Nabble - NetBSD

Re: Boot-Probleme mit dhcpd und named auf ALIX-Board

2009-11-25T13:53:30Z

P.S.:
Das Problem hat natuerlich nix mit dem r/o Filesystem und ALIX zu tun.

- Hubert (sehr gluecklich mit 'nem ALIX Board als Router:
10:52PM up 109 days, 13:54, 0 users, load averages: 0.00, 0.01, 0.00)

From forum: regional-de

Re: xsrc/41950: Xorg -configure is broken

2009-11-25T13:50:05Z

The following reply was made to PR xsrc/41950; it has been noted by GNATS.

From: "Jonathan A. Kollasch" <jakllsch@...>
To: gnats-bugs@...
Cc:
Subject: Re: xsrc/41950: Xorg -configure is broken
Date: Wed, 25 Nov 2009 21:45:10 +0000

The following patch fixes this:

Index: external/mit/xf86-video-radeonhd/dist/src/rhd_id.c
===================================================================
RCS file: /cvsroot/xsrc/external/mit/xf86-video-radeonhd/dist/src/rhd_id.c,v
retrieving revision 1.1.1.4
diff -u -r1.1.1.4 rhd_id.c
--- external/mit/xf86-video-radeonhd/dist/src/rhd_id.c 9 Nov 2009 01:43:57 -0000 1.1.1.4
+++ external/mit/xf86-video-radeonhd/dist/src/rhd_id.c 25 Nov 2009 21:41:07 -0000
@@ -112,7 +112,7 @@
# define RHD_DEVICE_MATCH(d, i) \
{ 0x1002, (d), PCI_MATCH_ANY, PCI_MATCH_ANY, 0, 0, (i) }
# define PCI_ID_LIST struct pci_id_match RHDDeviceMatch[]
-# define LIST_END { 0, 0, (~0), (~0), 0, 0, 0 }
+# define LIST_END { 0, 0, 0, 0, 0, 0, 0 }
#else
resRange res_none[] = { _END };

From forum: netbsd-bugs

Re: Boot-Probleme mit dhcpd und named auf ALIX-Board

2009-11-25T13:42:22Z

Hallo Thomas,

On Wed, 25 Nov 2009, Thomas Kaepernick wrote:
> /etc/rc.conf.d/named: 13: Syntax error: "}" unexpected
...

> /etc/rc.conf.d/dhcpd
> option domain-name "Zuhause";
...

Pilot error: In /etc/rc.conf.d werden Eintraege wie in /etc/rc.conf
erwartet, pro Dienst eine Datei und pro Datei normalerweise *nur* die
Zeilen fuer den Dienst. Damit kann dann ein Dienst sein eigenes Script
editieren, und nicht das Systemweite Config-File /etc/rc.conf
kaputtschiessen. Also z.B. in /etc/rc.conf.d/dhcpd sollte sowas wie ein
"dhcpd=yes" stehen. Die Config Files fuer named, dhcpd etc. bleiben in
/etc/named.conf, /etc/dhcpd.conf, etc.

Das erklaert dann auch die Fehlermeldungen: Deine /etc/rc.conf.d/* werden
als /bin/sh-Scripten interpretiert, und da setzt's dann aus.

Recommended reading: /etc/rc.subr, rc.conf(5).

- Hubert

From forum: regional-de

Re: Cambridge meet?

2009-11-25T13:34:26Z

On Wed, Nov 25, 2009 at 12:15:43PM +0000, Silke Scheler wrote:

> On Mon, Nov 16, 2009 at 12:08:58PM +0000, Stephen Borrill wrote:
> > It's been a couple of months since EuroBSDCon, elric@ is now in the UK and
> > would like to meet up, plus I've got stuff to give to abs.
> >
> > How are the following dates for everyone?
> >
> > 27th Nov
> > 4th Dec
> > 11th Dec
>
> Did we reach a consensus on this? To me it looked like the 11th
> might be the favored day?

I've a xmas curry on the 11th :-(
The date was either the 4th or 11th until that settled down!

David

--
David Laight: david@...

From forum: regional-london

Boot-Probleme mit dhcpd und named auf ALIX-Board

2009-11-25T13:18:37Z

Liebe Listener,

ich habe mir ein ALIX-Board mit einer CF-Karte eingerichtet und darauf
NetBSD 5.0 installiert. Das ganze Dateisystem dort ist bis auf ein
Verzeichnis readonly gemountet. Dorthin werden die Verzeichnisse /dev/
und /var kopiert und dann über die ursprünglichen Verzeichnis mit Nullfs
drübergelegt. Aber ich habe bei einigen Diensten Probleme und komme
nicht auf die Lösung.

Da ist ersten ein DHCP-Server. Dieser meint beim Starten, dass in seiner
Config-Datei eine unerwartete geschweifte Klammer auftritt. Ich habe die
Datei immer wieder kontrolliert. Ich sehe keine Klammer zuviel.

Zum anderen soll ein Nameserver laufen. Er beklagt beim Starten genau
das Gleiche. Die Datei root.cache, die auch fehlen soll, ist im
entsprechenden chroot vorhanden.

Was läuft hier falsch oder habe ich falsch gemacht?
Im Anschluß habe ich die Ausgabe beim Booten und die Config-Dateien
drangehängt

Gruß Thomas

Boot-Meldung:

Building databases: utmp, utmpx, services done
zone: not found
hint: not found
root.cache: cannot open `root.cache' (No such file or directory)
/etc/rc.conf.d/named: 13: Syntax error: "}" unexpected
...
zone: not found
hint: not found
root.cache: cannot open `root.cache' (No such file or directory)
/etc/rc.conf.d/named: 13: Syntax error: "}" unexpected
...
option: not found
option: not found
zone: not found
primary: not found
/etc/rc.conf.d/dhcpd: 12: Syntax error: "}" unexpected

Die Konfigurations-dateien dafür sind:

/etc/rc.conf.d/dhcpd
option domain-name "Zuhause";
option domain-name-servers ALIX.Zuhause;

zone Zuhause. {
primary 127.0.0.1;
}

zone 0.168.192.in-addr.arpa. {
primary 127.0.0.1;
}

default-lease-time 600;
max-lease-time 7200;

subnet 192.168.0.0 netmask 255.255.255.224 {
option domain-name "Zuhause.";
option domain-name-servers 192.168.0.1;
option routers 192.168.0.1;
option subnet-mask 255.255.255.224;
option broadcast-address 192.168.0.14;
pool {
range 192.168.0.4 192.168.0.9;
allow unknown-clients;
ignore client-updates;
}
pool {
range 192.168.0.2 192.168.0.3;
deny unknown-clients;
}
pool {
range 192.168.0.11 192.168.0.11;
deny unknown-clients;
}
}
host eins {
hardware ethernet 00:00:00:00:00:00;
fixed-address 192.168.0.2;
}

host Struppi {
hardware ethernet 11:11:11:11:11:11;
fixed-address 192.168.0.3;
}

host Haddock { hardware ethernet 22:22:22:22:22:22;
fixed-address 192.168.0.11;
}

/etc/rc.conf.d/named

zone "." {
type hint;
file "root.cache";
};

zone "localhost" {
type master;
file "localhost";
};

zone "127.IN-ADDR.ARPA" {
type master;
file "127";
};

zone
"1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.ip6.int"
{
type master;
file "loopback.v6";
};

zone
"1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.ip6.arpa"

{
type master;
file "loopback.v6";
};

zone "Herge" {
type master;
file "Zuhause";
};

zone "0.168.192.IN-ADDR.ARPA" {
type master;
file "192.168.0";
};

From forum: regional-de

Re: xsrc/41951: amd64 + radeondrm crash

2009-11-25T13:05:04Z

The following reply was made to PR xsrc/41951; it has been noted by GNATS.

From: "Jonathan A. Kollasch" <jakllsch@...>
To: gnats-bugs@...
Cc:
Subject: Re: xsrc/41951: amd64 + radeondrm crash
Date: Wed, 25 Nov 2009 21:04:05 +0000

This no longer happens on -current. (It ends up hard locking somewhat
often when running non-trivial GL apps however.)

From forum: netbsd-bugs

Re: chrooted named kills syslogd?

2009-11-25T13:02:00Z

At Tue, 24 Nov 2009 17:54:45 -0500, Jan Schaumann <jschauma@...> wrote:
Subject: Re: chrooted named kills syslogd?

>
> Matthias Scheler <tron@...> wrote:
> >
> > On 23 Nov 2009, at 22:58, Jan Schaumann wrote:
> > > I have a named running in named_chrootdir="/var/chroot/named"; it
> > > appears that if this named is running, syslogd hangs:
> >
> > It works fine on two systems I look after. Can you describe your
> > configuration more precisely, please? What version of NetBSD are you
> > using on what platform?
>
> NetBSD panix.netmeister.org 5.0 NetBSD 5.0 (PANIX-VC) #0: Tue Apr 28
> 20:51:45 EDT 2009
> root@...:/misc3/obj/misc2/devel/netbsd/5.0/src/sys/arch/amd64/compile/PANIX-VC
> amd64
>
> This is a domU. There's nothing out of the ordinary (that I'm aware of)
> in this build -- ie, that should be a plain vanilla 5.0 build.

The most obvious thing to do would be to find/build the debug version of
syslogd corresponding to the version you're having trouble with, and
then when it hangs attach to it with GDB and see where it's hung and
what state its variables are in.

You could try ktrace, but that may be a bit too fine-grained.

You could also just try running syslogd in the foreground ("-f"),
possibly along with "-d" to enable verbose debugging, and see what it
says just before it stops logging.

However I note that from a very high-level naive glance at the code that
the logerror() function could fail to ever report fatal errors due to
its attempt to avoid doing anything if it seems to have been called
recursively. This might be mitigated a wee bit for foreground debugging
if the following change were applied such that debugging messages can be
printed before logmsg() is attempted, which is the culprit which might
recursively call logerror() again:

Index: usr.sbin/syslogd/syslogd.c
===================================================================
RCS file: /cvs/master/m-NetBSD/main/src/usr.sbin/syslogd/syslogd.c,v
retrieving revision 1.84
diff -u -r1.84 syslogd.c
--- usr.sbin/syslogd/syslogd.c 13 Nov 2006 20:24:00 -0000 1.84
+++ usr.sbin/syslogd/syslogd.c 25 Nov 2009 20:57:41 -0000
@@ -1586,12 +1586,12 @@
else
(void)snprintf(buf, sizeof(buf), "syslogd: %s", tmpbuf);

- if (daemonized)
- logmsg(LOG_SYSLOG|LOG_ERR, buf, LocalHostName, ADDDATE);
if (!daemonized && Debug)
dprintf("%s\n", buf);
if (!daemonized && !Debug)
printf("%s\n", buf);
+ if (daemonized)
+ logmsg(LOG_SYSLOG|LOG_ERR, buf, LocalHostName, ADDDATE);

logerror_running = 0;
}

--
Greg A. Woods
Planix, Inc.

<woods@...> +1 416 218 0099 http://www.planix.com/

attachment0 (193 bytes) Download Attachment

From forum: netbsd-users

Re: Wild CPU usage times on NetBSD 5

2009-11-25T12:59:44Z

Date: Wed, 25 Nov 2009 10:12:54 -0800
From: buhrow@... (Brian Buhrow)
Message-ID: <200911251812.nAPICstK012464@...>

| Hello. In reading this message, it looks to me like you're getting
| stuck in the tstile bug, or, at least, something similar.

Thanks for the comment - possible - I'd seen the messages about that,
but mostly only superficially followed the discussion (on a "not relevant
here and nothing I can do about it" basis).

But ...

| The commonality
| between all your applications is the kernel, but it's also the filesystem.

Yes, maybe - they're all certainly accessing files on the same FFS.

| You didn't say, but I imagine, that the directories you have Magic Point
| look in,

Magicpoint could have, perhaps, yes - the directory containing its
image files tends to be quite big (ends up with lots of variant formats
of quite a lot of image files (that is originally diagrams from xfig in
most cases) for various reasons). The directory with the file magicpoint
is reading is nothing special (maybe a hundred files or so), but there's
no reason it would be touching that directory at all, just a read of a KB
or so from a file it already had open from there (the page source - it
contains references to the images as separate files, but everything else
is in the one file). I must try building an embedded magicpoint file (which
builds the images into the file itself) and see if that one ever encounters
the same kind of problem.

The problem with this theory is that I suspect that when I have seen this
happen, magicpoint has been (sometimes) accessing pages with no images
at all, which should give it no reason to be accessing the filesystem
any more than reading from a single open file (and not all that much).

| as well as Spam Assassin contain a large number of files,

That one, no, not many files in any directory spamassassin touches while
processing incoming mail - my mail directories contain huge numbers of
files, so if rcvstore, or procmail, both of which write in those were
having problems, that would make more sense, but spamassassin is just
reading a file in /tmp or, perhaps even via stdin - the procmailrc file
says "| /usr/pkg/bin/spamassassin -L", but I don't know for sure how
procmail implements that - its database directory contains just a half
dozen files (big ones, db format I think, most of them, but not many of them).

| just as your Internet junk directory does.

Only maybe my netbsd mail directory ("folder" if you like) contains anywhere
near that number of files... and while procmail & rcvstore would go there,
spamassassin certainly never would (it never even looks at the messages
that end up there, nor in other largish, but not that large, list
directories).

| Are you running softdeps or filesystem logging?

No, the filesystem (my home filesys) is a 100% default FFSv1.
It is about 30GB, and currently close to 90% full. It is a 16K/2K
filesys block size, and I probably changed the default file size to
get a more rational number of inodes than you'd normally get by default
(way too many for me on this filesys).

Looking at it now, dumpfs says "sblock FFSv2 fslevel 4" which
doesn't make much sense to me, I doubt I would have deliberately made
a FFSv2 filesys? On the other hand it also says "magic 11954 (UFS1)"
so now I'm just confused, I need to go do some more looking in the
filesys code - it's been a long time since I was near that stuff!

| This really sounds like a filesystem issue.

What's confusing about that is that I'd expect either just "missing" time,
(that is, real time much longer than measured cpu time, with no real
explanation what was happening while we're waiting) or high system time for
a filesystem related problem - what I'm seeing is a 100% busy CPU, all
recorded as being user time. That doesn't seem to fit - but as I don't
know what the tstile bug symptoms really are (I didn't pay that much
attention, I'm afraid, I'll go hunt in list archives now) I'm not sure.

Is there some way to check (while it is happening) whether this is what
it might be? (It hasn't happened to me again since the spamassassin
slowdown around 17:50-16:00 yesterday afternoon (Wednesday, it is early
Thurs here already) but I'm sure it will again, I see this quite a lot...)
If it happens during mail processing, of my index generation (that one
is rarer, as I don't rebuild the index all that often) I could do almost
anything to look (though my kernel has no diagnostic aids built in, no
DIAGNOSITC or DEBUG, and no ddb).

I can always build a new kernel though if that will help (I just switched,
at least for now, to a 5.0_STABLE as of 2009-11-26) - that's what I'm
running now (but only for the past few hours, certainly not long enough
to mean anything from not having seen the slowdown in this period).

| I realize you're not
| getting any i/o throughput, but I think I've seen similar things under
| NetBSD-5, and they just clear up without apparent explanation as well.

Yes, that certainly fits, there certainly is not no I/O, just I/O isn't
anything I'd think remarkable (nothing even close to what you'd see
unpacking a tar file, or probably even doing a routine compile).
During spamassassin, while it is in its CPU loop, systat can tell me
I/O rates like 1900 bytes/sec (yes, bytes, not KB or blocks or something),
though it will ocasionally peak at around 10-20 KB/sec for a very short
time (probably when the database is being searched)

Thanks for the clue, it is certainly something else to consider, I never
even considered looking in that direction.

kre

From forum: tech-kern

Re: Hair pinning with pf and NetBSD

2009-11-25T12:52:41Z

On Wed, Nov 25, 2009 at 02:39:56PM -0500, Brian A. Seklecki (CFI NOC) wrote:

> Joerg Sonnenberger wrote:
> >On Wed, Nov 25, 2009 at 10:03:17AM -0500, Brian A. Seklecki wrote:
> >>See attached diagrams. I think your customer is very very confused
> >>about TCP/IP, but anyway.
> >
> >Yeah. I think the easiest configuration is actually to just add the
> >public address of each machine as alias with /32 and use normal routing.
>
> 1) Create a lo1 or alias on lo0 inteface with a /32 of the public
> IP address
> 2) Create a static route on each machine, for the public /32, via the
> private /32 (in the /24 or whatever)

Actually easier, just create the route on the former NAT box. For the
clients in the private network, it will send a redirect ICMP message, so
they will learn the route automatically.

Joerg

From forum: current-users

Re: pkg/42373: misc/rpm would not build on Solaris 8: __db185_open

2009-11-25T12:45:02Z

The following reply was made to PR pkg/42373; it has been noted by GNATS.

From: Aleksey Cheusov <cheusov@...>
To: gnats-bugs@...
Cc: pkg-manager@..., gnats-admin@..., pkgsrc-bugs@...
Subject: Re: pkg/42373: misc/rpm would not build on Solaris 8: __db185_open
Date: Wed, 25 Nov 2009 22:41:07 +0200

> gmake[1]: Nothing to be done for `all'.
> gmake[1]: Leaving directory `/tmp/misc/rpm/work/rpm-2.5.4/build'
> gmake[1]: Entering directory `/tmp/misc/rpm/work/rpm-2.5.4/tools'
> cc -L/usr/pkg/lib -Wl,-R/usr/pkg/lib -L/usr/lib -Wl,-R/usr/lib -L/tmp/misc/rpm/work/rpm-2.5.4/lib -L/tmp/misc/rpm/work/rpm-2.5.4/build -L/tmp/misc/rpm/work/rpm-2.5.4/misc dumpdb.o ../lib/librpm.a -lintl -liconv -lnsl -lsocket -lz -lmisc -o dumpdb
> Undefined first referenced
> symbol in file
> __db185_open ../lib/librpm.a(dbindex.o)
> ld: fatal: Symbol referencing errors. No output written to dumpdb
> gmake[1]: *** [dumpdb] Error 1
> gmake[1]: Leaving directory `/tmp/misc/rpm/work/rpm-2.5.4/tools'
> gmake: *** [make-subdirs] Error 1
> *** Error code 2

The issue looks the same as pkg/37293

Does the following patch work for you?

Index: Makefile
===================================================================
RCS file: /cvsroot/pkgsrc/misc/rpm/Makefile,v
retrieving revision 1.68
diff -u -r1.68 Makefile
--- Makefile 22 Sep 2009 13:16:00 -0000 1.68
+++ Makefile 24 Oct 2009 11:07:06 -0000
@@ -26,6 +26,7 @@

# The rpm databases are kept under ${RPM_DB_PREFIX}/lib/rpm.
CONFIGURE_ENV+= varprefix=${RPM_DB_PREFIX:Q}
+LIBS+= ${BDB_LIBS:Q}

MAKE_FLAGS+= MANDIR=${PREFIX}/${PKGMANDIR}/man8
MAKE_FLAGS+= ROOT=${DESTDIR}

--
Best regards, Aleksey Cheusov.

From forum: pkgsrc-bugs

Re: pkg/42373: misc/rpm would not build on Solaris 8: __db185_open

2009-11-25T12:41:07Z

> gmake[1]: Nothing to be done for `all'.
> gmake[1]: Leaving directory `/tmp/misc/rpm/work/rpm-2.5.4/build'
> gmake[1]: Entering directory `/tmp/misc/rpm/work/rpm-2.5.4/tools'
> cc -L/usr/pkg/lib -Wl,-R/usr/pkg/lib -L/usr/lib -Wl,-R/usr/lib -L/tmp/misc/rpm/work/rpm-2.5.4/lib -L/tmp/misc/rpm/work/rpm-2.5.4/build -L/tmp/misc/rpm/work/rpm-2.5.4/misc dumpdb.o ../lib/librpm.a -lintl -liconv -lnsl -lsocket -lz -lmisc -o dumpdb
> Undefined first referenced
> symbol in file
> __db185_open ../lib/librpm.a(dbindex.o)
> ld: fatal: Symbol referencing errors. No output written to dumpdb
> gmake[1]: *** [dumpdb] Error 1
> gmake[1]: Leaving directory `/tmp/misc/rpm/work/rpm-2.5.4/tools'
> gmake: *** [make-subdirs] Error 1
> *** Error code 2

The issue looks the same as pkg/37293

Does the following patch work for you?

Index: Makefile
===================================================================
RCS file: /cvsroot/pkgsrc/misc/rpm/Makefile,v
retrieving revision 1.68
diff -u -r1.68 Makefile
--- Makefile 22 Sep 2009 13:16:00 -0000 1.68
+++ Makefile 24 Oct 2009 11:07:06 -0000
@@ -26,6 +26,7 @@

# The rpm databases are kept under ${RPM_DB_PREFIX}/lib/rpm.
CONFIGURE_ENV+= varprefix=${RPM_DB_PREFIX:Q}
+LIBS+= ${BDB_LIBS:Q}

MAKE_FLAGS+= MANDIR=${PREFIX}/${PKGMANDIR}/man8
MAKE_FLAGS+= ROOT=${DESTDIR}

--
Best regards, Aleksey Cheusov.

From forum: pkgsrc-bugs

Re: Linux NFS client, NetBSD server, locking problems?

2009-11-25T12:34:14Z

Matthias Scheler wrote:

> On Fri, Nov 20, 2009 at 09:13:29PM +0200, Teemu Rinta-aho wrote:
>>> Firefox works on Ubuntu 9.10 with an NFS home on a fellow workers desktop.
>> And (s)he uses locking?
>
> Yes, Linux does that by default.
>
>> Like this:
>>
>> server:/export/home /home nfs defaults 0 0
>
> Can you please try "defaults,tcp" instead of "defaults"?
>
> Kind regards
>

Done. Same old story. It worked for some time, then it stalled, had
to force quite, and when I try to restart it, it won't start.

These are the last lines from "strace firefox":

stat("/home/teemu/.mozilla/firefox/ppr8844q.default",
{st_mode=S_IFDIR|0700, st_size=2048, ...}) = 0
open("/home/teemu/.mozilla/firefox/ppr8844q.default/.parentlock",
O_WRONLY|O_CREAT|O_TRUNC, 0666) = 15
fcntl(15, F_GETLK

Note that fcntl is intentionally cut after F_GETLK, it is not
a copy-paste problem.

BR,
Teemu

From forum: current-users

Re: pkg/42378: build of fonts/t1lib with PKG_OPTION.t1lib=-x11 broken

2009-11-25T12:14:18Z

Wolfgang Stukenbrock <Wolfgang.Stukenbrock@...> writes:

> You need to split the PLIST file in order to separate the x11 parts to
> avoid warnings during package installation

Since there're few files involved, try to avoid creating separate file.

--
HE CE3OH...

From forum: pkgsrc-bugs

NetBSD Nightly Trouble Ticket Report

2009-11-25T12:01:04Z

There are 4896 non-confidential bugs in 50 categories.

Category critical serious non-crit TOTAL Median TTC
bin 18 222 339 579 27d 00:02:02
install 20 43 48 111 7m 23d 08:30:43
kern 247 869 518 1634 2m 8d 10:49:03
lib 4 72 93 169 19d 14:28:27
misc 5 38 134 177 9d 21:44:38
pkg 106 610 695 1411 11d 12:09:22
port-acorn26 0 2 2 4 12:28:42
port-acorn32 1 3 3 7 1y 4m 28d 05:06:42
port-alpha 18 36 18 72 5m 22d 13:02:30
port-amd64 6 11 7 24 3m 5d 21:57:28
port-amiga 2 7 8 17 1m 4d 23:54:21
port-arc 0 1 0 1 3y 4m 25d 19:05:35
port-arm 1 5 2 8 3m 6d 22:45:01
port-atari 2 2 2 6 15d 02:38:37
port-cats 0 2 0 2 3m 23d 06:15:06
port-evbarm 1 1 4 6 8m 9d 16:19:15
port-evbmips 0 3 3 6 2m 24d 18:32:41
port-evbppc 3 0 0 3 4m 19d 14:47:50
port-hp300 1 0 1 2 18d 05:04:38
port-hp700 1 1 0 2 7m 6d 11:44:41
port-hpcarm 3 1 2 6 4m 28d 01:42:18
port-hpcmips 5 6 9 20 25d 10:54:43
port-hpcsh 1 1 2 4 4m 1d 17:29:07
port-i386 35 83 68 186 6m 9d 02:46:17
port-ia64 0 0 1 1 n/a
port-ibmnws 1 0 0 1 n/a
port-iyonix 1 0 0 1 n/a
port-m68k 3 2 0 5 2m 3d 11:28:15
port-mac68k 4 14 7 25 21d 06:25:09
port-macppc 10 24 12 46 7m 24d 07:25:16
port-mips 1 2 1 4 6m 16d 10:47:33
port-mipsco 1 1 0 2 1y 11m 4d 00:07:00
port-newsmips 1 0 0 1 29d 02:42:14
port-next68k 0 0 3 3 1m 26d 17:28:26
port-playstation2 0 2 1 3 02:33:40
port-pmax 0 1 0 1 1m 1d 18:05:29
port-powerpc 1 5 3 9 3m 10d 13:40:34
port-prep 1 0 0 1 1y 3m 27d 03:56:26
port-sgimips 1 3 3 7 3m 28d 19:02:49
port-shark 0 3 2 5 3m 5d 09:45:56
port-sparc 16 21 11 48 2m 18d 13:56:45
port-sparc64 5 14 2 21 2m 7d 01:38:43
port-sun2 0 0 1 1 16d 22:09:29
port-sun3 0 1 3 4 1m 22d 03:04:54
port-vax 3 8 6 17 5m 24d 09:08:34
port-xen 2 6 4 12 24d 02:53:33
security 1 2 7 10 1m 23d 00:00:26
standards 1 10 28 39 1m 19d 15:22:20
toolchain 15 46 44 105 3m 10d 08:51:38
xsrc 7 32 28 67 12d 11:39:32
TOTAL 555 2216 2125 4896

From forum: netbsd-bugs

Re: Hair pinning with pf and NetBSD

2009-11-25T11:39:56Z

Joerg Sonnenberger wrote:
> On Wed, Nov 25, 2009 at 10:03:17AM -0500, Brian A. Seklecki wrote:
>> See attached diagrams. I think your customer is very very confused
>> about TCP/IP, but anyway.
>
> Yeah. I think the easiest configuration is actually to just add the

1) Create a lo1 or alias on lo0 inteface with a /32 of the public
IP address
2) Create a static route on each machine, for the public /32, via the
private /32 (in the /24 or whatever)

e.g.:

machine1$ sudo ifconfig lo1 create
machine1$ sudo ifconfig lo1 inet 129.144.100.100 netmask 255.255.255.255 up

machine2$ sudo route add -host 129.144.100.100 192.168.100.100

And vice-versa.

~BAS

> public address of each machine as alias with /32 and use normal routing.
>
> Joerg

From forum: current-users

Re: ACPI call for testing

2009-11-25T11:18:09Z

Hello. I find this patch seems to make no difference on any of the
machines I've tested it on. If a machine worked before the patch, it
worked after the patch. If a machine didn't work before the patch, it
still doesn't work after the patch is applied.
Here are some example acpi outputs for machines that give trouble:

Sun Microsystems Sun Fire X2200 M2 (Rev 50)
acpi0: X/RSDT: OemId <A M I ,OEMXSDT ,07000704>, AslId <MSFT,00000097>
With this machine, all but one revision of the ACPI BIOs hang with
NetBSD when ACPI is enabled and when multiple CPUs are in use.

Dell Computer Corporation OptiPlex GX400
This machine works fine with acpi enabled, except that the cd and dvd
players, which are configured as master and slave on the same ata bus,don't
work when acpi is enabled. When acpi is enabled, the cd shows up, but
takes a lot of lost interrupt errors and doesn't actually function. The
dvd player doesn't show up at all. when acpi is disabled, they show up as
separate working devices. This has been true since NetBSD-3.

-thanks
-Brian

From forum: current-users

Re: xsrc/41951: amd64 + radeondrm crash

2009-11-25T11:15:56Z

This no longer happens on -current. (It ends up hard locking somewhat
often when running non-trivial GL apps however.)

From forum: netbsd-bugs

Re: netbsd5-i386 is broken for me after this weekend's changes

2009-11-25T11:07:02Z

On Wed, 25 Nov 2009, Manuel Bouyer wrote:

> On Tue, Nov 24, 2009 at 04:58:55PM -0800, Hisashi T Fujinaka wrote:
>>> Can you enter ddb when this happens ?
>>> some combination of ps/l, tr/a and show all vnodes could point to
>>> the problem.
>>> What is the filesystem configuration ?
>>
>> I took some screen shots and they're in http://www.i8u.org/~htodd/logpix.tbz
>
> thanks. I think I'll have to try to reproduce it here to be able to debug
> it.
> I guess you have a null or union mount in this setup ? Can you give
> the details of your filesystem configuration ?

Does this help? I also filed a pr, just in case, kern/42377.

Thanks for your help on this.

htodd@kerry:/usr/src > mount
/dev/wd0a on / type ffs (NFS exported, local)
/dev/wd0f on /var type ffs (local)
/dev/wd0e on /usr type ffs (local)
/dev/wd0g on /home type ffs (NFS exported, local)
tmpfs on /tmp type tmpfs (local)
kernfs on /kern type kernfs (local)
ptyfs on /dev/pts type ptyfs (local)
/home/netbsd/netbsd-5/src on /usr/src type null (local)
/home/netbsd/netbsd-5/xsrc on /usr/xsrc type null (local)
/home/netbsd/pkgsrc/pkgsrc-2009Q3 on /usr/pkgsrc type null (local)

htodd@kerry:/usr/src > disklabel wd0
# /dev/rwd0d:
type: unknown
disk: NetBSD
label:
flags:
bytes/sector: 512
sectors/track: 63
tracks/cylinder: 16
sectors/cylinder: 1008
cylinders: 484521
total sectors: 488397168
rpm: 3600
interleave: 1
trackskew: 0
cylinderskew: 0
headswitch: 0 # microseconds
track-to-track seek: 0 # microseconds
drivedata: 0

16 partitions:
# size offset fstype [fsize bsize cpg/sgs]
a: 1024128 63 4.2BSD 1024 8192 0 # (Cyl. 0*- 1016*)
b: 8192016 1024191 swap # (Cyl. 1016*- 9143*)
c: 488397105 63 unused 0 0 # (Cyl. 0*- 484520)
d: 488397168 0 unused 0 0 # (Cyl. 0 - 484520)
e: 12288528 9216207 4.2BSD 2048 16384 0 # (Cyl. 9143*- 21334*)
f: 6144768 21504735 4.2BSD 2048 16384 0 # (Cyl. 21334*- 27430*)
g: 460747665 27649503 4.2BSD 2048 16384 0 # (Cyl. 27430*- 484520)

--
Hisashi T Fujinaka - htodd@...
BSEE(6/86) + BSChem(3/95) + BAEnglish(8/95) + MSCS(8/03) + $2.50 = latte

From forum: current-users

Re: pkg/42378: build of fonts/t1lib with PKG_OPTION.t1lib=-x11 broken

2009-11-25T11:05:03Z

The following reply was made to PR pkg/42378; it has been noted by GNATS.

From: Joerg Sonnenberger <joerg@...>
To: gnats-bugs@...
Cc:
Subject: Re: pkg/42378: build of fonts/t1lib with PKG_OPTION.t1lib=-x11
broken
Date: Wed, 25 Nov 2009 20:03:25 +0100

On Wed, Nov 25, 2009 at 06:10:04PM +0000, Wolfgang Stukenbrock wrote:
> You need to split the PLIST file in order to separate the x11 parts to
> avoid warnings during package installation

Care to use PLIST_VARS instead?

Joerg

From forum: pkgsrc-bugs

Re: netbsd5-i386 is broken for me after this weekend's changes

2009-11-25T10:59:28Z

On Tue, Nov 24, 2009 at 04:58:55PM -0800, Hisashi T Fujinaka wrote:
> >Can you enter ddb when this happens ?
> >some combination of ps/l, tr/a and show all vnodes could point to
> >the problem.
> >What is the filesystem configuration ?
>
> I took some screen shots and they're in http://www.i8u.org/~htodd/logpix.tbz

thanks. I think I'll have to try to reproduce it here to be able to debug
it.
I guess you have a null or union mount in this setup ? Can you give
the details of your filesystem configuration ?

--
Manuel Bouyer <bouyer@...>
NetBSD: 26 ans d'experience feront toujours la difference
--

From forum: current-users

qemu segmentation fault

2009-11-25T10:46:36Z

Hi!

I try to start qemu with the tap option:
qemu -m 256 -net nic -net tap debian.img

But then a segmentation fault occurs (the tap device was previously
created with "ifconfig tap0 create"). I use NetBSD 5.0_STABLE / amd64.

Any ideas?

--
Victor Dorneanu

Contact
- Web/Blog: http://dornea.nu

GnuPGP information
- KeyID = 0xD20870F4 (subkeys.pgp.net)
- Key fingerprint = DD6B 5E09 242F 7410 3F90 492A 4CBA FD13 D208 70F4

From forum: netbsd-users

Re: sysinst split project - The Configuration File

2009-11-25T10:17:27Z

On Tue, 24 Nov 2009 12:18:23 -0200
Silas Silva <silasdb@...> wrote:

> I posted it here: http://pastebin.com/f5ec7a9f
>
> Please, take a look and make suggestions.

I would put the "disklabel" (and "partitions") inside a "disks" or
"devices" entry. I think that will make it easier to implement your 2nd
TODO item (install on more then 1 disk) as well as other interesting
features in the future like installing on raidframe sets, CGDs, etc.

For example something like this:

devices {
disk1 {
disklabel {
a {
FStype: FFSv2
[...]
mount point: /
[...]
}
b {
FStype: swap
start: a
[...]
}
e {
FStype: raid_component
RAIDset: raid_db
}
[...]
}
}
disk2 {
[...]
}
raid_db {
level: 1
auto_config: yes
}
}

I don't know how you'll parse the configuration file or how its
contents will be internally represented; it may be possible to drop the
outer "devices" from this.

For the root password, you could handle the value as already encrypted,
like useradd does.

kind regards
dieter

From forum: tech-install

Re: sysinst split project - The Configuration File

2009-11-25T10:17:27Z

From forum: current-users

Re: Hair pinning with pf and NetBSD

2009-11-25T10:15:27Z

On Wed, Nov 25, 2009 at 10:03:17AM -0500, Brian A. Seklecki wrote:
> See attached diagrams. I think your customer is very very confused
> about TCP/IP, but anyway.

Yeah. I think the easiest configuration is actually to just add the
public address of each machine as alias with /32 and use normal routing.

Joerg

From forum: current-users

Re: Wild CPU usage times on NetBSD 5

2009-11-25T10:12:54Z

Hello. In reading this message, it looks to me like you're getting
stuck in the tstile bug, or, at least, something similar. The commonality
between all your applications is the kernel, but it's also the filesystem.
You didn't say, but I imagine, that the directories you have Magic Point
look in, as well as Spam Assassin contain a large number of files, just as
your Internet junk directory does. Are you running softdeps or filesystem
logging? This really sounds like a filesystem issue. I realize you're not
getting any i/o throughput, but I think I've seen similar things under
NetBSD-5, and they just clear up without apparent explanation as well.
Unfortunately, I'm not sure how to test for such misbehaviors.

-Brian
On Nov 25, 8:58pm, Robert Elz wrote:
} Subject: Wild CPU usage times on NetBSD 5
} While I run NetBSD 4 (or even earlier) most places, and everything
} I compile for pkgsrc is compiled for NetBSD 4, on my laptop I have a
} NetBSD 5 (5.0.1 actually) kernel installed (and 4.99.38 userland -
} it was the current version of -current when I acquired it, in (I
} think it was) December 2007).
}
} Everything works correctly (well, except the azalia (or hdaudio),
} which I suspect probably needs the magic Dell GPIO pin hack,
} which I don't have - but also don't care about much).
} (The laptop is a Dell Precision M4300 - or called XPS M4300 sometimes).
}
} However, predicting CPU usage for any of several of my fairly
} common tasks is wild - sometimes things go quick, and sometimes
} they're slow. This isn't EST related, while changing the est
} frequency alters the times (obviously, if the CPU is running
} faster, things happen quicker) but the maximum variation there is
} about a factor of 2
}
} I first noticed this with spamassassin (I filter my mail locally
} as I process it, not as it arrives at my mail server - why is not
} relevant here). Sometimes it takes a second or so (give or take)
} to process each message, other times it takes 20 seconds (this is
} basically just bayyesian filtering, and the built in rules, all
} network based checks are disabled - I sometimes fetch mail online, then
} handle the filtering, generating replies, ... without a network
} connection, then connect again to allow replies to be transmitted.)
} I'm also using the spamassassin script (perl code) not spamd - that's
} just laziness on my part (I set it up this way, and can't be bothered
} changing it). The problem isn't confined to spamassassin or I would
} alter it, just to see if it makes a difference.
}
} At first glance you'd think the variation in processing time would relate
} to the actual messages being filtered, but it doesn't. During some
} periods (and this is unpredictable) every message gets processed in a
} small number of seconds (sure they vary, but usually from < 1 sec to about 2
} or 3 secs). When things are slow the times also vary, but from
} more like 10 secs to 20 secs or so (per message). Every message,
} big or small, gets either fast or slow processing - depending on whether
} the system is in its "go slow" mode or not.
}
} This is a dual core CPU, usually when spamassassin is working, nothing
} else significant is happening (the other CPU is just idling, doing
} nothing at all.) That's fine, I'm not asking to spread the load...
}
} That was OK, I could live with that (and have done for ages now), then
} I noticed another process with similar bi-polar behaviour.
}
} I have a BIG directory into which I save stuff I download from the
} internet (I should be more intelligent, and impose some hierarchy,
} but again I'm lazy, I would if starting again, but adding hierarchy now
} would mean going and moving many thousands of files, by hand...)
}
} Of course, I don't expect operations on that directory to be very quick,
} and that's OK, it doesn't get used all that much (and when it does, it
} is fast enough - usually).
}
} To manage it, I have a script that reads it, and builds a private
} HTML page that is an index (actually several indexes, including one
} of the recently added/changed files, which are the ones I'm most likely
} to want, another by file type (pdf, html, ...) - it also converts the
} file names from my own private encoding of arbitrary (reasonable) names
} into "unix safe" file names - that is, for example, the item I want to
} see in the index might be
}
} Fred's rebuttal of "cat -v considered harmful" (version 1)
}
} (no, there isn't such a thing, that I know of...) but in the filesystem,
} while that filename would work, and the browser could handle it OK,
} when I want to copy the file using regular sh utilities, I'd need
} something like apb's quote command (or my more primitive sh function
} that's similar) and use it - which makes life much more difficult, so
} instead, the file name is (that is, it would be, if it existed)
}
} Fred,,s_rebuttal_of_,oQcat_-v_considered_harmful,cQ_,opversion_1,cp.pdf
}
} which is a string I can cut/paste into any sh command to copy the
} file, or whatever - my script converts that into the string I want
} to see in my index (as above - aside from the char changes, also note
} the .pdf is removed). That's just for background, the conversion is
} done mostly in sh funcs, using $(var%str} type operations, with "echo"
} to write to the output file - ie: essentially everything is built into
} the shell, there's very little use of external commands (ls to get
} the file names, and sort to order things when needed, but neither of
} those matters much here). There are tens of thousands of files to
} be manipulated (I've been collecting for years...) so naturally, all
} this takes a while.
}
} OK, it's my problem, I'm the idiot with the huge directory... The issue
} for here is that while the time it takes to generate the index should just
} grow slowly (as new stuff gets added and so the index gets slightly
} bigger) that isn't what actually happens - the time oscillates wildly
} between slow, but acceptable, and glacial (say 5 minutes or so, and about
} an hour) - this looks to be just about the same kind of slowdown that
} spamassassin gets (somewhere around a factor of 10).
}
} Both of those tasks essentially run in the background, and each occupies
} just one CPU (this is not a fault in one core, which CPU is used varies,
} it sometimes even switches during the run of a process) so I can more
} or less just ignore the problem, and do other stuff while those are
} processing.
}
} But I also see the same problem running magicpoint (which I use when I'm
} teaching) - most of the time switching from one page to the next goes
} just as you'd expect (essentially undetectable delay - I have all the
} images, when there are any, pre-cached, so magicpoint doesn't need to
} go running prcoesses for scaling etc - when that happens, things do
} slow down, but that's not the issue here.) I am seeing the same behaviour
} with magicpoint as with spamassassin and my indexing sh script, sometimes
} it takes 15-20 seconds to change to a new page (including when the new
} page is just text). Now this is an X application, so it is interacting
} with the X server (the other two aren't), but the X server itself seems
} still highly responsive while this is happening (I can create new
} windows to try and monitor what's going on ... while waiting for my
} new magicpoint page to appear, which all works just fine).
}
} Now I get to explain why all this is being sent to tech-kern - the only
} thing that I can see in common between these 3 applications is the kernel,
} and the hardware, one is perl, one is sh (it is a NetBSD /bin/sh
} script, not ksh, bash, or anything else), the other is C++ - nothing
} very common there (I guess they all use libc, but so does everything else
} running which isn't being affected, as far as I can tell).
}
} What they have in common is that they each are, at least for a short time,
} fairly intensive CPU users - it smells as if something is detecting a
} high CPU using process, and just making it go slower - somehow.
}
} According to systat, the CPU time is all being spent in user mode, not
} in the kernel (and not niced, though that wouldn't matter, there's no
} competition of note, even a nice -19 process would get as much cpu time
} as it requests). There are no unusual number of interrupts (the
} interrupt rate seems quite low actually), there's very little I/O
} (even during spamassassin, its index files aren't tiny, but it seems
} to be fairly effective in not abusing them), during the magicpoint case
} (usually) there's essentially no competition at all (the network isn't
} usually even configured, I kill dhclient before it gets a chance to
} configure anything - not that it would usually succeed then anyway).
}
} I can't find a particular pattern to when this likely to strike - usually
} soon after a boot all is fast, but it isn't a case of "start fast, then
} get slow" as if something was being consumed (what it would be that would
} affect CPU times I can't guess) - after a period of slowness, for no
} conceivable reason, everything can just revert to "normal" again (with
} no reboots, or anything). Sometimes the system runs ages in "fast" mode,
} other times it can be slow within 15 minutes after booting (I reboot
} before I teach - suspend doesn't work for me - that is, (ACPI) suspend works
} fine, but the X server can never (re-)start after a suspend, it's an
} nv video chip - again, I don't mind that, booting is fast enough - in any
} case, the X server needs to learn the existence of the external VGA plug
} for the projector, and restarting it seems to be the only way to make it
} do that - rebooting is not that much slower than just restarting X.)
}
} Since I can't find anything in common between the applications that I
} can blame (rationally) what's left is the kernel and the hardware (the
} cpu itself). The hardware is certainly possible, but it is kind of
} hard to tell how the hardware can tell what process is running, and slow
} down just that process, while allowing others to continue seemingly unchanged.
} I guess it is possible that it might detect a period of intense activity
} without any context switches, and deliberately decide to go slow, but
} that's kind of stretching rationality. Other processes seem unaffected,
} and which CPU (of the two cores in the one chip) is unrelated to the
} problem - once it is in "slow" mode, both CPU's run those processes slowly
} (even if the cpu in use switches while the process is running), while
} other processes seem unaffected (X server, xterms, systat when I decide to try
} and work out what's happening, ...).
}
} If it isn't the CPU, what's left is the kernel - it certainly knows what
} process is running, and how much cpy time it is consuming, and it has
} control of the stats, so while everything tells me the cpu time is all
} user time, if it is the kernel, it could be lying...
}
} Does anyone have any idea what this might be, or even what else I should
} try doing to debug this - I can try a current kernel, but I don't really
} want to run that, the 64 bit time change is mostly invisible, but to process
} accounting it isn't, and with a 32 bit sa, and 32 bit summary files,
} processing 64 bit time acct files just makes a mess ... I have a hack
} in my startup scripts that leave accounting off if they detect a 5.99.*
} kernel is running, so testing is fine, but I actually make use of the
} accounting stuff - in fact I am going to include some "lastcomm" output
} below as an example of what I see (the unpredictable nature of this makes
} it difficult to know whether "ran current for 4 hours and nothing went
} slow" would mean anything or not, if that happened.)
}
} Also, if anyone else sees anything like this, that would be useful to
} know (but unless we start seeing lots of "me too" messages, which I doubt,
} there's no need to tell me that it does not happen to you!)
}
} My current guess might be something along the lines of some internal state
} causing very poor cache utilisation, or something like that, but I don't
} know how to measure that.
}
} Thanks,
}
} kre
}
} If you don't want boring details, stop reading now, the remaining half of
} this message isn't interesting... On the other hand, if there was something
} you needed to know about my setup, what's below might answer that.
}
} For info, this is what "cpuctl identify 0" tells me ...
}
} cpu0: Intel Core 2 (Merom) (686-class), 1995.13 MHz, id 0x6fd
} cpu0: features 0xbfebfbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR>
} cpu0: features 0xbfebfbff<PGE,MCA,CMOV,PAT,PSE36,CFLUSH,DS,ACPI,MMX>
} cpu0: features 0xbfebfbff<FXSR,SSE,SSE2,SS,HTT,TM,SBF>
} cpu0: features2 0xe3bd<SSE3,DTES64,MONITOR,DS-CPL,VMX,EST,TM2,SSSE3,CX16,xTPR,PDCM>
} cpu0: features3 0x20100000<XD,EM64T>
} cpu0: "Intel(R) Core(TM)2 Duo CPU T7250 @ 2.00GHz"
} cpu0: I-cache 32KB 64B/line 8-way, D-cache 32KB 64B/line 8-way
} cpu0: L2 cache 2MB 64B/line 8-way
} cpu0: ITLB 128 4KB entries 4-way
} cpu0: DTLB 256 4KB entries 4-way, 32 4MB entries 4-way
} cpu0: Initial APIC ID 0
} cpu0: Cluster/Package ID 0
} cpu0: Core ID 0
} cpu0: family 06 model 0f extfamily 00 extmodel 00
}
} (cpu 1 is just the same, of course). I run it as i386, not amd64.
}
} machdep.est.frequency.target = 1600
} machdep.est.frequency.current = 1600
} machdep.est.frequency.available = 2200 2000 1800 1600 1400 1200
}
} I run at 1600 (set in sysctl.conf), rather than 2000, as this is a fairly
} warm environment, and at high speeds, things seem to get quite hot...
}
} The "2200" in the "available" output looks like a bug, attempting to
} set that as "target" works, but "current "just goes to 2000 (this is a
} 2GHz processor, not 2.2GHz). I booted a FreeBSD live CD, just for fun
} one day, and its equivalent sysctl didn't mention 2200, it told me,
} correctly I think, that 2000 was the fastest available speed. But
} that's a very minor bug...
}
} uanme -a result ...
}
} NetBSD epsilon.noi.kre.to 5.0.1 NetBSD 5.0.1 (EPSILON-1.2-20090731) #8: Sat Aug 1 02:22:55 ICT 2009 kre@...:/usr/obj/5/kernels/EPSILON i386
}
} Last, here's a lastcomm | grep perl, with some blocks of similar
} lines edited out to make the result smaller, and with some annotations
} added to explain what I see. perl run on ttyp2 is spamassassin run
} in my office (at work I fetch mail manually when I have time to think
} about it), perl on no tty at or just after 20 mins past an hour is
} mail processing - spamassassin - run from cron, that happens only when
} I'm home (the script tests, and just exits otherwise) which fetches
} mail when (mostly) the laptop is just sitting around doing nothing much
} else... Any other perl is most likely unrelated.
}
} perl - kre ttyp2 1.67 secs Wed Nov 25 18:54 (0:00:11.81)
} perl - kre ttyp2 1.61 secs Wed Nov 25 18:54 (0:00:11.67)
} perl - kre ttyp2 1.59 secs Wed Nov 25 18:54 (0:00:11.70)
} [ lines elided here, nothing different, this is the most recent
} incoming mail processing I did, as of when I created this output,
} all fast ... ]
} perl - kre ttyp2 1.58 secs Wed Nov 25 18:50 (0:00:11.67)
} perl - kre ttyp2 1.59 secs Wed Nov 25 18:49 (0:00:11.67)
} perl - kre ttyp2 1.64 secs Wed Nov 25 18:49 (0:00:12.08)
} perl - kre ttyp2 12.83 secs Wed Nov 25 18:06 (0:00:23.11)
} perl - kre ttyp2 12.92 secs Wed Nov 25 18:06 (0:00:23.12)
} perl - kre ttyp2 12.86 secs Wed Nov 25 18:05 (0:00:23.17)
} [ more lines deleted, this lot, around 18:00, was the previous
} time I fetched mail ]
} perl - kre ttyp2 13.16 secs Wed Nov 25 18:01 (0:00:23.31)
} perl - kre ttyp2 26.19 secs Wed Nov 25 18:00 (0:00:36.44)
} perl - kre ttyp2 13.28 secs Wed Nov 25 18:00 (0:00:23.42)
} perl - kre ttyp2 13.12 secs Wed Nov 25 17:59 (0:00:23.25)
} perl - kre ttyp2 26.30 secs Wed Nov 25 17:59 (0:00:36.53)
} perl - kre ttyp2 13.33 secs Wed Nov 25 17:58 (0:00:23.42)
} perl - kre ttyp2 14.05 secs Wed Nov 25 17:58 (0:00:24.41)
} perl - kre ttyp2 12.95 secs Wed Nov 25 17:58 (0:00:23.20)
} perl - kre ttyp2 13.06 secs Wed Nov 25 17:57 (0:00:23.19)
} perl - kre ttyp2 12.70 secs Wed Nov 25 17:57 (0:00:22.81)
} perl - kre ttyp2 13.95 secs Wed Nov 25 17:56 (0:00:24.08)
} perl - kre ttyp2 13.03 secs Wed Nov 25 17:56 (0:00:23.14)
} perl - kre ttyp2 12.97 secs Wed Nov 25 17:56 (0:00:23.11)
} perl - kre ttyp2 13.80 secs Wed Nov 25 17:55 (0:00:23.98)
} perl - kre ttyp2 12.94 secs Wed Nov 25 17:55 (0:00:23.09)
} perl - kre ttyp2 10.08 secs Wed Nov 25 17:54 (0:00:20.19)
} perl - kre ttyp2 13.62 secs Wed Nov 25 17:54 (0:00:23.73)
} perl - kre ttyp2 5.61 secs Wed Nov 25 16:52 (0:00:15.83)
} perl - kre ttyp2 4.66 secs Wed Nov 25 16:52 (0:00:12.11)
} perl - kre ttyp2 3.67 secs Wed Nov 25 16:51 (0:00:13.77)
} perl - kre ttyp2 3.42 secs Wed Nov 25 16:51 (0:00:13.55)
} perl - kre ttyp2 4.06 secs Wed Nov 25 16:51 (0:00:13.67)
} [ quite a few lines deleted, this (16:30-16:52) was soon after a reboot,
} after I returned to my office after teaching today, and fetched
} several hours worth of mail - all processed in fast mode ]
} perl - kre ttyp2 1.56 secs Wed Nov 25 16:32 (0:00:11.66)
} perl - kre ttyp2 1.70 secs Wed Nov 25 16:31 (0:00:11.92)
} perl - kre ttyp2 1.66 secs Wed Nov 25 16:31 (0:00:12.23)
} perl - kre ttyp2 1.70 secs Wed Nov 25 16:31 (0:00:11.83)
} perl - kre ttyp2 1.61 secs Wed Nov 25 16:31 (0:00:11.69)
} perl - kre ttyp2 13.33 secs Wed Nov 25 16:30 (0:00:25.41)
} [ No idea why that singleton at 13 secs, whether that was "slow mode"
} or just one weird message spamassassin had problems with, that one,
} the very first message processed in this block, is odd - but I'd
} treat it as an outlier, and ignore it...
} Between 13:00 and 14:30 and 15:00 and 16:00 I was teaching,
} the system shutdown and rebooted three times, shutdown, move,
} teach, shutdown, move, teach, shutdown, move to office, and reboot
} magicpoint was (sometimes) very slow - especially late in the
} first class ]
} perl - kre __ 1.66 secs Wed Nov 25 12:20 (0:00:01.69)
} perl - kre __ 1.61 secs Wed Nov 25 12:20 (0:00:01.66)
} perl - kre __ 1.70 secs Wed Nov 25 12:20 (0:00:01.77)
} perl - kre __ 1.23 secs Wed Nov 25 12:20 (0:00:01.25)
} perl - kre __ 1.72 secs Wed Nov 25 12:20 (0:00:01.84)
} perl - kre __ 1.25 secs Wed Nov 25 12:20 (0:00:01.27)
} [lots deleted during this period, this was the laptop at home,
} processing mail an hour's worth at a time - the system had been
} running for about 38 hours or so, it stayed home all Tuesday,
} without rebooting during that period - there would have been
} slow processing periods during that, but none from when the
} process accounting file was regenerated by the nightly script]
} perl - kre __ 1.62 secs Wed Nov 25 10:20 (0:00:01.66)
} perl - kre __ 1.61 secs Wed Nov 25 10:20 (0:00:01.64)
} perl - kre __ 1.61 secs Wed Nov 25 10:20 (0:00:01.61)
} perl - kre __ 1.92 secs Wed Nov 25 10:20 (0:00:01.97)
} [more deleted]
} perl - kre __ 1.72 secs Wed Nov 25 09:20 (0:00:01.77)
} perl - kre __ 1.62 secs Wed Nov 25 09:20 (0:00:01.67)
} perl - kre __ 1.67 secs Wed Nov 25 09:20 (0:00:01.70)
} perl - kre __ 1.66 secs Wed Nov 25 09:20 (0:00:01.72)
} perl - kre __ 2.34 secs Wed Nov 25 09:20 (0:00:02.34)
} perl - kre __ 1.23 secs Wed Nov 25 09:20 (0:00:01.25)
} perl - kre __ 1.56 secs Wed Nov 25 09:20 (0:00:01.58)
} perl - kre __ 1.69 secs Wed Nov 25 09:20 (0:00:01.72)
} perl - kre __ 1.59 secs Wed Nov 25 09:20 (0:00:01.64)
} perl - kre __ 1.61 secs Wed Nov 25 08:21 (0:00:01.66)
} perl - kre __ 1.25 secs Wed Nov 25 08:21 (0:00:01.27)
} perl - kre __ 2.56 secs Wed Nov 25 08:21 (0:00:02.73)
} perl - kre __ 1.83 secs Wed Nov 25 08:21 (0:00:01.92)
} perl - kre __ 2.34 secs Wed Nov 25 08:21 (0:00:02.36)
} perl - kre __ 1.25 secs Wed Nov 25 08:21 (0:00:01.27)
} perl - kre __ 1.64 secs Wed Nov 25 08:21 (0:00:01.66)
} perl - kre __ 1.59 secs Wed Nov 25 08:21 (0:00:01.64)
} perl - kre __ 1.64 secs Wed Nov 25 08:21 (0:00:01.66)
} perl - kre __ 1.62 secs Wed Nov 25 08:21 (0:00:01.67)
} perl - kre __ 1.23 secs Wed Nov 25 08:21 (0:00:01.25)
} [more deleted]
} perl - kre __ 1.59 secs Wed Nov 25 07:20 (0:00:01.61)
} perl - kre __ 1.61 secs Wed Nov 25 07:20 (0:00:01.64)
} perl - kre ttyp5 1.39 secs Wed Nov 25 07:10 (0:00:01.41)
} perl - kre ttyp5 1.39 secs Wed Nov 25 07:08 (0:00:01.42)
} perl - kre ttyp5 1.56 secs Wed Nov 25 07:08 (0:00:01.61)
} [those ttyp5 perls have to be something else, no idea what now,
} ... irrelevant]
} perl - kre __ 2.27 secs Wed Nov 25 06:20 (0:00:03.02)
} perl - kre __ 1.69 secs Wed Nov 25 06:20 (0:00:01.70)
} perl - kre __ 1.59 secs Wed Nov 25 06:20 (0:00:01.62)
} perl - kre __ 1.66 secs Wed Nov 25 06:20 (0:00:01.69)
} perl - kre __ 1.64 secs Wed Nov 25 06:20 (0:00:01.67)
} perl - kre __ 1.59 secs Wed Nov 25 06:20 (0:00:01.64)
} [lots more deleted]
} perl - kre __ 1.58 secs Wed Nov 25 03:22 (0:00:01.61)
} perl - kre __ 1.64 secs Wed Nov 25 03:22 (0:00:01.66)
} perl - kre __ 1.64 secs Wed Nov 25 03:22 (0:00:01.77)
} perl - kre __ 1.61 secs Wed Nov 25 03:22 (0:00:01.62)
} perl - kre __ 1.67 secs Wed Nov 25 03:22 (0:00:02.33)
} [previous mail processing would have been 02:20 approx, in the
} previous day's accounting file]
}
} spamassassin is run via procmail, but neither procmail, nor formail
} (which procmail also runs) nor the nmh rcvstore (which procmail sometimes
} runs) ever seem to run particularly slowly - not that they do very much
} (but some of the messages - in particular netbsd list messages - go into
} a very big mail directory, it currently has 70K messages approx, and
} probably another couple of thousand ,N messages - nmh "deleted" (rmm'd)
} but not yet gone from the filesys - just finding the next message number
} to use, must take some cpu time (which would be rcvstore cpu time.)
}
} Here's a complete ttyp2 process list (from lastcomm) during a part of the
} "slow" period above) ...
}
} procmail - kre ttyp2 0.09 secs Wed Nov 25 17:55 (0:00:24.19)
} procmail -F kre ttyp2 0.00 secs Wed Nov 25 17:55 (0:00:24.03)
} perl - kre ttyp2 13.80 secs Wed Nov 25 17:55 (0:00:23.98)
} formail - kre ttyp2 0.03 secs Wed Nov 25 17:55 (0:00:00.03)
} procmail - kre ttyp2 0.03 secs Wed Nov 25 17:55 (0:00:00.09)
} formail - kre ttyp2 0.03 secs Wed Nov 25 17:55 (0:00:00.03)
} procmail - kre ttyp2 0.06 secs Wed Nov 25 17:55 (0:00:00.47)
} rcvstore - kre ttyp2 0.31 secs Wed Nov 25 17:55 (0:00:00.34)
} formail - kre ttyp2 0.03 secs Wed Nov 25 17:55 (0:00:00.03)
} procmail - kre ttyp2 0.09 secs Wed Nov 25 17:55 (0:00:00.16)
} formail - kre ttyp2 0.03 secs Wed Nov 25 17:55 (0:00:00.03)
} procmail - kre ttyp2 0.09 secs Wed Nov 25 17:55 (0:00:23.30)
} procmail -F kre ttyp2 0.00 secs Wed Nov 25 17:55 (0:00:23.14)
} perl - kre ttyp2 12.94 secs Wed Nov 25 17:55 (0:00:23.09)
} formail - kre ttyp2 0.03 secs Wed Nov 25 17:55 (0:00:00.03)
} procmail - kre ttyp2 0.12 secs Wed Nov 25 17:54 (0:00:44.31)
} rcvstore - kre ttyp2 0.09 secs Wed Nov 25 17:55 (0:00:00.11)
} procmail -F kre ttyp2 0.00 secs Wed Nov 25 17:54 (0:00:20.22)
} perl - kre ttyp2 10.08 secs Wed Nov 25 17:54 (0:00:20.19)
} procmail -F kre ttyp2 0.00 secs Wed Nov 25 17:54 (0:00:23.78)
}
} Note that not all incoming messages are sent to spamassassin, stuff like
} mailing lists (like the netbsd lists) are just filed away, I only check for
} spam on mail that is likely to in fact be spam. Note that only perl seems
} affected by the slowdown - unless it is just that the others use so little time
} that even at 10 times as slow they aren't detectable - that's possible
} here, but if everything was 10 times as slow as "normal" in a slow period,
} I think I'd have noticed more other processes going slow, than the three I do
} observe - it is just those three I am aware of.
}
}
}
>-- End of excerpt from Robert Elz

From forum: tech-kern

Re: pkg/42378: build of fonts/t1lib with PKG_OPTION.t1lib=-x11 broken

2009-11-25T10:10:04Z

The following reply was made to PR pkg/42378; it has been noted by GNATS.

From: Wolfgang Stukenbrock <Wolfgang.Stukenbrock@...>
To: gnats-bugs@...
Cc:
Subject: Re: pkg/42378: build of fonts/t1lib with PKG_OPTION.t1lib=-x11 broken
Date: Wed, 25 Nov 2009 19:08:09 +0100

Hi again - sorry for the incomplete patch for report

You need to split the PLIST file in order to separate the x11 parts to
avoid warnings during package installation

Here the complete changes needed to get a working setup.

changes to PLIST:
--- PLIST 2009/11/25 17:25:24 1.1
+++ PLIST 2009/11/25 17:26:43
@@ -1,10 +1,7 @@
@comment $NetBSD: PLIST,v 1.10 2005/09/07 17:46:28 reed Exp $
bin/type1afm
-bin/xglyph
include/t1lib.h
-include/t1libx.h
lib/libt1.la
-lib/libt1x.la
share/doc/${PKGBASE}/Makefile
share/doc/${PKGBASE}/Makefile.in
share/doc/${PKGBASE}/Tee.eps

new file PLIST.x11:

@comment $Id: PLIST.x11,v 1.1 2009/11/25 17:30:12 wgstuken Exp $
bin/xglyph
include/t1libx.h
lib/libt1x.la

changes for options.mk:

--- options.mk 2009/11/25 17:15:59 1.1
+++ options.mk 2009/11/25 17:28:34
@@ -12,4 +12,8 @@
.include "../../x11/libXp/buildlink3.mk"
.include "../../x11/libXpm/buildlink3.mk"
.include "../../x11/libXt/buildlink3.mk"
+# need to set PLIST, if we need the x11 parts too
+PLIST_SRC= ${PKGDIR}/PLIST ${PKGDIR}/PLIST.x11
+.else
+CONFIGURE_ARGS+= --without-x
.endif

gnats-admin@... wrote:

> Note: There was a bad value `' for the field `Class'.
> It was set to the default value of `sw-bug'.
>
> Thank you very much for your problem report.
> It has the internal identification `pkg/42378'.
> The individual assigned to look at your
> report is: pkg-manager.
>
>
>>Category: pkg
>>Responsible: pkg-manager
>>Synopsis: build of fonts/t1lib with PKG_OPTION.t1lib=-x11 broken
>>Arrival-Date: Wed Nov 25 17:25:00 +0000 2009
>>
>

From forum: pkgsrc-bugs

Support for S3C2440 SD Controller

2009-11-25T10:00:39Z

Hi there,
I have now managed to make a driver for the S3C2440 SD Controller,
together with some very preliminary interface to the S3C2440 DMA
controller.
The SD driver supports only SD memory cards, and has only been tested
with 512MB, 1GB, and 2GB non-SDHC cards, and a 4GB SDHC card.
DMA transfers are supported, while wide bus transfers are NOT supported.

The patch[1] applies cleanly against the netbsd-5 branch in CVS.

Comments and critique is very welcome.

Cheers,
Paul

[1]: http://xpg.dk/files/File/netbsd/mini2440-20091125.patch

From forum: port-arm

Re: Improve caching

2009-11-25T09:37:59Z

On Wed, Nov 25, 2009 at 04:28:55PM +0100, Matthias Kretschmer wrote:

I think the first step should be stabilize CLOCKPRO and make it actually
possible to run busy servers with the alternate paging policies we already
have in the tree. If CLOCKPRO didn't eat it fairly quickly on large
systems under load, from the literature it looks like it would be a better
alternative than the ad-hoc-modified CLOCK we have by default now.

Once we have the paging algorithms already in the tree stable and working,
then, it seems to me, would be the right time to add more.

Thor

From forum: tech-kern

pkg/42378: build of fonts/t1lib with PKG_OPTION.t1lib=-x11 broken

2009-11-25T09:25:00Z

Note: There was a bad value `' for the field `Class'.
It was set to the default value of `sw-bug'.

>Number: 42378
>Category: pkg
>Synopsis: build of fonts/t1lib with PKG_OPTION.t1lib=-x11 broken
>Confidential: no
>Severity: non-critical
>Priority: medium
>Responsible: pkg-manager
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Wed Nov 25 17:25:00 +0000 2009
>Originator: W. Stukenbrock
>Release: NetBSD 4.0
>Organization:

Dr. Nagler & Company GmbH

>Environment:

System: NetBSD test-s0 4.0 NetBSD 4.0 (NSW-WS) #2: Fri Nov 6 11:14:49 CET 2009 wgstuken@s012:/export/NetBSD-4.0/N+C-build/.OBJDIR_amd64/export/NetBSD-4.0/src/sys/arch/amd64/compile/NSW-WS amd64
Architecture: x86_64
Machine: amd64
>Description:
When x11 options is deselected in fonts/t1lib all package
references to the x11 stuff is ignored, but the package is still
build with X support.
The configure argument --without-x is missing ...
>How-To-Repeat:
Try to compile t1lib just after installing a system with the
opton x11 deselected. It will fail to compile.
>Fix:
The followin patch will fix this problem.
It adds --without-x to the configure arguments in option.mk
if the x11 references are not included.

--- options.mk 2009/11/25 17:15:59 1.1
+++ options.mk 2009/11/25 17:16:47
@@ -12,4 +12,6 @@
.include "../../x11/libXp/buildlink3.mk"
.include "../../x11/libXpm/buildlink3.mk"
.include "../../x11/libXt/buildlink3.mk"
+.else
+CONFIGURE_ARGS+= --without-x
.endif

>Unformatted:

From forum: pkgsrc-bugs

Re: buildlink3 and cmake?

2009-11-25T09:11:46Z

Thomas Klausner writes:
- On Tue, Nov 24, 2009 at 11:19:56PM -0600, Eric Schnoebelen wrote:
- > I'm trying to package a software suite that uses cmake for
- > configuration.
- >
- > From examination, it appears that the cmake support (USE_TOOLS+=
- > cmake) doesn't pay attention to the buildlink directories. Is
- > this by design?
- >
- > Or have I missed something?
-
- You'll have to provide more information than that -- cmake in pkgsrc
- works together with buildlink3, as you can see when looking at
- e.g. the kde4 packages.

Thanks for the reference to a working packaging using buildlink
and cmake. I'm taking a look at the KDE4 packages to see how
they make use of cmake.

What I'm not seeing is the include file search paths and library
search paths referencing the buildlink directories on the
compilation and link-editing command lines.

Hence, more, or incorrect, libraries are being drawn in.

Thanks,
Eric

--
Eric Schnoebelen eric@... http://www.cirr.com
"Fix reason firmly in her seat, and call to her tribunal every fact,
every opinion. Question with boldness even the existence of a God;
because, if there is one, He must more approve of the homage of
reason, than that of blindfolded fear." -- Thomas Jefferson

From forum: tech-pkg

Re: Improve caching

2009-11-25T09:06:48Z

On Wed, 25 Nov 2009, Matthias Kretschmer wrote:

> Hi,
>
> > There was some discussion about this [1]. And I have found this page
> > [2] which describes optional page cache replacement algorithms in Linux.
> >
> > [1] http://marc.info/?l=netbsd-tech-kern&m=111806501516943&w=2
> > [2] http://linux-mm.org/AdvancedPageReplacement
>
> maybe it would be nice to exchange something like LRU/CLOCK by a
> randomized paging algorithm. Deterministic online algorithm are at best
> k competitive (where k is the size of the fast memory in number of
> pages). There are known randomized online algorithms which are 2H_k
> competitive against oblivious adversaries (where H_k is the k-th
> harmonic number).

Another option that might be worth investigating is to move pages
immediately to the head of the active list when they take a protection
(reference/write) fault. This might result in a closer approximation to
true LRU. But it may also result in increased overhead.

Another thing to consider is moving buffer cache pages that are otherwise
unreferenced to the inactive list immediately after each copy operation
completes. That should give preference to replacing those pages over
replacing pages that are mapped into process address spaces.

Be prepared to make detailed performance measurements or, knowing the
history of these mailing lists, you are likely to be flamed.

Eduardo

From forum: tech-kern

Re: Installing and booting and OF3 machine from a single empty internaldisk, not working

2009-11-25T08:34:02Z

I wrote:

> (it looks OF could be confused if MSDOSFS has some certain parameters)

After several tests, it looks OF3 on my G4 AGP doesn't like FAT16
but it can load ofwboot.xcf properly from FAT12.
On the other hand, it looks OF2 on my Apus2000 doesn't like FAT12
but it can load ofwboot.xcf properly from FAT16.

Oh well.

> > then
> > install with the install kernel, and then boot the machine as you have
> > done.
>
> It would be trivial to make sysinst prepare such MSDOSFS
> with fdisk partition. (I did it for NetBSD/arc)

I've created a test install CD ISO image based on netbsd-5-0 branch:
ftp://ftp.NetBSD.org/pub/NetBSD/misc/tsutsui/iso/macppccd-5.0.1_PATCH-OF3test.iso.gz
ftp://ftp.NetBSD.org/pub/NetBSD/misc/tsutsui/iso/macppc-OF3test-20091125.diff

Note:
- this is not well-tested and could coruppt your disks
(i.e. use it at your own risk)
- this won't work on OF2 machines
(no version check and no usr/mdec/ofwboot due to ramdisk overflow)
- not sure what happens if your disk has HFS with Apple partition map

- see NetBSD guide for general sysinst info
http://www.NetBSD.org/docs/guide/en/chap-exinst.html
- on MBR partitions menu, select "b: Use the entire disk"
- on Disklabel partitions menu, select
"a: Set sizes of NetBSD partitions"
and change partition or swap size if necessary
- MSDOSFS boot partition is prepared automatically, but
don't change it in the disklabel editor
- to boot from disk, command like "boot ultra0:,ofwboot.xcf" should work
(check "devalias" on OF and use an appropriate device for your disk)

Please let me know how it goes on you OF3 machines.
---
Izumi Tsutsui

From forum: port-macppc

Re: Intel I55 chipset supported?

2009-11-25T08:30:49Z

On Wed, Nov 25, 2009 at 03:36:29PM +0100, Ignatios Souvatzis wrote:
>
> "integrated 10/100/1000 Ethernet" (might be: Support for the Intel?
> 82578DC Gigabit Network Connection)

I can't dig up this datasheet immediately -- Intel's website is being
extremely slow -- but from the FreeBSD and Linux drivers it looks as if
this is a follow-on to the 82567 integrated Ethernet which, unfortunately,
uses the programming interface of the newer 82575 instead of that of the
traditional Intel gig chips which 'wm' supports.

We don't have a driver for this. It might work with NDIS, with the usual
caveats that NDIS probably doesn't work with a 64-bit kernel, could maybe
be fixed by borrowing code from FreeBSD, etc.

The FreeBSD driver for this family of chips is awful.

> The alternative offer is Dell Optiplex 760 MT.
> Chipset is Intel Q43 w/ICH10D
> Integrated Ethernet is 82567LM

This will work.

Thor

From forum: tech-kern

Re: Intel I55 chipset supported?

2009-11-25T08:30:49Z

From forum: port-i386

Re: Does firefox 3.5 on netbsd-4/i386 work?

2009-11-25T08:09:25Z

[moving thread from pkgsrc-users to tech-pkg]

The following happens on netbsd-4:

On Tue, 24 Nov 2009 12:40:35 +0100
Tobias Nygren <tnn@...> wrote:

> # ldd /usr/pkg/lib/xulrunner/xulrunner-bin | grep fontconf
> -lfontconfig.1 => /usr/pkg/lib/libfontconfig.so.1
> -lfontconfig.1 => /usr/X11R6/lib/libfontconfig.so.1

I believe the problem is caused by run path ordering.
Usually the run path for pkgsrc binaries is /usr/pkg/lib:/usr/X11R6/lib,
but mozilla binaries get it the other way around, so the wrong
fontconfig library is picked up.

In the work log I see the following:

WARNING: [transform-gcc] passing unknown option -R/usr/X11R6/lib

It comes from gtk+-2.0.pc:

# pkg-config --libs gtk+-2.0
-Wl,-R/usr/pkg/lib -R/usr/X11R6/lib -Wl,-R/usr/X11R6/lib -L/usr/pkg/...

Removing the "-R/usr/X11R6/lib" from the .pc file makes the duplicate
fontconfig reference go away in xulrunner. I guess this happens because
the pkgsrc wrappers don't pay attention to the -R flag, rendering any
ordering done behind the scene incomplete.

From the WARNING message I assume that fixing the .pc file is the
correct approach, but I'd like to hear a second opinion on this.

Also, could someone using native X11 on netbsd-5 please check if
libxul.so has the duplicate libfontconfig.so in ldd(1) output?
If so I think we need a recursive revbump on gtk2 after this is
fixed. :-(

Kind regards,
-Tobias

From forum: tech-pkg