Hi, Are you able to verify that the interrupts are not being delivered? Were you able to see that in /proc/interrupts?
As far as working around the issue if the interrupts are the cause then you may want to work with the Xen maintainers to see what can be done. I have not seen reloading the igb driver resolve this issue when we encountered it in the past. Thanks. - Alex On Mon, May 28, 2018 at 11:49 PM, Kojedzinszky Richárd <kojedzinszky.rich...@euronetrt.hu> wrote: > Dear Alex, > > Thanks for your response. Somehow we are expecting interrupts as a cause, > but we've also tried pci=nomsi kernel setting, also without any success. I > still dont understand why the driver stucks on resetting the hardware. > > I've also tried removing the igb driver and loading it again, without > success. Should such an operation reset the interface and bring it back > alive? > > Thanks, > Kojedzinszky Richárd > Euronet Magyarorszag Informatika Zrt. > > On Mon, 28 May 2018, Alexander Duyck wrote: > >> Date: Mon, 28 May 2018 10:39:40 -0700 >> From: Alexander Duyck <alexander.du...@gmail.com> >> To: Kojedzinszky Richárd <kojedzinszky.rich...@euronetrt.hu> >> Cc: e1000-devel@lists.sourceforge.net >> Subject: Re: [E1000-devel] random igb timeouts >> >> >> On Mon, May 28, 2018 at 12:09 AM, Kojedzinszky Richárd >> <kojedzinszky.rich...@euronetrt.hu> wrote: >>> >>> Dear developers, >>> >>> We have a X8DTU Supermicro system with onboard lan gigabit interfaces, we >>> run debian stretch on it, we use it for virtualisation with xen. >>> Unfortunately, we periodically end up with igb tx timeouts on the host, >>> the >>> driver starts resetting the interface without success, the only solution >>> is >>> to reset the host. >> >> >> When you say you are seeing timeouts I assume you mean dev_watchdog >> messages, and not the driver Tx hang messages. >> >> I have seen similar issues in the past and it is due to MSI-X >> interrupts being handled correctly by either Dom0 or Xen. You might >> try monitoring /proc/interrupts for the interface in question and see >> if interrupts are being delivered or not. You should see at least 1 >> interrupt every 2 seconds on the interface if it is up. If the >> interrupts are the issue then you might check with the Xen community >> to see if they are aware of any interrupt issues. The other option is >> to look at disabling MSI-X interrupts in the system to force the >> device to use either MSI or legacy IRQ. >> >> Thanks. >> >> - Alex ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ 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
