Run your test instruments on Windows? No thanks. / General Chatting / Forums

Forums

4hv.org :: Forums :: General Chatting

« Previous topic | Next topic »

Run your test instruments on Windows? No thanks.

previous 1 2 3 next

Move Thread

LAN_403

2Spoons

Thu Nov 10 2011, 04:34AM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

That's a valuable skill you are learning there! Make sure you put it in your resume.

Patrick

Thu Nov 10 2011, 05:01AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Is this all ill get to do with my EE degree ? Fix crap devices that should have been made right the first time?

Steve Conner

Thu Nov 10 2011, 07:57AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Well, this all makes me happy that I didn't use XP embedded.

I saw a LeCroy scope that ran Windows 2000 and decided to start asking for a login user name and password! But it didn't have a keyboard to enter them, even if anybody in the lab had known what they were.

Our instruments have the registry, indeed all of the OS files, read-only. No matter how bad you screw them up, you can always get back to a known good state by cycling the power.

Carbon_Rod

Thu Nov 10 2011, 09:34AM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

@2Spoons
Usually, modern exploits use a multi-vector approach. It is quite common to have reinfection routines given that windows update is usually a trusted process by most firewalls, and payloads will evolve over time to better suit the targeted machine.

@Patrick
You may find your "EE degree" will get you pigeonholed into a maintenance career if you are not careful.

@Steve
Unfortunately, a "read only partition" is usually just a suggestion to the hardware.
Typically write protected storage is the only way to ensure a power cycle will purge the problem code. Yet in true exotic fashion, some modern boards are vulnerable to having their security-module or GPU re-purposed to bypass the storage devices and hide from the OS.

The new secure boot model Microsoft has suggested is actually very old (1990s IBM), and has some inherent design problems that could make secure recovery impossible. The EFF group is actually making a very important observation about open hardware.

Steve Conner

Thu Nov 10 2011, 10:30AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Carbon_Rod: The OS is stored as an image in flash ROM that isn't even accessible to the OS, as it's not mounted and not in a format the OS has a filesystem driver for anyway. When you turn the power on, the bootloader copies it into RAM and it executes from there.

It needs to do that anyway because the flash isn't fast enough to run code out of, but it makes it pretty bombproof as a side effect. Every time you turn the thing on you get a factory fresh OS "install".

You can't be careful enough where Microsoft junk is concerned!

Dr. Slack

Fri Nov 11 2011, 08:22AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

The time to market logic behind having a standard operating system on an instrument, at least on a high-end complex instrument, is inescapable. Why that OS has to be windows rather than Linux is a different argument. Just as you wouldn't write in assembly code but compiler, and wouldn't place FPGA gates by hand but would use VHDL (or something higher these days), you wouldn't write your own OS, comms stack, etc etc for an instrument.

We are wall to wall windows equipped in the labs, and we are just introducing several new embedded windows based products. What it means is I can run a hardware module on my bench with exactly the same drivers that are in the instrument, and then write my own debug code to use the same API, maybe from Matlab or Python, whichever is more approrpiate. When an instrument plays up, I can run debug tools on it from my stick. If I'm home and get a call, I can remote desktop into any instrument anywhere in the world. When I'm working on the instrument, I can save results to a printer, or a company NAS box.

There's an argument that says you get these extra facilities for "free" when you put a standasrd OS on the customer box. That's true, as long as you define free as ff'ing expensive, restrictive, flaky and slow. We spend a lot of our time trying to work around the limitations of the OS. Yes sir, you do have to wait three minutes before it's ready, and that's just the easy one to explain.

Why windoze rather than 'nux? I guess that just as nobody got fired for buying IBM, an IT manager that undertook to maintain a linux facility would be taking a huge gamble. If a new windows version causes problems, he can blame microshaft. If a new linux distro causes problems, his reputation is on the line, he has got to fix it. Arguments about which OS is more likely to drop him in the brown stuff are moot, we are talking career insurance here.

Windows on isntruments is a logical evil that was inevitable, like derivative financial products.

Steve Conner

Fri Nov 11 2011, 08:44AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

When we were designing our instrument platform, we spent a good deal of time arguing over whether we should use Linux or WinCE. What did it in the end was, the Linux distro for the single board computer module we intended to use had no window manager, it was console only. The vendor waved their hands and said it would be easy to bolt on the window manager from the Angstrom project or whatever. I never believe a salesman who tells me that, if it was easy why didn't they do it themselves?

The situation was similar for other vendors of SBCs based on the XScale chip that we'd identified as the best match for our project.

CE on the other hand has a graphics and windowing API pretty much identical to desktop Windows. And when you plug it into a desktop Windows PC it pops up as "My Mobile Device", which is very useful. You can drag and drop the results files to your desktop, or if you're a developer, run Visual Studio and debug the code on the instrument. (I did provide another way to get results out of the thing, in case we ever come across a user who doesn't have Windows.)

You do have to pay a licence fee, but it's Â£4 per unit.

Carbon_Rod

Fri Nov 11 2011, 08:29PM

Registered Member #65 Joined: Thu Feb 09 2006, 06:43AM
Location:
Posts: 1155

@Steve Yep, having to write the GUI to support cross platform GTK/Qt code can be a task.
However, I typically use the VNC style loopback protocol to dump a full JPG of the "headless" virtual desktop session, rescale it to match the output dimensions, and drop it strait into the screen's buffer. It took about 2 weeks to implement, and works on just about any SPI based screen (no kernel driver updates needed).

The "pushed" refresh rates are good enough to play Doom...
I would like to see a GPL version of the adapter daemon for LCD/LED/E-Ink at some point.

Sales people usually rip-off whatever they see as a market, and generally never show anything useful.

Cheers,
=)

Conundrum

Sat Nov 12 2011, 07:34AM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

Having wasted many an hour on the monstrosity known as Windows, I can sympathize.

Microsoft may be the world leaders in desktop software, but they really need to work on customer support.
Crashing the OS because someone didn't check for a buffer overflow is IMHO unacceptable.

-A

Pinky's Brain

Sat Nov 12 2011, 10:03AM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

Win Ce is a defensible choice, Windows proper? No way, you're just introducing way too many additional failure modes. For what? Slightly easier development? These are boutique items, you don't go cheap on components ... don't go cheap on development either. Management might forgive you, but as a customer I'd curse you after the first blue screen/virus/etc.

previous 1 2 3 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob