]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoFix an inconsistency between what FreeBSD NIS and AMD think the
Mark Murray [Wed, 6 Feb 2002 20:42:36 +0000 (20:42 +0000)]
Fix an inconsistency between what FreeBSD NIS and AMD think the
AMD map name should be.

22 years agoEnable TCP_WRAPPERs for the NIS server. The protection afforded is
Mark Murray [Wed, 6 Feb 2002 20:39:36 +0000 (20:39 +0000)]
Enable TCP_WRAPPERs for the NIS server. The protection afforded is
not massive, but usable.

22 years agoMake LINT compile after fruitless attempts to get the authors
Julian Elischer [Wed, 6 Feb 2002 19:35:37 +0000 (19:35 +0000)]
Make LINT compile after fruitless attempts to get the authors
to fix their code.

ata stuff:
Change name of ar_attach to not colide with existing ar_attach in if_ar.c.
usb stuff:
Create a dummy function to satisfy a call to it when in DEBUG mode.

22 years agoReadded the svc_create() and the registering of the local transport,
Alfred Perlstein [Wed, 6 Feb 2002 19:15:34 +0000 (19:15 +0000)]
Readded the svc_create() and the registering of the local transport,
now it is fixed. This should get us a working keyserv again, since
it depends on local transport for key exchange.

Since we do not have any KEYFILE name hardcoded anymore, set the
umask that way that the keyserver socket can be created with with
the appropriate permissions.

Re-add the accidently removed signal(SIGPIPE, SIG_IGN); to the code
which makes sense to avoid SIGPIPE when a disconnect on rpc socket
occurs.

Submitted by: mbr

22 years agoMove the AF_LOCAL check to the AF_INET check since portmap protocol
Alfred Perlstein [Wed, 6 Feb 2002 19:14:02 +0000 (19:14 +0000)]
Move the AF_LOCAL check to the AF_INET check since portmap protocol
v. 2 only needs this. That also makes it shorter and simpler.

Submitted by: mbr

22 years agoRemove NO_WERROR, now that WARNS=n is gone.
Mark Murray [Wed, 6 Feb 2002 18:46:48 +0000 (18:46 +0000)]
Remove NO_WERROR, now that WARNS=n is gone.

22 years agoComment out the WARNS= so as to not trample all over the GCC3 work.
Mark Murray [Wed, 6 Feb 2002 18:14:59 +0000 (18:14 +0000)]
Comment out the WARNS= so as to not trample all over the GCC3 work.

22 years agoBackout -Werror protection. No need to walk all over the upcoming
Mark Murray [Wed, 6 Feb 2002 18:13:12 +0000 (18:13 +0000)]
Backout -Werror protection. No need to walk all over the upcoming
GCC3 changeover.

22 years agoNeither .depend nor ${DEPENDFILE} depend on _SUBDIR.
Ruslan Ermilov [Wed, 6 Feb 2002 18:08:31 +0000 (18:08 +0000)]
Neither .depend nor ${DEPENDFILE} depend on _SUBDIR.
(One step closer to _SUBDIR <-> _SUBDIRUSE merge.)

22 years agoBack out revision 1.23 (SUBDIR_CHANGE).
Ruslan Ermilov [Wed, 6 Feb 2002 18:05:53 +0000 (18:05 +0000)]
Back out revision 1.23 (SUBDIR_CHANGE).

Not objected to by: bde, eivind

22 years agoFix sentence in comment
Alexey Zelkin [Wed, 6 Feb 2002 18:00:11 +0000 (18:00 +0000)]
Fix sentence in comment

22 years agoUpdate description of 'ac' subdirectory
Alexey Zelkin [Wed, 6 Feb 2002 17:50:05 +0000 (17:50 +0000)]
Update description of 'ac' subdirectory

22 years agoAdd descirptions of scripts in this directory
Alexey Zelkin [Wed, 6 Feb 2002 17:49:13 +0000 (17:49 +0000)]
Add descirptions of scripts in this directory

22 years agoAdd script that checks contents of .ent files (sorting and addresses)
Alexey Zelkin [Wed, 6 Feb 2002 17:47:18 +0000 (17:47 +0000)]
Add script that checks contents of .ent files (sorting and addresses)

22 years agoList of committers was moved out of handbook. Teach 'ac' about new location.
Alexey Zelkin [Wed, 6 Feb 2002 17:09:08 +0000 (17:09 +0000)]
List of committers was moved out of handbook. Teach 'ac' about new location.
Also teach it about complex entries in access file (or correctly to say
simplify them)

22 years agoo Remove __P
Warner Losh [Wed, 6 Feb 2002 16:51:09 +0000 (16:51 +0000)]
o Remove __P
o Use proper prototypes
o remove register

22 years agoRemove some unused variables, mark unused parameters as unused and change
Warner Losh [Wed, 6 Feb 2002 16:47:06 +0000 (16:47 +0000)]
Remove some unused variables, mark unused parameters as unused and change
names of variables that shadow globally declared variables.  This should
help people doing later WARNS= fixes.

22 years agoo __P removal
Warner Losh [Wed, 6 Feb 2002 16:38:40 +0000 (16:38 +0000)]
o __P removal
o remove register
o use strict prototypes

22 years agoRemove files that shouldn't have been imported.
Jacques Vidrine [Wed, 6 Feb 2002 16:31:41 +0000 (16:31 +0000)]
Remove files that shouldn't have been imported.
These will be surgically removed from the repository.

Approved by: markm
(with a CVS Meister hat on)

22 years agoremove -M from RPCCOM, it slipped in with the tirpc upgrade.
Alfred Perlstein [Wed, 6 Feb 2002 16:13:04 +0000 (16:13 +0000)]
remove -M from RPCCOM, it slipped in with the tirpc upgrade.

Submitted by: mbr

22 years agoANSIfy and remove some dead code.
Dag-Erling Smørgrav [Wed, 6 Feb 2002 15:26:07 +0000 (15:26 +0000)]
ANSIfy and remove some dead code.

Sponsored by: DARPA, NAI Labs

22 years agoApply the following mechanical transformations in preparation for
Dag-Erling Smørgrav [Wed, 6 Feb 2002 13:30:31 +0000 (13:30 +0000)]
Apply the following mechanical transformations in preparation for
ansification and constification:

    s{\s+__P\((\(.*?\))\)}{$1}g;
    s{\(\s+}{\(}g;
    s{\s+\)}{\)}g;
    s{\s+,}{,}g;
    s{(\s+)(for|if|switch|while)\(}{$1$2 \(}g;
    s{return ([^\(].*?);}{return ($1);}g;
    s{([\w\)])([!=+/\*-]?=)([\w\(+-])}{$1 $2 $3}g;
    s{\s+$}{\n};g

Also add $FreeBSD$ where needed.

MFC after: 1 week

22 years agoSet WFORMAT=0, overlooked in previous commits to libexec/.
Kris Kennaway [Wed, 6 Feb 2002 11:07:55 +0000 (11:07 +0000)]
Set WFORMAT=0, overlooked in previous commits to libexec/.

Reported by: jhay

22 years agoForced commit. Previous commit was in response to:
Dag-Erling Smørgrav [Wed, 6 Feb 2002 10:23:02 +0000 (10:23 +0000)]
Forced commit.  Previous commit was in response to:

PR: kern/33583

22 years agoDo a bit more of prototype cleanup.
Ruslan Ermilov [Wed, 6 Feb 2002 07:34:17 +0000 (07:34 +0000)]
Do a bit more of prototype cleanup.

22 years agoUse the stock [3.1-snap] version of this.
David E. O'Brien [Wed, 6 Feb 2002 05:10:06 +0000 (05:10 +0000)]
Use the stock [3.1-snap] version of this.

22 years agoA slightly more general version of rev 1.2 that might be more acceptable
David E. O'Brien [Wed, 6 Feb 2002 05:07:17 +0000 (05:07 +0000)]
A slightly more general version of rev 1.2 that might be more acceptable
to the GCC maintainers.

22 years agoMerge BDE's printf format error fixes (rev 1.2) into GCC 3.1-snap.
David E. O'Brien [Wed, 6 Feb 2002 05:07:01 +0000 (05:07 +0000)]
Merge BDE's printf format error fixes (rev 1.2) into GCC 3.1-snap.

22 years agoUse the stock GCC 3.1-snap version of this.
David E. O'Brien [Wed, 6 Feb 2002 05:05:49 +0000 (05:05 +0000)]
Use the stock GCC 3.1-snap version of this.

22 years agoA slightly different version of rev 1.2 that is closer to something the
David E. O'Brien [Wed, 6 Feb 2002 05:04:25 +0000 (05:04 +0000)]
A slightly different version of rev 1.2 that is closer to something the
GCC maintainers might actually accept.

22 years agoMerge rev 1.2 (printf format error fixes) into GCC 3.1-snap.
David E. O'Brien [Wed, 6 Feb 2002 05:03:43 +0000 (05:03 +0000)]
Merge rev 1.2 (printf format error fixes) into GCC 3.1-snap.

22 years agoUse the stock [3.l-snap] version of this.
David E. O'Brien [Wed, 6 Feb 2002 05:02:18 +0000 (05:02 +0000)]
Use the stock [3.l-snap] version of this.

22 years agoUse the stock [3.1-snap] file.
David E. O'Brien [Wed, 6 Feb 2002 05:01:51 +0000 (05:01 +0000)]
Use the stock [3.1-snap] file.

22 years agoUse the stock file for now.
David E. O'Brien [Wed, 6 Feb 2002 05:01:29 +0000 (05:01 +0000)]
Use the stock file for now.

The changes from GCC 2.95.3 to 3.1 are large and it is hard to figure
out how to apply our changes.  Once BDE can properly test GCC 3.1 he'll
add back in our changes.

22 years agoMerge the stock gcc-3.1_snap file onto HEAD.
David E. O'Brien [Wed, 6 Feb 2002 05:00:48 +0000 (05:00 +0000)]
Merge the stock gcc-3.1_snap file onto HEAD.

22 years agoMerge rev 1.2 (-fformat-extensions) and rev 1.7 (optimization downgrade)
David E. O'Brien [Wed, 6 Feb 2002 04:59:39 +0000 (04:59 +0000)]
Merge rev 1.2 (-fformat-extensions) and rev 1.7 (optimization downgrade)
into GCC 3.1-snap.

22 years agoInstall PROTO.localhost-v6.rev. Umm, it seems namedb/Makefile
Hajimu UMEMOTO [Wed, 6 Feb 2002 04:57:25 +0000 (04:57 +0000)]
Install PROTO.localhost-v6.rev.  Umm, it seems namedb/Makefile
is not used.

22 years agoWe use the stock [3.1-snap] file.
David E. O'Brien [Wed, 6 Feb 2002 04:49:08 +0000 (04:49 +0000)]
We use the stock [3.1-snap] file.

22 years agoInstall PROTO.localhost-v6.rev.
Hajimu UMEMOTO [Wed, 6 Feb 2002 04:46:07 +0000 (04:46 +0000)]
Install PROTO.localhost-v6.rev.

Reported by: Scott Allendorf <scott-allendorf@uiowa.edu>
Forgot by: me (ume)

22 years agoFixes for alpha pmap on SMP machines:
John Baldwin [Wed, 6 Feb 2002 04:30:26 +0000 (04:30 +0000)]
Fixes for alpha pmap on SMP machines:

- Create a private list of active pmaps rather than abusing the list of all
  processes when we need to look up pmaps.  The process list needs a sx lock
  and we can't be getting sx locks in the middle of cpu_switch()
  (pmap_activate() can call pmap_get_asn() from cpu_switch()).  Instead, we
  protect the list with a spinlock.  This also means the list is shorter
  since a pmap can be used by more than one process and we could (at least
  in thoery) dink with pmap's more than once, but now we only touch each
  pmap once when we have to update all of them.
- Wrap pmap_activate()'s code to get a new ASN in an explicit critical section
  so that when it is called while doing an exec() we can't get preempted.
- Replace splhigh() in pmap_growkernel() with a critical section to prevent
  preemption while we are adjusting the kernel page tables.
- Fixes abuse of PCPU_GET(), which doesn't return an L-value.
- Also adds some slight cleanups to the ASN handling by adding some macros
  instead of magic numbers in relation to the ASN and ASN generations.

Reviewed by: dfr

22 years agoMerge revs 1.{2,15} (FREEBSD_NATIVE); 1.{5,13} (ELF, and objformat support);
David E. O'Brien [Wed, 6 Feb 2002 02:53:38 +0000 (02:53 +0000)]
Merge revs 1.{2,15} (FREEBSD_NATIVE); 1.{5,13} (ELF, and objformat support);
1.16 (allow specification of TOOLS_PREFIX); 1.21 ("GCC_OPTIONS")
into GCC 3.1-snap.

Note that rev 1.20 was fixed in the stock GCC 3 sources.

22 years agoMake cron actually build without defining LOGIN_CAP.
Rob Braun [Wed, 6 Feb 2002 02:00:07 +0000 (02:00 +0000)]
Make cron actually build without defining LOGIN_CAP.

Reviewed by: jkh

22 years agoDon't attach to Dell PERC2/QC cards that have a firmware rev of 1.x. This
Scott Long [Wed, 6 Feb 2002 01:34:09 +0000 (01:34 +0000)]
Don't attach to Dell PERC2/QC cards that have a firmware rev of 1.x.  This
check is complicated by the fact that the Adaptec 5400S cards claim to use
1.x firmware also.  PERC2/QC 1.x firmware is not compatible with this driver
and will cause a system hang.

MFC after: 3 days

22 years agoAllow the kern.maxusers boot tuneable to be set to 0 (previously only
Matthew Dillon [Wed, 6 Feb 2002 01:19:19 +0000 (01:19 +0000)]
Allow the kern.maxusers boot tuneable to be set to 0 (previously only
the kernel config's maxusers could be set to 0 for autosizing to work).
Reviewed by: rwatson, imp
MFC after: 3 days

22 years agobalance parens.
Alfred Perlstein [Tue, 5 Feb 2002 23:56:29 +0000 (23:56 +0000)]
balance parens.

Submitted by: mbr

22 years agoFor RPC v2 we need to fill in ret->xp_raddr too. INET_6
Alfred Perlstein [Tue, 5 Feb 2002 23:46:37 +0000 (23:46 +0000)]
For RPC v2 we need to fill in ret->xp_raddr too. INET_6
is not supported in portmap, so we don't have to care about it.

Submitted by: mbr

22 years agoFix cc -Wall, fix rcsid warnings, add missing prototypes,
Alfred Perlstein [Tue, 5 Feb 2002 23:43:43 +0000 (23:43 +0000)]
Fix cc -Wall, fix rcsid warnings, add missing prototypes,
change prototypes to be the same as in the original sun tirpc code.
Remove ()P macro in a file where the mayority had ()P already removed.
Add them if the mayority use ()P macros.

Submitted by: mbr
Requested by: bde

22 years agoCorrectly identify the Intel 82830 AGP bridge.
Benno Rice [Tue, 5 Feb 2002 23:13:25 +0000 (23:13 +0000)]
Correctly identify the Intel 82830 AGP bridge.

22 years agoDo not call addrinfo on local transport adresses, if returning a
Alfred Perlstein [Tue, 5 Feb 2002 23:12:51 +0000 (23:12 +0000)]
Do not call addrinfo on local transport adresses, if returning a
server handle (for reuse or whatever). We just return now a handle
connected to the local rpcbind.

Do not try to call checkcache, if host = NULL;

Submitted by: mbr

22 years agoSwitch to a self-starting allocation scheme.
Dag-Erling Smørgrav [Tue, 5 Feb 2002 22:15:16 +0000 (22:15 +0000)]
Switch to a self-starting allocation scheme.

22 years agoReindent, and add parentheses to return statements. Some functions in
Dag-Erling Smørgrav [Tue, 5 Feb 2002 22:13:51 +0000 (22:13 +0000)]
Reindent, and add parentheses to return statements.  Some functions in
ftp.c and http.c now have exceedingly long lines due to deep nesting;
this will be corrected by reorganizing the code in a later revision.

22 years agoFix the race between the stat() and the mkdir().
Dave Zarzycki [Tue, 5 Feb 2002 21:55:12 +0000 (21:55 +0000)]
Fix the race between the stat() and the mkdir().
Reviewed by: jkh

22 years agoUpdate the physio man page to better reflect reality.
Josef Karthauser [Tue, 5 Feb 2002 21:32:57 +0000 (21:32 +0000)]
Update the physio man page to better reflect reality.

PR: docs/30603
Reviewed by: phk
MFC after: 3 days

22 years agoEliminate a redundant loop.
Dag-Erling Smørgrav [Tue, 5 Feb 2002 21:32:16 +0000 (21:32 +0000)]
Eliminate a redundant loop.

22 years agoFix a race with free'ing vmspaces at process exit when vmspaces are
Alfred Perlstein [Tue, 5 Feb 2002 21:23:05 +0000 (21:23 +0000)]
Fix a race with free'ing vmspaces at process exit when vmspaces are
shared.

Also introduce vm_endcopy instead of using pointer tricks when
initializing new vmspaces.

The race occured because of how the reference was utilized:
  test vmspace reference,
  possibly block,
  decrement reference

When sharing a vmspace between multiple processes it was possible
for two processes exiting at the same time to test the reference
count, possibly block and neither one free because they wouldn't
see the other's update.

Submitted by: green

22 years agoFix minor disorder in functions declared in extern.h
Warner Losh [Tue, 5 Feb 2002 21:07:47 +0000 (21:07 +0000)]
Fix minor disorder in functions declared in extern.h

22 years agoo Move externs to extern.h
Warner Losh [Tue, 5 Feb 2002 21:06:56 +0000 (21:06 +0000)]
o Move externs to extern.h
o Use new-style prototypes exclusively rather than the old foo() style.
o Use new-style function definitions.
o remove register
o make functions passed to signal have the right signature.
o do minor const poisoning.

22 years agoLet the number of timecounters follow hz, otherwise people with
Poul-Henning Kamp [Tue, 5 Feb 2002 20:44:56 +0000 (20:44 +0000)]
Let the number of timecounters follow hz, otherwise people with
HZ=BIGNUM will strain the assumptions behind timecounters to the
point where they break.

This may or may not help people seeing microuptime() backwards messages.

Make the global timecounter variable volatile, it makes no difference in
the code GCC generates, but it makes represents the intent correctly.

Thanks to: jdp
MFC after: 2 weeks

22 years agostyle: return(x) -> return (x)
Alfred Perlstein [Tue, 5 Feb 2002 19:31:16 +0000 (19:31 +0000)]
style: return(x) -> return (x)

Submitted by: mbr

22 years agoFix the credential handling code.
Alfred Perlstein [Tue, 5 Feb 2002 19:30:30 +0000 (19:30 +0000)]
Fix the credential handling code.

In NetBSD, Solaris, xprt->xp_p2 pointed directly to the credentials,
in FreeBSD xprt->xp_verf.oa_base was a pointer to a struct cmessage,
which is defined as follow:

struct cmessage {
        struct cmsghdr cmsg;
        struct cmsgcred cmcred;
};

The credentials were submitted the right way and xprt->xp_p2 pointed to them.
But cb_verf.oa_flavor was still empty. There was an assignment missing
in svc_recv() in svc_vc.c:

msg->rm_call.cb_verf.oa_flavor = AUTH_UNIX;

Also

+       if (addr.ss_family == AF_LOCAL) {
+               xprt->xp_raddr = *(struct sockaddr_in *)xprt->xp_rtaddr.buf;
+               xprt->xp_addrlen = sizeof (struct sockaddr_in);
+       }

was missing. But the first seems not to be needed:

I guess in rpc.yppasswdd there was a typo:

- transp>xp_verf.oa_flavor != AUTH_UNIX) {
+ rqstp->rq_cred.oa_flavor != AUTH_UNIX) {

This little fix does fix the breakage in rpc.yppasswdd :-)

+       if (msg.msg_controllen == 0 ||
+           (msg.msg_flags & MSG_CTRUNC) != 0)
+               return (-1);

We cannot set the cb_verf.oa_length in svc_recv() of svc_vc.c,
the credentials get overwritten then, and that's bad.

Submitted by: mbr

22 years agoThe clnt_create(KEYSERVSOCK, CRYPT_PROG, CRYPT_VERS, "unix") hacks
Alfred Perlstein [Tue, 5 Feb 2002 19:26:04 +0000 (19:26 +0000)]
The clnt_create(KEYSERVSOCK, CRYPT_PROG, CRYPT_VERS, "unix") hacks
were removed and replaced them with clnt_tp_create, now the af_local
support is fixed.

I also removed the hack how rpcinfo contacted rpcbind, now we can
relay on clnt_tp_create create the client-handle for us.  Only
rpcbind itself needs a hardcoded socket-path.

Submitted by: mbr

Also add $FreeBSD

22 years agoFix the breakage in rpc.yppasswd. Readded the svc_create() and
Alfred Perlstein [Tue, 5 Feb 2002 18:51:13 +0000 (18:51 +0000)]
Fix the breakage in rpc.yppasswd. Readded the svc_create() and
the registering of the "unix" transport, now it is fixed.

Everywhere, rq_cred is taken to look what authentification we have.
We can not be sure that transp>xp_verf.oa_flavor is also filled in.
This seems to be the same for all sun source. they take the flavor
of rq_cred, instead of transp.

Submitted by: mbr

22 years agoThe hardware checksum code makes the assumption that a packet routed out
Archie Cobbs [Tue, 5 Feb 2002 18:27:30 +0000 (18:27 +0000)]
The hardware checksum code makes the assumption that a packet routed out
a particular Ethernet interface will actually be delivered by (only) that
device driver. This is not necessarily true when ng_ether(4) is used.

To word around this, while a ng_ether(4)'s "upper" hook is connected,
turn off all hardware checksum, fragmentation, etc., features for that
interface.

PR: kern/31586
MFC after: 1 week

22 years agotab in pw_mkdb's arg for style.
Alfred Perlstein [Tue, 5 Feb 2002 18:27:06 +0000 (18:27 +0000)]
tab in pw_mkdb's arg for style.

22 years agoConstify things to unbreak world.
Alfred Perlstein [Tue, 5 Feb 2002 18:25:59 +0000 (18:25 +0000)]
Constify things to unbreak world.

Submitted by: David Wolfskill <david@catwhisker.org>

22 years agoAdd a manpage for VOP_REVOKE(9).
Brian Feldman [Tue, 5 Feb 2002 16:08:49 +0000 (16:08 +0000)]
Add a manpage for VOP_REVOKE(9).

Sponsored by: DARPA, NAI Labs

22 years agoDocument that VOP_GETATTR(9) does not actually expect to hold an
Brian Feldman [Tue, 5 Feb 2002 15:26:28 +0000 (15:26 +0000)]
Document that VOP_GETATTR(9) does not actually expect to hold an
exclusive lock on entry.

Sponsored by: DARPA, NAI Labs

22 years agoDon't use non-signal-safe functions (exit(3) in this case) in
Sheldon Hearn [Tue, 5 Feb 2002 15:20:02 +0000 (15:20 +0000)]
Don't use non-signal-safe functions (exit(3) in this case) in
signal handlers.  In this case, use _exit(2) instead, following
the call to shutdown(2).

This fixes rare telnetd hangs.

PR: misc/33672
Submitted by: Umesh Krishnaswamy <umesh@juniper.net>
MFC after: 1 month

22 years agoAdd support for the Aztech 2320 chip.
Thomas Gellekum [Tue, 5 Feb 2002 12:09:23 +0000 (12:09 +0000)]
Add support for the Aztech 2320 chip.

Reviewed by: cg
Obtained from: NetBSD (partly)

22 years agoFix -F with logfiles rotated on time only.
Peter Pentchev [Tue, 5 Feb 2002 09:33:07 +0000 (09:33 +0000)]
Fix -F with logfiles rotated on time only.

Approved by: silence on -audit
MFC after: 1 month

22 years agoThis commit was generated by cvs2svn to compensate for changes in r90238,
Sheldon Hearn [Tue, 5 Feb 2002 09:17:24 +0000 (09:17 +0000)]
This commit was generated by cvs2svn to compensate for changes in r90238,
which included commits to RCS files with non-trunk default branches.

22 years agoFix obscure bug in code loading support that could cause parsing of
Sheldon Hearn [Tue, 5 Feb 2002 09:17:24 +0000 (09:17 +0000)]
Fix obscure bug in code loading support that could cause parsing of
functions in multi-line files to fail, depending on the state of the
stack.  The fix is sanctioned by the vendor and will appear in bc-1.07.

PR: bin/34601
Reported by: mi
Obtained from: Phil Nelson <phil@cs.wwu.edu>

22 years agoThree times lucky: <stddef.h>, not <sys/param.h>
Dag-Erling Smørgrav [Tue, 5 Feb 2002 08:01:32 +0000 (08:01 +0000)]
Three times lucky: <stddef.h>, not <sys/param.h>

22 years agoOops, the correct header to include for NULL is <sys/param.h>.
Dag-Erling Smørgrav [Tue, 5 Feb 2002 07:53:00 +0000 (07:53 +0000)]
Oops, the correct header to include for NULL is <sys/param.h>.

22 years ago- Do not hang if the resource allocation fails.
Seigo Tanimura [Tue, 5 Feb 2002 06:52:56 +0000 (06:52 +0000)]
- Do not hang if the resource allocation fails.
- Add another quirk entry of SB AWE64.

PR: kern/32530
Submitted by: Magnus Backstrom <b@etek.chalmers.se>

22 years agoANSIfy and constify.
Dag-Erling Smørgrav [Tue, 5 Feb 2002 06:49:11 +0000 (06:49 +0000)]
ANSIfy and constify.

Sponsored by: DARPA, NAI Labs

22 years agoMove arguments in prototype out of application namespace.
Dag-Erling Smørgrav [Tue, 5 Feb 2002 06:31:58 +0000 (06:31 +0000)]
Move arguments in prototype out of application namespace.

Sponsored by: DARPA, NAI Labs

22 years agoMakes __istype() an inline (and a non inlined) function to avoid C++
Rob Braun [Tue, 5 Feb 2002 06:21:34 +0000 (06:21 +0000)]
Makes __istype() an inline (and a non inlined) function to avoid C++
compile issues.  std::isspace(' ') was expanding to std::(!!_maskrune...)
which would cause a C++ compile error.  Making __istype() an inline
causes the expansion to be std::__istype() instead, which is valid.

Reviewed by: jkh

22 years ago#include <sys/types.h> for NULL (hidden by Linux-PAM header pollution)
Dag-Erling Smørgrav [Tue, 5 Feb 2002 06:20:27 +0000 (06:20 +0000)]
#include <sys/types.h> for NULL (hidden by Linux-PAM header pollution)

Sponsored by: DARPA, NAI Labs

22 years ago#include cleanup.
Dag-Erling Smørgrav [Tue, 5 Feb 2002 06:08:26 +0000 (06:08 +0000)]
#include cleanup.

Sponsored by: DARPA, NAI Labs

22 years agoMFen (1.272 --> 1.275).
Jun Kuriyama [Tue, 5 Feb 2002 02:23:14 +0000 (02:23 +0000)]
MFen (1.272 --> 1.275).

22 years agoGet rid of the twisted MFREE() macro entirely.
Matthew Dillon [Tue, 5 Feb 2002 02:00:56 +0000 (02:00 +0000)]
Get rid of the twisted MFREE() macro entirely.

Reviewed by: dg, bmilekic
MFC after: 3 days

22 years agoGive the next BIND-importing schmuck some assistance.
Jacques Vidrine [Mon, 4 Feb 2002 21:28:45 +0000 (21:28 +0000)]
Give the next BIND-importing schmuck some assistance.

22 years ago+ A variety of 23XX changes:
Matt Jacob [Mon, 4 Feb 2002 21:04:25 +0000 (21:04 +0000)]
+ A variety of 23XX changes:
disable MWI on 2300

based on function code, set an 'isp_port' for the 2312- it's a
separate instance, but the NVRAM is shared, and the second port's
NVRAM is at offset 256.

+ Enable RIO operation for LVD SCSI cards. This makes a *big* difference
as even under reasonable load we get batched completions of about 30
commands at a time on, say, an ISP1080.

+ Do 'continuation' mailbox commands- this allows us to specify a work
area within the softc and 'continue' repeated mailbox commands. This is
more or less on an ad hoc basis and is currently only used for firmware
loading (which f/w now loads substantially faster becuase the calling
thread is only woken when all the f/w words are loaded- not for each
one of the 40000 f/w words that gets loaded).

+ If we're about to return from isp_intr with a 'bogus interrupt' indication,
and we're not a 23XX card, check to see whether the semaphore register is
currently *2* (not *1* as it should be) and whether there's an async completion
sitting in outgoing mailbox0. This seems to capture cases of lost fast posting
and RIO interrupts that the 12160 && 1080 have been known to pump out under
extreme load (extreme, as in > 250 active commands).

+ FC_SCRATCH_ACQUIRE/FC_SCRATCH_RELEASE macros.

+ Endian correct swizzle/unswizzle of an ATIO2 that has a WWPN in it.

MFC after: 1 week

22 years agoMove the pseudofs, procfs and linprocfs modules out from the fs directory.
Dag-Erling Smørgrav [Mon, 4 Feb 2002 20:16:50 +0000 (20:16 +0000)]
Move the pseudofs, procfs and linprocfs modules out from the fs directory.
Keeping them there seemed like a good idea at the time, but it annoys bde
and confuses people who do not understand how MODULES_OVERRIDE works.

22 years agoNew release note: ldd(1) can be used on shared libraries.
Bruce A. Mah [Mon, 4 Feb 2002 20:02:07 +0000 (20:02 +0000)]
New release note:  ldd(1) can be used on shared libraries.

Updated release note:  BIND 8.3.1-REL.

22 years agoMatch the requirements of the new kernel structures.
Søren Schmidt [Mon, 4 Feb 2002 19:24:43 +0000 (19:24 +0000)]
Match the requirements of the new kernel structures.

22 years agoMajor update of the ATA RAID code, part 1:
Søren Schmidt [Mon, 4 Feb 2002 19:23:40 +0000 (19:23 +0000)]
Major update of the ATA RAID code, part 1:

Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.

A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.

Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.

Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.

Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.

If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.

Misc fixes that I had lying around for various minor bugs.

Sponsored by: Advanis Inc.

22 years agogetgrouplist now takes `gid_t' instead of `int'.
Jacques Vidrine [Mon, 4 Feb 2002 19:19:34 +0000 (19:19 +0000)]
getgrouplist now takes `gid_t' instead of `int'.

22 years agohex.c is new in libisc with 8.3.1-REL.
Jacques Vidrine [Mon, 4 Feb 2002 19:18:23 +0000 (19:18 +0000)]
hex.c is new in libisc with 8.3.1-REL.

22 years agoResolve conflicts after import of ISC BIND 8.3.1-REL.
Jacques Vidrine [Mon, 4 Feb 2002 19:15:33 +0000 (19:15 +0000)]
Resolve conflicts after import of ISC BIND 8.3.1-REL.

22 years agoThis commit was generated by cvs2svn to compensate for changes in r90209,
Jacques Vidrine [Mon, 4 Feb 2002 19:12:46 +0000 (19:12 +0000)]
This commit was generated by cvs2svn to compensate for changes in r90209,
which included commits to RCS files with non-trunk default branches.

22 years agoImport of ISC BIND 8.3.1-REL.
Jacques Vidrine [Mon, 4 Feb 2002 19:12:46 +0000 (19:12 +0000)]
Import of ISC BIND 8.3.1-REL.

22 years agoChange EPERM to EOPNOTSUPP when failing pseudofs_setattr() arbitrarily.
Robert Watson [Mon, 4 Feb 2002 18:21:59 +0000 (18:21 +0000)]
Change EPERM to EOPNOTSUPP when failing pseudofs_setattr() arbitrarily.

Quoth the alfred: The latter would be better.

22 years agoReturn EPERM instead of 0 in the un-implemented pseudofs_setattr().
Robert Watson [Mon, 4 Feb 2002 18:09:29 +0000 (18:09 +0000)]
Return EPERM instead of 0 in the un-implemented pseudofs_setattr().
Conceivably, it should even return EOPNOTSUPP.

22 years agoExplicitly declare (gcc internal) functions.
Mark Murray [Mon, 4 Feb 2002 17:59:25 +0000 (17:59 +0000)]
Explicitly declare (gcc internal) functions.

Submitted by: ru

22 years agoo Scatter vn_start_write() and vn_finished_write() through ACL code so
Robert Watson [Mon, 4 Feb 2002 17:58:15 +0000 (17:58 +0000)]
o Scatter vn_start_write() and vn_finished_write() through ACL code so
  that it interacts properly with snapshotting.

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

22 years agoNew release notes: ngctl(8) write, ps(1) -W deorbit burn.
Bruce A. Mah [Mon, 4 Feb 2002 17:50:01 +0000 (17:50 +0000)]
New release notes:  ngctl(8) write, ps(1) -W deorbit burn.

MFCs noted:  dumpdev loader tunable.

22 years agoCorrected an argument to in6_pcbnotify().
Hajimu UMEMOTO [Mon, 4 Feb 2002 17:40:36 +0000 (17:40 +0000)]
Corrected an argument to in6_pcbnotify().

Obtained from: KAME
MFC after: 1 week

22 years agoIn tcp_respond(), correctly reset returned IPv6 header. This is essential
Hajimu UMEMOTO [Mon, 4 Feb 2002 17:37:06 +0000 (17:37 +0000)]
In tcp_respond(), correctly reset returned IPv6 header.  This is essential
when the original packet contains an IPv6 extension header.

Obtained from: KAME
MFC after: 1 week