Tomasz Kapela
Tomasz Kapela
Another solution would be to include the lane index into the persistent mutex. The lane index effectively identifies the outermost transaction and could be used to do this. Since mutexes...
The example shows why we have to traverse all locks held by the transaction - because our transactions are flattened. For each new lock we do a _foreach_ on the...
I see two issues with using recursive mutexes: 1. I don't think there are recursive rwlocks on linux 2. You have to unlock a recursive mutex exactly the same number...