When a user runs the "ethtool -p ethX" to identify the physical interface by flashing the appropriate interface LED, and then removes the cable at that interface, the keyboard locks up. This is naturally annoying to customers with multi-port systems, who wish to use ethtool -p so that they can disconnect/connect interface cables to appropriate ports. I am told that this is because "the RTNL lock is held by ethtool, and the link [down] message needs the RTNL", which would, I think, make this a kernel issue. This has been known about within INTEL for over 3 years, but only recently come to my attention. I have verified it on a 2.6.30 system using a default kernel configuration and the e1000e driver.
(switched to email. Please respond via emailed reply-to-all, not via the bugzilla web interface). On Wed, 9 Sep 2009 00:06:10 GMT bugzilla-daemon@bugzilla.kernel.org wrote: > http://bugzilla.kernel.org/show_bug.cgi?id=14147 > > Summary: System unresponsive if I unplug network cable while > ethtool -p is running > Product: Networking > Version: 2.5 > Kernel Version: 2.6.30 > Platform: All > OS/Version: Linux > Tree: Mainline > Status: NEW > Severity: normal > Priority: P1 > Component: Other > AssignedTo: acme@ghostprotocols.net > ReportedBy: david.graham@intel.com > Regression: No > > > When a user runs the "ethtool -p ethX" to identify the physical interface by > flashing the appropriate interface LED, and then removes the cable at that > interface, the keyboard locks up. This is naturally annoying to customers > with > multi-port systems, who wish to use ethtool -p so that they can > disconnect/connect interface cables to appropriate ports. > > I am told that this is because "the RTNL lock is held by ethtool, and the > link > [down] message needs the RTNL", which would, I think, make this a kernel > issue. > > This has been known about within INTEL for over 3 years, but only recently > come > to my attention. I have verified it on a 2.6.30 system using a default kernel > configuration and the e1000e driver. > hah. cute.
I reproduced this issue on F14 and e1000e driver.
Why was this assigned to e1000e owner (bruce)? It is a kernel bug that is basically a design issue, should probably be assigned to network core owner. Not sure if it will ever be fixed, but would be nice.
This has been fixed for almost a year now.
Thanks