Atomic I/O letters column #157
Originally published 2014, in PC & Tech Authority(in which Atomic magazine is now a section)
Reprinted here January 13, 2015 Last modified 18-Jan-2015.
The silent death
Do modern PCs still have beep error codes? There was a small power cut last night and my PC hasn't worked since (so this is yet another of your "I'm typing this on my phone, because..." letters). I've taken every component out of the motherboard except for the CPU and the power connections, and when I turn the machine on now the fans still spin, but there's no error beep for no-RAM, no-video-card, et cetera.
Any ideas on what's dead, besides "everything"?
J.
Answer:
Yes, you should get BIOS error
beeps if you power up a
motherboard without some essential component(s) connected to it. If there are no beeps
then either you don't have a PC speaker
plugged into (or built into) the motherboard, or something major is fried, or more temporarily
non-functional (unplugged, temporarily shorted out by a stray screw...).
It might be something wrong with the power supply, which is the component most likely to get barbecued by mains power problems. This is by design; a PSU is meant to sacrifice itself to save the rest of the PC. Passing through mains weirdness to the other components is one of the things that crappy off-brand maybe-not-even-legal PSUs can be expected to do. If the fans are spinning, though, then the PSU is probably OK.
So, most probably, you've got a dead CPU, a dead motherboard, or possibly even both. If nothing else in the computer is old and flaky (no five-year-old drives, either spinning or SSD...), so you can't justify a whole new machine, then I'd just get a new CPU and mobo and turn the old ones into wall decorations.
Thumb-drive surgery
I trod on my flash drive - it had to be the 128GB one not the 4GB one, didn't it? - and snapped the connector off. The little circuit board's delaminated where the connector was soldered on, but the traces that went to the connector are clearly visible past the destroyed part of the board, and I reckon I could solder something onto them.
Is there a way to put a new connector on this thing, or will that be an express train to data corruption?
Christian
Answer:
Yes, you can fix it, and yes, it'll probably be fine, if ugly. As long as the mechanical
damage really is only to the USB-connector portion of the
board, and no cracks
have gone further into the land of microscopic surface-mount components, then it really
does just need a new connector. USB connectors only have
four contacts, too, so a novice
can solder them quite successfully. (As compared with, say, a
DVI connector, which uses 24
contacts, plus another five for the optional cross-shaped analogue contacts. Some many-pinned
connectors can be fixed near-magically with
drag-soldering tricks, but
usually hand-soldering them is about as fiddly as you'd expect.)
The easiest way to do this, especially with a drive with damage to the USB-plug traces, is to sacrifice a USB cable with an A plug on it (the same gender the drive had before). Chop the non-A-plug end off, strip the wires, figure out which one goes to which contact with your $10 yellow plastic multimeter in continuity mode, and then solder the wires onto whatever device has lost its plug.
[Any low-powered soldering iron is suitable for this job. 25 watts is plenty, a 15-watt
"soldering pencil" is probably even better. I do not recommend you purchase one of the
suspiciously-cheap mains-powered irons that abound on eBay, unless
electric shocks and/or burning down your house excite you. Here
in Australia you can get a perfectly good mains iron for
less than $AU15; in the
States you can get an actual
brand-name Weller iron
[A note on wire strippers:
This kind
If you don't care about keeping the busted thumb drive, then you can improvise something without having to solder. Little alligator clips, syringe-type circuit test clip leads, paper clips... whatever's necessary to temporarily connect the circuit-board terminals to the USB socket of a computer for long enough to recover your data.
If you solder a cable onto the circuit board, though, you can keep using the thumb drive. Once you know it's working, just drown the board in hot-melt glue or some similar substance to protect the connection. (Although it is a great Half-Arsed Electrical Repairer tradition to use half a mile of vinyl electrical tape in this situation, please don't. You could use far-more-expensive self-amalgamating silicone repair tape, though. Old-style self-amalgamating tape is nasty sticky rubber, but the modern silicone stuff is slick, and has a thousand uses.)
The management, as usual, disclaims all responsibility if this advice results in you shorting a USB connector and blowing up your USB controller. That's about the worst that could happen, though, unless you take really extreme steps to deliberately electrocute yourself.
Config, schmonfig
I've got a new PC (but with Windows 7 x64, cos like all right thinking people I will not have Win8 in the house) and now I can install stuff in Program Files (x86), but some of the stuff I install gets error messages if it tries to make any files of its own, like config files, in the same folder.
Windows pops up a do-you-really-want-to-do-this thing when I install the programs, but there's no similar notification when the programs try to do stuff, it just fails every time.
I thought it was a permissions problem, but digging through account permissions and folder permissions only shows me that there are indeed some things my administrator account can't do, but they're all greyed out and I can't change them!
Don't make me smash this thing. It has an SSD and a GeForce 750. It's really nice. It's just allergic to boringprogram.cfg files.
Keith
Answer:
You've got User Account
Control turned on, and your old programs weren't installed in a way that UAC could
track. (Did you just copy a folder or two into Program Files, by any chance?) UAC thinks
those programs may be doing something underhanded. But even then, as you say, it should
tell you, not just deny those programs entry to the nightclub and swear it never
even saw them.
UAC is a good idea, and by and large works, though it can also be a good idea to turn UAC off when you're setting up a new scratchbuilt computer, to avoid endless "yes, mother, I meant to install that one TOO" experiences. If UAC totally blocks some software that matters to you, though, just turn it off for good. You can still restrict the account permissions of your everyday login and get reasonable security, provided you actually read the requesters when things install.
(This is the big problem with many security systems that pop up "do you really want to..." requesters. The less experienced the user, and thus the more they actually need to be protected by that security system, the more likely it is that the miserable schemozzle of computer-security "training" these days will actually have just trained them to always click OK without reading. Especially if there's a promise of dancing pigs.)
An impenetrable Cat5 jungle
At my work the server room has a huge network patch panel, the backside of which is a horrorshow. Multiple cable colours, multiple cable LENGTHS, all just plugged in all over the place.
The front of the patch panel looks fine, though – neatly and correctly labelled, and everything works. I presume whoever wired the back just used every cable that came to hand before buying a reel of one colour, or something. But is this likely to be a problem in the future? We've had about the same number of network problems in this office that I've seen in every other job I've had (not very many), and none were traceable to the patch panel, but is it a ticking bomb situation?
T.
Answer:
It's probably fine.
The idea of the standard patch panel is that the back of it is wired up precisely once, when it's installed. Every single socket is supposed to be connected to the appropriate other socket, even if nowhere near all of them are yet needed. Then, the panel's pushed into its rack or into the corner or into an Ikea bookshelf (depending on your business's scale...) and the back of it is never touched again. The wiring behind the panel should be neat and tidy, but provided it doesn't break network rules (wrong cable grade, kinked conductors, outrageously long, that sort of thing) it can be as much of a spaghetti-explosion as you like, and still work fine.
It's hardly best practice to have random Medusa-hair cables all over the place behind the patch panel, but not so much for functional reasons as because old cables can be unreliable, network devices dangling from said cables can damage them, and passers-by may hook their foot in a loop of cable and cause a slapstick disaster.
This is also why many sexy super-organised zillion-cable network setups are entirely composed of unlabeled blue cables. Nobody needs to be able to tell them apart by sight. That sort of thing is strictly amateur league.
So this is another of those things that's only a worry insofar as it indicates that the person responsible for the ugly rack wiring might have done something just as ugly, but more likely to cause problems, somewhere else.
The h4xx0r-f0n3
I was sitting in a cafe the other day, and as I always do I tried to connect my phone to any local free WiFi, of which there wasn't any. The phone connected to an unencrypted network, but the browser just showed one of those "give us money for a magic code" pages, I think from the small hotel next door. OK, no problem, no surfing for me, get on with lunch.
Except after I finished lunch, I noticed that I now had new email downloaded to the phone. I definitely didn't get it via 3G or any other phone network, but there it was. But the Internet was still clearly blocked, because I hadn't gone next door with however much money they wanted.
How did this happen? Do I have a magic telephone?
Harry
Answer:
The network you connected to did not block your Internet access as well as it was meant
to.
The "gateway" device that was supposed to block you until you typed a code into the "captive portal" correctly redirected port-80 HTTP requests to that portal page. But it passed port-110 POP e-mail traffic without complaint. It might also pass port-25 SMTP e-mail upload traffic, though any such activities on an unencrypted public network are not a great idea security-wise.
Holes like this can be exploited in a number of ways. You could, for instance, "tunnel" an encrypted SSH connection through some port that turns out to be un-blocked, and connect to the Internet quite securely via the server - like perhaps your home PC - at the other end of the tunnel. Or, for a clumsy but more straightforward solution, you could connect to a remote machine via VNC or Remote Desktop or something and just use a browser on that machine.
(My own tough-as-nails NSA-grade security-through-obscurity protection for the VNC server running on my main computer, beyond using a proper password, is to have it listen on a non-standard, high-numbered port.)
A properly configured pay-gateway should block all ports for non-paying users, but dumb ones that only block port 80, or let through any port higher than 999, are not hard to find.
Note that doing any of this stuff is now quite spectacularly illegal in many jurisdictions. Just accidentally checking your mail isn't likely to get you indefinitely detained as a cyber-terrorist, but deliberately exploiting such loopholes can, in the unlikely event that you're caught, get you in big trouble even if you haven't been downloading movies and/or sending death threats to world leaders.
