Miscellaneous Operating Systems/Hardware

Hey, the only way I know about these releases is when I see them here. When Nekochan gets a new subsystem, that means that 2 1/2 hours ago a new version came out

Don't forgot to put it in a closed room with an armed guard in the front

And NO USB !!!

Guess SGI has got that one covered....

Why has everyone been complaining about the security features of IRIX?

Irix at some point had a reputation for not being that secure, I think it had to do with the fact that a ton of stuff was open in the default install, whether that reputation still stands today is a bit moot since the system is pretty much EOL for all intents and purposes and the damage in "mindshare" was already done.

Also early releases of Irix were awful, and that also affected the perception of the OS. A shame since it ended up being a nice Unix variant.

Compared to contemporary UNIXes I don't think the 4D1-3.x or 4.0.5 releases were very bad. SunOS 4 might have been a bit better, but 3.x wasn't too bad. Early IRIX 5 releases were, though. I recall one bug (I think it was in Objectserver) that was particularly pernicious and compromised security.

SGI was building systems for people who wanted to throw them on LANs and not worry about them, so they were pretty much wide open. Unfortunately this translated into them keeping things open for a long time after they branched into HPC/servers so they didn't get too many disgruntled "I want to plug it in and have it work instead of having to worry about all this password junk" types.

True.

However, the perception did a lot of damage. I remember some sysadmins being completely hostile to the idea of having Irix machines into the network of the school I was at the time. Probably due more to FUD than reality, but I assume such a negative view must have hurt sales of SGI systems, esp. when the internet was taking off.

Anyhow, all major operating systems seem to have had growing pains at some point in their development. Early releases of Irix 5 were utter sh*t, and same goes for the first releases of Solaris.

Only moron, crazy-minded or envious will be intreresting in hacking not a production box from outside.
Or did someone put his eye on nekochan already?
If you'll stop and pull out all useless services like time and e.t.c., and keep only really using (like ftpd, sshd, httpd, e.t.c.) and better to use stable versions with security patches applied services, well then you may say that the box is not just critical hole from outside view.
Of course if you get ssh accounts to anyone from your friends, wait presents But in this war case almost all UNIXes are under disgrace if there are just installed out from a box.

And don't forget the Faraday cage.

Just turn it off, pull the cables, and weld some steel plate over the back just in case.

I remember when my college moved from SunOS 4.1.4 to Solaris 2.can't_remember. I think it was 2.6 (1997-1998 year). First halfway decent Solaris 2 release, and I hated it.

Solaris has gotten better, and I've gotten more used to it, but 5 years after first release to get a decent product is pretty bad.

I try I just got 5.3.0 up and running today along with a SVN pull of xcache-1.3. So far so good - I'll move the archive over to the usual place.

NetBSD has support for IP6, IP10, IP12, IP20, IP22 and IP32. OpenBSD supports more of newer machines, Linux has some IP26 and IP28 (along with IP22, IP32 and IP30?) support, and Russ Cox's Plan9 sources contain SGI Power code. Graphics aside, the opensource world supports most of the SGI machines already. The graphics being the most interesting part in most cases, the allure of IRIX will remain unparalleled, but for the server-class or headless machines, there's plenty you can still do with them -- even a 4D/20 will run the NetBSD-current

It would be nice to see this all consolidated a bit, though. Maybe someday the OpenBSD guys will import the support NetBSD has for the older machines, or the other way around.

When I spoke to Art Grabowski years ago, picking up my Indy from him, he told me someone in .se had code for the Indy it in his attic... They are kinda lazy with pushing the code to the trunk the OBSD'ers .

Any more current comments on the status of OpenBSD/sgi?

The OpenBSD port page says things are working fine (including X11), while FAQ page on X configuration mentions for sgi "no X servers, only X clients." (Apr 09 version).

Comments? You found the wrong topic: viewtopic.php?f=15&t=16721761&p=7307390&

That's clearly an error in the FAQ. Framebuffer support is currently still limited to the O2 machines. However there is promising work behind the scene to support VPro boards on Octane and Fuel systems, which will be commited soon.

As for non-framebuffer hardware support, Origin 350 are now working since a few hours , and there is ongoing SMP work targeting the Octane and Origin 200 systems.

O350 "working" meaning what?

Is PCI probing supported? Does NUMA work? Linux has been "working" on O350 for quite a long time now - in a single-CPU configuration with no PCI support (and therefore no way to access any storage - making it completely and totally useless).

Hey, bri, don't be so down on it; at least you can see how fast the delay loop is running! Unless they haven't figure out how to bring up the serial ports either...

Meaning that an Origin 350 is now used to build the OpenBSD/sgi snapshots.


Yes. Which also means the on-board SCSI and gigabit Ethernet controllers are recognized and working.


Multiple-o350 configurations could not be tested for lack of hardware (the only multiple-node system which has been tested, to the best of my knowledge, is a dual Origin 200 system).

I'm going to have to check this out. Thanks!