headers
Mitja Muženič | 2 Jan 2004 16:16

Re: FW: fix for bridge performance on if_sis

Not yet, I'm still waiting for new hardware. My existing Soekris
hardware doesn't use shared interrupts for sis. New box should be here
soon after the holidays, I'll try to reproduce this and report.

Regards, Mitja

> -----Original Message-----
> From: Markus Friedl [mailto:markus <at> openbsd.org] 
> Sent: Tuesday, December 30, 2003 1:11 PM
> To: Mitja Mu?eni?
> Cc: tech <at> openbsd.org
> Subject: Re: FW: fix for bridge performance on if_sis
> 
> 
> have you been able to reproduce this?
> 
> On Sun, Dec 28, 2003 at 07:59:34PM +0100, Mitja Mu?eni? wrote:
> > Posted this on misc <at> , but no answer. Then I realized it 
> could go here -
> > it has a patch attached, after all. :-)
> > 
> > 
> > Regards, Mitja
> > 
> > -----Original Message-----
> > From: owner-misc <at> openbsd.org 
> [mailto:owner-misc <at> openbsd.org] On Behalf
> > Of Mitja Mu?eni?
> > Sent: Saturday, December 27, 2003 12:54 AM
> > To: misc <at> openbsd.org
(Continue reading)

Permalink | Reply | 0 comments |
headers
The TenDRA Project | 3 Jan 2004 03:11

Calling all boat anchors!

The TenDRA project (http://www.ten15.org) has come a long way in the vibrant two
years that have passed. Development has been quite steady and as a result of
this a lot of excellent changes have been brought into TenDRA to help make it
one of the most modern compilers available.

We're now at the point where more hardware is needed in order to bring
TenDRA up to speed with modern platforms. While none of us have been blessed
with a trust fund, we've managed to scrape together as much hardware as
possible to help us fix a lot of the issues we've come across in test-builds
on platforms such AIX, FreeBSD, HP-UX, Irix, Linux, OSF1, OpenBSD, SCO, Solaris,
SunOS, Ultrix and Unixware.

The limited amount of money the TenDRA project has is better spent on improving
its build, development and regression cluster. We ask all of you to please
donate your unused hardware to the TenDRA project to assure that it has a brighter
future (and possibly one day assure a brighter future for you :)

Please keep in mind, if your favorite operating system is not listed or you
wish for more emphasis to be put into your platform we would be more than happy
to target your donated hardware for this goal.  Despite the amount of time such
tasks may take, the TenDRA team will eventually get around to it.

Currently, we support the PowerPC, HP-PA, MIPS, IA32, Alpha and SPARC CPU's.
A full list can be found here http://www.ten15.org/support/

Our donations page contains the complete list of hardware we're currently 
looking for: http://www.ten15.org/support/

Please CC us on all replies, as it is too difficult for us to monitor all replies.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Ishikodzume | 3 Jan 2004 19:07

6to4 and OpenBSD

Hi,

This is probably more of a 'user' question, but i figured i was much
less likely to get a decent response on the users list.

I'm wondering, how does one set up 6to4 in OpenBSD? I'm not looking for
a detailed explanation, just a point in the right direction.
For a start, all the tutorials and docs i have seen for 6to4 on BSD, use
the stf interface. However, this does not seem present in the GENERIC
kernel, if in the kernel at all - I downloaded the kernel source, and
looked through conf/GENERIC, and there isn't even a commented out option
for stf.

--
# for IPv6
pseudo-device   gif     4       # IPv[46] over IPv[46] tunnel (RFC1933)
#pseudo-device  faith   1       # IPv[46] tcp relay translation i/f
--
...is all i can see.

All the mailing list posts i see on this are years old, and do not
mention 6to4 unless in the context of a question which always goes
unanswered. I'm thinking maybe someone knows something new now.

The only tutorials i can find for 6to4 are either in Linux, for which i
cannot get the same techniques to work with the gif interface (could be
my own fault, though - if someone has got it to work, please say and
i'll persevere), or NetBSD... which explicitly uses the stf interface in
the instructions.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jun-ichiro itojun Hagino | 3 Jan 2004 20:27

Re: 6to4 and OpenBSD

> This is probably more of a 'user' question, but i figured i was much
> less likely to get a decent response on the users list.
> 
> I'm wondering, how does one set up 6to4 in OpenBSD? I'm not looking for
> a detailed explanation, just a point in the right direction.
> For a start, all the tutorials and docs i have seen for 6to4 on BSD, use
> the stf interface. However, this does not seem present in the GENERIC
> kernel, if in the kernel at all - I downloaded the kernel source, and
> looked through conf/GENERIC, and there isn't even a commented out option
> for stf.

	6to4 (stf interface) is intentionally omitted from kame -> openbsd
	merge due to its security issues.  see
	draft-savola-v6ops-6to4-security-03.txt as well as warnings on
	netbsd manpage.

itojun
Permalink | Reply | 3 comments |
headers
Nedko Arnaudov | 4 Jan 2004 12:31
Picon

shared libraries without SONAME

Is it ok to have shared libraries without SONAME ?
Linking such libraries by directly supplying them in command line and
not by -l option causes produced executables to contain NEEDED entries
that contain paths in them, like this:

NEEDED      /usr/local/lib/libvorbis.so.2.0

I've actually tracked this to be caused by libtool that does not supply
-soname option to linker when creating shared libraries.

While absolute paths may have some reason (really ?), realtive paths in
NEEDED entries are very bad. One may wonder why someone will link by
directly supplying path to library ? Well, adding libraries to xxx_LDADD
in Makefile.am (automake) files is legal AFAIK. This is used when
package(program) contains shared libraries that are used by executables
in this package. If these libraries are currently in say ../mylibs this
this will cause the NEEDED entries in executable linked against them to
contain ../mylibs path. After this, one installs the binary in say
/usr/local/bin and then runs it with current directory being user home
directory. Of course there is no $HOME/../mylibs/libxxxx.so.xx so binary
cannot load its libraries. All this will never happen if libraries it is
linked against have SONAME entries. I've reproduced this with 3.4 and
I'm not aware if this is a bug caused by recent switch to ELF. Also may
be this behaviour is by design. Please give me some info, I've fixed my
libtool, however I'll be happy if next release of libtool (1.6) and the
one from OpenBSD ports will supply -soname option to linker when
creating shared libraries. If this behaviour is not a bug but a feature,
please let me know, so I can fix the software and not libtool. Also in
"not a bug" case any directions about how to  change "path/library in
xxx_LDADD" to something that will fit in such shared libraries design
(Continue reading)

Permalink | Reply | 0 comments |
headers
Andrew Snare | 4 Jan 2004 13:57

Re: 6to4 and OpenBSD

On Jan 3, 2004, at 8:27 PM, Jun-ichiro itojun Hagino wrote:

> 	6to4 (stf interface) is intentionally omitted from kame -> openbsd
> 	merge due to its security issues.  see
> 	draft-savola-v6ops-6to4-security-03.txt as well as warnings on
> 	netbsd manpage.

It's not easy to find the document you quoted here since the draft 
apparently expired without being revised or graduating to RFC status. 
All the Google hits amount to a placeholder saying as much.

The NetBSD man page mentions some security concerns, mainly to do with 
the risk posed by packets with spoofed addresses. However, these issues 
affect any router (based on a cursory look through the list) and are 
usually addressed by any semi-reasonable packet-filter.

  - Andrew
Permalink | Reply | 2 comments |
headers
Jun-ichiro itojun Hagino | 4 Jan 2004 14:55

Re: 6to4 and OpenBSD

> > 	6to4 (stf interface) is intentionally omitted from kame -> openbsd
> > 	merge due to its security issues.  see
> > 	draft-savola-v6ops-6to4-security-03.txt as well as warnings on
> > 	netbsd manpage.
> It's not easy to find the document you quoted here since the draft 
> apparently expired without being revised or graduating to RFC status. 
> All the Google hits amount to a placeholder saying as much.

	then look for -02.

itojun
Permalink | Reply | 0 comments |
headers
Matthew Jacob | 4 Jan 2004 22:05
Favicon

Re: Slow disk reads with ISP1020 on Alphaserver

First of all, the drives are negotiating Fast/Wide (10MHz/wide- see
lower in your report).

The read/write numbers do seem slow given the drives you're using.
Permalink | Reply | 0 comments |
headers
Drain Fade | 4 Jan 2004 22:36
Gravatar

Why no chrooted sftp ?

I've looked everywhere and have wondered why there is no native chrooting
of atleast sftp ? I run chrooted ftp for my users and love it, and the
only reason I keep it is because I don't want these people to have access
to my system unchrooted with shell access. Anyone know why this hasn't
been developed into openssh ?
Permalink | Reply | 1 comment |
headers
Sacha Ligthert | 4 Jan 2004 22:55

Re: Why no chrooted sftp ?

> Anyone know why this hasn't been developed into openssh ?

Just like many other "features" end users take for granted from ftpd, the 
sftp subsystem isn't as advanced as the default ftpd. And compared it 
lacks many features.

But feel free to submit your patches. Somebody is bound to pick them up 
:-)

Sacha

Disclaimer: no developer.

--
This sig has been ROT26 encrypted.
Reading it is unlawful under the terms of the DMCA.
Permalink | Reply | 0 comments |

Gmane