Re: Upgrade 6.1 -> 6.2: No /mnt/etc/myname

[Steven McDonald]

This is a complete guess, but is /etc/myname a symbolic link? If it is
a symlink to an absolute path, that is unlikely to exist in the bsd.rd
filesystem and would cause this error.

If that's not it, please clarify what you mean by the "file is there".
What command did you run, and what was its output? What sort of file is
it? And is sd0 the correct disk?

Re: Blank screen after boot with Radeon HD 5450

[Steven McDonald]

On Sun, 28 May 2017 22:54:22 +0200
Steven McDonald <ste...@steven-mcdonald.id.au> wrote:

> Re-reading your mail, it sounds like you installed the card into an
> existing system. In that case, my question becomes "have you run
> fw_update?"

Just noticed the part of your mail where you showed you had the
firmware installed. Sorry for the noise, brain must be in slow motion
tonight.

Re: Blank screen after boot with Radeon HD 5450

[Steven McDonald]

Re-reading your mail, it sounds like you installed the card into an
existing system. In that case, my question becomes "have you run
fw_update?"

Re: Blank screen after boot with Radeon HD 5450

[Steven McDonald]

On Sun, 28 May 2017 15:37:08 -0400
Maximilian Pichler  wrote:

> After the installation of a VisionTek Radeon 5450 graphics card my
> machine gives a blank screen after booting. It still shows the normal
> system messages (full dmesg below), with the last visible one being
> "scsibus4 at softraid0: 256 targets". Then the monitor (connected via
> DisplayPort) goes into power save mode. This is on OpenBSD 6.1 and
> amd64.

Have you tried booting more than once? I've had this problem on a
Radeon HD 6850 after a fresh install in the past, and after a reboot it
worked. I suspect the cause was that fw_update ran too late for the
firmware to load correctly (and I no longer have access to that
hardware to test).

Re: file systems

[Steven McDonald]

On Fri, 26 May 2017 11:35:49 -0300
Friedrich Locke  wrote:

> Hi folks,
> 
> does anybody here run OBSD with a file system bigger than 10TB ?
> How much time boot takes to bring the system up (i mean fsck) ?
> Are you using ffs2 ? With softdep ?
> 
> Thanks.

This depends heavily on how you plan to use the filesystem. I have an 8
TB filesystem (FFS2 with softdep) that takes several hours to fsck,
mostly in phase 2 ("Check Pathnames") because of a very large number of
hard links.

Re: What does it mean this error when I try install a package?

[Steven McDonald]

On Mon, 17 Apr 2017 11:02:37 +
"C. L. Martinez"  wrote:

> pkg_add -v python-2.7

There is no package called python-2.7. The package you want is called
python-2.7.13p0. You have a few options:

 1. pkg_add python, then select the version you want.
 2. pkg_add python-2.7.13p0
 3. pkg_add -z python-2.7 (fuzzy matching, see pkg_add(1))

Re: It is possible to start xenodm on Radeon HD 8730M (or Intel HD Graphics 4000) ?

[Steven McDonald]

On Sat, 8 Apr 2017 15:04:18 +0500
"dmitry.sensei"  wrote:

> "ATI Radeon HD 8730M" rev 0x00 at pci1 dev 0 function 0 not configured
> inteldrm0 at pci0 dev 2 function 0 "Intel HD Graphics 4000" rev 0x09
> drm0 at inteldrm0

Your Intel card is supported via DRM, but your Radeon one is not. The
problem is that Xorg is trying to use the Radeon card.

You can try creating the file /etc/X11/xorg.conf with the contents
below to force Xorg to use the Intel card.

Section "Device"
Identifier "default device"
Driver "intel"
EndSection

Section "Screen"
Identifier "default screen"
Device "default device"
EndSection

Re: perl fctnl woes

[Steven McDonald]

Hi Tilo,

On Tue, 11 Aug 2015 17:11:13 +0200
Tilo Stritzky lf...@gmx.de wrote:

 Am I doing something silly here?  Or is there a bug?
 I see the same result on i386 and amd64. Same for /dev/sound.
 This works on a 5.5 release, but not on later releases or current.

I've been able to reproduce this with a simple C program on amd64
-current. I've tracked it down to the recent audio(4) subsystem rewrite,
which removed support for the FIOASYNC ioctl(2) (used internally by
fcntl(2)). However, this implies it should work on 5.6 and 5.7. I don't
have any such systems with audio devices available to test; are you
sure you're seeing the same behaviour on those releases?

Below is a diff that fixes the problem for me. I'm not sure if the
EINVAL on set is needed, as other drivers which don't support FIOASYNC
(drm(4), hotplug(4), random(4) and systrace(4)) appear to just ignore
that ioctl completely. I'm sure someone more experienced than I will
know the answer to that.

 Incidentally, are there perl bindings for the native sound interface
 sio_open(3)?

I wrote some low-level Perl bindings a few weeks ago:

  https://github.com/stevenjm/perl-Audio-Sndio

They're not on CPAN yet because they're not finished, mainly because
I'm not yet sure if writing a higher-level interface in pure Perl or
extending the XS code is a better approach. The low-level bindings are
usable as is, and probably a better option than talking to /dev/audio
directly.

Index: audio.c
===
RCS file: /cvs/src/sys/dev/audio.c,v
retrieving revision 1.138
diff -u -p -r1.138 audio.c
--- audio.c 29 Jul 2015 21:13:32 -  1.138
+++ audio.c 13 Aug 2015 10:35:08 -
@@ -1542,6 +1542,11 @@ audio_ioctl(struct audio_softc *sc, unsi
case FIONBIO:
/* All handled in the upper FS layer. */
break;
+   case FIOASYNC:
+   /* No async mode, so set is an error, unset is a noop. */
+   if (*(int *)addr)
+   error = EINVAL;
+   break;
case AUDIO_PERROR:
mtx_enter(audio_lock);
*(int *)addr = sc-play.xrun / (sc-pchan * sc-bps);

Re: Dual Booting OpenBSD vs Windows7

[Steven McDonald]

On Mon, 29 Jun 2015 16:11:20 +0330
Mohammad BadieZadegan mbzade...@gmail.com wrote:

 When I choose My OpenBSD at boot state it display me:
 *Loading.*
 *ERR M*

ERR M means that the pbr successfully loaded a file into memory, but
it wasn't a valid ELF executable. Most probably, this means that the
pbr you're using refers to the wrong inode.

You said something about a second system; did you copy openbsd.pbr
from one system to another? installboot patches the pbr with the
location of /boot, so you need to create openbsd.pbr on the same system
you're using it on.

Perl bindings for libsndio

[Steven McDonald]

Hi misc@,

I've begun working on some Perl bindings for libsndio. The reason I
started was to be able to interface with sndiod's MIDI thru boxes
directly from Perl, but I've implemented XS bindings for the complete
libsndio API:

  https://github.com/stevenjm/perl-Audio-Sndio

It's still far from finished (I've put notes in the README indicating
what I'm still planning to do), but I've been able to listen to some
music with a Perl one-liner already. I've put it up in case it's
useful/interesting to anyone else. Also, this was my first time using
XS, so feedback would be welcome.

There's some minimal POD documentation inline here:

  
https://github.com/stevenjm/perl-Audio-Sndio/blob/master/ext/lib/Audio/Sndio/Bindings.pm

Thanks,
Steven.

Re: cp from 4 different home folders without overwriting files with different content

[Steven McDonald]

On Sun, 28 Jun 2015 17:39:18 -0500
Chris Bennett chrisbenn...@bennettconstruction.us wrote:

 But I don't want to overwrite any files with same name but different
 content.

You could try GNU cp (gcp in the coreutils package) with the -n option:

   -n, --no-clobber
  do not overwrite an existing file (overrides a previous -i
  option)

Re: Lock FVWM2 with nosuid mount option

[Steven McDonald]

On Sat, 23 May 2015 05:12:58 -0400
ertetlen barmok ertetlenbar...@safe-mail.net wrote:

 There is no solution for locking an OpenBSD Desktop while using FVWM2
 and nosuid mount option enabled for all mountpoints in /etc/fstab? 

xlock is setgid auth, so you could add your user to the auth group.
Whether or not this is a good idea is another matter, and it's likely
to be a worse one than allowing binaries to run setuid.

There are going to be many other things broken on a system with nosuid
used for all mounts. Why do you want / and /usr to be mounted nosuid?
You're already implicitly trusting the OpenBSD codebase by running
OpenBSD, so what's wrong with trusting OpenBSD's setuid executables?

Re: building userland failed with Unable to remove directory /./var/tmp: Directory not empty

[Steven McDonald]

On Sat, 14 Mar 2015 12:18:13 +0100
Harald Dunkel ha...@afaics.de wrote:

 pax: Unable to remove directory /./var/tmp: Directory not empty ***

/var/tmp was replaced with a symlink to /tmp between 5.6 and 5.7.
Compiling from source isn't a supported way to upgrade from one to the
other; this is well documented in the FAQ:

  http://www.openbsd.org/faq/faq5.html#BldBinary

Re: OpenBSD as base OS for Virtualization

[Steven McDonald]

On Sat, 14 Mar 2015 13:44:47 +0200
Ruslanas Gžibovskis rusla...@lpic.lt wrote:

 So question is:
 What Virtualization solutions OpenBSD support?

Anything that runs as a userspace process in non-accelerated mode and
has been ported to OpenBSD (QEMU, for example). There is no support for
Xen, nor hardware acceleration (VT-x or AMD-V).

 I know about jails 100% are available on OpenBSD.

Jails aren't available on OpenBSD. There is chroot; whether or not that
is sufficient for your needs depends on your goal.

 Maybe VirtualBox,  is it native support?
 Maybe bhyve? I know it's quite new, but it sounds cool. :)

VirtualBox might work; if by native support you mean using hardware
virtualisation, then no. bhyve is a FreeBSD (not OpenBSD) project.

 I am searching for something like Solaris Containers.

 My requirements, are to support WiFi host or forward my PCIexpress 1
 so VM. VM - be able to run Debian.

This seems to be contradictory. If you want to use OpenBSD to run a
completely different OS as a guest, you want something quite different
from Solaris Containers.

You should consider what you're trying to accomplish and then figure
out which tools serve that job best, rather than starting with the
notion of using virtualisation and trying to fit that to your goal.

 Sorry for bad English,

No problem, it wasn't all that bad. :)

Re: Broadcom NIC issues

[Steven McDonald]

On Fri, 13 Mar 2015 15:48:02 +
Miguel Barbosa Gonçalves m...@mbg.pt wrote:

 I recently installed OpenBSD 5.6 on an amd64 platform. Everything went
 smoothly. After installation, at the first boot, OpenBSD updated the
 firmware of some devices. I found this strange...

OpenBSD runs fw_update(8) on first boot. fw_update simply downloads
firmware packages and installs firmware onto the filesystem (not
directly into the devices that use it) for drivers that need to load it
at runtime. Linux has a similar firmware-loading mechanism, but it
typically ships the firmware embedded in the kernel.

 I had to reinstall this machine with CentOS and now the NIC does not
 work.

Did the machine work with CentOS previously? It seems extremely
unlikely that fw_update would be able to break Linux's use of the
hardware, since that firmware is loaded on every boot by the relevant
driver in both operating systems.

 I reinstalled OpenBSD again and it works. I tried to reflash the NIC's
 firmware and the installer does not recognize the NIC. At the moment,
 the machine only works with OpenBSD...

Some details as to specifically what you did and what failed, as well as
a dmesg, would be useful here. All I can say with the information given
is that, if your Broadcom NIC requires non-free firmware to be loaded by
the driver, the OpenBSD installer would not be able to use it because
it does not include non-free firmware.

If fw_update was able to run on first boot, though, it sounds like your
NIC is usable without firmware. Again, a dmesg would help (I'm not even
sure which of the three Broadcom NIC drivers in OpenBSD you're using).

goals.html: Clarify GPL policy

[Steven McDonald]

goals.html disagrees with policy.html as to the policy regarding
inclusion of GPL code. Since that part of policy.html was updated more
recently (2014 vs. 2003), here's a diff that makes goals.html more
agreeable:

Index: goals.html
===
RCS file: /cvs/www/goals.html,v
retrieving revision 1.80
diff -u -p -r1.80 goals.html
--- goals.html  2 Nov 2014 23:56:57 -   1.80
+++ goals.html  22 Feb 2015 07:54:25 -
@@ -30,8 +30,8 @@ and priorities, but it is possible to cl
 a href=http://cvsweb.openbsd.org/cgi-bin/cvsweb;directly on the web!/a
 p
 liIntegrate good code from any source with a href=policy.htmlacceptable
-copyright (ISC or Berkeley style preferred, GPL acceptable as a last
-recourse but not in the kernel, NDA never acceptable)/a.
+copyright (ISC or Berkeley style preferred, GPL not acceptable for new
+code, NDA never acceptable)/a.
 We want to make available source code that anyone can use for ANY
 PURPOSE, with no restrictions.
 strongWe strive to make our software robust and secure, and
 encourage

pfsync(4) typo (sychronisation)

[Steven McDonald]

Spotted a missing n in pfsync(4):

Index: share/man/man4/pfsync.4
===
RCS file: /cvs/src/share/man/man4/pfsync.4,v
retrieving revision 1.31
diff -u -p -r1.31 pfsync.4
--- share/man/man4/pfsync.4 29 Apr 2010 08:45:44 -
1.31 +++ share/man/man4/pfsync.41 Feb 2015 07:07:49 -
@@ -29,7 +29,7 @@
 .Os
 .Sh NAME
 .Nm pfsync
-.Nd packet filter state table sychronisation interface
+.Nd packet filter state table synchronisation interface
 .Sh SYNOPSIS
 .Cd pseudo-device pfsync
 .Sh DESCRIPTION

Re: *ERROR* radeon_cp: Failed to load firmware radeon-r300_cp

[Steven McDonald]

On Fri, 21 Nov 2014 06:30:16 +0100
John Smith hufflep...@bsdmail.com wrote:

 Is there anyway I can prevent this seemingly harmless error? I don't
 use X and didn't select the X packages during the installation
 process.

You can probably do it by disabling the drm and/or radeondrm driver with
config(8):

  http://www.openbsd.org/faq/faq5.html#config

I've not had cause to use config(8) in this way myself, so YMMV.

devtree: A utility for printing device trees

[Steven McDonald]

Hi misc,

I've written a small utility for pretty-printing a tree of system
devices based on dmesg(8) output. It's nothing fancy, but my apropos(1)
and web searches didn't bring up anything to do the job. I thought it
might be of interest to other newcomers to OpenBSD like myself who are
exploring how the system fits together, so I've put it up on the web:

  http://www.sjm.so/projects/openbsd_devtree.shtml

It also has the potential to provide an easy way to diff the hardware
in two different systems, or on the same system running two different
OpenBSD versions, since it sorts nodes alphabetically when printing
them.

I'm not sure if it'd be worth making a port for it, given that it's a
single Perl file + man page, but if there's enough interest I'd be
happy to try my hand at that.

Re: weird ksh '~' substitution - ksh bug?

[Steven McDonald]

On Fri, 16 Dec 2011 09:47:54 +0100
LEVAI Daniel l...@ecentrum.hu wrote:

 $ cd ~/stuff/
 ksh: cd: /home/daniell/~/stuff - No such file or directory
 
[snip]
 It seems  ~  is substituted in the first case

Just a correction here: I don't think it's being substituted. Rather,
the lack of a leading / is causing it to be interpreted as a relative
path, and you are already working in /home/daniell/.

As for your actual question, I'll wait for some of the more experienced
OpenBSDers to come along with an answer and go back to lurking the
list. :)