[MaulingMonkey]

Part of the interview for my last job involved interviewing me about debugging, which I thought was neat. "What's wrong with this code?" All the samples were relatively simple - less than 30 lines each?

- Double delete, caused by not following the rule of 3 in C++, which would likely crash - Multithreading code missing volatile or memory barriers (which segued into a discussion of it's disassembly) - Nonvirtual delete through base pointer

Extremely realistic - it's all stuff I've seen and fixed in the wild (this was for a non-entry gamedev job involving plenty of C++)

[ori_b]

Note that unless you are coding in Java, volatile is not a memory barrier, and both the CPU and compiler may reorder things around them.

[MaulingMonkey]

Yes - perhaps I should have written that as "volatile AND memory barriers". Either way, worth mentioning, as it's a common misunderstanding.

Further complicating matters is the fact that Microsoft's compilers will treat volatile as a memory barrier - for x86 code. But not for ARM! https://msdn.microsoft.com/en-us/library/jj204392.aspx

[jedmeyers]

Same thing might be with jvm. On x86 'concrete machine' volatile might be a memory barrier, but this behavior is not guaranteed by the definitions of Java 'abstract machine'.