Hello Helin, On Thu, Oct 16, 2014 at 2:43 AM, Zhang, Helin <helin.zhang at intel.com> wrote:
> Hi Thomas > > > > Yes, your proposal it the perfect one, also the most complicated one. I > was thinking of that one as well, but we did not have enough time for that > in our 1.8 timeframe. > > In the long run, I agree with you to implement EAL function to access PCI > config space directly. I will try to put it in our plan as soon as > possible, if no objections. > > > > For now, I think the quickest and easiest way might be to write out a > script of using ?setpci?, the Linux command. It is harmless for our code > base, and we can remove it when we have better choice. What do you think? > > > > Thank you very much for the great comments on this topic! I really like it! > Did you make any progress on this ? Actually, looking at Stephen patches ( http://dpdk.org/dev/patchwork/patch/3024/), I think we could go with this approach once both uio and vfio are fine. -- David Marchand
