The trickiest bug I've encountered

http://mokafive.tumblr.com/post/41912162643/the-case-of-the-mystery-data-blocks

by kenjackson • 11 years ago

183 65 11 years ago

mokafive.tumblr.com

digi_owl3• 11 years ago

When a dev goes "why would anyone do that?!?" you know you are in for a bad day...

dredmorbius• 11 years ago

When it's the QA manager saying it, leave.

slazaro• 11 years ago

Don't ever report bugs or UI annoyances to the Adobe team then. It's so infuriating.

tonetheman• 11 years ago

I might agree with at least one thing the MS guy said. I really do not trust the OS so I would be zeroing out memory. That is me, I am an un-trusting soul.

kleer001• 11 years ago

That is me, I am an un-trusting soul.

Same here. I think it's a rare person that's been living and working with computers for more than 20 years that doesn't develop an innate distrust of them or at the very least a subtle set of supersticious about how they work, how they're supposed to work, and the best way to get your work done.

ikeboy2• 11 years ago

I really do not trust the OS so I would be zeroing out memory.

If you don't trust the OS, don't use it. If you use it, you're giving it access to all your files, full stop.

randyrand• 11 years ago

Sound simple but reality is much different. What if the intended users of your software use it?

banachtarski• 11 years ago

How do you trust the OS to zero out the memory when you tell it to then? You're being facetious.

ww5202• 11 years ago

Zeroing your own pages won't help. What if the OS reads in your super secret file into its cache pages which you have no control of. And those cache pages somehow got written to another process's page.

raverbashing• 11 years ago

Sure, when are you doing that?

Remember, the bug "can cause arbitrary host OS pages to be written to the target file in some circumstances"

something like

   do_encryption(origin, destination)
   save_to_disk(destination)

You're clearing origin right before save_to_disk?

Oh I'm so sorry, destination had only been allocated for you, writing to it caused some other pages to be evicted and other processes being scheduled and the bug to be triggered and your data is still there. You lost

nathanb6• 11 years ago

In my day job, I have to write reams of complicated code that can slow down the system or make maintenance more annoying...just because a user can do something (even though doing that is unsupported).

That's what it means to write business-class software. Nobody worth having as a customer is going to build their business on your platform if your attitude whenever something goes wrong is "you shouldn't have been doing that in the first place".

(I am surprised to hear this story though. I actually found a bug in the NTFS buffer cache a few years ago which was introduced in (as I recall) Windows Server 2012. Maybe the Server organization are way more on the ball than the consumer OS organization, which is definitely possible. But they took it seriously and fixed it in a patch.)

angersock• 11 years ago

There's a balance, though. Oftentimes, you serve the business folks better by preventing them from doing something really silly, and instead letting them optimize their workflow not to include silliness.

Unfortunately, it's tricky to tell ahead of time which features are bad ideas and which aren't.

tjradcliffe2• 11 years ago

It's pretty easy at first order: features that users want are good ideas. I agree at second order users want a lot of features that solve their problems in ways that are less clever than one might like, but in the vast majority of cases the right answer from development is, "Users know what they need better than I do, so if they ask for something, I'll do my best to implement it even if I don't totally understand why."

That answer exposes the deeper answer, too: "My job as a developer is to understand user needs so that our software can help them fulfill them, so if I don't understand why someone needs a feature I should dig in further before implementing it so I don't implement the wrong thing or the right thing in the wrong way." Not always possible given shedule constraints, though.

Admitedly, those rules don't seem to apply in this case, since if the OS allows you to do something that corrupts data, that's a problem no matter why the user wants to do it. If it can corrupt data, the OS shouldn't allow you to do it, end of story.

derefr2• 11 years ago

Nobody worth having as a customer is going to build their business on your platform if your attitude whenever something goes wrong is "you shouldn't have been doing that in the first place".

My favorite set of APIs is AWS. You know why? Because they've realized they hold two very weighty sticks that they can use when designing, and they've put them in place all over.

1. They can make any arbitrary message to the API cost the user money every time they send it, to disincentivize using that part of the API thoughtlessly. That's whether or not they expect this to be an actual revenue stream at the rates people are charged for reasonable usage.

2. They can put a "soft cap" on any arbitrary resource, so that you have to phone them and get the cap raised if you want more than [some reasonable number] of something. This likewise disincentivizes bad designs that use a nigh-infinite number of costly somethings to accomplish tasks that could be just as easily accomplished some more idiomatic, less costly way.

AWS doesn't prevent you from doing stupid things... but it makes you really not want to. I love it.

click170• 11 years ago

Have you ever tried to setup AWS IAM permissions for a user pursuant to the principle of least privilege? Because Amazon's APIs are about as far from friendly as you can get in this respect.

Their docs make it easy to make the mistake of thinking that fine-grained controls are available for most things, but when it comes to really important things like being able to segregate a production and Dev VPC, their APIs basically force you to grant permissions to everything or nothing.

Some examples of things I've hit: Not being able to restrict a user to only change a specific routing table Not being able to restrict a user to only change a specific elastic NIC

I'm consistently surprised at what's missing from their API and couldn't disagree more about being happy with it.

digi_owl3• 11 years ago

No wonder everyone wants to be in the *aaS business. That sounds like the analog telco days, where long distance were billed harshly to get people to keep their calls short. Thus getting the most out of a low number of wires.

ksk• 11 years ago

When you reported the bug did you include a repro? I would appreciate it if someone could post it here.

To be specific, I wanted to know the specific conditions the article talks about under which this causes an issue. Flush will cause writes (obviously) so issuing writes to a file which is pending a flush is an interesting scenario.

vollmond• 11 years ago

> We reported the bug to Microsoft and included a short program that easily reproduced the issue.

Now we just need a copy of it.

j_s• 11 years ago

Am I correct in believing that from a high-level view, this NTFS error resembles the ext4 bug currently on the front page?

https://news.ycombinator.com/item?id=9576917

Edit: That must be why this post from 2013 (update the title!) discussing an issue fixed in 2011 hit the front page...

jimrandomh• 11 years ago

Similar, but the NTFS one was worse in that it was potentially a security vulnerability, since the corrupted data that it wrote inappropriately into the middle of files came from elsewhere in the filesystem, rather than zeroes.

(Software is hard and often buggy, filesystems included. Check your backups! Even if your filesystem is perfect, I spilled water on my laptop yesterday, and you could, too.)

mikeash2• 11 years ago

If data isn't backed up, it doesn't exist. If it's not backed up offsite, it's not backed up. These are words to live or die by.

I don't think it's likely that I'll spill water on your laptop, though.

vermooten• 11 years ago

How about fixing the white-text-on-a-pale background bug while you're at it?

Diamons• 11 years ago

I love the blog design

snarkyturtle• 11 years ago

I hate it, skinny text, low contrast, non-consistent background. It's everything wrong with modern web design. If, say, as you scrolled down the gradient went away, I'd be fine with it, but since it's fixed it's distracting.

joshuapants• 11 years ago

I care less about the gradient than the text and contrast. It's a bit distracting but not to the point where it alone really hinders me.

Benjamin_Dobell• 11 years ago

Agreed! I genuinely wanted to read the article but found myself squinting just to try read the text on that awful background. It was excruciating so I just gave up.

freebish• 11 years ago

In retrospect, I wonder if there was a better way to report that bug. Clearly the first person didn't get it, at all.

perlgeek• 11 years ago

Public disclosure as a security bug (information leak)?

beat2• 11 years ago

Sounds vaguely similar to the trickiest bug I ever had (and the first really hard bug I ever dealt with on my own). Mine was in AIX 3.2.5, circa 1995. We were buffering a latency-prone data stream between read and write processes using shared memory buffers. The original design used shmat(), which was limited to three buffers total on AIX. I rewrote the buffering using mmap() to create memory-mapped anonymous file buffers - the number was effectively unlimited and could be tuned via configuration. Around 100 buffers gave optimum performance, with a 300%+ throughput improvement in real-world conditions - a huge win on a huge problem!

Then it blew up in production. Like hard crashes shortly after starting. Upon investigation, I found that entire pages of mapped memory were being overwritten with nulls, more or less randomly (4096 bytes at a time).

Turned out the bug was in mmap() due to the order patches were applied on various servers. The dev/qa servers were patched at different times than the production servers. That, that was hairy. And for a junior programmer to have to explain this to the tech leads and IBM support - I don't even know how many times I heard variants on "What's wrong with your code, really? mmap() isn't buggy!"

Ah, the 1990s. No Stack Overflow. No ssl (not even http). We did network programming by wrapping raw sockets in C and writing stream parsers in lex and yacc. Kids these days, they don't know hacking!

tjradcliffe2• 11 years ago

I once #defined strlen to return a short. This was in the days of the 16-bit to 32-bit transition on Windows, and I was running a project to eliminate spurious warnings from our code. There were several hundred places where the return value of strlen was being assigned to a short, every one of which created a warning.

I was a senior developer but it was my first commerical job--I'd been in academia previously. I ran the change by the the most senior technical people in the company, all the way up to the guy who'd written the original application as employee #1. He OK'd it because, as he said, "strlen returns as short on the Mac anyway, and since our code runs on Mac as well as Windows it's a limitation we have to respect anyways."

A few years later the company stopped supporting the Mac.

A few years after that (and well after I'd left the company) one user site started getting crashes. I heard later that it took four senior devs a week to track down the cause, and much head-scratching because strlen was documented to return an int. They eventually found my #define, along with a comment that this eliminated so-many hundred warnings in the build, and that the change had been approved by the most senior people (I wasn't totally naive).

It turned out the problem at the specific site having the issue was users putting the entire contents of files into what amounted to tool-tips. It was totally unexpected user behaviour, but they'd found a place they could cache some useful data and we let them do it, so it should have worked.

Today I'd write a script that auto-edited all the cases where the problem occured, and regression test the hell out of it, but yeah: the '90's were a different time!

unwind• 11 years ago

I'd just like to point out that at least in the draft C89 standard (http://port70.net/~nsz/c/c89/c89-draft.html#4.11.6.3), strlen() had the same prototype as it has today: in other words it returns size_t. Not int.

huhtenberg• 11 years ago

raw sockets

Raw _TCP_ sockets probably. Streaming over raw RAW sockets - now, that'd be pretty damn hardcore indeed :)

beat2• 11 years ago

Oh yeah. That's sort of insane coding.

Still, it's weird to think that 20 years ago, just having TCP was kind of a luxury. I don't ever want to have to manage socket timeouts again.

zrm• 11 years ago

circa 1995

How about this one from just last year: I'm writing an application that encrypts packets using datagram TLS with OpenSSL. Take a raw packet, send it through OpenSSL, it comes out the same on the other end and everything looks fine.

Then I try load testing it and start getting mangled packets. Application data where there should be IP headers and vice versa. So get out the debugger and find out that the data going into SSL_write() on one end isn't the same as the data coming out of SSL_read() on the other end. No TLS error, just mangled data.

So I install the OpenSSL symbols and discover that the data going into the cipher is exactly the same as the data that gets decrypted on the other side, hence no HMAC verification failure. But OpenSSL by default compresses data before encrypting it, and decompression output is not the data that was originally compressed.

TLS requires records to be delivered reliably. Datagram TLS, by contrast, is like UDP. Packets can be lost. And if they are, you can't use stateful compression or the lost data creates a hole in the decompressor's stream and corrupts the output. But OpenSSL was doing exactly that. So disable compression and the problem disappears instantly. (After three days in a debugger.)

jmhodges• 11 years ago

FYI, using TLS compression makes you susceptible to the CRIME attack. I think I have a ticket with OpenSSL for them to turn that off by default, but I don't think they've done it, yet. Glad you got there accidentally!

meshko• 11 years ago

The experience of dealing with Microsoft's support sounds exactly the same as what I had to deal with. In my situation there was no workaround possible short of rebooting the system (it was a kernel resource leak), so after spending 3 months iterating through different ways to reproduce the problem (including "we don't support Windows on VMWare, after asking me to send them a VM image to reproduce the problem), going through 3 levels of support, I got to the people who were able to get me a fix. Alas, it was a private patch which was only available upon request, which didn't help much as I was working for an ISV.

zamalek• 11 years ago

Getting Microsoft to fix bugs is hard. We had a bug in the .Net runtime. Took ages to reach even agreement on the memory dumps: probably because bugs in the .Net runtime are like unicorns, even I had a level of disbelief that .Net was to blame.

Once you prove it, it's completely free. Ultimately memory dumps are the way to go when it comes to MSFT bug reports, if you can catch their bug red-handed and snap it to disk things go real smooth.

sseveran• 11 years ago

Yeah, I found a great one in .net. If you used HttpWebRequest to fetch a url that had an invalid GZIP header the app would hard crash. This is because the request and decompression was being handled on a threadpool thread. An exception would be thrown and no one would be around to catch. Fortunately this was in a web crawler so it took quite a while to build a repro case and an exact diagnoses. I believe it was fixed 4 or 5 years after I reported it in 2008.

koyote• 11 years ago

How did you report the bug?

I always found it incredibly difficult to report anything to Microsoft and usually only get a 'we have reported it to engineers and it will be fixed some day' message.

ww5202• 11 years ago

Ahhh, the fun with caching in the file system. Incidentally one of trickiest bugs I encountered also had to do with the Windows Cache Manager.

It was the interaction between the Cache Manager and the Memory Manager in managing the rare transient state ModifiedNoWrite of the cache pages when dealing with reentry IO read requests. The cache page status became Modified when its content was filled in from disk, but it's marked as NoWrite to avoid being flushed out by the Memory Manager. The physical page backing the cache page can't be reused since it's dirty (Modified) but the Memory Manager can't flush it out (NoWrite). Slowly over time as more pages are read, the system would run out of physical pages.

The Cache Manager was supposed to change the cache page status back to Standby after the read returned from lower layer. But with reentry IO read requests, it won't do it when upper layer IO request buffer passed straight down. The work around was to allocate a separate buffer to interact with the Cache Manager and copied the content back to the upper layer IO request buffer, incurring an extra copy.

At the end I intimately knew about the Windows Cache Manager and Memory Manager more than I needed to know.

alexdowad• 11 years ago

To me, stuff like this is a powerful argument for using open-source software (or at least software which you can access the source for), whenever possible. When things go very wrong, at least you can dig into the source code and try to fix it.

MaulingMonkey• 11 years ago

It's certainly an argument for having source access. For OS / "platform" level bugs it's no panacea however - you may very well need to keep your workaround, either because you can't rely on end users upgrading to the latest version, or simply need a stopgap until you finish the patch, submit the patch upstream, address all the issues brought up in the review, have it integrated into the next stable release, have it trickle downstream into the stable releases of individual distros, etc etc etc...

dba7dba• 11 years ago

I think the trickiest bug that I heard of was USAF F22 jets losing all computer systems (navigation, communication etc etc) as it crossed date line while flying to Hawaii for the first deployment on the island.

The flight of 4 jets were able to return to US mainland only because the accompanying tanker was able to guide them back.

chris_wot• 11 years ago

Yeah, well Microsoft never managed to find why Liberation fonts caused a BSOD:

https://news.ycombinator.com/item?id=5468390

beagle3• 11 years ago

Ah, reminds me of a Windows bug that I chased in Win2000 Server. (For all I know it might still be there - I've stopped writing Windows code in 2007).

When you wrote to a file (using WriteFile or fwrite()), it first extended the file length, and then committed the buffer. This is supposed to be atomic - that is, you should never be able to see the length already extended but the data not yet there. And it apparently was atomic if both reads and writes came from the local machine, or both came from the network - however, if the write was on the local machine but the read was from the network, locking was missing, and it WAS possible to read zeros instead of the real data (but only because of a race condition - reading the same file again later would give the expected answer)

Tried to get Microsoft to at least confirm this bug, to no avail - there was no one interested in talking to a lone freelance developer back then.

The trickiest bug I've encountered

Related Stories