Re: HP DL145 G2?

2005-07-26 Thread Johan M:son Lindman
On Monday 25 July 2005 16.52, Mike Shaw wrote:
 Hey folksI'm about to build another obsd server for some pseudo-mission
 critical work, and HP is kind of our standard now. I've verified with
 someone off list that a DL140's run well, but for performance and
 philosophical reasons I'm choosing AMD...looking at a DL145 G2 2Ghz SATA.

 I saw some troubles on the archives regarding this, but I wanted to verify
 the latest:

 * Are the broadcom nics reliable at this point?
 * I'm assuming amd64 OpenBSD is ready for prime time.
 * Any potential gotchas?

Currently the DL145 G2 is not well supported for various reasons.
There's an issue with the pci shitz on this server which makes OBSD unable to 
detect cards seated in the expansion slots.

Since this server is based on Nvidia (nForce4) chipset I guess it's not prio 
one to support it due to lack of documentation etc.

So before you shop HP stuff make sure it's not fitted with Nvidia motherboards 
as in the case with DL 145G2.


Regards
Johan M:son



Re: carp failover on DSL and Cable connection?

2005-07-26 Thread Stephen Marley
On Mon, Jul 25, 2005 at 08:57:06PM -0700, Jonathan Walther wrote:
 You could run ospfd (or quagga) on each host. (You'll need to use gif
 or gre tunnels to give a multicast capable link over the vpns). Make
 the dsl tunnel the lower cost route and ospf will change the routing
 tables to use the other link if it goes down. When it comes back up,
 ospfd will switch the routing table back to the lower cost route. I use
 precisely this method to provide a backup to a 100Mb WAN link using
 ipsec/adsl.
 
 Thank you Stephen!  This is exactly what I was looking for.  One
 question; does this solution drop any connections during the change of
 the routing table?  For my application, that isn't a problem, but it is

Nothing is explicitly dropped, but the behaviour depends on how long you
set the router dead time to and how the application behaves. The default
dead time is 40sec, but I use 10secs in my setup. TCP/IP is able to
handle some packet loss and routing table changes without dropping
connections.

-- 
stephen



Re: MySQL socket problem (solved)

2005-07-26 Thread Uwe Dippel
On Thu, 21 Jul 2005 13:22:43 +0300, Tomas wrote:

 BTW Edd, I liked your trick :)

Me, too !!

maybe something wrong still on my side, though;
because the trick only works with an additional

chown _mysql:_mysql /var/www/var/run/mysql/

in my case, otherwise I get 

050726 16:57:22  mysqld started
050726 16:57:22 Can't start server : Bind on unix socket: Permission denied
050726 16:57:22 Do you already have another mysqld server running on socket: 
/var/run/mysql/mysql.sock ?
050726 16:57:22 Aborting
050726 16:57:22 /usr/local/libexec/mysqld: Shutdown Complete
050726 16:57:22  mysqld ended



Re: Disable IPv6 on 3.7

2005-07-26 Thread Russell J. Wood
On Tue, Jul 26, 2005 at 03:05:37AM +0200, knitti wrote:
 On 7/26/05, Russell J. Wood [EMAIL PROTECTED] wrote:
  On Mon, Jul 25, 2005 at 08:42:29PM -0400, Brad wrote:
   Go ahead if you want to use a custom un-supported system.
  
  Thanks, I will.
  
   What is it that you think you're gaining from this?
  
  A system without IPv6.
  
 
 you won't. you'll get a kernel without IPv6. and a broken system.
 
 --knitti

My system works fine with IPv6 disabled...

- Russell



Re: MySQL socket problem (solved)

2005-07-26 Thread Edd Barrett
 chown _mysql:_mysql /var/www/var/run/mysql/

Because mine was on my laptop (which isnt connected to the network
when I run mysql. Infact I hardly ever run mysql), I took the shortcut
'mysqld_safe --user=root' to start it.

You are right, the mysql user will have to have access

Edd



Re: Disable IPv6 on 3.7

2005-07-26 Thread Adam Papai
knitti said:
  What is it that you think you're gaining from this?

 A system without IPv6.

 you won't. you'll get a kernel without IPv6. and a broken system.

knitti's right.

But why do you so much against Ipv6?


-- 
Adam Papai
D i g i t a l Influence
E-mail: [EMAIL PROTECTED]
Phone: +36 30 33-55-735



Re: Disable IPv6 on 3.7

2005-07-26 Thread Henning Brauer
* Russell J. Wood [EMAIL PROTECTED] [2005-07-26 02:36]:
 Yes, one can by commenting out `OPTION INET6' in the kernel
 configuration.

the question was specifically (and for good reasons) for doing so 
WITHOUT compiling a custom kernel.

 And one would want to do that if they don't use, IPv6, since it's
 pointless fat otherwise.

another wrong myth.
it just doesn't make a difference.

-- 
BS Web Services, http://www.bsws.de/
OpenBSD-based Webhosting, Mail Services, Managed Servers, ...
Unix is very simple, but it takes a genius to understand the simplicity.
(Dennis Ritchie)



spamd greylisting, masking on /24

2005-07-26 Thread Stuart Henderson
I seem to remember seeing a patch to spamd that makes greylisting only 
look at the first /24 of the address, but I can't find it after fairly 
extensive searching with google/marc. Does anyone have a copy they 
could point me at?


The whitelists on puremagic.com (on which greylisting.org's lists are 
based) don't list networks with a common spool unless more than a /24 
is involved (there are some /24 listed with other factors requiring 
whitelisting, e.g. unique sender addresses per delivery attempt).




Re: Create my own shell?

2005-07-26 Thread Jon Drews
On 7/25/05, Jon Drews [EMAIL PROTECTED] wrote:
 On 7/25/05, Abel Talaversn Estevez [EMAIL PROTECTED] wrote:
  I need to create a particular but simple shell for a firewall running 
  OpenBSD
  3.6. The idea is create a user whose shell is a very limited one. 
 Hi:
 
   Operating ksh in restricted mode may fulfill your needs. 

Oops - this is not true. I set up an account with rksh (ksh -r) and it
is possible for the user to still switch shells. For the details on
this see:
Practical Unix  Internet Security, 3rd Edition by Simson Garfinkel,
Gene Spafford, Alan Schwartz. The relevant material is on pages 576 to
578.

Basically the restricted shell can be subverted and they advise using chroot.


-- 
Kind regards,
Jonathan



Re: Did anybody hear this??

2005-07-26 Thread Jurjen Oskam
On Mon, Jul 25, 2005 at 10:05:32PM -0700, Bruno Delbono wrote:

  how much truth is actually in this article???
 It makes a lot of sense and is right on. What I take out of this article is
 that having one single firewall (can be any type: network, application etc.)
 at the perimeter doesn't stop hackers.

It does look like the before situation in the article is one where there
is only one firewall that separates the LAN from the Internet, and
everything on the LAN is treated equally, workstations and servers alike.

Generally, that is a bad situation. So, the advice to put different types
of machines into different (protected) networks is good. Many people
wouldn't go as far as entirely eliminating the outside firewall though; although
he says that the desktops run secure OSes he also mentions Active
Directory. Some would say those two terms don't go well together. :-)

 I don't see what really alarmed you? The author makes excellent points and I 
 agree with the him.

I also agree, except for the part of eliminating the externally facing firewall
entirely.

-- 
Jurjen Oskam



Updating packages.

2005-07-26 Thread Kevin MacPherson
Hello,

I'm trying to go from stable to current.
I have the kernel and userland in place but having trouble updating my desktop.
When trying to compile gnome I'm getting errors on the dependencies conflicting 
with older versions of themselves?
Is there anyway to tell the make command to force reinstall the dependencies?

After an evening of googling I still can't find the answer.


Best Regards,
Kevin MacPherson



Re: Updating packages.

2005-07-26 Thread Edd Barrett
On 26/07/05, Kevin MacPherson [EMAIL PROTECTED] wrote:
 Hello,
 
 I'm trying to go from stable to current.
 I have the kernel and userland in place but having trouble updating my 
 desktop.
 When trying to compile gnome I'm getting errors on the dependencies 
 conflicting with older versions of themselves?
 Is there anyway to tell the make command to force reinstall the dependencies?
 
 After an evening of googling I still can't find the answer.
 
 
 Best Regards,
 Kevin MacPherson
 
 

Hello,

Take a look at pkg_add -r  ;)

OpenBSD favors pkg's over ports.

Either that or pkg_delete -q /var/db/pkg/* then complile your stuff.

Edd



Re: Did anybody hear this??

2005-07-26 Thread Terry Tyson
On Tue, Jul 26, 2005 at 03:20:05PM +0200, Jurjen Oskam wrote:

snip

 It does look like the before situation in the article is one where there
 is only one firewall that separates the LAN from the Internet, and
 everything on the LAN is treated equally, workstations and servers alike.
 
 Generally, that is a bad situation. So, the advice to put different types
 of machines into different (protected) networks is good.

I only have one firewall but it is three legged, the DMZ box and the
LAN are seperate. Is this what you mean by different (protected)
networks?

Terry



Re: Did anybody hear this??

2005-07-26 Thread Spruell, Darren-Perot
From: Terry Tyson [mailto:[EMAIL PROTECTED]
  Generally, that is a bad situation. So, the advice to put 
 different types
  of machines into different (protected) networks is good.
 
 I only have one firewall but it is three legged, the DMZ box and the
 LAN are seperate. Is this what you mean by different (protected)
 networks?

I take it as meaning avoiding the crunchy on the outside, chewy in the
middle architecture that only perimeter security gives you. 

Depending on your network and the assets and information located on the LAN,
you may find that seperating services by access level gives you benefit. For
example, say you have financial users, financial servers, HR users, HR
servers, standard internal servers, and regular end users / trained monkey
staff. Even though they are technically all on the LAN, you can protect your
financial servers from the places and people on the LAN that don't need
access to them by placing/protecting them such that only your financial
users that DO need access to them can reach them. Ditto for the HR
systems/people. As for the standard network services servers, since
everybody needs to access them, you have a less restrictive policy around
them. Real segmentation of the LAN works for this kind of thing, via VLANs
or whatever.

DS



Re: Did anybody hear this??

2005-07-26 Thread Jurjen Oskam
On Tue, Jul 26, 2005 at 11:20:35AM -0500, Terry Tyson wrote:

 I only have one firewall but it is three legged, the DMZ box and the
 LAN are seperate. Is this what you mean by different (protected)
 networks?

Everything depends on your particular situation and needs, but the
general idea is that servers shouldn't be wide open to the clients.
In your case, if that one firewall is compromised, all attached networks
are exposed. This might or might not be something you should worry
about. It all depends on your needs.

-- 
Jurjen Oskam



chrooted httpd and directory

2005-07-26 Thread Vlad Ciubotariu
Which path should I use in a directory directive in the config file
for a chrooted  httpd?

In both cases below, changing Deny to Allow achieves the desired effect.

thanks!

Directory /var/www/users/*
AllowOverride FileInfo AuthConfig Limit Options
Options MultiViews Indexes SymLinksIfOwnerMatch IncludesNoExec
ExecCGI
Limit GET POST OPTIONS PROPFIND
Order allow,deny
Deny from all
/Limit
Limit PUT DELETE PATCH PROPPATCH MKCOL COPY MOVE LOCK UNLOCK
Order deny,allow
Deny from all
/Limit
/Directory


Directory /users/*
AllowOverride FileInfo AuthConfig Limit Options
Options MultiViews Indexes SymLinksIfOwnerMatch IncludesNoExec
ExecCGI
Limit GET POST OPTIONS PROPFIND
Order allow,deny
Deny from all
/Limit
Limit PUT DELETE PATCH PROPPATCH MKCOL COPY MOVE LOCK UNLOCK
Order deny,allow
Deny from all
/Limit
/Directory



Anyone know of a mavell based dual gigE copper card

2005-07-26 Thread Bill Chmura
From what everyone told me last time, the SK stuff is good.  So I can
fit my network together with a few dual cards, trunk the smaller stuff
together and then be on my way.  Trouble is I cannot find (for the life
of me) anything dual based on the marvell stuff.

The obsd man page
http://www.openbsd.org/cgi-bin/man.cgi?query=skapropos=0sektion=0manpath=OpenBSD+Currentarch=i386format=html

For dual it only lists the SK-9822 SK-NET GE-T dual port, copper
adapter, which from threads I read is now realTek chips in the newer
revs.

I've tried contacting Marvell for info on products made using them, but
no answer yet.  I've searched, prodded, poked and cursed and I still
have not found one.

Thoughts or suggestions?

I appreciate the advice from the last round... I am using much of it.


-- 

Bill Chmura



Other Sharp Zauri?

2005-07-26 Thread STeve Andre'
   Hopefully this more general question is appropriate here.   I've 
noticed sales for the Zaurus SL-5500.  Geeks.com has a sale on
them right now.

   My question is if it makes sense for efforts to support any of
the other units besides the C3000 and C3100.  I know very little
about the Zaurus in general, hence this question.  I'm not asking
for any kind of a time-table for anything, just general feasibility
for support for some of the cheaper units.

Thanks,

STeve Andre'



Re: Anyone know of a mavell based dual gigE copper card

2005-07-26 Thread Henning Brauer
* Bill Chmura [EMAIL PROTECTED] [2005-07-26 19:48]:
 For dual it only lists the SK-9822 SK-NET GE-T dual port, copper
 adapter, which from threads I read is now realTek chips in the newer
 revs.

huh? that was linksys or dlink or netgear or one of the usual bandits.
The SysKonnect stuff was and is sk based (take a minute to guess what 
sk stands for after all :) )

-- 
BS Web Services, http://www.bsws.de/
OpenBSD-based Webhosting, Mail Services, Managed Servers, ...
Unix is very simple, but it takes a genius to understand the simplicity.
(Dennis Ritchie)



isakmpd stops forwarding data over enc0

2005-07-26 Thread Sean Knox
Sometime this morning, our openbsd firewall/VPN server entered a state
where it stopped forwarding encrypted traffic over the enc0 interface. Incoming
roadwarrior connections establish tunnels fine, but nothing is sent over
enc0. There have been no isakmpd or pf configuration changes. There's
nothing in the logs that seems indicate a problem.

any ideas?

thanks,
Sean



Re: Updating packages

2005-07-26 Thread Dimitri Yioulos
On Tuesday July 26 2005 11:09 am, Edd Barrett wrote:
 Today 11:09:49 am

 On 26/07/05, Kevin MacPherson [EMAIL PROTECTED] wrote:
  Hello,
 
  I'm trying to go from stable to current.
  I have the kernel and userland in place but having trouble updating my
  desktop. When trying to compile gnome I'm getting errors on the
  dependencies conflicting with older versions of themselves? Is there
  anyway to tell the make command to force reinstall the dependencies?
 
  After an evening of googling I still can't find the answer.
 
 
  Best Regards,
  Kevin MacPherson

 Hello,

 Take a look at pkg_add -r  ;)

 OpenBSD favors pkg's over ports.

 Either that or pkg_delete -q /var/db/pkg/* then complile your stuff.

 Edd

Outta http://www.openbsd.org/ports.html:

# setenv PKG_PATH ftp://ftp.openbsd.org/pub/OpenBSD/3.6/packages/i386/
# pkg_add ${PKG_PATH}packageyouwant.tgz
 
This takes care of dependencies.

Dimitri



Bridge Interface stop ICMP redirects?

2005-07-26 Thread Steve Williams

Hi,

I had a great design redesign and secure a client's network. Once I got 
on site, there was a little surprise for me ... I can't re number a 
router that I had hoped to.


Right now, there are multiple gateways on one network (all the gateways 
are plugged into one switch).  They have a default gateway (OpenBSD) 
that sends ICMP redirects for computers to access the (non-default) 
gateways.  This was done to avoid having to maintain routes on many 
network devices, and is causing problems with some applications.


I have a nice new Dell server with 4 NIC's in it that I want to hang 
each gateway off of.  Unfortunately, one of them cannot be renumbered.  
The main network is 192.168.11.0/24.  The default gateway for the 
network is the OpenBSD server at 192.168.11.20/32.  The router that 
can't be changed is 192.168.11.1/32.  I figure I can still plug the 
192.168.11.1/32 router into it's own NIC and set up bridging for it.  
But I am wondering if the OpenBSD box is going to be smart enough to NOT 
send ICMP redirects, knowing that it is on the other end of a bridge.


This is so screwed up, and it's in production, so I do not have a lot of 
downtime to play around with things.


I envision something like:
OpenBSD   3.7-current with 4 NIC's

Internal Network (em0)   192.168.11.20 (default gateway for 
192.168.11.0/24)

Internet Connection  (em1)   PPPoE
Hospital Traffic (em2)   (transparent bridge) to 192.168.11.1, dest 
IP A.B.C.D/32
Government Traffic   (em3)   10.2.60.4 dest 
IP E.F.G.H/32


Right now the OpenBSD box has a route...
route add A.B.C.D/32 192.168.11.1

As far as I can figure, this would still have to be there...to forward 
the packets to the next hop.  It's just that the next hop is local..


Humm... maybe a route add -ifp em2 A.B.C.D/32 192.168.11.1 would solve 
the problem?


Do I have a hope of getting this to work?  Are ICMP redirects still 
going to be issued?  I guess the joy of OpenBSD is that I can always 
block them with pf!!, but that just plain sounds wrong!


Thanks,
Steve



Re: Other Sharp Zauri?

2005-07-26 Thread Stefan Johnson
On 7/26/05, Greg Thomas [EMAIL PROTECTED] wrote:
 On 7/26/05, STeve Andre' [EMAIL PROTECTED] wrote:
 Hopefully this more general question is appropriate here.   I've
  noticed sales for the Zaurus SL-5500.  Geeks.com has a sale on
  them right now.
 
 
 I've been thinking about one myself even if I have to stick with
 Linux.  How much are they now, something like $120?
 
 One thing that scared me off is really short battery life.  Anyone have one?
 
 Greg
 
 

I have an SL-5500.  It works great, but the battery life can be bad.  When I
was still in college, I could use it for a day of classes, but would
have to charge it as soon as I got home.  If you plan on doing anthing
that uses the CF slot (wireless networking maybe?) the battery life
drops drastically, so keep the charger with you if you can.

Stefan



Re: Bridge Interface stop ICMP redirects?

2005-07-26 Thread Steve Williams

Hi,

That sounds great!  Thanks very much for pointing that out, I would 
never have thought about sysctl to control that...


Cheers,
Steve

Spruell, Darren-Perot wrote:


From: Steve Williams [mailto:[EMAIL PROTECTED]
 

The main network is 192.168.11.0/24.  The default gateway for the 
network is the OpenBSD server at 192.168.11.20/32.  The router that 
can't be changed is 192.168.11.1/32.  I figure I can still plug the 
192.168.11.1/32 router into it's own NIC and set up bridging for it.  
But I am wondering if the OpenBSD box is going to be smart 
enough to NOT 
send ICMP redirects, knowing that it is on the other end of a bridge.
Do I have a hope of getting this to work?  Are ICMP redirects still 
going to be issued?  I guess the joy of OpenBSD is that I can always 
block them with pf!!, but that just plain sounds wrong!
   



Right. You can configure a kernel variable using this sysctl:

net.inet.ip.redirect

sysctl(3) gives more information about what this controls. sysctl.conf(5)
may be of use to you.

DS 




Re: MySQL socket problem (solved)

2005-07-26 Thread Jonathan Schleifer
Another way is this:

# MySQL
if [ -x /usr/local/bin/mysqld_safe ]; then
echo -n ' mysqld'
rm -f /var/www/var/run/mysql/mysql.sock
/usr/local/bin/mysqld_safe  /dev/null 
sleep 10
ln /var/run/mysql/mysql.sock /var/www/var/run/mysql/mysql.sock
fi

That's my rc.local for starting mysql. Works just fine here :)
This way, it's /var/run/mysql.sock inside and outside the chroot. But
you have to recreate the hardlink if mysql restarts.

-- 
Jonathan



rdr question

2005-07-26 Thread GV
Hi list,

is it possible to have the following:

rdr on $ext_if proto tcp from any to any port 80 - $server

re-written as:

rdr on $ext_if proto tcp from any to domain.com port 80 - $server

where $server an internal web server and domain.com a specific domain 
name?

In general I would like to have one static IP where more than one domains are 
registered and for each domain a different internal web server should serve 
the incoming requests!

Thanks

George



Re: rdr question

2005-07-26 Thread Stuart Henderson

--On 27 July 2005 00:27 +0200, GV wrote:


is it possible to have the following:

rdr on $ext_if proto tcp from any to any port 80 - $server

re-written as:

rdr on $ext_if proto tcp from any to domain.com port 80 - $server

where $server an internal web server and domain.com a specific
domain  name?

In general I would like to have one static IP where more than one
domains are  registered and for each domain a different internal web
server should serve  the incoming requests!


No, you need some kind of 'reverse-proxy' to do this type of thing 
(maybe pound, tinyproxy 1.70, or squid in accelerator-mode). It would 
run on either the PF box or another box that you rdr to.




Apache icons inside chroot

2005-07-26 Thread Chris Zakelj
I've done the googling and turned up empty :(  I'm trying to get the 
included icons to show when someone does a directory view, but 
everything I try comes back with:


[Wed Jul 27 01:35:57 2005] [error] [client 192.168.0.3] (13)Permission 
denied: access to /icons/movie.gif failed because

search permissions are missing on a component of the path
192.168.0.3 - - [27/Jul/2005:01:35:57 +] GET /icons/movie.gif 
HTTP/1.1 403 225


in the error and access logs, respectively.  I'm almost certain it's 
because I'm not accounting for the chroot properly (the icons live in 
/var/www/icons by default).  The section of httpd.conf that addresses it 
reads thus:


---
Alias /icons/ /var/www/icons/ 


Directory /var/www/icons
   Options Indexes MultiViews
   AllowOverride None
   Order allow,deny
   Allow from all
/Directory

---

I've also tried using /icons/, /icons, and ../icons, all with 
negative results.  The files inside /var/www/icons are all mode 444, and 
the directory itself is mode 644, so I'm not sure what permission it 
needs that it doesn't already have.  Where should I look next?




openbsd rpc/xdr

2005-07-26 Thread Gustavo Rios
Hey folks,

i am doing efforts in order to learn about xdr/rpc. So, i decided to
read some code in src/lib/libc/rpc. I found it to be a little heavy,
cause there too many function invocation overhead between the caller
and the real function that do the job.

So, i wonder if anybody knows an alternative implementation for
xdr/rpc? Just like there are for stdio functionalities.

Of course, my requirement is that it runs on our OS of choice.

Thanks a lot.



Re: Did anybody hear this??

2005-07-26 Thread Siju George
On 7/26/05, Bruno Delbono [EMAIL PROTECTED] wrote:
 +++ Siju George [Tue Jul 26, 2005 at 10:18:56AM +0530]:
 
  how much truth is actually in this article???
 
 It makes a lot of sense and is right on. What I take out of this article is
 that having one single firewall (can be any type: network, application etc.)
 at the perimeter doesn't stop hackers.
 
 I don't see what really alarmed you? 


Thanks for the reply Bruno. Just the thing whether this is the current
trend. eliminating firewalls and going for an alternative like he
mentioned?

kind regards

Siju

The author makes excellent points and I
 agree with the him.  Now SMB's might traditionally fit better with these
 articles, bigger enterprises tends to differ as many roles (for the users
 anyway) are well defined and access (incoming, outgoing) for internal 
 external.
 
 -Bruno



Re: Did anybody hear this??

2005-07-26 Thread Chris Kuethe
On 7/26/05, Siju George [EMAIL PROTECTED] wrote:
 On 7/26/05, Bruno Delbono [EMAIL PROTECTED] wrote:
  +++ Siju George [Tue Jul 26, 2005 at 10:18:56AM +0530]:
 
   how much truth is actually in this article???
 
  It makes a lot of sense and is right on. What I take out of this article is
  that having one single firewall (can be any type: network, application etc.)
  at the perimeter doesn't stop hackers.
 
  I don't see what really alarmed you?
 
 
 Thanks for the reply Bruno. Just the thing whether this is the current
 trend. eliminating firewalls and going for an alternative like he
 mentioned?

You completely missed the point.

The point was that the crunchy on the outside, chewy on the inside
security model is wrong. A single perimeter firewall tends to allow
the inside network to be woefully unsecure and this is something to be
avoided. Or, put another way, the single greatest failing of a
firewall is that it allows people to continue behaving unsafely.

Think about it: if every host you control is set up to survive contact
with an evil host, then it doesn't matter much if someone out there
tries to break in, or someone brings in a virus-laden laptop or
whatever else. So maybe the elimination of the firewall is a
worthwhile pursuit so long as you keep an eye toward properly bolting
down your empire.

CK

-- 
GDB has a 'break' feature; why doesn't it have 'fix' too?



Re: Anyone know of a mavell based dual gigE copper card

2005-07-26 Thread Brad
On Tue, Jul 26, 2005 at 08:06:59PM +0200, Henning Brauer wrote:
 * Bill Chmura [EMAIL PROTECTED] [2005-07-26 19:48]:
  For dual it only lists the SK-9822 SK-NET GE-T dual port, copper
  adapter, which from threads I read is now realTek chips in the newer
  revs.
 
 huh? that was linksys or dlink or netgear or one of the usual bandits.
 The SysKonnect stuff was and is sk based (take a minute to guess what 
 sk stands for after all :) )

That was the Linksys EG1032 cards. 32-bit PCI single port cards. rev 2
uses Marvell chipset and rev 3 uses RealTek. It was even more fun fixing
support for the cards since Linksys used the same PCI ID code for both
cards.

Thanks to Daniel Polak who got in contact with the appropriate people from
SysKonnect. I will be receiving 3 newer SysKonnect cards for testing and
developement sometime very soon. He managed to score some hardware too.

I also had some offers from other individuals to purchase hardware. I'm
very appreciative of these offers. Of course there is no point to taking
these offers when I was able to get hardware from the vendor itself.

I will be looking for other stuff in the future. I can always use the help
of kind and helpful individuals on this list.



Re: openbsd rpc/xdr

2005-07-26 Thread Artur Grabowski
Gustavo Rios [EMAIL PROTECTED] writes:

 Hey folks,
 
 i am doing efforts in order to learn about xdr/rpc. So, i decided to
 read some code in src/lib/libc/rpc. I found it to be a little heavy,
 cause there too many function invocation overhead between the caller
 and the real function that do the job.

If I read correctly, it seems that you don't like fuction calls.
Why are functions bad? You prefer a macro and inline hell?

//art