From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jesse Brandeburg Date: Fri, 15 Apr 2016 17:10:10 -0700 Subject: [Intel-wired-lan] [next PATCH S35 11/14] i40e/i40evf: Refactor receive routine In-Reply-To: References: <1460639974-2556-1-git-send-email-harshitha.ramamurthy@intel.com> <1460639974-2556-12-git-send-email-harshitha.ramamurthy@intel.com> <20160415111913.00007d33@unknown> Message-ID: <20160415171010.00002329@unknown> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: intel-wired-lan@osuosl.org List-ID: On Fri, 15 Apr 2016 12:21:24 -0700 Alexander Duyck wrote: > On Fri, Apr 15, 2016 at 11:19 AM, Jesse Brandeburg > wrote: > > On Fri, 15 Apr 2016 10:25:37 -0700 > > Alexander Duyck wrote: > >> If due to only the size I really think this should probably be split > >> into two patches. One for the VF and one for the PF. That way we > >> should only be looking at about 1000 lines of change per patch instead > >> of 2000 which becomes a bit unwieldy. If nothing else it makes the > >> reviews easier to read as we don't end up with a novel with review > >> comments scattered throughout. > > > > Thanks for all your comments Alex, they're really useful. I can split > > the patches into (at least) two. > > > > As far as the comments go, most seem to be tweaks/tuning, and unless I > > missed it not any bugs yet (but I know the patch was huge) > > > > So, would it be at all reasonable to proceed with the code basically > > unchanged except for splitting it up, if no bugs are noticed during > > review? Then I will follow up with a cleanup series? > > The only big issue I saw is the RXE bit check being dropped > completely. If you can incorporate that into the cleanup_headers > function then I would say the rest is mostly just performance bits > that can be cleaned up later. > > - Alex Alex, The patches will be coming from Harshitha, but in the meantime I sent them to you after I broke them all apart and added a check for RXE. I couldn't use cleanup_headers as it doesn't have access to the receive descriptor, so rather than adding an argument I just put it inline. v2: split receive refactor patches into pieces that are smaller. added check for RXE (receive error) bit being set in the descriptor. Thanks again!