Re: [RFC PATCH 0/5] Introduce low-latency one-shot rx interrupt into DPDK with polling/interrupt switch control example

[Ivan Boule <ivan.boule-pdR9zngts4EAvxtiuMwx3w@public.gmane.org>]

Hello Danny,


On 01/20/2015 10:53 AM, Danny Zhou wrote:
> DPDK interrupt notification/handling mechanism is based on UIO with
> below limitation:
> 1) It is designed to handle LSC interrupt only with inefficient
> suspended pthread wakeup procedure (e.g. UIO wakes up LSC interrupt
> handling thread which then wakes up DPDK polling thread). In this way,
> it introduces non-deterministic wakeup latency for DPDK polling thread
> as well as packet latency if it is used to handle Rx interrupt.
> 2) UIO only supports a single interrupt vector which has to been shared
> by LSC interrupt and interrupts assigned to dedicated rx queues.
>
> This patchset includes below features:
> 1) Enable one-shot rx queue interrupt in ixgbe PMD for PF as well as VF.
> 2) Build on top of the VFIO mechanism instead of UIO, so it could support
> up to 64 interrupt vectors for rx queue interrupts.
> 3) Have 1 DPDK polling thread handle per Rx queue interrupt with a
> dedicated VFIO eventfd, which eliminates non-deterministic pthread wakeup
> latency in user space.
> 4) Demonstrate interrupts control APIs and userspace NAIP-like
> polling/interrupt switch algorithms in L3fwd-power example.

In the patch 3/5:

- the function eal_intr_process_rx_interrupts() systematically returns 
0. It should be declared as "void" (and get rid of the return 0).

- the function eal_intr_handle_rx_interrupts() loops again when 
epoll_wait() returns with errno == EINTR.
   It should also return in this case.

- the function rte_eal_wait_rx_intr() should not invoke rte_panic() but 
instead return "-errno" upon system call failure.

With these fixes:
Acked-By: Ivan Boule <ivan.boule-pdR9zngts4EAvxtiuMwx3w@public.gmane.org>

-- 
Ivan Boule
6WIND Development Engineer