
While the mix binary format support was nice to have in the FreeBSD MI
header as a frame work, it just clutters up too much and makes the FreeBSD
MI header more different from the FSF/GNU stock one than it needs to be.

23 years agoPush as much as was easy of the mixed ELF & a.out support for the i386
obrien [Fri, 2 Mar 2001 02:42:42 +0000 (02:42 +0000)]
Push as much as was easy of the mixed ELF & a.out support for the i386
into the i386 MD FreeBSD header.  Also comment tweaking, continuation line
style changes, and other minor changes to make this closer to the latest
FSF/GNU 3.0 sources (to reduce my headache when 3.0 is imported).

23 years agoAdd comments regarding enabling IPv6
gshapiro [Fri, 2 Mar 2001 02:10:51 +0000 (02:10 +0000)]
Add comments regarding enabling IPv6

Submitted by: ume

23 years agoThere were a few changes missed when this file was converted to
iedowse [Fri, 2 Mar 2001 00:40:06 +0000 (00:40 +0000)]
There were a few changes missed when this file was converted to
newbus in revision 1.19. As a result, lnc was, I believe, broken
for all PCI cards. The softc fields `lnc_btag' and `lnc_bhandle'
were not initialised, `rap', `rdp' and `bdp' were initialised to
the wrong values, and the size of the DMA ring memory was calculated
incorrectly.

Paul Richards has further cleanups in the pipeline, but this at
least is enough to make the driver usable with VMware.

Approved by: paul

23 years agoSpell deprecated correctly (dufault)
imp [Fri, 2 Mar 2001 00:36:13 +0000 (00:36 +0000)]
Spell deprecated correctly (dufault)
Suggest mergemaster in the 4.x-stable -> current upgrade (fenner)
correct path to hints file (fenner)

Submitted by: dufault, fenner.

23 years agoAdd the LINPROCFS option.
des [Thu, 1 Mar 2001 23:13:12 +0000 (23:13 +0000)]
Add the LINPROCFS option.

Forgotten by: peter

23 years agoDo not need to define "VERSION" here -- we do it on the command line.
obrien [Thu, 1 Mar 2001 23:07:18 +0000 (23:07 +0000)]
Do not need to define "VERSION" here -- we do it on the command line.

23 years agoNew release notes: CMI8338/8738, CS4281 sound driver, ISO-9660 Unicode,
bmah [Thu, 1 Mar 2001 22:11:00 +0000 (22:11 +0000)]
New release notes:  CMI8338/8738, CS4281 sound driver, ISO-9660 Unicode,
ida crashdumps, login exports PAM environment variables, wall -g,
make.conf CPUTYPE, CVS_RSH default now ssh.

Updated release notes:  ahc(4) updates, ICMP and TCP RSTs, random(4),

MFCs noted:  rc.syscons.

23 years agoCorrectly emulate linux_connect. For nonblocking sockets, the behavior
jlemon [Thu, 1 Mar 2001 21:44:40 +0000 (21:44 +0000)]
Correctly emulate linux_connect.  For nonblocking sockets, the behavior
is to return EINPROGRESS, EALREADY, (so_error ONCE), EISCONN.  Certain
linux applications rely on the so_error (normally 0) being returned in
order to operate properly.

Tested by: Thomas Moestl <tmoestl@gmx.net>

23 years agoFree lock before returning from process_worklist_item.
mckusick [Thu, 1 Mar 2001 21:43:46 +0000 (21:43 +0000)]
Free lock before returning from process_worklist_item.

Obtained from: Constantine Sapuntzakis <csapuntz@stanford.edu>

23 years agoReviewed by: jlemon
adrian [Thu, 1 Mar 2001 21:00:17 +0000 (21:00 +0000)]
Reviewed by: jlemon

An initial tidyup of the mount() syscall and VFS mount code.

This code replaces the earlier work done by jlemon in an attempt to
make linux_mount() work.

* the guts of the mount work has been moved into vfs_mount().

* move `type', `path' and `flags' from being userland variables into being
  kernel variables in vfs_mount(). `data' remains a pointer into
  userspace.

* Attempt to verify the `type' and `path' strings passed to vfs_mount()
  aren't too long.

* rework mount() and linux_mount() to take the userland parameters
  (besides data, as mentioned) and pass kernel variables to vfs_mount().
  (linux_mount() already did this, I've just tidied it up a little more.)

* remove the copyin*() stuff for `path'. `data' still requires copyin*()
  since its a pointer into userland.

* set `mount->mnt_statf_mntonname' in vfs_mount() rather than in each
  filesystem.  This variable is generally initialised with `path', and
  each filesystem can override it if they want to.

* NOTE: f_mntonname is intiailised with "/" in the case of a root mount.

23 years agoAllocate vm_page_array and vm_page_buckets from the end of the biggest chunk
gallatin [Thu, 1 Mar 2001 19:21:24 +0000 (19:21 +0000)]
Allocate vm_page_array and vm_page_buckets from the end of the biggest chunk
of memory, rather than from the start.

This fixes problems allocating bouncebuffers on alphas where there is only
1 chunk of memory (unlike PCs where there is generally at least one small
chunk and a large chunk).  Having 1 chunk had been fatal, because these
structures take over 13MB on a machine with 1GB of ram. This doesn't leave
much room for other structures and bounce buffers if they're at the front.

Reviewed by: dfr, anderson@cs.duke.edu, silence on -arch
Tested by: Yoriaki FUJIMORI <fujimori@grafin.fujimori.cache.waseda.ac.jp>

23 years agoThe 'X' permission symbol is specified by POSIX.2, so don't say that it
ben [Thu, 1 Mar 2001 19:03:01 +0000 (19:03 +0000)]
The 'X' permission symbol is specified by POSIX.2, so don't say that it
isn't.

Obtained from: NetBSD
Confirmed by: ru, wollman

23 years agoTurn on interrupt-entropy harvesting for all/any mass storage devices
markm [Thu, 1 Mar 2001 17:09:09 +0000 (17:09 +0000)]
Turn on interrupt-entropy harvesting for all/any mass storage devices
I could find. I have no doubt missed a couple.

Interrupt entropy harvesting is still conditional on the
kern.random.sys.harvest_interrupt sysctl.

23 years agoFix brokeness in rev. 1.6: this manpage appeared in NetBSD 1.4.
ru [Thu, 1 Mar 2001 17:09:01 +0000 (17:09 +0000)]
Fix brokeness in rev. 1.6: this manpage appeared in NetBSD 1.4.

Obtained from: NetBSD

23 years agoNo longer an option. Config(8) is whining over LINT.
markm [Thu, 1 Mar 2001 16:46:39 +0000 (16:46 +0000)]
No longer an option. Config(8) is whining over LINT.

23 years agoAdd code to turn on the entropy harvesting sysctl's as early as possible
dougb [Thu, 1 Mar 2001 13:19:49 +0000 (13:19 +0000)]
Add code to turn on the entropy harvesting sysctl's as early as possible
during the boot process. We're turning it on by default, based on the
actual presence of a configured ethernet card, and/or ppp/tun devices.
Of course, it's easy to disable in rc.conf.

23 years agoPut back several lines that I erroneously deleted in my previous commit. This
sobomax [Thu, 1 Mar 2001 13:08:07 +0000 (13:08 +0000)]
Put back several lines that I erroneously deleted in my previous commit. This
should fix recent segfault of `pkg_info -e'.

Submitted by: Garrett Rooney <rooneg@electricjellyfish.net>, knu

23 years agoUpdate the BUGS section.
ru [Thu, 1 Mar 2001 12:58:54 +0000 (12:58 +0000)]
Update the BUGS section.

The first bug is fixed in tmac.doc, revision 1.21.
The second bug fixed in doc-common, revision 1.33.

23 years agoFixed undesirable hyphenation on the dash of a flag argument.
ru [Thu, 1 Mar 2001 12:46:29 +0000 (12:46 +0000)]
Fixed undesirable hyphenation on the dash of a flag argument.
Example:

.Sh SYNOPSIS
.Nm natd
.Fl unregistered-only
.Fl proxy-only
.Fl deny-incoming
.Fl use-sockets
.Fl same-ports

Ported from: mdocNG

23 years agoFixed two bugs with -tag lists.
ru [Thu, 1 Mar 2001 12:13:04 +0000 (12:13 +0000)]
Fixed two bugs with -tag lists.

If the -tag list definition didn't have a -width modifier,
the first .It call was supposed to set the width depending
on the first argument type; if it is a macro name, use the
macro's width value; otherwise, use width value of `No'.

The following two lists should produce identical output:

.Bl -tag -width Er -compact
.It Er EINVAL
invalid argument
.El
.Bl -tag -compact
.It Er EINVAL
invalid argument
.El

If the outermost -tag list definition did't have a -width
modifier, the .It elements of inner lists might not work
(producing a list where each successive element `walks' to
the right).

Example:

.Bl -tag -compact
.It outer
.It outer
.Bl -tag -compact
.It inner
.It inner
.El
.It outer
.It outer
.El

Ported from: mdocNG

23 years agomdoc(7) police: misc fixes.
ru [Thu, 1 Mar 2001 11:06:56 +0000 (11:06 +0000)]
mdoc(7) police: misc fixes.

23 years agoCorrect English used.
asmodai [Thu, 1 Mar 2001 09:37:38 +0000 (09:37 +0000)]
Correct English used.

23 years agoNumerous formatting/content fixes.
ru [Thu, 1 Mar 2001 09:11:16 +0000 (09:11 +0000)]
Numerous formatting/content fixes.

23 years agoHook Warner's resource_* manpages to the build.
asmodai [Thu, 1 Mar 2001 09:06:57 +0000 (09:06 +0000)]
Hook Warner's resource_* manpages to the build.
This includes appropriate MLINKS for the other functions contained
within them.

23 years agoRemove leading . from .Nd description.
asmodai [Thu, 1 Mar 2001 09:02:01 +0000 (09:02 +0000)]
Remove leading . from .Nd description.

23 years agoFix typo: the the -> to the.
asmodai [Thu, 1 Mar 2001 09:01:43 +0000 (09:01 +0000)]
Fix typo: the the -> to the.

23 years agoh_errno is declared in netdb.h.
imp [Thu, 1 Mar 2001 06:41:43 +0000 (06:41 +0000)]
h_errno is declared in netdb.h.

23 years agoInclude -DDEFSHELL=1 in Makefile.dist so people boostrapping will
imp [Thu, 1 Mar 2001 06:26:32 +0000 (06:26 +0000)]
Include -DDEFSHELL=1 in Makefile.dist so people boostrapping will
have a sane default.

Also add $FreeBSD$ while I'm here.

Pointed out by: bde

23 years agoInclude <unistd.h> for getopt's interface rather than declare things
imp [Thu, 1 Mar 2001 06:22:46 +0000 (06:22 +0000)]
Include <unistd.h> for getopt's interface rather than declare things
individually here.

23 years agooptarg and optind are declared in unistd.h, so don't redundantly
imp [Thu, 1 Mar 2001 06:20:51 +0000 (06:20 +0000)]
optarg and optind are declared in unistd.h, so don't redundantly
declare them here.

23 years agoMAXPATHLEN includes the trailing NUL.
imp [Thu, 1 Mar 2001 06:03:17 +0000 (06:03 +0000)]
MAXPATHLEN includes the trailing NUL.

Correct array sizes to reflect this.
Correct NUL termination after strncpy.

# Didn't to strncpy -> strlcpy change.

23 years agoMAXPATHLEN contains the trailing NUL.
imp [Thu, 1 Mar 2001 06:00:22 +0000 (06:00 +0000)]
MAXPATHLEN contains the trailing NUL.

23 years agoMAXPATHLEN is enough.
imp [Thu, 1 Mar 2001 05:52:38 +0000 (05:52 +0000)]
MAXPATHLEN is enough.

Also, snprinf is guaranteed by the new ansi standard to NUL terminate the
string, so we don't need to do that ourselves.

23 years agoAll temperature scales are centigrades. Use "Celsius" instead.
dcs [Thu, 1 Mar 2001 05:51:46 +0000 (05:51 +0000)]
All temperature scales are centigrades. Use "Celsius" instead.

23 years agoMAXPATHLEN already accounts for the trailing NUL, so no need to add one in.
imp [Thu, 1 Mar 2001 05:49:36 +0000 (05:49 +0000)]
MAXPATHLEN already accounts for the trailing NUL, so no need to add one in.
In addition, since we pass readlink MAXPATHLEN - 1, we would have never have
used that extra byte anyway.

23 years agoMAXPATHLEN includes the trailing NUL character, so there's no need to add 1
imp [Thu, 1 Mar 2001 05:47:28 +0000 (05:47 +0000)]
MAXPATHLEN includes the trailing NUL character, so there's no need to add 1
to it for the size of path.

23 years agoRemove redundant declaration of optind. It has been declared in unistd.h
imp [Thu, 1 Mar 2001 05:46:03 +0000 (05:46 +0000)]
Remove redundant declaration of optind.  It has been declared in unistd.h
for ages.

23 years agoo Add support for wall -g. This will send a message to all members of
imp [Thu, 1 Mar 2001 05:43:12 +0000 (05:43 +0000)]
o Add support for wall -g.  This will send a message to all members of
  a given group.
o Minor code style cleanups while I'm here

Reviewed by: bde, kris, markm, audit@

23 years agoHide the definition of struct __sFILEX and add the needed
deischen [Thu, 1 Mar 2001 05:22:14 +0000 (05:22 +0000)]
Hide the definition of struct __sFILEX and add the needed
lock definitions to it.  flockfile state is now allocated
along with the rest of FILE.  This eliminates the need for a
separate allocation of flockfile state as well as eliminating
the mutex/lock used to serialize its allocation.

23 years agos/fstat/_fstat/
deischen [Thu, 1 Mar 2001 04:59:01 +0000 (04:59 +0000)]
s/fstat/_fstat/

Approved by: phantom

23 years agoDocument the resource_*_value and resource_query_string families of
imp [Thu, 1 Mar 2001 04:28:40 +0000 (04:28 +0000)]
Document the resource_*_value and resource_query_string families of
functions.  These man pages likely need some markup work, so I've
specifically not added them to the Makefile yet.

23 years agoDon't build/install sendmail related items if NO_SENDMAIL is set.
gshapiro [Thu, 1 Mar 2001 03:51:18 +0000 (03:51 +0000)]
Don't build/install sendmail related items if NO_SENDMAIL is set.

Submitted by: ru

23 years agoMove installation of sendmail helpfile into etc/sendmail/Makefile. It is
gshapiro [Thu, 1 Mar 2001 03:50:28 +0000 (03:50 +0000)]
Move installation of sendmail helpfile into etc/sendmail/Makefile.  It is
a user configurable file and it would benefit from mergemaster.

23 years agoIn ascpoll, return revents, not 0.
jlemon [Thu, 1 Mar 2001 03:21:16 +0000 (03:21 +0000)]
In ascpoll, return revents, not 0.

23 years agoEliminate the use of the getenv_int stuff we'd been using (with a bitmap
mjacob [Thu, 1 Mar 2001 02:21:36 +0000 (02:21 +0000)]
Eliminate the use of the getenv_int stuff we'd been using (with a bitmap
for selecting unit). Instead, use the resource hints mechanism.

One unfortunate situation here is that there is no resource_quad_value
function- which is what I needed for WWN boot time replacement. Worse-
you can't store the hint as just plain

hint.isp.0.nodewwn="0x50000000aaaa0001"

because this gets interpreted as an int- incorrectly because it can't
be converted to an int. I can't even get this as a string. To work
around this particular case for nodewwn && portwwn setting, this
rather grotesque form will be used:

hint.isp.0.nodewwn="w50000000aaaa0001"
hint.isp.0.portwwn="w50000000aaaa0002"

At the same time, if we have no hinted WWN, set the default WWN (which, btw,
gets overridden if the card has valid NVRAM, which is usual) to
0x400000007F000009ull (which translates to NAA == IPv4, 127.0.0.9).

Eliminate more printf's and replace them either with device_printf or
isp_prt calls.

23 years agoGo to a default port and default node wwn model. Eliminate isp_name
mjacob [Thu, 1 Mar 2001 02:15:58 +0000 (02:15 +0000)]
Go to a default port and default node wwn model. Eliminate isp_name
and isp_unit and just store the device_t, fer gosh sakes.... Include
sys/bus.h for use by isp_pci.c.

23 years agoFinally eliminate as many of the printf calls as possible (still leaving
mjacob [Thu, 1 Mar 2001 02:14:54 +0000 (02:14 +0000)]
Finally eliminate as many of the printf calls as possible (still leaving
ones where we have a CAM path) and replacing them with calls to isp_prt.,

Eliminate isp_unit references- we no longer have an isp_unit- we now
have an isp_dev that device_get_unit can work with.

23 years agoStylistic changes mirroring ones done in the FSF/GNU 3.0 development sources.
obrien [Wed, 28 Feb 2001 22:37:32 +0000 (22:37 +0000)]
Stylistic changes mirroring ones done in the FSF/GNU 3.0 development sources.

23 years agoHave coff be a default ibcs2 loader if we have ibcs2 support enabled.
jkh [Wed, 28 Feb 2001 22:28:00 +0000 (22:28 +0000)]
Have coff be a default ibcs2 loader if we have ibcs2 support enabled.

23 years agoThe kernel did not hold a vnode reference associated with the
iedowse [Wed, 28 Feb 2001 20:54:28 +0000 (20:54 +0000)]
The kernel did not hold a vnode reference associated with the
`rootvnode' pointer, but vfs_syscalls.c's checkdirs() assumed that
it did. This bug reliably caused a panic at reboot time if any
filesystem had been mounted directly over /.

The checkdirs() function is called at mount time to find any process
fd_cdir or fd_rdir pointers referencing the covered mountpoint
vnode. It transfers these to point at the root of the new filesystem.
However, this process was not reversed at unmount time, so processes
with a cwd/root at a mount point would unexpectedly lose their
cwd/root following a mount-unmount cycle at that mountpoint.

This change should fix both of the above issues. Start_init() now
holds an extra vnode reference corresponding to `rootvnode', and
dounmount() releases this reference when the root filesystem is
unmounted just before reboot. Dounmount() now undoes the actions
taken by checkdirs() at mount time; any process cdir/rdir pointers
that reference the root vnode of the unmounted filesystem are
transferred to the now-uncovered vnode.

Reviewed by: bde, phk

23 years agoThe secure temporary directory is always created, so always remove it.
jhay [Wed, 28 Feb 2001 19:21:44 +0000 (19:21 +0000)]
The secure temporary directory is always created, so always remove it.

23 years agoShuffle netgraph mutexes a bit and hold a reference on a node
julian [Wed, 28 Feb 2001 18:49:09 +0000 (18:49 +0000)]
Shuffle netgraph mutexes a bit and hold a reference on a node
from the function that is calling the destructor.

23 years agoUse _PATH_DEV.
obrien [Wed, 28 Feb 2001 17:50:29 +0000 (17:50 +0000)]
Use _PATH_DEV.

Reviewed by: grog

23 years agoMerge in strtoul.3 rev 1.11 mdoc(7) police changes:
obrien [Wed, 28 Feb 2001 17:46:56 +0000 (17:46 +0000)]
Merge in strtoul.3 rev 1.11 mdoc(7) police changes:
.Nd line broken in rev 1.10
use .Bx for ``BSD''

23 years agoEliminate mdocNG warnings caused by misplaced or extraneous macro calls.
ru [Wed, 28 Feb 2001 17:38:53 +0000 (17:38 +0000)]
Eliminate mdocNG warnings caused by misplaced or extraneous macro calls.

23 years agoUse EV_SET() macro to initialize kevent structure.
jlemon [Wed, 28 Feb 2001 16:16:31 +0000 (16:16 +0000)]
Use EV_SET() macro to initialize kevent structure.

23 years agoAdd missing prefix for nano, I presume this was missing 'cos of the
dwmalone [Wed, 28 Feb 2001 16:00:32 +0000 (16:00 +0000)]
Add missing prefix for nano, I presume this was missing 'cos of the
prefix recusrion bug. Remove units which can be derived using prefixes
and other units.

Add new prefixes for 2^10, 2^20, etc...

23 years agoGive units a bit of a cleanup:
dwmalone [Wed, 28 Feb 2001 15:57:38 +0000 (15:57 +0000)]
Give units a bit of a cleanup:

1) Fix a bug where the recursion on prefixes where if recusrion failed
then the next attempt started too far along the string.
2) Up some internal limits that we were about to hit off.
3) Change the constness and signedness of some things and complete missing
prototypes.
4) Remove the readerror function 'cos it could print out the wrong
wrong file name. Replace it with more useful error messages.
5) Use sizeof when passing buffers to fgets.
6) Warn about empty prefix definitions.

Reviewed by: iedowse

23 years agoUse the new EV_SET macro to insure that all fields in struct kevent
jlemon [Wed, 28 Feb 2001 15:47:47 +0000 (15:47 +0000)]
Use the new EV_SET macro to insure that all fields in struct kevent
are correctly initialized before use.  This should fix the problem
with DNS.

Pointy hat to:  me

23 years agoPrepare for mdocNG.
ru [Wed, 28 Feb 2001 13:18:04 +0000 (13:18 +0000)]
Prepare for mdocNG.

23 years agomdoc(7) police: prepare for mdocNG.
ru [Wed, 28 Feb 2001 13:11:51 +0000 (13:11 +0000)]
mdoc(7) police: prepare for mdocNG.