Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
'struct rtrie' has no member named 'lock' #245
Compilation error on ARM (apparently):
If anyone gets this, the temporal workaround is to replace
Here's my explanation:
The second argument to the
This is desired because we'd rather crash the kernel (which would be an immediately identifiable bug) than let a concurrence problem slip by (which would be a serious error that could be extremely difficult to even identify, let alone track).
The thing is, depending on compilation options, the condition might or might not be evaluated.
And here is the catch:
The fact that it's not crashing means that the condition is very rarely being evaluated, which in turn means that concurrence is very rarely being validated during runtime. This is the core problem that needs fixing.
(The check validates code, not input, and the code seems fine, so there's not much a user can do to break something.)
If anyone is paranoid, all one needs to do to completely zeroize any chances of a concurrence problem is to never edit the EAM table while traffic is being translated. Preventing these two things from happening at the same time is the only purpose of these protections.
Update: This was reported by someone who's messaging via the dev mailing list, so credits will remain anonymous until further notice.