> trick there, which is _specifically_ to avoid grabbing the lock in the
> top-level list traversal code, does not work correctly.
Seems, Andrea is right, the trick was too clever to be correct (read, silly 8))
The problem is that irqs add skbs and modify links of the queue,
so that they can be in inconsistent state from the viewpoint of another cpu.
You are right too. This bug is impossible to observe without strong
reordering both by compiler and cache machine. But it is bug yet.
Alexey
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/