Are there operating systems or versions of an OS that are known as stable and secure such that you can put it on and not worry about updating it once it's configured?
that sounds like a good idea and I've seen a lot of linux spinoffs that are lightweight and secure like snappy, but is it safe forever even if I have it sitting on a network that could potentially be accessed by outsiders?
@EwokNightmares Would there be a service exposed to the network?
like an ssh server, or something like that?
If there isn't, your attack surface is basically zero
If you even take a current minimal distribution, turn off all services and block all ports except whatever custom software you're running, you can be pretty assured you'll be safe. However, I'm not a security expert, you'd probably want to ask this question on security.stackexchange
Potentially, but these are ideally data collection computers that collect off hardware and then transmit that data to a central server/database periodically
@Giskard42 Thanks for this advice. It's interesting to think of attack surfaces and such.
Eh, well the data on them have critical importance legally in case of loss of life, but the system itself should not have any role in safety/human interaction.
The worst that would happen is someone gets into the computer and replaces/interferes or destroys the data. The second worst thing that could happen is someone gets on and steals the data, but that isn't as big a deal.
Hmm. With an up-to-date linux distro, ssh key login, and the typical hardening stuff, I think you'll be fine, but if the data is important to the business I'd almost want to ask a security expert
However, can you set up your system such that even if someone were to break into the computers no data could be tampered with?
@Giskard42 No I haven't. The encryption is something I didn't consider but would be great if the OS can take care of that for me. I will have to put this past security experts, so I just wanted a good first shot on my own.
If whatever you're collecting data from can send the data pre-encrypted, then you don't have to worry about how low your attack surface is at all
Any tampering would be apparent and reversible
But I guess that doesn't help you against destroying the data
If the "central server/database periodically" part could be changed to sending constantly, then you can make your life a bit easier by putting the security requirement on the presumably-secure central server
I don't know if that's possible, just throwing out ideas
@Giskard42 The idea is that the data is stored locally for some months as a backup, but it is continuously transmitted to a server as well. Maybe the local data could be encrypted too.
@Giskard42 Good point. I haven't considered what the server/database should be but I think that thing can be handed over to security experts. I am looking at raspberry pis as cheap data collection nodes and so I am worried about the security and reliability on those for now. I might even use redundant computers at these locations.
@EwokNightmares If you're using Rpis and it's not a huge amount of data, first thing that comes to mind would be to use two Rpis, have one run normally connected to the data system and the network, and another that
@Giskard42 Hopefully I won't have to worry about that part. The real problem in hardware I worry about is that the reliability requirement (constant up time and data integrity) is very high, without any consideration of malicious attacks.
One of these would operate like a multi channel scalar if you heard of those. You could have 20 detectors putting out pulses and then every second, an array of 20 buckets is created with a time stamp, so even less than half a kb /sec even.
Yeah, so even just an Atmega directly connected to the detectors, logging everything live to an SD card.
Our product has similar requirements; if our system turns off a pump for about 5 minutes, $40000 in damage occours
Internet connected, logging data to a central server
We have a cheap ATtiny watchdog on the board that just constantly monitors what the other chips are doing, and if anything goes awry, it turns on the pump
@Giskard42 It seems so simple. I thought I would need to have a vxworks real time system or a national instruments compactRIO to meet reliability "requirements' although none exist formally.
@Giskard42 That's pretty funny. I wonder if they would complain if it was an Arduino board?
Yeah, I'm not going to pretend I'm a better engineer than the people who make those things, but at the same time they don't worry about a lot of stuff I might need to be worrying about like EMC
I don't see a reason why they would need a precise clock..
anyways, surely on broken ones the oscillator won't startup ;) but if you link a datasheet maybe someone can point you to the thing that is necessary for it to happen that you forgot
Yeah, if I step through the code with JTAG clock, it powers up fine, waits the required time, then switches to fast RC clock, then waits forever for main clock to be ready
SAM4S series (always specify the entire series at least!) to my knowledge doesn't have the PLL Conf Erratum, which is unlikely on SAM4 to begin with, so can't be that
You can ask @jippie himself. But presumably, @BartekBanachewicz, he means that the one cannot directly derive from the other in an unknown way, which may introduce an offset between the two ground points which then gets shorted out.
Of course that offset will also invalidate the transistor operation
@BartekBanachewicz If they relate to each other and there's an offset, say 1V, one ground will be at 1V with respect to the other, thus connecting their "grounds" will short out that 1V and create problems, so you can't connect them. If they are either already guaranteed to be same ground (One battery with a high-side regulator of some sort, for example) or they are 100% floating, the connection can be made without risk and in the case of floating it has to for the return path
Installer for scanner software, specifically, only scanners: "Only run this option if your printer isn't working properly". Yes. It's not working properly. It only scans. I thought it was just a scanner, but now I find out it's actually a printer, so, no it is not at all working properly. Never was.
Hey guys, I've been tasked from my lab supervisor with finding a programmer to set up a basic BT or BLE library for arduino for the ESP-WROOM-32. I know someone who is qualified but he asked for a list of specifications and I dont think either of us knows enough about bluetooth to quite make that.
Basically if we could send json packets to and from our esp32 board (partnered with a phone), that we should be able to basically do anything we need. Translating that to what needs to happen on the BT side is an issue.
