On Fri, 2016-10-21 at 13:55 +0200, Paolo Abeni wrote:
> Avoid using the generic helpers.
> Use the receive queue spin lock to protect the memory
> accounting operation, both on enqueue and on dequeue.
>
> On dequeue perform partial memory reclaiming, trying to
> leave a quantum of forward allocate
On Fri, 2016-10-21 at 15:34 +0200, Paolo Abeni wrote:
> Updating sk_rmem_alloc would still need an atomic operation,
> because it is touched also by the error queue path: we will end up
> adding an atomic operation (or two, when reclaiming the fwd allocated
> memory) inside the critical section. T
On Fri, 2016-10-21 at 05:24 -0700, Eric Dumazet wrote:
> On Fri, 2016-10-21 at 13:55 +0200, Paolo Abeni wrote:
> > Avoid using the generic helpers.
> > Use the receive queue spin lock to protect the memory
> > accounting operation, both on enqueue and on dequeue.
> >
> > On dequeue perform partial
On Fri, 2016-10-21 at 13:55 +0200, Paolo Abeni wrote:
> Avoid using the generic helpers.
> Use the receive queue spin lock to protect the memory
> accounting operation, both on enqueue and on dequeue.
>
> On dequeue perform partial memory reclaiming, trying to
> leave a quantum of forward allocate
Avoid using the generic helpers.
Use the receive queue spin lock to protect the memory
accounting operation, both on enqueue and on dequeue.
On dequeue perform partial memory reclaiming, trying to
leave a quantum of forward allocated memory.
On enqueue use a custom helper, to allow some optimizat
