Submitting just the PMD for integration makes sense. I will remove all the references to nfp_uio.
My doubt is with documentation. Working with the NFP PMD will not be possible without nfp_uio. We could modify the documentation saying it is possible to use igb_uio, but this is not the right thing to do (pci mask will be wrong). So, would it be acceptable to submit a new PMD without any documentation by now? I prefer this for the sake of integration than giving wrong or incomplete documentation. Thanks On Wed, Oct 21, 2015 at 8:40 PM, Alejandro Lucero < alejandro.lucero at netronome.com> wrote: > > > On Wed, Oct 21, 2015 at 5:03 PM, Thomas Monjalon < > thomas.monjalon at 6wind.com> wrote: > >> 2015-10-21 16:57, Alejandro Lucero: >> > I understand interest for not having another UIO driver does exist. We >> > could maintain an external nfp_uio by now till either we get rid of it >> or >> > we definitely find out it is really needed. any chance to accept >> nfp_uio by >> > now? >> >> No, there are some work currently to get rid of igb_uio. >> So there are little chances to accept nfp_uio one day. >> Please take the first step of integrating your PMD without link interrupt. >> Later we'll be able to discuss how to mitigate the interrupt issue. >> > > Ok. I will create a new patchset version without nfp_uio. > > By the way, that work with igb_uio is about the patches to > pci_uio_generic? I thought there was some reticence from the maintainer for > adding pci bus master there. > > >