Good morning, > -----Original Message----- > From: Jeroen Van den Keybus [mailto:jeroen.vandenkey...@gmail.com] > Sent: Tuesday, April 29, 2014 1:52 PM > To: e1000-devel@lists.sourceforge.net > Subject: [E1000-devel] Descriptor writeback settings of i217 > > I'm using a patched e1000e driver for control of a realtime process. > > Basically the code that's responsible for setting up the link on > various e1000 derivatives is simply kept and e1000_xmit_frame / > e1000_intr_msi are called directly. The e1000e (82574) family of cards > has worked fine but I found that the more recent i217 does not do so > well in this area. Whereas a 82574 starts transmitting on the wire > after about 6 us, the i217 only does so after 20 (on an idle system), > which ruins the realtime process. > > While looking into this issue, I noticed that the default RXDCTL > setting isn't touched by e1000e netdev/ich8lan and stays at > 0x00010000, leading to RX descriptors being written back as soon as > one cache line is filled (or one of the RDTR / RADV timers expires). > Is this intentional ? > > Also, the default TXDCTL the i217 ends up with is 0x0141001f. Bit 22 > is reserved per the Intel docs but is set nevertheless in ich8lan. > What does it do ? Additionally, HTHRESH is zero whereas the docs > mandate a nonzero value if PTHRESH is nonzero (is 0x1f now). > > And, of course, would anyone now why the i217 takes so long to fetch > its descriptors and corresponding packet ? > > I apologize if this is too off-topic, but if anyone out there has > practical knowledge of these chips, he or she is probably on this > list. > > Thanks, > > > Jeroen. > . > > ------------------------------------------------------------------------------ > "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE > Instantly run your Selenium tests across 300+ browser/OS combos. Get > unparalleled scalability from the best Selenium testing platform available. > Simple to use. Nothing to install. Get started now for free." > http://p.sf.net/sfu/SauceLabs > _______________________________________________ > E1000-devel mailing list > E1000-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/e1000-devel > To learn more about Intel® Ethernet, visit > http://communities.intel.com/community/wired
When a system with an i217 is idle, the interconnect between the MAC and the PHY (Kumeran) enters a power-saving state (K1). It takes approximately 15us to exit this state. There are examples of disabling the K1 state in the driver, and should be a relatively simple modification for you to try out and see if this fixes your problem. Hope this helps, Dave Ertman ------------------------------------------------------------------------------ "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE Instantly run your Selenium tests across 300+ browser/OS combos. Get unparalleled scalability from the best Selenium testing platform available. Simple to use. Nothing to install. Get started now for free." http://p.sf.net/sfu/SauceLabs _______________________________________________ E1000-devel mailing list E1000-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/e1000-devel To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
