Re: Please compile ports-mgmt/pkg for i486

2013-11-21 Thread Stanisław Halik
On Wed November 20 2013 20:15:34 Dimitry Andric wrote:
  http://llvm.org/bugs/show_bug.cgi?id=11212
  http://llvm.org/bugs/show_bug.cgi?id=17792
  Got bitten by that with -march=i486...
 Those bugs are about other CPU's, not specifically the Geode, or i486.
 The version of clang in head, stable/10 and stable/9 produces regular
 nops for -march=i486, which is the default setting if you don't pass any
 arch.  So how did you get bitten, then?  Were you using 9.1-RELEASE, by
 any chance?

It might be PEBKAC wrt the samba and cups ports - updating geode by putting CF 
into a VM with USB passhtrough... Perhaps switched to -march=native by 
accident.

Was using RELENG_9 from Feb 2013, after update, Nov 18 2013 RELENG_10.

-- 
xmpp:stha...@jabster.pl
sip:stha...@misaki.pl
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


pkg(1) ipv6 address not listening

2013-11-21 Thread Stanisław Halik
Hey,

pkg(1) doesn't like ipv6 from two different subnets:

| ^C15720: connect(6,{ AF_INET6 [2001:41c8:112:8300::50:1]:80 },28) ERR#4 
'Interrupted system call'

It's up fine, just not listening. Could you fix it, please?

-sh

-- 
xmpp:stha...@jabster.pl
sip:stha...@misaki.pl
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


Re: pkg(1) ipv6 address not listening

2013-11-21 Thread Stanisław Halik
Hey,

On Fri November 22 2013 07:15:00 Matthew Seaman wrote:
  pkg(1) doesn't like ipv6 from two different subnets:
  | ^C15720: connect(6,{ AF_INET6 [2001:41c8:112:8300::50:1]:80 },28) ERR#4
 
 pkg(8) works fine over IPv6 for me.

Failing command is:

% pkg update

using default repo configuration. The above line was producted by:

% truss -f pkg update

Unable to access pkg.freebsd.org IPv6 server over two subnets.
Tested with `nc -v'

2001:470:600d:dead:beae:c5ff:fee1:4407/48
2a01:4f8:192:50e3::/64

1001 ~ % ssh ananke.laggy.pk cat /etc/pkg/FreeBSD.conf
FreeBSD: {
  url: pkg+http://pkg.FreeBSD.org/${ABI}/latest;,
  mirror_type: srv,
  enabled: yes
}

cheers,
-sh

-- 
xmpp:stha...@jabster.pl
sip:stha...@misaki.pl
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


Re: pkg(1) ipv6 address not listening

2013-11-21 Thread Stanisław Halik
On Fri November 22 2013 07:30:02 Matthew Seaman wrote:
 Are you running the ports-mgmt/pkg-devel port or did you install from
 GitHub sources?  (I deduce this because the config format with curly
 braces does not work with pkg-1.1.4...)

Using pkg as bundled with RELENG_10.

# uname -sr
FreeBSD 10.0-BETA3

 Can't reproduce with 1.1.4 [..]

Is geolocated DNS in use? Note that the following server IP fails:

9273: connect(6,{ AF_INET6 [2001:41c8:112:8300::50:1]:80 },28) ERR#4 
'Interrupted system call'

That is, pkg0.bme.freebsd.org

Interesting... It works from Hurricane Electric... Apologies for 
misinformation.

It doesn't work from hetzner.de native IP:

07:43:30.110332 IP6 (flowlabel 0x8c886, hlim 64, next-header TCP (6) payload 
length: 40) 2a01:4f8:192:50e3::.36954  2001:41c8:112:8300::50:1.80: Flags 
[S], cksum 0x67c9 (incorrect - 0x98d3), seq 914261876, win 65535, options 
[mss 1440,nop,wscale 6,sackOK,TS val 14546988 ecr 0], length 0

The checksum bit is probably due to offload...

cheers,
-sh

-- 
xmpp:stha...@jabster.pl
sip:stha...@misaki.pl
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


Re: pkg(1) ipv6 address not listening

2013-11-21 Thread Stanisław Halik
good Sir,

Apologies!

It of course works as it should, wrong `route-to' applied, causing SYN packets 
to get stuck.

Apologies once again for wasting your valuable time.

cheers,
-sh

-- 
xmpp:stha...@jabster.pl
sip:stha...@misaki.pl
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


Re: Please compile ports-mgmt/pkg for i486

2013-11-20 Thread Stanisław Halik
On Wed November 20 2013 11:16:03 you wrote:
 Ah yes, we did attempt to fix those long nops on Geode once and for all,
 can you please confirm that pkg built with recent clang does not cause
 SIGILL for you?

Not unless -march=geode :(

Please see:

http://llvm.org/bugs/show_bug.cgi?id=11212
http://llvm.org/bugs/show_bug.cgi?id=17792

Got bitten by that with -march=i486...

-- 
xmpp:stha...@jabster.pl
sip:stha...@misaki.pl
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


Re: random_harvestq eats much on geode-lx alix2c3

2013-11-20 Thread Stanisław Halik
On Wed November 20 2013 12:58:28 freebsdonline wrote:
 I did not deployed yet 10.x (I'm waiting for nanobsd.sh to correct
 problems with pkgng) for now I am running it on VirtualBox, but I hope
 it will also run well on Alix hardware. Your problems were on 10.x or 11.x?

10.x. It's a new issue, works fine with 9.x.


xmpp:stha...@jabster.pl
sip:stha...@misaki.pl
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


Re: random_harvestq eats much on geode-lx alix2c3

2013-11-19 Thread Stanisław Halik
On Tue November 19 2013 23:01:37 Mark Robert Vaughan Murray wrote:
  rand_harvestq
 Take a look at random(4) - sysctls to turn off harvesting are documented
 there.

Thanks! Turning off software interrupt collection brings the cpu usage next to 
nothing.

 That is quite a busy harvestq - could you please give me some more details
 of what that box is and what it was doing at the time (numbers would be
 good!)

The box is a home firewall with pf/altq. HZ=1000, despite poor hardware. It's 
alix2c3. Uses polling and a long, bloated pf.conf. Also, scrubs packets a bit.

I'd be glad to attach more detail, but what do you ask for? Also, can text 
attachments be put here? For posterity's sake, better not to link to them.

cheers,
-sh

-- 
xmpp:stha...@jabster.pl
sip:stha...@misaki.pl
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


random_harvestq eats much on geode-lx alix2c3

2013-11-18 Thread Stanisław Halik
Hey,

random_harvestq eats much, much CPU on alix2c3:

CPU: Geode(TM) Integrated Processor by AMD PCS (498.06-MHz 586-class CPU)
glxsb0: AMD Geode LX Security Block (AES-128-CBC, RNG) mem 
0xefff4000-0xefff7fff irq 9 at device 1.2 on pci0

Could you please add a sysctl/loader knob for it, or a way to throttle 
collection?

Here's top output:

14 root  1 -16- 0K 8K -6:12  15.97% rand_harvestq

cheers,
-sh

-- 
xmpp:stha...@jabster.pl
sip:stha...@misaki.pl
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org