On 06/17/2015 05:17 PM, Tantilov, Emil S wrote: >> -----Original Message----- >> From: Alexander Duyck [mailto:alexander.du...@gmail.com] >> Sent: Wednesday, June 10, 2015 6:26 PM >> Subject: Couple of issues in ixgbe and ethregs >> >> First I am seeing an issue where switching from the out-of-tree driver >> to the in kernel driver for ixgbe w/ x540 is causing some sort of link >> issue. I can switch back to the out-of-tree driver and I am able to get >> link, but if I try to use the in kernel driver after unloading the >> out-of-tree I just cannot get link again. I saw this with a net-next >> pull from about a week ago and the 4.0.3 out of tree ixgbe driver. I >> actually have the port connected back-to-back with the other port of the >> adapter. The driver on the client side is just the standard RHEL7.X. > The link loss is due to the out-of-tree driver powering down the PHY, but > the upstream driver doesn’t know to power it back up. This issue is resolved > upstream by the following patch: > https://git.kernel.org/cgit/linux/kernel/git/davem/net-next.git/commit/drivers/net/ethernet/intel/ixgbe?id=961fac889c0f2e1930092b6de00043cdd1cb2942 > >> Second it looks like ethregs-1.20.0 is reading the SW_FW_SYNC register, >> mislabeled as the GSSR register. This is triggering the SW semaphore to >> be taken by ethregs and after that the driver cannot load as it returns >> EEPROM read errors. This should really be disabled or placed behind the >> -c option since the register has read side effects. > I have removed it altogether and will post a new update to SF. > > Thanks, > Emil >
Thanks for looking into this. - Alex ------------------------------------------------------------------------------ _______________________________________________ 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
