From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bowers, AndrewX Date: Mon, 9 Sep 2019 21:19:17 +0000 Subject: [Intel-wired-lan] [PATCH v2] i40e: fix potential RX buffer starvation for AF_XDP In-Reply-To: <20190909165538.8125-1-jeffrey.t.kirsher@intel.com> References: <20190909165538.8125-1-jeffrey.t.kirsher@intel.com> Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: intel-wired-lan@osuosl.org List-ID: > -----Original Message----- > From: Intel-wired-lan [mailto:intel-wired-lan-bounces at osuosl.org] On > Behalf Of Jeff Kirsher > Sent: Monday, September 9, 2019 9:56 AM > To: intel-wired-lan at lists.osuosl.org > Cc: Karlsson, Magnus > Subject: [Intel-wired-lan] [PATCH v2] i40e: fix potential RX buffer starvation > for AF_XDP > > From: Magnus Karlsson > > When the RX rings are created they are also populated with buffers so that > packets can be received. Usually these are kernel buffers, but for AF_XDP in > zero-copy mode, these are user-space buffers and in this case the > application might not have sent down any buffers to the driver at this point. > And if no buffers are allocated at ring creation time, no packets can be > received and no interrupts will be generated so the NAPI poll function that > allocates buffers to the rings will never get executed. > > To rectify this, we kick the NAPI context of any queue with an attached > AF_XDP zero-copy socket in two places in the code. Once after an XDP > program has loaded and once after the umem is registered. > This take care of both cases: XDP program gets loaded first then AF_XDP > socket is created, and the reverse, AF_XDP socket is created first, then XDP > program is loaded. > > Fixes: 0a714186d3c0 ("i40e: add AF_XDP zero-copy Rx support") > Signed-off-by: Magnus Karlsson > --- > drivers/net/ethernet/intel/i40e/i40e_xsk.c | 5 +++++ > 1 file changed, 5 insertions(+) Tested-by: Andrew Bowers