Random Tech Questions you want answered.

I used a G and later a GL for years, and loved both. They were great hardware in 2003.

In 2021, not so much.

dejanzie wrote:

Thanks for the advice guys, I gave the advice to our friends to look for a real extender instead. I am going to hold on to the WRT54GL, for old times' sake :-)

I should have read your original post better. If you just need range extending I recommend the TP-Link extenders. I have a couple of the AC750 models operating in bridge mode (one for my wifi less PC and one for my sh*tty Samsung HDTV that had to be power cycled multiple times per day sometimes to get the wifi in it to work).

I can’t stress enough that keeping that kind of ancient hardware around is just a security disaster waiting to happen. The last official release of the software, along with the hardware, has *579* known security issues, each of which will have an exploit built into every hacking toolkit known to science. And the open source stuff, last I could find was from 2016, so it might only have a hundred or so known flaws. Still just absolutely crazy to give it space in your house, much less use it...

Robear wrote:

I can’t stress enough that keeping that kind of ancient hardware around is just a security disaster waiting to happen. The last official release of the software, along with the hardware, has *579* known security issues, each of which will have an exploit built into every hacking toolkit known to science. And the open source stuff, last I could find was from 2016, so it might only have a hundred or so known flaws. Still just absolutely crazy to give it space in your house, much less use it...

Depends on the specific situation. My parents are still using one (with DD-WRT) as a wired->wifi range extender. There's a much more modern router/firewall between it and the outside world, which should limit the attack possibilities to just someone within wifi range of their house, which is on a pretty quiet, isolated suburban street. I'm not overly concerned.

Wardriving is thing...

merphle wrote:
Robear wrote:

I can’t stress enough that keeping that kind of ancient hardware around is just a security disaster waiting to happen. The last official release of the software, along with the hardware, has *579* known security issues, each of which will have an exploit built into every hacking toolkit known to science. And the open source stuff, last I could find was from 2016, so it might only have a hundred or so known flaws. Still just absolutely crazy to give it space in your house, much less use it...

Depends on the specific situation. My parents are still using one (with DD-WRT) as a wired->wifi range extender. There's a much more modern router/firewall between it and the outside world, which should limit the attack possibilities to just someone within wifi range of their house, which is on a pretty quiet, isolated suburban street. I'm not overly concerned.

If that's their trusted network, you might want to think about replacing it. They're not vulnerable to anyone in the world, but they're vulnerable to anyone within a mile or so that has a good enough antenna; those little Pringles cantennas, for instance, allow fairly amazing range.

I like to put the WiFi on a separate network segment for untrusted devices, so that even if the AP is hacked, they don't have an easy way to attack the machines I actually care about. If that's too complex to set up, then I strongly recommend making sure the AP is getting patches.

Anyone aware of anything Firefox has done recently that would cause me to suddenly stop being able to load videos for some reason? I have two tabs open from a guitar chords website that have had the little "back-and-forth" running animation on their tabs in my window all morning, and it looks like because that page has those little video ads that want to run (and usually do, without audio) but aren't able to run, and then I tried to watch something on Amazon Prime but after clicking Play I'm just getting an endless black screen (the tab also has that loading back-and-forth blip on it). GiantBomb's videos won't run either. But youtube is playing fine?

I realize now this happened to me sometime last week too, and I ended up just closing out and reloading my browser, so I can do that again of course, but does anyone have an idea why this might be happening? I'm not running an ad blocker (I do have Firefox's built in protection stuff, but I've switched that off for some of these stuck pages, reloaded the tabs, and it didn't make a difference), and I haven't installed any other new extensions in the last few weeks before this started happening.

mrlogical wrote:

Anyone aware of anything Firefox has done recently that would cause me to suddenly stop being able to load videos for some reason?

I'm having a kinda-related issue and it's been happening for 3-4 weeks. I use Chrome, though.

I'll have a tab open in chrome and it will stop rendering what it on it. I notice it most when listening to music in the browser - the progress bar will stop moving, but the music keeps playing, even going to the next track. None of the interface works on the tab and a reload of the tab doesn't fix it. If I recall correctly, the reload doesn't do anything or loads a black page. Closing the tab and reopening it fixes it.

While I notice it most while listening to music, it can happen to any tab. It will happen a couple times a day on my music player tab and a couple times a week on another tab. It may be media related, I dunno.

It started happening after the latest big update I allowed Windows to push out. I've reinstalled Chrome and my video drivers. No go. I've just lived with it and haven't searched to see if it's some kind of known issue.

I don't run may plugins. I did disable all of them at one point and it didn't fix it.

If you do get a fix going, please post back in here. Maybe we have the same underlying root cause.

-BEP

Disable all of your extensions / plugins and see if it still happens. On Firefox even if you haven't installed any new ones, the existing ones can be updated and cause weird things. If you haven't even restarted the browser already, at least do that. I have seen some strange Firefox behavior sometimes when an update is pending restart.

It must be them wardrivers attacking your home network with a Pringles can from a mile away.

merphle wrote:

It must be them wardrivers attacking your home network with a Pringles can from a mile away.

Why would you sneer at things that actually happen in the real world? That isn't imaginary, merphle.

I actually used to wardrive back when WEP was still in heavy use. I had software on a Nintendo DS Lite that could break WEP in about 10 minutes, so I'd take my lunch break, park my van up close to an apartment building, break into someone's network, and then surf for 30 minutes on my laptop before heading to my next job. Normally I just read threads here so I wasn't really doing anything destructive with it.

Malor wrote:
merphle wrote:

It must be them wardrivers attacking your home network with a Pringles can from a mile away.

Why would you sneer at things that actually happen in the real world? That isn't imaginary, merphle.

Does it happen? Yes, certainly.

But it's not common or problematic enough for me to worry about. There are things much higher on my priority list, anyway.

Back in my stupid days, I 100% built a wifi antenna out of an old wok and connected it up to sniff out WEP passwords from neighbors and use their internet to feed into my router.

Like I said, I was stupid and didn't have enough money back then.

astralplaydoh wrote:

Back in my stupid days, I 100% built a wifi antenna out of an old wok and connected it up to sniff out WEP passwords from neighbors and use their internet to feed into my router.

Like I said, I was stupid and didn't have enough money back then.

Hah, during my immediate post college days I used a linux live environment to do the same thing. I think it had a tool called aircrack or something similar. It was basically just to have internet until Comcast could install.

EvilDead wrote:
astralplaydoh wrote:

Back in my stupid days, I 100% built a wifi antenna out of an old wok and connected it up to sniff out WEP passwords from neighbors and use their internet to feed into my router.

Like I said, I was stupid and didn't have enough money back then.

Hah, during my immediate post college days I used a linux live environment to do the same thing. I think it had a tool called aircrack or something similar. It was basically just to have internet until Comcast could install.

The worst/best thing I ever did of this sort was logging into the wifi router of a place I was staying at on vacation using the default password and changing the channel they were using off of the default channel which was being used by like 20 other networks in the same area. Their wifi was completely unsecured (this was probably 2004 or so) and still using the routers default SSID. It was a WRT54G of course and by this time I had setup about half a dozen of them for friends and family.

I was actually pretty paranoid about network security back then and used a white list for wifi access. As more and more devices started getting wifi (hell my washer and dryer have wifi now) that got to be very annoying to maintain.

Update on my 3800X RMA. So I got an e-mail on Monday letting me know that AMD had received my faulty 3800X and that I could expect a replacement in 5 business days. Today a package arrived and not only was it a replacement CPU but they sent me a brand new 3800X in box with the cooler as well. I still have the cooler that came with the original 3800X purchased a year ago.

So now I have a decision to make. I could open the box up and switch out the 3600X I bought as a replacement with this new 3800X OR I noticed that a local shop has several 5800X chips for 560 bucks Canadian, which is a sale price (regular 619.00) for another week or so.

Do I purchase the 5800X and sell the brand new never opened 3800X for slight discount? I think I may just keep the 3600X as a backup but I could sell that too. I am pretty sure I am going to sell the extra cooler I have from my original CPU. I am sure I could probably get 25 bucks for it.

If you sell both chips, you can probably get pretty close to paying for the 5800X completely.

That chip is very fast, and you'll probably notice the speed improvement if you do anything that requires computational muscle, but it runs very hot. Be sure you've got adequate cooling for it.

The 5900X is more cost-effective if you can get one; four more cores for a hypothetical $100 extra.

Malor wrote:

If you sell both chips, you can probably get pretty close to paying for the 5800X completely.

That chip is very fast, and you'll probably notice the speed improvement if you do anything that requires computational muscle, but it runs very hot. Be sure you've got adequate cooling for it.

The 5900X is more cost-effective if you can get one; four more cores for a hypothetical $100 extra.

I have the blacked out Noctua NH-D15. The 5800X as I mentioned is on sale for 569.00 and the 5900X which the site says they are currently out of stock of they have listed at 769.00. Not sure it's worth 200.00 more dollars.

Well, I have one, and it's a damn fast chip. Unless you're intending to do real work with the machine, and that work can use wide multicore algorithms, it will probably be at least a couple more years before 12 cores vs. 8 makes much difference.

By then, you might be able to pick up a used 5900, or maybe even a 5950.

I picked up the 5800X and installed it. So far I am glad I picked it up. I mostly do audio editing right now as far as workstation stuff is concerned. The 3800X was plenty fast so was 3600X for that matter, but I might do some Video editing at some point in the future, and the 5800X will help. I am also hoping at some point I will be able to get a 3080 or something and be able to use the resizable bar feature.

Now to sell the new 3800X I got from the RMA to replace my defective CPU.

Someone on Ars upgraded from a 3600 to a 5800 or 5900, and was saying that the whole machine felt notably snappier in routine use, which surprised him. And since you're now faster in per-core performance than probably any Intel chip at any price, you shouldn't have any trouble running anything that comes out for quite a long time. At the rate Intel has been improving, I'd say ten years, but AMD is more competitive now, and you might "only" get five out of it. (Heh, in the 90s, you could spend a thousand dollars on a CPU and be thoroughly obsoleted in a year. Not so much anymore.)

I will be able to get a 3080 or something and be able to use the resizable bar feature.

edit: I thought this was AMD-only, but I was wrong. Redacting everything related to that. The AMD chipset is the first one to support it, but it looks like everyone else is jumping on board, and it will work with pretty much any of the newest GPUs and motherboards.

Sorry for the (temporary) bad info.

Why don't BIOSes have a simple native utility to clone one physical drive to another? It surely can't be that complicated, especially for a true bit-for-bit copy, and would avoid potential issues with cloning a live, running system disk.

I believe that would be "out of scope" for what a BIOS does, which is to check HW and keep track of HW addresses for the system. If you added a bitwise copy engine, you'd need to have a RAM manager active, as well as disk drivers, the IO bus drivers would need to be active, plus you need error checking, a kernel to handle calls and interrupts, a driver library, etc. In other words, you'd need parts that are actually at the OS level in order for this to operate. The disk interfaces are only designed to handle calls from an OS, not stand independently with full functionality. (Otherwise, you could simply throw a cable between two disks and press a button for a copy.)

Alright, those are fair points, but some BIOSes seem to have at least minimal aspects of this capability already - I'm looking at the BIOS FlashBack utility, which is able to read a binary file from a filesystem of a USB drive.

(nb: I'm not complaining or anything; just voicing out a thought experiment)

Pretty sure what Flashback does is to provide enough of an operating system to copy a single known file from a USB drive that's formatted in one way to a disk that is formatted in one way, using a special fixed address physical port. That's a far more limited task than the one you proposed, but I can see why you might think that they are similar. They are, but they differ in the complexity of the tasks you are asking the system to do.

(That's not to say that a hacker could not substitute a bootstrapping payload, but that's a different topic.)

Edit - As an analogy, think of moving a small plant from one pot to another using a provided spade. Easy, right? Now, think of "your job is to move any plant (including Redwood trees and mushrooms and aquatic plants) from wherever the customer wants, to wherever they want". Which one will require more tools, planning and effort?

A blind block copy could absolutely be done in the BIOS, as it has enough OS-type facilities to handle that. Having enough intelligence to interpret at least a couple of filesystems (FAT and NTFS) in a read-only mode means that they could absolutely do block transfers from one device to another. All the basic stuff to read blocks from a drive is there, as part of the filesystem code, so they'd need slightly updated logic to write blocks. From there, it's just a matter of batching up lots of reads and writes. It should even fit inside the BIOS's image, which is very limited in size, as a block-style copy really doesn't take much code.

But imagine the havoc that would ensue; customers would get this wrong. They would, absolutely inevitably, sometimes copy the wrong way, and they would blame the motherboard maker. And a simple block-level copy is usually not enough; you also need to adjust the filesystem(s) to take up the additional space, and there are a lot of filesystems out there. The code to do all this is open source, but it's big, and the licensing to put it into a BIOS would be very messy, perhaps impossible. Independently rewriting code to handle all those filesystems is way way way out of the scope of what any motherboard BIOS team could hope to tackle.

So, we're left with the current setup: you can pay for commercial software to do this, or you can do it yourself with something like Partimage. You can even do it yourself from the Linux command line, using 'dd' and then various command-line resizing programs. The Raspberry Pi image works this way, for instance; you do a dd-style copy of the image to the flash drive, and when it boots up, it auto-resizes its own filesystem to fit whatever drive you put it on.

(I'm not sure how good NTFS resizing is, in Linux; you might need to do the dd-style copy, and then once Windows boots, resize the filesystem using Disk Manager.)

tl;dr: the actual block copy is easy and could be done in the BIOS, but customers would mess this up and be angry. Worse, that's only like 5% of the code required to meet user expectations. Best for them not to touch it with a ten-meter pole.

edit: oh, originally, you were saying that it's complex to do this from a live system drive.... the whole point of the Linux tools is that the disk isn't live, that you're typically booting from a USB key of some kind, with the OS image shut down and quiescent. It reduces the scope of the problem a great deal.

second edit: also, the GPT partition table layout is apparently at the end of the drive, so if you're copying an image to a bigger drive, I believe that needs to be relocated. MBR-style partitioning is at the start of the drive, so that's easy, but AFAIK GPT takes some intelligence, not just a blind block copy. (I haven't messed with direct drive imaging in quite a few years.)

Thank you both for the great insight. I appreciate you taking the time to indulge my curiosity.

Just to clarify, the BIOS does not have access to the filesystem drivers or to the Master Boot Record, so it would have no way to identify the partition targets or set up even block data movement between them. That requires an OS, which is why a small Linux setup on a USB stick can be so useful for this kind of stuff. But the BIOS itself would need to be rewritten with what amounts to large portions of an OS in order to do this. It might seem like they are small - after all, you can do it with a Linux stick - but that’s still at least an order of magnitude bigger than most BIOSes.

Robear wrote:

Just to clarify, the BIOS does not have access to the filesystem drivers or to the Master Boot Record, so it would have no way to identify the partition targets or set up even block data movement between them. That requires an OS, which is why a small Linux setup on a USB stick can be so useful for this kind of stuff. But the BIOS itself would need to be rewritten with what amounts to large portions of an OS in order to do this. It might seem like they are small - after all, you can do it with a Linux stick - but that’s still at least an order of magnitude bigger than most BIOSes.

AFAIK, the BIOS has full access to the hardware. It has to, because it configures most of it before the OS even boots. (eg, it has to set up the USB controllers, the SATA ports, the NVMe ports, and then scan all of them to find boot records.)

In fact, it has more access to the hardware than the OS. It can make changes to anything on the system, even with the OS booted and running. The BIOS runs in an invisible, super-privileged ring (System Management Mode) that can't be detected or controlled by the OS. This is why UEFI viruses are so dangerous, and so hard to get rid of. (and, fortunately, quite uncommon. At least, that's the present belief... they can be damnably hard to spot.)

Anything that an OS can do, the BIOS also can, plus more besides. It's just a matter of whether someone writes the code, and whether there's enough room. The current flash chips are getting very large. I don't know what the physical size is for the flash chip on this board, for instance, but the updates are 32 megs. You can pack a ton of code into 32 megs; the base Linux kernel, before it loads any modules, is only 14. Modules add a ton more code, obviously; the initrd on Debian, which provides all the modules that are considered boot-necessary, is another 83 megs. But a BIOS would only need a small subset of those, since it would know what hardware it had.

If the flash chip is indeed 32 megs, then there's enough room for the Linux kernel plus another 18 megs, which could cover a large spectrum of base functionality. If motherboard BIOS writers wanted to write a disk copy utility, they absolutely could. It's just.... they wouldn't bother, because it's high-risk (could damage customer data) and low reward (most customers wouldn't care if it was there.)

UEFI is complex enough to be considered an OS; it's far more capable than DOS, and DOS can do anything, provided someone's willing to write the code.