The collected works of robespierre - Page 13

jan-jaap wrote:

Code: Select all

dvhtool -d <EFS image filename> --print-partitions | grep "Part# 10" | cut -d ',' -f3 | awk '{ print $2*512 }'

I was going to respond with a portable version, but there is no portable 'od' or 'hexdump' with correct handling of other-endianness. The best I found was 'xxd' (always uses big-endian), which is not part of POSIX.

I also found that the "O2 Demos 1.1.1 for IRIX 6.3 Including R10000.img" and "Cosmo Software May 1996 for IRIX 6.2, 6.3, 6.4 and 6.5.img" were too short.

(HD)SDI are less robust than analog CVBS because they are digital, without any error correction. It's even less practical for a fixed installation like surveillance that has to work unattended. And coaxial is expensive and difficult to pull.

pentium wrote: with some form of a PDS slot.

The only information on the web is that it is a factory test connector.
On the face of it that's a bit odd, as Apple usually used card-edge fingers for that purpose on all of their hardware. Why they would need 96 pins for burn-in is a mystery.

PRMan was available on Irix for a long time, but I'm not surprised that it's no longer sold. The interface and REYES rendering was cloned by several third-party tools so you could use pretty much any of them the same way.
The first was Blue Moon Render Tools (BMRT). Its author then formed a company to sell an improved version called Entropy. They both were served with lawsuits from Pixar and were discontinued, but the Internet is forever and source code is easily found.
The other clones were Render Dot C, which was a commercial product but can be found as source and Irix binary, and I think there were a few others.
The latest free-software REYES renderers are Pixie and Aqsis.

There are several Maya users on this forum who should be able to tell you what's missing in Maya 6.5, but remember that it is very old. But even by then the main platform for A|W was Windows, so performance on Irix is not that good. Same problem with Softimage XSI.

The 3d tools with the best performance in Irix are Lightwave 5.6, Softimage 3.9 or 4.0, Houdini 5 or 6, Alias Studio Tools 9, Mirai 1.2, and there are many others.

All of the LaserWriter II printers use a Canon SX engine, which is capable of 300 dpi. There are no printers based on this engine that can print in true 600 dpi because the marking engine can't do it: it scans its laser in 300 lines per inch. The IIf and IIg boards have an enhancement process called FinePrint that tries to modulate the laser power to reduce aliasing on outlines, similar to the LaserJet III's enhancement feature called RET. There were other SX printers that rasterized at 600 dpi and then converted to a lower resolution on output, this was yet another way to improve the 300 dpi print quality.

Apple used the Am29000 in its printers too, starting with the Personal LaserWriter NTR and many of the subsequent models.

SGI-branded CRTs are probably your only option for stereoscopic use on SGI workstations. 120 Hz LCDs exist now, but their timing is different.

I have one, but it is currently in use and can't be taken apart.
I believe it is stored in the 28F008, but this might be wrong.

If all you want is to change the MAC, this can be done from software by copying the driver and editing it. Then you force the real driver to unload and load the copy.

Then I think you need to write to the flash. In these machines the flash is writeable if the CPU maps its physical region. So if you open /dev/mem at 0xFFF0,0000 and dump it to a file, and search for the MAC address (which will be "prop-encoded"), you could change it by writing a different address back (also "prop-encoded"). The property is readable from Open Firmware at
dev /enet .properties
called "local-mac-address". I don't know how to set this from Open Firmware, it may not be easy to do.

It goes without saying that writing to the flash can brick the machine. There is also an area of a 28F008 that is treated as NVRAM and it might be in that area, although it is not in any of the exposed nvram variables.

Another idea is that if you wrote FCode that set the MAC address at boot time, it might or might not be honored by the system (it depends on order of execution of the boot options). You would put the FCode to do this in the nvram variable "nvramrc". It would be permanent in the sense of always being run before loading the OS, but without changing the firmware directly.

You would do something similar to (check this):

or perhaps this: [fixed a bug]

I don't quite understand what "bare metal provisioning" is, but SGI had a product called RoboInst that automated a large part of the install process.

You could automate the earliest stage of installation by using a terminal server connected to each machine's console port.

I don't think any of the SGI discs are multimode (except for Octane Soundtrack). So you don't need cues, a regular ISO image is enough.
Toast .image files (or Nero .NRG) is just an ISO with a 512 byte trailer.

The LaserWriter formatter boards are designed to be debugged using the same tools as Macintosh models, such as the TechStep. They have a "sad mac" diagnostic mode that can be triggered using similar external signals.
https://mac68k.info/wiki/display/mac68k/Diagnostic+Mode

At one time I was going to fish some ST-ST multimode through the wall and use it to connect my two concentrators into a dual loop.
So I acquired some passive ST-FSD couplers because my WS-C1400 use FSD cables (apparently they also can use SC port cards but I've never seen them). The couplers were very hard to source, maybe not available any more.

There are 4 possible ways to install two VPros in an XIO carrier. The fastfan logic doesn't know that they are wide cards, it only knows which XIO slot they connect to. Starting in the upper left and reading clockwise, they are QA, QB, QC, and QD: it's a card connected to QC that activates fastfan. So when the bottom card is facing one way it will be recognized in slot QD; in the other direction it is in slot QC and turns on fastfan.

A converter for Suns is here: http://kentie.net/article/sunkbd/
The SGI interface is similar (dual asynchronous serial), but uses higher voltages.

edit: actually it works the wrong way round.

OK, I was able to restart and test whether this works:
Short answer: it doesn't.
The escaped double-quotes in the version supra cause problems with csh (it complains they don't balance). With ksh they work ok. You can force them to work in tcsh with the backslash_quote shell variable. But using single-quotes is a cleaner solution (but see infra ).
Setting the " local-mac-address" property in the enet package works from the Open Firmware prompt, but vagaries of the probing process overwrite the property if done from nvramrc.
You can get an idea of the process from Oracle docs.
http://docs.oracle.com/cd/E19620-01/805 ... index.html
(although banner and suppress-banner don't seem to be defined in Apple's FCode, so there appears to be no way to override the probe steps)

I still couldn't get it to work from nvramrc, so I wrote a test:

If you drop to Open Firmware console by holding Command-Option-O-F during the boot chime, you can see that the nvramrc has already run:

But there was no property to use then:

If we run it again now, the property is there:


The problem is that when nvramrc runs, there is no mac-address property yet, because the enet device hasn't been probed. If we probe it ourselves the device tree gets messed up when OF does it again (tried that, OSX hangs on boot). What we need is to force some code to be run later, after probing is complete (when the console would be able to do it interactively).
The way that you continue booting OSX from the OF console is typing mac-boot. You can see what it does, funny enough, with the see word:

A clue! interesting. We can change what happens when mac-boot runs since it is dynamic. It runs after all the probing and device init is done, after the console exits. But we still want that original code to run when we are finished for the OS to load.

(Notice the switch from single-quoting to double-quoting for the shell's sake.)
I tested this and it works.

ClassicHasClass wrote: Nicely done!

Thanks, it was fun (and a dozen restarts). Apple OpenFirmware docs aren't much use, but Sun's OpenBoot 3 docs are very helpful.

C and C++ don't allow access to carry bits at all, which is inconvenient. There are some C compilers with extensions that allow for overflow to be trapped, in case the architecture detects overflow, which is not universal. ( http://clang.llvm.org/docs/LanguageExte ... c-builtins )
What most people do is use a pre-existing library like GMP with machine-dependent assembler files.

It could be argued that much of mathematics is not particularly important for programmers, although I am frequently amazed at how little number theory many of them know. For instance, a problem like "calculate the sum of the multiples of either 3 or 5 less than 1000" will always be written as a loop (or, if you're really smart, a list comprehension), even though that quantity has a closed form. One programmer was very proud to have written a solution in terms of a Horn Clause solver; it can be calculated rather easier on the back of a cocktail napkin.

This is the very same function as the GHOST vulnerability from last May.

Aren't all USB keyboards crap by design though? the HID protocol doesn't even allow N-key rollover.

I don't think that just any XIO card can be installed reversed; the design of the Octane quad XIO bay is that the original (MGRAS) carrier frame can have XIO cards installed on it in four positions, two on each side. So the XIO connectors are reflected on the left side vis a vis the right. In that setup, any single-slot XIO card is recognized in the same slot as its position in the carrier frame (which is labeled A through D on the actuator). The double-slot cards like MXI have their XIO connector on the right, so they appear in either QA or QC (the latter forcing fastfan).
VPro have double-slot cards that are fat instead of wide, so they can be installed in more different ways: by turning the card 180° it moves its connector from one slot to the mirror-image beside it. This is only true for VPro cards in a VPro carrier.

"Invalid argument" repeated dozens of times isn't clue enough?
/dev/dsk/dks0d1vol is a cooked device, of no use for performance testing. You should be using the raw device (as the man page describes)

Irix 6.3 was the version released with the O2 in 1996. Not sure what to make of purple flashes, it sounds like the display cable could be bad or your monitor isn't compatible.
The unit turning itself on and the power button not working are common problems on the O2 (and some other SGI machines). The cause is the automatic boot circuitry that takes its input from the real-time clock. You can tell it to turn itself on at a later time, then shut down the machine: when the time arrives, it turns back on again. I don't know that there is any way to fix this particular problem, but there are certain things to try if it really bothers you.

It looks like your entire SCSI bus is borked. (errors on ID 5, 6, 7, and 1, and likely all the others scrolled off screen). This can happen if the cable is plugged in backwards. (Have you checked that your 50 to 68 pin adapter isn't backwards?)

SPARC has fixed-size windows (8 in, 8 local, and 8 out), but some other archs have variable window size, like Am29000.
Also interesting is the Mill architecture, which uses an operand "belt" of most recent instruction results, and can select any 16 to drop into the callee, which is similar to an arbitrary register window.

josehill wrote: I'm ever so slightly freaked out that your bathroom has the exact same style of tub, vanity, and fixtures as my childhood home once had.

Now that you mention it, they do look very familiar. I think they were common for houses built in the early '70s.

I hope you aren't spending hours setting up your workspace as the root user? There's no point! Simply create a new account and will have all the defaults.

If they really do chkconfig desktop off, then it will prevent any users from having the Indigo Magic desktop stuff unless they have written custom init files to run those services. The less fascist way to do this for any individual user is to touch .disableDesktop in their home directory. So what I would probably do is to (as Titox said) chkconfig desktop on, then touch .disableDesktop in the Flame user's directory (which disables it for just him), and create a new user for general (non Flame) usage.
The exact interactions of the init scripts are detailed here (for the old Irix 5.3): https://www.talisman.org/unix/hack-x-on-sgi.shtml

Logging in as root is just lazy and can cause a lot of problems. With 6.5 you can use sudo, which I prefer to su because you keep your own shell. Shell sectarianism is very strong.

The clear plastic piece is part of the CPU module, and you should leave it alone for now.
Wrangling PCI cards can be a tight fit, but the card simply lifts out horizontally straight out of the PCI connector. If there is a protruding connector on the I/O panel you give it a slight twist. The PCI riser assembly (is that what you mean?) lifts straight out of the IP32 mainboard once its single retaining screw is removed.

Recent versions (like, the last 15 years) of Illustrator no longer use Postscript document format, they have some XML crap. You need to Export as an Illustrator 5 or 6 file from the new version.

Also, it's worth saying that EPS is mainly used for embedding logos and things, not as a format that allows editing. You can sometimes open EPS and edit it, but it doesn't always work. Originally it had no layers groups etc. Not totally sure if that was changed in later versions.

Hi, nice to see someone else taking an interest in the O2 hardware.
The yellow wires are the main logic supply, where most of the power is. Unlike, say, an Intel box, the O2 regulates its logic supply in the PSU and does not have any point-of-load regulators. If they are as high as you measured, then that supply circuit has failed (most likely from a shorted pass transistor, but there are other failure modes in a switcher you need an oscilloscope to find). Good luck.

An OVP circuit would need to contain a pretty beefy SCR, or Transzorb type varistor component. I never saw one but they could be hiding somewhere. The LED is also controlled by logic, not directly on the rail (reflecting its 3 color states), so it could stay off due to failed logic.

I think it's an aftermarket CDROM drive, they shipped with Toshiba XM-3501B units. The Archive CTD-8000 DDS2 were stock.
Let the forum know if you have success with audio DAT transfers. I found that the DATplayer software really liked to hang.

chkconfig mediad on?

Indigo and Indy need different kernels and sysgen files. I think that different drivers are also needed for Starter Indigo vs GR2 (XS, XZ, Elan).

While researching keyboards at deskthority.net, I learned that PBT was used for the key caps, while ABS was used for the space bar (for its greater toughness). ABS is the material for computer cases and suffers the most from bromine yellowing.