Windows XP SP3 Causing Router Crashes

KrispyBytes writes "Windows XP SP3 has been named as the culprit causing home routers to go into a crash and reboot cycle. One router maker has released firmware updates to fix the problem, but has not yet revealed what is actually different about XP SP3's networking stack or UPnP behaviour that causes the problem. Router maker Billion Managing Director Raaj Menon said "as Microsoft plans to make Windows XP SP3 an automatic upgrade this month, the number of affected routers may increase significantly.""

Before anyone goes on a MS rant

[Gregb05]

A computer on the network should not be able to crash the router. This is a problem with the manufacturing of the routers, not anything in particular with SP3. This problem could have arisen in any OS. The fact that it appeared with SP3 is irrelevant. I return you to your MS bashing.

Blaming the wrong programmers

[Jeffrey Baker]

Shouldn't the title of this post be "Shitty router programming causing router crashes"? It should matter what type of garbage come off the wire, the router must be able to handle it all without error.

Re:What if I was the other way around?

[vux984]

If an upgrade to a router caused Windows to enter a reboot cycle would we be blaming the router manufacturer or Microsoft?

Would anyone notice?

Kidding aside, my first thought was this is CLEARLY a router problem. Even if SP3 is completely defective and sending out complete garbage to the router, the router should cope better than going into a 'crash and reboot cycle'.

Re:Not surprising

[Anonymous Coward]

Lets not jump to blame this on Windows. It could be that Windows isn't doing anything wrong, just something the router should be able to handle, but can't. We can point fingers when we know what the actual issue causing the router problems is.

Re:Before anyone goes on a MS rant

[Wavebreak]

How do you know it's not? Until we see some actual analysis of this, there's no way to know if XP is at fault or if it's just shitty routers doing what they do. The latter seems far more likely to me, considering how absolutely shit most of the routers out there are, especially with stock firmware.

Buggy Routers

[John Hasler]

Any router that can be crashed by anything that any of the computers connected to it do is seriously buggy. This is not Microsoft's fault.

Re:Before anyone goes on a MS rant

[Jugalator]

True but it takes two to tango

Yes, it wouldn't have happened without the "help" of SP3 in this case. That being said, with the relevant information not released here, it's not certain SP3 is doing anything inherently wrong according to the networking standards. Testing SP3 on all hardware configs is additionally nothing one can expect Microsoft of doing.

Re:Blaming the wrong programmers

[Jugalator]

Agreed. If SP3 can do this unintentionally, imagine what the series of communicated data with the routers can do if a malicious writer now reverse-engineer whatever SP3 is doing, and would spread a time-triggered virus, for example. These kind of hardware issues are never good.

Re:Before anyone goes on a MS rant

[John Hasler]

> It is the manufactures fault that thier crashing, but this bug wouldnt be seen if xp was
> behaving correctly.

Nonsense. Any router that can be crashed by anything that a computer connected to it does has a critical bug and should be recalled immediately.

Re:Before anyone goes on a MS rant

[jellomizer]

However... If they find out what causes the router to crash with SP3 then all it will take is someone to duplicate the information sent and crash the router again and again. If the router crashes is has to be the fault of the router not of the OS, as other routers don't crash. As well as a poorly designed website. If your web browser crashes from a badly made website then it is the web browsers faults. Your argument only really holds true in cases of custom designed software where the sender of the data will need to agree to send the data in the correct format as well the receiver agrees to get the data in the correct format. And still even in that case a good program will be able to atleast say something is wrong, vs. it crashing.

Re:What if the router ran Linux?

[pembo13]

There hardly any comments yet. Most are defending Microsoft. Who is this "we" that are flaming msft?

Re:Buggy Routers

[Lehk228]

not quite, any router that can be crashed by bad packets is buggy, i don't see the need for routers to be designed to anticipate other bad behavior, such as dumping 110 V AC down all 8 pins

/pedantic

Re:Same as Vista

[Enderandrew]

I agree. Quite frankly, given that only one small manufacturer is reporting issues, and that they were able to resolve the issue on their end, makes me believe the issue is more with the router.

Correlation != causation

[catscan2000]

It sounds like the Billion router's firmware had a really bad bug that happened to be poked by Windows XP SP3. Unless if this was in a third party library or some external code that they were using, I wouldn't be surprised if this was limited to just Billion routers.

XP SP3 didn't _cause_ the bug; it merely happened to recreate a condition that triggered a bug inside the router to crash itself. :-)

Exactly

[biolysis]

Your point is the relevant one. Regardless of what OS did this, the fact is that no computer should ever be able to crash a router period. The incessant MS bashing reaches absurd levels sometimes.

I'm sorry, broadband router manufacturer who?

[urbanriot]

Billion? Never heard of'em. My Linksys router isn't complaining...

Re:What if I was the other way around?

[Opportunist]

Problem is, though, he would (as usually) be pissed at the wrong party. That MS is the wrong party this time might be divine justice, but it still means that the wrong side gets heat for something someone completely different fubar'ed.

Re:Before anyone goes on a MS rant

[Kjella]

Between having a security hole that allows denial-of-service attacks and sending some slightly mangled packets, I'd go about 98-2% to the router manufacturers on this one. That is assuming the packets are actually mangled, which isn't proven. I wouldn't care how broken a web page was, if Firefox crashes then it's Firefox's fault (or extensions/plugins, but that's a different issue). If you came and said "While it's true that the browser shouldn't crash in any circumstances, Apache would certainly deserve bashing if it's a result of them violating the specs and sending out mangled web pages." I think people would laugh. Assume the input data is crap, that applies equally everywhere and any software that can't handle that is poor software.

Re:Oh brother...

[spazdor]

is a coincidence, or just completely made up

Unlikely, given that the OP mentions that at least one manufacturer has fixed the problem with a firmware update. You can't really write software to fix a problem until you've figured out what the problem is.

You're right though, a properly hardened router will keep ticking regardless of what's plugged into it. Mostly. [fiftythree.org]

Re:Before anyone goes on a MS rant

[Anonymous Coward]

Lets make this easier and flat out assume XP's UPnP implementation was intentionally designed to crash the largest number of routers on the planet in a clever bid to raise Vista sales.

Even with this unlikely assumption in play it would still be 100% the fault of the router for crashing.

maybe, maybe not

[frovingslosh]

I agree that no router should crash based just on packets it passes. But there are a few issues here. If SP3 is causing something akin to a DOS, and a router's tables are filling up due to bad packets, it might very reasonably decide that things are so bad that the best thing for it to do is a reset. We don't blame the router maker when an external DOS attack interrupts Internet access, why blame it if the DOS is from Microsoft software on the inside?

And there is also the potential issue of this being UPNP related. UPNP is a completely bogus thing, but Microsoft strong armed the industry to support it and it's in most routers and many users don't know to disable it. UPNP could certainly give ways to cause this issue, and I only hold the router itself responsible to the extent that it supports this blasphemy.

Not MS to blame

[Opportunist]

As much as I hate defending the Redmond Computer Virus (tm), that's the router's fault.

Now, if SP3 created nonstandard packets that most routers still swallow but a router drops because they don't work to spec, blame MS. If the router replied with a bogus message to said nonstandard packet that locked up XP, blame MS. But a router HAS TO be able to accept a bogus packet. It may drop it, report it or if it feels like it send it on a roundtrip in hope that some machine can figure out what it's about, but it may NEVER crash due to it.

I hope I don't have to mention the security implications of this.

Re:Works for me, and probably for you

[oldspewey]

DD-WRT (any version)

The dd-wrt forums [dd-wrt.com] have nothing on XP SP3 problems ... have you reported the issue?

Re:No lies, just a bad track record.

[biolysis]

"Apparently sarcasm and humor is flamebait nowadays."

It is when there's no sarcasm or humor. Like in your post.

You came off as just another anti-MS troll with no idea what they're talking about. If it was intended to be sarcastic or humorous, you missed badly on both.

NAPT != Firewall

[Luke-Jr]

uPNP configures port forwarding for a NAPT (aka NAT) router. NAPT/NAT is *not* a firewall, and should not be treated like one. Its sole purpose is to translate addresses and ports (Network Address and Port Translating) between the internal and external networks. It is not meant to protect computers on either end from each other. uPNP facilitates the NAPT job by giving applications an easy way to automate the needed port forwarding for the WAN->LAN direction. If you want a firewall, get a real firewall.

apples to oranges

[spazdor]

Did the hardware manufacturers all just write flawless Linux drivers and buggy Windows ones?

Or did Linux developers just go a step further than Windows did, and take it upon themselves to make sure that hardware works properly on their OS?

What's worse

[Sycraft-fu]

It sounds like the packet isn't bogus. MS chose to implement an optional part of the DHCP spec (vendor information). As per the DHCP spec if your device doesn't implement the optional parts, you just ignore them, not crash.

So this isn't MS sending a bogus packet, or even doing a "Windows own spec," thing. They are properly following the DHCP spec, and this POS can't handle it. I mean I'll give someone a tiny bit of credit if the problem is due to bad data. Not a lot, it's still a bug that needs to be fixed, but at least it was something unexpected. However when you are crashing because you didn't deal with part of a spec, well then you get zero sympathy.

Re:Tinfoil hat time!

[anotherone]

Even if that were true (it isn't) it would STILL be the router's fault.

Re:Nah...

[cnettel]

If you can crash the router, you have a possible DDoS attack. If you can do it on the WAN port, it would certainly be a flaw in the device. Depending on the crashing behavior, it is also possible that this is actually an exploitable path that could be used to permanently reflash the router for malevolent purposes.

Re:Before anyone goes on a MS rant

[postbigbang]

Everyone wants a bulletproof, but quickly reacting app. Sometimes you can't have both. You can build parsers that vet web pages for sanity sake (or just look for malware as some plugins do), but they'll slow down even the fastest clocking machines; the insertion loss of the parser will be like putting a foot on the garden hose.

Routers and layer 2/3 bridges have to react at wire speed, and therefore have lean, racing engine code with only the barest of exception handlers. Inside the code are lots of routines that have to react to protocol changes related to table building. Screw up those tables even legally (according to the obscurities of even well-known protocols) and the routing/bridging device will behave badly, even to the point of apparently not working. It's happened before, and will happen again. Is it XP3? No one knows yet.

The next update of will likely fix the problem; likely it would arrive before a Microsoft fix, and it would be more effective to fix the crashing device than go after all possible XP SP3 users. Sadly, once in the 'wild', it's the router vendor's problem rather than Microsoft's, no matter who is to blame for the original mistake.

RTFRFC

[Krunch]

The client sends a DHCPREQUEST packet that contains options it is requesting from the server. The server is then supposed to craft a DHCPACK packet containing values for the options requested by the client.

http://www.ietf.org/rfc/rfc2132.txt [ietf.org]

Standards Shell Game?

[Zero__Kelvin]

The consensus seems to be that the router is at fault if SP3 crashes it. There is a major factor that people are overlooking here. uPNP is a M$ "standard." So here is the possible, and I would even assert likely, scenario:

M$ creates the uPNP standard, then revises it, then revises it again. To the extent that it is a standard, different versions of the "standard" are made available to different router designers, based on how close they come to touching their palms to the floor when bending over for M$. Now, those who handed over their first born have the newer tweaked standard available, and if they comply their router doesn't crash. In the meantime, other router companies have a different/older standard, to which they comply fully. Of course, SP3 makes use of the newer, less widely disseminated standard. Doing so causes implementations that haven't "paid up" to crash.

Yes, this definitely sounds like a scenario imagined by a guy who wears a tin-foil hat to those who don't know the M$ history, haven't read the M$ internal documents known as the Halloween Documents [catb.org], etc. To people who know the history and understand how M$ works, this is a likely though unproven scenario.

Re:Not surprising

[lyml]

Yes, lets not blame Windows XP SP3... after all, this has happened with other OS's and other versions of Windows, right?

Oh, wait, this has only happened with XP SP3 machines....

So, again, what's the probability that it isn't Windows XP SP3 doing something wrong?

I'd say 100% seeing as that it is the router that is crashing.

Think of it this way, if writing this post made your computer crash, would it be my computers fault, or would it be yours?

Re:Not surprising

[RobertM1968]

I'd say 100% seeing as that it is the router that is crashing.

Hmmm... what is the margin of error on that statistic?

Oh, nevermind... I came up with the same figure anyway. ;-)

Think of it this way, if writing this post made your computer crash, would it be my computers fault, or would it be yours?

Well, I'd like to blame it on your computer... and wonderfully, I can - and not be deemed insane. Things like that happen on /. all the time (like the guy I was responding to).

=)

Re:Not surprising

[jibjibjib]

In fact, it's definitely the case that it's something the router should be able to handle. If a router receives a packet that causes it to crash, it's a flaw in the router's software, no matter whether the packet is malformed or not.