]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoAdd typedef of sa_family_t to <sys/un.h>, per POSIX and X/Open. Hide
Mike Barcroft [Sat, 20 Apr 2002 02:26:43 +0000 (02:26 +0000)]
Add typedef of sa_family_t to <sys/un.h>, per POSIX and X/Open.  Hide
portions of <sys/un.h> when a standard has been requested.

22 years agoAdd sa_family_t type to <sys/_types.h> and typedefs to <netinet/in.h>
Mike Barcroft [Sat, 20 Apr 2002 02:24:35 +0000 (02:24 +0000)]
Add sa_family_t type to <sys/_types.h> and typedefs to <netinet/in.h>
and <sys/socket.h>.  Previously, sa_family_t was only typedef'd in
<sys/socket.h>.

22 years agoRemove <ctype.h> - not needed
Andrey A. Chernov [Sat, 20 Apr 2002 02:18:54 +0000 (02:18 +0000)]
Remove <ctype.h> - not needed

22 years agoPrevent sign extension (again)
Andrey A. Chernov [Sat, 20 Apr 2002 02:04:58 +0000 (02:04 +0000)]
Prevent sign extension (again)

22 years agoAllow space between -a and its argument. Honour locale collating order
Tim J. Robbins [Sat, 20 Apr 2002 01:55:19 +0000 (01:55 +0000)]
Allow space between -a and its argument. Honour locale collating order
by using strcoll() instead of strcmp().

PR: 36270
Reviewed by: mike

22 years agoCrank WARNS.
Juli Mallett [Sat, 20 Apr 2002 01:49:10 +0000 (01:49 +0000)]
Crank WARNS.

Cast sizeof() to (int), as it's being compared against an int, not a size_t.
If i is changed to a size_t, it means the logic must be slightly changed later
in the flow, where --i is checked to be >= 0.  I am not sure I want to make a
logic change to account for clearing up a warning, when an aesthetic one will
keep from modifying the logic.

Other harmless casts, that I think I've made in the right directions.

Make gpbc() an inline function, rather than an obfuscated macro, make its
scratch space local, rather than global.  The previous macro used a dirty
hack (logical AND in place of a conditional) which would lead GCC to throw
a fit (rightly so) as the logical check, as well as the incrementation of
a variable, were not used for anything.

const'ify a few places where gcc3 yells. xstrdup() some global consts in
places where we xstrdup() when not using consts, but tried to assign them
to non-consts before.

Don't use execv(2) if we don't have the kind of arguments it wants.

Reviewed by:    asmodai obrien tjr
Submitted by:   tjr (a gcc3 build log)

22 years agoDon't put a line break in string literals. GCC 3.1 complains and GCC
Marcel Moolenaar [Sat, 20 Apr 2002 01:42:56 +0000 (01:42 +0000)]
Don't put a line break in string literals. GCC 3.1 complains and GCC
3.2 drops the ball.

22 years agoImprove style consistency of vfs_syscalls.c by converting the style used
Robert Watson [Sat, 20 Apr 2002 01:37:08 +0000 (01:37 +0000)]
Improve style consistency of vfs_syscalls.c by converting the style used
in various extattr_*() calls to match the rest of the file.  Originally,
these bits at the end looked more like style(9).  This patch was submitted
by green by way of the TrustedBSD MAC tree, and I fixed a few problems
with it on the way through.  Someone with more time on their hands should
convert the entire file to style(9); this commit is for diff reduction
purposes.

Submitted by: green
Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoDon't put a line break in string literals. GCC 3.1 complains and GCC
Marcel Moolenaar [Sat, 20 Apr 2002 01:35:45 +0000 (01:35 +0000)]
Don't put a line break in string literals. GCC 3.1 complains and GCC
3.2 drops the ball.

22 years agoIt is easier for me to debug with -I's at the rear.
David E. O'Brien [Sat, 20 Apr 2002 01:27:28 +0000 (01:27 +0000)]
It is easier for me to debug with -I's at the rear.

22 years agoSpelling fix for comment.
Robert Watson [Sat, 20 Apr 2002 01:14:25 +0000 (01:14 +0000)]
Spelling fix for comment.

22 years agoComment out and mark broken the hea and hfa devices until someone has
Mike Barcroft [Sat, 20 Apr 2002 00:51:30 +0000 (00:51 +0000)]
Comment out and mark broken the hea and hfa devices until someone has
time to fix them.

22 years ago- Nuke small sgi ifdef bits.
Andrew R. Reiter [Sat, 20 Apr 2002 00:39:32 +0000 (00:39 +0000)]
- Nuke small sgi ifdef bits.

22 years agoBump WARNS to 4.
Juli Mallett [Sat, 20 Apr 2002 00:20:54 +0000 (00:20 +0000)]
Bump WARNS to 4.

22 years agoIn the case where we can't open /dev/tty, fall back to -t behaviour.
Juli Mallett [Fri, 19 Apr 2002 23:58:00 +0000 (23:58 +0000)]
In the case where we can't open /dev/tty, fall back to -t behaviour.

Submitted by: fenner

22 years agoBump WARNS to 4.
Juli Mallett [Fri, 19 Apr 2002 23:46:45 +0000 (23:46 +0000)]
Bump WARNS to 4.

22 years agoUse `The .Nm utility'
Philippe Charnier [Fri, 19 Apr 2002 23:44:58 +0000 (23:44 +0000)]
Use `The .Nm utility'

22 years agoCompound two fprintf(3)s into 1 using string concatenation.
Juli Mallett [Fri, 19 Apr 2002 23:41:56 +0000 (23:41 +0000)]
Compound two fprintf(3)s into 1 using string concatenation.

22 years agoFix compiling of acpica when debugging is enabled. In the previous
Mike Barcroft [Fri, 19 Apr 2002 23:36:38 +0000 (23:36 +0000)]
Fix compiling of acpica when debugging is enabled.  In the previous
revision, two getenv()s were accidentally changed to use testenv().

Pointy hat to: mux

22 years agoAfter 3 months...
Juli Mallett [Fri, 19 Apr 2002 23:28:54 +0000 (23:28 +0000)]
After 3 months...

Merge xargs(1) with that of xMach.

Bring in xargs(1) changes to add -L and -I as per the Single Unix Specification
version 3.  Proper exit status numbers are implemented, and the manual page has
been updated to reflect reality.

The code has been ANSIfied, and a new file has been added to xargs(1) to do the
substring substitution as SUSv3 requires.

Traditional behaviour should not be affected, use of -J should be deprecated
in favor of the more portable -I (though -J has been left, for now).

Submitted by: me, tjr (the exit status stuff)
Obtained from: xMach

22 years agoUse `The .Nm utility'. Introduce options with well known sentence.
Philippe Charnier [Fri, 19 Apr 2002 23:23:28 +0000 (23:23 +0000)]
Use `The .Nm utility'. Introduce options with well known sentence.

22 years agoAdd .Pp before enumerating options
Philippe Charnier [Fri, 19 Apr 2002 23:18:18 +0000 (23:18 +0000)]
Add .Pp before enumerating options

22 years agoUse `The .Nm utility'
Philippe Charnier [Fri, 19 Apr 2002 23:06:29 +0000 (23:06 +0000)]
Use `The .Nm utility'

22 years agoClean up:
Alfred Perlstein [Fri, 19 Apr 2002 22:58:09 +0000 (22:58 +0000)]
Clean up:

Comment run_filter() to explain what it does.

Remove chatty comments.

void busdma_swi() { }  -> void busdma_swi(void) { }

22 years agoMove tx(4) driver to sys/dev/tx. BTW split hardware structures and constants
Semen Ustimenko [Fri, 19 Apr 2002 22:43:57 +0000 (22:43 +0000)]
Move tx(4) driver to sys/dev/tx. BTW split hardware structures and constants
into if_txreg.h.

MFC after: 1 week

22 years agoMake this build on 4.x machines again (building a -current kernel on a
Warner Losh [Fri, 19 Apr 2002 22:28:09 +0000 (22:28 +0000)]
Make this build on 4.x machines again (building a -current kernel on a
-stable machine via the old-school methods):

Use __FreeBSD_version in preference to __FreeBSD__ >= N where possible.
Define a single variable mythread which is set to curproc or curthread
depending on the OS version (with a comment saying it is a white lie on
4.x since it really is a proc).

NB: __FreeBSD__ is the OS level of the host machine, not the target,
and should never be used, if possible, as __FreeBSD__ >= N.

22 years ago- Nuke some more #ifdef sun related sections.
Andrew R. Reiter [Fri, 19 Apr 2002 21:38:43 +0000 (21:38 +0000)]
- Nuke some more #ifdef sun related sections.

22 years ago- Nuke some #ifdef sun4m code; it hasn't been updated in 4 years and is not
Andrew R. Reiter [Fri, 19 Apr 2002 20:02:03 +0000 (20:02 +0000)]
- Nuke some #ifdef sun4m code; it hasn't been updated in 4 years and is not
  being used.

22 years agoPrevent sign extension on characters with 8bit set
Andrey A. Chernov [Fri, 19 Apr 2002 19:43:46 +0000 (19:43 +0000)]
Prevent sign extension on characters with 8bit set

22 years agoChange lpd to recognize the '-s' parameter as a preferred synonym
Garance A Drosehn [Fri, 19 Apr 2002 19:33:24 +0000 (19:33 +0000)]
Change lpd to recognize the '-s' parameter as a preferred synonym
for what is currently the '-p' parameter.  '-s' is what NetBSD
used (and they implemented it before I added -p in FreeBSD), and
it also matches the '-s' option in syslogd.  Someone in OpenBSD
land had also talked about adding a '-s' option, but it hasn't
happened yet.

MFC after: 5 days

22 years agoRemove a safety-setting line which is unnecessary now that the previous
Garance A Drosehn [Fri, 19 Apr 2002 18:46:10 +0000 (18:46 +0000)]
Remove a safety-setting line which is unnecessary now that the previous
line is using strlcpy instead of strncpy.

MFC after: 4 days

22 years agoFix 'deamon' -> 'daemon' in a comment.
Garance A Drosehn [Fri, 19 Apr 2002 18:36:56 +0000 (18:36 +0000)]
Fix 'deamon' -> 'daemon' in a comment.

Obtained from: NetBSD, OpenBSD
MFC after: 4 days

22 years agoAdd a little detail to the syslog-msg that comes up when lpd can not
Garance A Drosehn [Fri, 19 Apr 2002 18:28:35 +0000 (18:28 +0000)]
Add a little detail to the syslog-msg that comes up when lpd can not
execute a given filter.

MFC after: 4 days

22 years agoAdd the previously committed usb(4) manual pages to the build.
Tom Rhodes [Fri, 19 Apr 2002 18:24:03 +0000 (18:24 +0000)]
Add the previously committed usb(4) manual pages to the build.

PR: 27919
MFC after: 3 days

22 years agoAdd usb(4) manual pages.
Tom Rhodes [Fri, 19 Apr 2002 18:18:11 +0000 (18:18 +0000)]
Add usb(4) manual pages.

PR: 27919
Obtained from: NetBSD
MFC after: 3 days

22 years agoAdd .asm as an alias for .s. .asm is common in contribed sources.
David E. O'Brien [Fri, 19 Apr 2002 18:01:03 +0000 (18:01 +0000)]
Add .asm as an alias for .s.  .asm is common in contribed sources.

Helps with: gcc31 build.

22 years ago- Change KM_ macro calls to the appropriate function call.
Andrew R. Reiter [Fri, 19 Apr 2002 17:45:22 +0000 (17:45 +0000)]
- Change KM_ macro calls to the appropriate function call.
- Nuke KM_ macros from port.h

  This is a leadin step towards cleaning up this code as I wait for some
  ATM cards and a ATM switch to arrive.

22 years ago- Remove KM_ macro calls and replace with the real function we're calling.
Andrew R. Reiter [Fri, 19 Apr 2002 17:43:11 +0000 (17:43 +0000)]
- Remove KM_ macro calls and replace with the real function we're calling.

  As a note, this driver needs the same updating as the hfa driver was
  just given; removing these macros since I will be nuking them from
  netatm.

22 years agoAdd __FBSDID, this file was unmodified so it was missed in the initial sweep.
Juli Mallett [Fri, 19 Apr 2002 17:33:12 +0000 (17:33 +0000)]
Add __FBSDID, this file was unmodified so it was missed in the initial sweep.

22 years agoBring OpenBSD m4(1) off of the OPENBSD vendor branch, and add the -s option,
Juli Mallett [Fri, 19 Apr 2002 17:26:21 +0000 (17:26 +0000)]
Bring OpenBSD m4(1) off of the OPENBSD vendor branch, and add the -s option,
$FreeBSD$ identifiers, and fix initialisation to stderr to happen in a function
as stderr is not the same in CURRENT as in OpenBSD.

Reviewed by: obrien

22 years agoMerge differences and correct the manual page.
Juli Mallett [Fri, 19 Apr 2002 17:16:23 +0000 (17:16 +0000)]
Merge differences and correct the manual page.

22 years agoNew release notes: my driver.
Bruce A. Mah [Fri, 19 Apr 2002 17:07:28 +0000 (17:07 +0000)]
New release notes:  my driver.

MFCs noted:  viapm(4), ng_etf(4) (fixed spelling), libstand splitfs.

22 years agoAdd a reminder that commits to the security branches must go through the
Jacques Vidrine [Fri, 19 Apr 2002 16:22:30 +0000 (16:22 +0000)]
Add a reminder that commits to the security branches must go through the
Security Officer.

22 years agoIn sendfile(), use the vn_rdwr() helper function, rather than manually
Robert Watson [Fri, 19 Apr 2002 13:46:24 +0000 (13:46 +0000)]
In sendfile(), use the vn_rdwr() helper function, rather than manually
constructing a struct aio and invoking VOP_READ() directly.  This cleans
up the code a little, but also has the advantage of making sure almost
all vnode read/write access in the kernel goes through the helper
function, meaning that instrumentation of that helper function can impact
almost all relevant read/write operations.  In this case, it permits us
to put MAC hooks into vn_rdwr() and not modify uipc_syscalls.c (yet).

In general, if helper vn_*() functions exist, they should be used in
preference to direct VOP's in system call service code.

Submitted by: green
Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoDivorce proc0 and proc1 credentials earlier; while this isn't technically
Robert Watson [Fri, 19 Apr 2002 13:35:53 +0000 (13:35 +0000)]
Divorce proc0 and proc1 credentials earlier; while this isn't technically
needed in the current code, in the MAC tree, create_init() relies on the
ability to modify the credentials present for initproc, and should not
perform that modification on a shared credential.  Pro-active diff
reduction against MAC changes that are in the queue; also facilitates
other work, including the capabilities implementation.

Submitted by: green
Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoRecord libtelnet.a death.
Ruslan Ermilov [Fri, 19 Apr 2002 12:39:09 +0000 (12:39 +0000)]
Record libtelnet.a death.

22 years agoExplicitly specifies sin6_family & sin6_len of sockaddr_in6 for raw socket
SUZUKI Shinsuke [Fri, 19 Apr 2002 11:39:54 +0000 (11:39 +0000)]
Explicitly specifies sin6_family & sin6_len of sockaddr_in6 for raw socket
destination.
(Currently lack of their specification does not lead to any problem, because
 kernel does not check the consistency between actual address and its
 address family / length on raw socket.

 However kernel should always check their consistency and stop sending packets
 if there is a contradiction.  Considering backward compatibility of
 programs, I just fixed rtsol now;  I'd like to fix the kernel behavior later.)

Reviewed by: ume
MFC after: 3 days

22 years agoMention the UCONSOLE kernel option removal.
Ruslan Ermilov [Fri, 19 Apr 2002 11:37:34 +0000 (11:37 +0000)]
Mention the UCONSOLE kernel option removal.

Prompted by: sheldonh

22 years agomdoc(7) police: polishing.
Ruslan Ermilov [Fri, 19 Apr 2002 11:24:02 +0000 (11:24 +0000)]
mdoc(7) police: polishing.

22 years agoBuild modules on ia64:
Marcel Moolenaar [Fri, 19 Apr 2002 09:44:50 +0000 (09:44 +0000)]
Build modules on ia64:
o  Make the cam, cd9660 lomac and sound modules i386 and alpha
   specific due to link problems (@gprel relocation when @ltoff
   is required).  Once resolved, these can be moved back to the
   generic list.
o  Build linprocfs only on those architectures that have the
   linux module.
o  Make the sppp module i386 and alpha specific due to compile
   problems (pointers as switch cases). Once resolved, this can
   be moved back to the generic list.
o  Build all i386 specific modules, with the exception of those
   mentioned above as being moved from the generic list to the
   i386 list and those with dependencies on the linux module (aac)
   or i386 dependent (ar, apm, atspeaker, fpu, gnufpu, ibcs2,
   linux, ncv, nsp, netgraph, oltr, pecoff, s3, sbni, stg and
   vesa).
o  Don't build acpi as a module yet. It most be ported first.
   Once ported, it can be added to the ia64 list.
o  Don't build ipfilter yet due to compile errors (osreldate.h
   not found).

22 years agoPush the .info file contents out when we're done so people can
Poul-Henning Kamp [Fri, 19 Apr 2002 09:38:45 +0000 (09:38 +0000)]
Push the .info file contents out when we're done so people can
read the description while the megs shift from one place to another.

22 years agoMake kernel dumps work with GEOM.
Poul-Henning Kamp [Fri, 19 Apr 2002 09:24:12 +0000 (09:24 +0000)]
Make kernel dumps work with GEOM.

Notice that if the device on which the dump is set is destroyed for
any reason, the dump setting is lost.  This in particular will
happen in the case of spoilage.  For instance if you set dump on
ad0s1b and open ad0 for writing, ad0s* will be spoilt and the dump
setting lost.  See geom(4) for more about spoiling.

Sponsored by: DARPA & NAI Labs.

22 years agoMake life easier for reference-vector generatorts in tools/regression/geom
Poul-Henning Kamp [Fri, 19 Apr 2002 09:21:27 +0000 (09:21 +0000)]
Make life easier for reference-vector generatorts in tools/regression/geom
by including a FreeBSD friendly CVS identifier in the XML output.

Sponsored by: DARPA & NAI Labs.

22 years agosuser is Giant safe, so optimize a pointless case.
Poul-Henning Kamp [Fri, 19 Apr 2002 09:20:13 +0000 (09:20 +0000)]
suser is Giant safe, so optimize a pointless case.

22 years agoAdd a SI_DUMPDEV flag for devices.
Poul-Henning Kamp [Fri, 19 Apr 2002 09:19:37 +0000 (09:19 +0000)]
Add a SI_DUMPDEV flag for devices.

Sponsored by: DARPA & NAI Labs.

22 years agoRespect setting of NM to allow cross-building.
Marcel Moolenaar [Fri, 19 Apr 2002 09:04:53 +0000 (09:04 +0000)]
Respect setting of NM to allow cross-building.

22 years agoLocalize it, LC_CTYPE
Andrey A. Chernov [Fri, 19 Apr 2002 09:02:16 +0000 (09:02 +0000)]
Localize it, LC_CTYPE

22 years agoSome fixes for Symbol card.
Atsushi Onoe [Fri, 19 Apr 2002 08:16:59 +0000 (08:16 +0000)]
Some fixes for Symbol card.
- The version string doesn't need to start with 'V'.
- Symbol firmware doesn't support ROAMING_MODE nor MICROWAVE_OPEN.
Obtained from: NetBSD

22 years agoUse LC_ALL to pick collate
Andrey A. Chernov [Fri, 19 Apr 2002 08:16:01 +0000 (08:16 +0000)]
Use LC_ALL to pick collate

Noticed by: tjr

22 years agoStyle.
Ruslan Ermilov [Fri, 19 Apr 2002 07:43:50 +0000 (07:43 +0000)]
Style.

22 years agoFix the type of .include.
Ruslan Ermilov [Fri, 19 Apr 2002 07:42:41 +0000 (07:42 +0000)]
Fix the type of .include.

22 years agoFix the bsd.port.mk breakage caused by sys.mk,v 1.60 commit.
Ruslan Ermilov [Fri, 19 Apr 2002 07:40:34 +0000 (07:40 +0000)]
Fix the bsd.port.mk breakage caused by sys.mk,v 1.60 commit.

Submitted by: fenner

22 years agoAllocate sufficient pages to hold the bootinfo block and stop
Marcel Moolenaar [Fri, 19 Apr 2002 06:43:09 +0000 (06:43 +0000)]
Allocate sufficient pages to hold the bootinfo block and stop
hardwiring the location.

22 years agoRemove the bootinfo kludge. We get the address of the bootinfo
Marcel Moolenaar [Fri, 19 Apr 2002 06:35:55 +0000 (06:35 +0000)]
Remove the bootinfo kludge. We get the address of the bootinfo
block from the loader.

22 years agojust merged cosmetic changes from KAME to ease sync between KAME and FreeBSD.
SUZUKI Shinsuke [Fri, 19 Apr 2002 04:46:24 +0000 (04:46 +0000)]
just merged cosmetic changes from KAME to ease sync between KAME and FreeBSD.
(based on freebsd4-snap-20020128)

Reviewed by: ume
MFC after: 1 week

22 years agoDo some minor editing of the entires (the date of the smmsp user
Warner Losh [Fri, 19 Apr 2002 04:22:35 +0000 (04:22 +0000)]
Do some minor editing of the entires (the date of the smmsp user
wasn't April 4, but Feb 17th).  Update the updating proceedure to use
the new mergemaster -p flag.  Add a footnote telling users how to cope
if their mergemaster doesn't have a -p flag.

22 years ago o Move the acquisition of Giant from vm_fault() to the point
Alan Cox [Fri, 19 Apr 2002 04:20:31 +0000 (04:20 +0000)]
 o Move the acquisition of Giant from vm_fault() to the point
   after initialization in vm_fault1().
 o Fix some style problems in vm_fault1().

22 years agoMove the sysV variable substitution up with the BSD feature that replaced it.
David E. O'Brien [Fri, 19 Apr 2002 04:08:35 +0000 (04:08 +0000)]
Move the sysV variable substitution up with the BSD feature that replaced it.

22 years ago o Remove vm_map_growstack() from ia64's trap_pfault().
Alan Cox [Fri, 19 Apr 2002 02:08:33 +0000 (02:08 +0000)]
 o Remove vm_map_growstack() from ia64's trap_pfault().
 o Remove the acquisition and release of Giant from ia64's trap_pfault().
   (vm_fault() still acquires it.)

22 years agoClean up usage message: don't put optional parameter's in brackets.
Robert Watson [Fri, 19 Apr 2002 01:42:55 +0000 (01:42 +0000)]
Clean up usage message: don't put optional parameter's in brackets.
Update copyright date.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs
Extracted from: green

22 years agoWhen exec'ing a set[ug]id program, make sure that the stdio file descriptors
Jacques Vidrine [Fri, 19 Apr 2002 00:45:29 +0000 (00:45 +0000)]
When exec'ing a set[ug]id program, make sure that the stdio file descriptors
(0, 1, 2) are allocated by opening /dev/null for any which are not already
open.

Reviewed by: alfred, phk
MFC after: 2 days

22 years agoFix the breakage of tagged queueing that the busdma integration
Søren Schmidt [Thu, 18 Apr 2002 19:11:45 +0000 (19:11 +0000)]
Fix the breakage of tagged queueing that the busdma integration
introduced. Since its now only possible to have one DMA control
block at a time, we move the setup to dmastart instead.

22 years agoAdd a reference to the kenv(2) manpage.
Maxime Henrion [Thu, 18 Apr 2002 18:49:41 +0000 (18:49 +0000)]
Add a reference to the kenv(2) manpage.

22 years agoDon't list pam_unix in the session chain, since it does not provide any
Dag-Erling Smørgrav [Thu, 18 Apr 2002 17:40:27 +0000 (17:40 +0000)]
Don't list pam_unix in the session chain, since it does not provide any
session management services.

Sponsored by: DARPA, NAI Labs

22 years agoDe-__P, use ansi function definitions
Warner Losh [Thu, 18 Apr 2002 17:14:09 +0000 (17:14 +0000)]
De-__P, use ansi function definitions

22 years agoConnect the kenv.2 manpage to the build.
Maxime Henrion [Thu, 18 Apr 2002 16:46:23 +0000 (16:46 +0000)]
Connect the kenv.2 manpage to the build.

22 years agoAdd a manpage for the kenv(2) syscall that Chad David kindly
Maxime Henrion [Thu, 18 Apr 2002 16:45:57 +0000 (16:45 +0000)]
Add a manpage for the kenv(2) syscall that Chad David kindly
wrote for me.

Submitted by: davidc

22 years agoFix the fdisk(8) manual page.
Tom Rhodes [Thu, 18 Apr 2002 16:25:18 +0000 (16:25 +0000)]
Fix the fdisk(8) manual page.
s/partition/slice/ in fdisk.8
fix a bug in fdisk(8) where it did not honer -t when using -I

PR: 36563 and 35688
Submitted by: bde
Reviewed by: bde

22 years agoBack out previous commit, which was not at all approved by the maintainer
Dag-Erling Smørgrav [Thu, 18 Apr 2002 16:20:18 +0000 (16:20 +0000)]
Back out previous commit, which was not at all approved by the maintainer
of this code.  I very much doubt that "the FreeBSD way" really means "make
it as unreadable and unmaintable as possible", and I would like Makefile
style (which is not currently documented anywhere except in the minds of
bde and ru) to be discussed and agreed upon in the appropriate forum
before any further commits of this kind happen.

22 years agoAdd the PCI ID for an upcoming variant of a soon-to-be released series of
Scott Long [Thu, 18 Apr 2002 15:47:31 +0000 (15:47 +0000)]
Add the PCI ID for an upcoming variant of a soon-to-be released series of
cards.

22 years agoThis style is *way* outside the relm of any other of our Makefiles
David E. O'Brien [Thu, 18 Apr 2002 15:13:06 +0000 (15:13 +0000)]
This style is *way* outside the relm of any other of our Makefiles
(other than the single sys.mk).  Adjust to smell like FreeBSD.

Reviewed & prefered by: bde

22 years agoCleanup comments.
Alfred Perlstein [Thu, 18 Apr 2002 14:54:26 +0000 (14:54 +0000)]
Cleanup comments.

Remove all current poll(2) extensions except POLLINIGNEOF as they are now
unused.

Submitted by: bde

22 years agoRemove support for using soon to be retired "special" poll(2) ops.
Alfred Perlstein [Thu, 18 Apr 2002 14:52:28 +0000 (14:52 +0000)]
Remove support for using soon to be retired "special" poll(2) ops.
Replace with kevent(2) ops.

This is untested, but the code would rot even further if this wasn't
applied.  I've chosen to apply this to prompt some cleanup.

Submitted by: bde

22 years agoCleanup of logic, flow and comments.
Alfred Perlstein [Thu, 18 Apr 2002 14:47:34 +0000 (14:47 +0000)]
Cleanup of logic, flow and comments.

Submitted by: bde

22 years agoWork around an Intel 21143 chip bug.
Stephen McKay [Thu, 18 Apr 2002 14:40:20 +0000 (14:40 +0000)]
Work around an Intel 21143 chip bug.

Rev 1.56 of if_dc.c removed calls to mii_pollstat() from the dc_tick()
routine.  dc_tick() is called regularly to detect link up and link down
status, especially when autonegotiating.

The expectation was that mii_tick() (which is still called from dc_tick())
would update status information automatically in all cases where it would
be sensible to do so.

Unfortunately, with authentic 21143 chips this is not the case, and
the driver never successfully autonegotiates.  This is because (despite
what it says in the 21143 manual) the chip always claims that link is not
present while the autonegotiation enable bit is set.  Autonegotation takes
place and succeeds, but the driver tests the link bits before it switches
off the autonegotiation enable bit, and success is not recognised.

The simplest solution is to call dcphy_status() more often for MII_TICK
calls by dropping out of the switch statement instead of exiting when
we are autonegotiating and link appears to not be present.  When
autonegotiation succeeds, dcphy_status() will note the speed and fdx/hdx
state and turn off the autonegotiation enable bit.  The next call to
dcphy_status() will notice that link is present, and the dc driver code
will be notified.

Macronix chips also use this code, but implement link detection as
described in the manual, and hence don't need this patch.  However, tests
on a Macronix 98715AEC-C show that it does not adversely affect them.

This could be done better but is the minimal effective change, and most
closely mimics what was happening prior to rev 1.56 of if_dc.c.  (Actually
I also deleted a small amount of unnecessary code while I was in the area.)

Reviewed by: wpaul

22 years agoRemoved gratuitous LIBCOMPATDIR assignments.
Ruslan Ermilov [Thu, 18 Apr 2002 13:24:35 +0000 (13:24 +0000)]
Removed gratuitous LIBCOMPATDIR assignments.

22 years agoFixed some style bugs:
Bruce Evans [Thu, 18 Apr 2002 13:02:17 +0000 (13:02 +0000)]
Fixed some style bugs:
- ifdefs around vendor sccsid were left reversed after switching to
  __FBSDID().
- vertical whitespace after some of the non-FALLTHROUGH cases was lost.
- too much vertical whitespace before prototypes.

22 years agoBacked out the style bugs in rev.1.17:
Bruce Evans [Thu, 18 Apr 2002 12:53:39 +0000 (12:53 +0000)]
Backed out the style bugs in rev.1.17:
- don't use unusual indentation for 39 lines of declarations when only 2
  of the lines benefit from it.
- don't use __DECONST().  This was the one use of it in the tree, and it
  was just wrong.  It was used to hide the warning about tgetnum() having
  the wrong prototype (missing a `const') due to libncurses being
  misconfigured.  libncurses has been fixed, so the original code now
  compiles cleanly with WARNS=4.

22 years agoDo not reset MAKEFILE when reading ".depend" as this rather
Ruslan Ermilov [Thu, 18 Apr 2002 12:04:34 +0000 (12:04 +0000)]
Do not reset MAKEFILE when reading ".depend" as this rather
eliminates the usefulness of ${MAKEFILE}.

Obtained from: NetBSD
MFC after: 1 week

22 years agoFixed bugs in previous revision:
Ruslan Ermilov [Thu, 18 Apr 2002 10:58:14 +0000 (10:58 +0000)]
Fixed bugs in previous revision:

Added NOOBJ if anyone even attempts to "make obj" here.
Revert to installing files with mode 644 except README.
Make this overall look like a BSD-style Makefile rather
than roll-your-own (this is not a bug).

For the record.  Previous revision also fixed the breakage
introduced by the sys.mk,v 1.60 commit: bsd.own.mk is no
longer automatically included from sys.mk.

Reported by: jhay

22 years agoUse ${FILES} and <bsd.prog.mk> rather than roll-your-own.
Dag-Erling Smørgrav [Thu, 18 Apr 2002 10:07:36 +0000 (10:07 +0000)]
Use ${FILES} and <bsd.prog.mk> rather than roll-your-own.

22 years agoOptimize for i486 better (-m486 is just another deprecated
Ruslan Ermilov [Thu, 18 Apr 2002 09:37:15 +0000 (09:37 +0000)]
Optimize for i486 better (-m486 is just another deprecated
synonym for -mcpu=i486).

PR: i386/37212
Submitted by: Matthias Andree <matthias.andree@web.de>
MFC after: 3 days

22 years agoInstall bsd.init.mk.
Ruslan Ermilov [Thu, 18 Apr 2002 07:10:39 +0000 (07:10 +0000)]
Install bsd.init.mk.

Submitted by: "Peter S. Housel" <housel@acm.org>

22 years agoReplaced exists() tests with two equivalent defined().
Ruslan Ermilov [Thu, 18 Apr 2002 07:01:35 +0000 (07:01 +0000)]
Replaced exists() tests with two equivalent defined().

LIBDIR is defined in bsd.own.mk but sys.mk no longer
includes bsd.own.mk as of revision 1.60.

22 years agoINFODIR is defined in bsd.own.mk but sys.mk no longer includes
Ruslan Ermilov [Thu, 18 Apr 2002 06:55:32 +0000 (06:55 +0000)]
INFODIR is defined in bsd.own.mk but sys.mk no longer includes
bsd.own.mk as of share/mk/sys.mk,v 1.60.

I did not notice this because I tested with DESTDIR=/foo/5.0,
and the "exists(/foo/5.0)" test apparently succeeded.

Reported by: fenner

22 years agoThis Makefile.inc depended on the (wrong) order of inclusion
Ruslan Ermilov [Thu, 18 Apr 2002 06:46:34 +0000 (06:46 +0000)]
This Makefile.inc depended on the (wrong) order of inclusion
of makefiles: bsd.own.mk was included before ../Makefile.inc,
effectively hiding the (wrong) LIBCOMPATDIR assignment here.
share/mk/sys.mk,v 1.60 (and assorted share/mk fixes) fixed
this order, revealed this bug, and broke "make release" and
"make installworld" with either of -DCOMPAT1X or -DCOMPAT2?.

Reported by: jhay

22 years agoIn rare cases, we might want to include bsd.own.mk directly.
Ruslan Ermilov [Thu, 18 Apr 2002 06:37:48 +0000 (06:37 +0000)]
In rare cases, we might want to include bsd.own.mk directly.
So protect it from being included twice.

22 years agoAdd a comment documenting a race condition in vm_fault(): Specifically, a
Alan Cox [Thu, 18 Apr 2002 03:55:50 +0000 (03:55 +0000)]
Add a comment documenting a race condition in vm_fault(): Specifically, a
modification is made to the vm_map while only a read lock is held.

22 years agoSince WITNESS doesn't just do mutexes, remove "mutex" from the WITNESS
Robert Watson [Thu, 18 Apr 2002 03:44:44 +0000 (03:44 +0000)]
Since WITNESS doesn't just do mutexes, remove "mutex" from the WITNESS
comment in GENERIC config files of appropriate platforms.  For whatever
reason, powerpc didn't use WITNESS in GENERIC.

22 years agoWitness doesn't just track mutexes, so don't say mutexes specifically. It
Robert Watson [Thu, 18 Apr 2002 03:41:49 +0000 (03:41 +0000)]
Witness doesn't just track mutexes, so don't say mutexes specifically.  It
also tracks sxlocks, etc.