Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Any progress on fixing the openpty() issue with sysinst?

15 views
Skip to first unread message

Adrian Christiansen

unread,
Jul 27, 2015, 12:33:30 PM7/27/15
to
Hi guys!,

Was so long since I installer 6.1.4 on my SPARCstation IPX that I had
forgotten all about this. Thought it was me that didn't understand the
disklabels on SPARC systems, but when I tried 7.0_RC1 I found out
about PR install/46812. That was for 6.0_BETA2.

Maybe it is a recent regression and I'm confusing it with some other
issue I had with 6.1.4. But I'm having that issue now so I wanted to
hear about it.

Cheers!,
Adrian.

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-...@muc.de

Martin Husemann

unread,
Jul 27, 2015, 12:59:05 PM7/27/15
to
On Mon, Jul 27, 2015 at 06:33:19PM +0200, Adrian Christiansen wrote:
> disklabels on SPARC systems, but when I tried 7.0_RC1 I found out
> about PR install/46812. That was for 6.0_BETA2.

That PR is closed, so it should be all fixed - if you see it somehow,
we need more details from your failure (like exact install media you
use and when it does happen).

Martin

Adrian Christiansen

unread,
Jul 27, 2015, 1:37:17 PM7/27/15
to
Ah, I see. Saw the comment about running ./MAKEDEV all in INSTALL.txt just now.

Install process:
RARP with boot.net loaded over tftp.
Kernel from NFS and filsystem being bootfs.tgz extracted with tar xzpf
into shared directory.

The openpty() error is showed at every disklabel operation in sysinst,
so they fail. Doing it by hand works fine. sysinst can't be used to
install system even with a prepared disk.

So I suspect I both missed that the bug was closed and that I was
supposed to makedev? :/

If so I'm sorry for this.

Martin Husemann

unread,
Jul 27, 2015, 1:43:48 PM7/27/15
to
On Mon, Jul 27, 2015 at 07:37:02PM +0200, Adrian Christiansen wrote:
> Install process:
> RARP with boot.net loaded over tftp.
> Kernel from NFS and filsystem being bootfs.tgz extracted with tar xzpf
> into shared directory.

The kernel is GENERIC?

This should work both with and without "MAKEDEV all" on the NFS server IMHO -
I'll have a look.

Adrian Christiansen

unread,
Jul 27, 2015, 2:00:53 PM7/27/15
to
Yes. This is GENERIC from binary/kernel/netbsd-GENERIC.gz. Recreated
the steps with the latest 7.0_RC2 snapshot from nyftp. When selection
upgrade in the sysinst menu, I get these two error messages in the
status bar:

openpty() failed
sh -c 'cat /tmp/disktab.* > /tmp/disktab'

openpty() failed
/sbin/fsck_ffs -p -q /dev/rsd0a

It then prompts me about being unable to determine filsystem type and
if I want to go ahead and try mounting the drive anyway, which also
fails:

openpty() failed
/sbin/mount /dev/sd0a /targetroot

I can't ./MAKEDEV all from the installation, there's no MAKEDEV in /dev?

2015-07-27 19:43 GMT+02:00 Martin Husemann <mar...@duskware.de>:
> On Mon, Jul 27, 2015 at 07:37:02PM +0200, Adrian Christiansen wrote:
>> Install process:
>> RARP with boot.net loaded over tftp.
>> Kernel from NFS and filsystem being bootfs.tgz extracted with tar xzpf
>> into shared directory.
>
> The kernel is GENERIC?
>
> This should work both with and without "MAKEDEV all" on the NFS server IMHO -
> I'll have a look.
>

Martin Husemann

unread,
Jul 28, 2015, 2:51:10 PM7/28/15
to
I am not sure what goes wrong, the instfs.tgz contains an already populated
dev/ wich contains the needed pty files:

crw-rw-rw- 1 root wheel 21, 0 Jul 28 13:43 ptyp0
crw-rw-rw- 1 root wheel 21, 1 Jul 28 13:43 ptyp1

In this variant no /dev tmpfs will be created and no MAKEDEV run is needed.

Could you please boot again, but go to S)hell instead of installer and
then do:

ls -l /dev/pty*

and compare to the above output?

Thanks,

Adrian Christiansen

unread,
Jul 28, 2015, 4:34:52 PM7/28/15
to
Hi Martin!,

I found out what I did wrong. So not only did I read a closed PR, I
also setup the nfs-server wrong, so root wasn't mapped to root on the
server... what I'm saying is that I've wasted your time by not
checking if I can actually write to the remote filesystem.

This is the output using the instfs.tgz (this isn't mentioned in INSTALL.html?)
# ls -l /dev/pty*
crw-rw-rw- 1 root wheel 21, 0 Jul 28 11:43 /dev/ptyp0
crw-rw-rw- 1 root wheel 21, 1 Jul 28 11:43 /dev/ptyp1

This is the rootfs.tgz
# ls -l /dev/pty*
crw-rw-rw- 1 root wheel 21, 0 Jul 28 11:43 /dev/ptyp0
crw-rw-rw- 1 root wheel 21, 1 Jul 28 11:43 /dev/ptyp1

Once again I'm really sorry for wasting your time with this stupid
mistake of mine.


2015-07-28 20:50 GMT+02:00 Martin Husemann <mar...@duskware.de>:
> I am not sure what goes wrong, the instfs.tgz contains an already populated
> dev/ wich contains the needed pty files:
>
> crw-rw-rw- 1 root wheel 21, 0 Jul 28 13:43 ptyp0
> crw-rw-rw- 1 root wheel 21, 1 Jul 28 13:43 ptyp1
>
> In this variant no /dev tmpfs will be created and no MAKEDEV run is needed.
>
> Could you please boot again, but go to S)hell instead of installer and
> then do:
>
> ls -l /dev/pty*
>
> and compare to the above output?
>
> Thanks,
>
0 new messages