ChallengeThe race condition in the TCP stack

time to read 3 min | 463 words

Occasionally, one of our tests hangs. Everything seems to be honky dory, but it just freezes and does not complete. This is a new piece of code, and thus is it suspicious unless proven otherwise, but an exhaustive review of it looked fine. It took over two days of effort to narrow it down, but eventually we managed to point the finger directly at this line of code:


In certain cases, this line would simply not read anything on the server. Even though the client most definitely sent the data. Now, given that TCP is being used, dropped packets might be expected. But we are actually testing on the loopback device, which I expect to be reliable.

We spent a lot of time investigating this, ending up with a very high degree of certainty that the problem was in the TCP stack somewhere. Somehow, on the loopback device, we were losing packets. Not always, and not consistently, but we were absolutely losing packets, which led the server to wait indefinitely for the client to send the message it already did.

Now, I’m as arrogant as the next developer, but even I don’t think I found that big a bug in TCP. I’m pretty sure that if it was this broken, I would have known about it. Beside, TCP is supposed to retransmit lost packets, so even if there were lost packets on the loopback device, we should have recovered from that.

Trying to figure out what was going on there sucked. It is hard to watch packets on the loopback device in WireShark, and tracing just told me that a message is sent from the client to the server, but the server never got it.

But we continued, and we ended up with a small reproduction of the issue. Here is the code, and my comments are below:

This code is pretty simple. It starts a TCP server, and listens to it, and then it reads and writes to the client. Nothing much here, I think you’ll agree.

If you run it, however, it will mostly work, except that sometimes (anywhere between 10 runs and 500 runs on my machine), it will just hang. I’ll save you some time and let you know that there are no dropped packets, TCP is working properly in this case. But the code just doesn’t. What is frustrating is that it is mostly working, it takes a lot of work to actually get it to fail.

Can you spot the bug? I’ll continue discussion of this in my next post.

More posts in "Challenge" series:

  1. (06 May 2022) Spot the optimization–solution
  2. (05 May 2022) Spot the optimization
  3. (06 Apr 2022) Why is this code broken?
  4. (16 Dec 2021) Find the slow down–answer
  5. (15 Dec 2021) Find the slow down
  6. (03 Nov 2021) The code review bug that gives me nightmares–The fix
  7. (02 Nov 2021) The code review bug that gives me nightmares–the issue
  8. (01 Nov 2021) The code review bug that gives me nightmares
  9. (16 Jun 2021) Detecting livelihood in a distributed cluster
  10. (21 Apr 2020) Generate matching shard id–answer
  11. (20 Apr 2020) Generate matching shard id
  12. (02 Jan 2020) Spot the bug in the stream
  13. (28 Sep 2018) The loop that leaks–Answer
  14. (27 Sep 2018) The loop that leaks
  15. (03 Apr 2018) The invisible concurrency bug–Answer
  16. (02 Apr 2018) The invisible concurrency bug
  17. (31 Jan 2018) Find the bug in the fix–answer
  18. (30 Jan 2018) Find the bug in the fix
  19. (19 Jan 2017) What does this code do?
  20. (26 Jul 2016) The race condition in the TCP stack, answer
  21. (25 Jul 2016) The race condition in the TCP stack
  22. (28 Apr 2015) What is the meaning of this change?
  23. (26 Sep 2013) Spot the bug
  24. (27 May 2013) The problem of locking down tasks…
  25. (17 Oct 2011) Minimum number of round trips
  26. (23 Aug 2011) Recent Comments with Future Posts
  27. (02 Aug 2011) Modifying execution approaches
  28. (29 Apr 2011) Stop the leaks
  29. (23 Dec 2010) This code should never hit production
  30. (17 Dec 2010) Your own ThreadLocal
  31. (03 Dec 2010) Querying relative information with RavenDB
  32. (29 Jun 2010) Find the bug
  33. (23 Jun 2010) Dynamically dynamic
  34. (28 Apr 2010) What killed the application?
  35. (19 Mar 2010) What does this code do?
  36. (04 Mar 2010) Robust enumeration over external code
  37. (16 Feb 2010) Premature optimization, and all of that…
  38. (12 Feb 2010) Efficient querying
  39. (10 Feb 2010) Find the resource leak
  40. (21 Oct 2009) Can you spot the bug?
  41. (18 Oct 2009) Why is this wrong?
  42. (17 Oct 2009) Write the check in comment
  43. (15 Sep 2009) NH Prof Exporting Reports
  44. (02 Sep 2009) The lazy loaded inheritance many to one association OR/M conundrum
  45. (01 Sep 2009) Why isn’t select broken?
  46. (06 Aug 2009) Find the bug fixes
  47. (26 May 2009) Find the bug
  48. (14 May 2009) multi threaded test failure
  49. (11 May 2009) The regex that doesn’t match
  50. (24 Mar 2009) probability based selection
  51. (13 Mar 2009) C# Rewriting
  52. (18 Feb 2009) write a self extracting program
  53. (04 Sep 2008) Don't stop with the first DSL abstraction
  54. (02 Aug 2008) What is the problem?
  55. (28 Jul 2008) What does this code do?
  56. (26 Jul 2008) Find the bug fix
  57. (05 Jul 2008) Find the deadlock
  58. (03 Jul 2008) Find the bug
  59. (02 Jul 2008) What is wrong with this code
  60. (05 Jun 2008) why did the tests fail?
  61. (27 May 2008) Striving for better syntax
  62. (13 Apr 2008) calling generics without the generic type
  63. (12 Apr 2008) The directory tree
  64. (24 Mar 2008) Find the version
  65. (21 Jan 2008) Strongly typing weakly typed code
  66. (28 Jun 2007) Windsor Null Object Dependency Facility