Hi Aditya, Thanks for the information. Could you please provide the full dmesg output from boot to the time you see those spikes. I am curious about the error messages you are seeing.
Do you know if you are running the latest BIOS? Maybe check for updates. Do you have AMT enabled? Maybe try enabling/disabling it and see if it would make a difference. It is enabled on my system. Also if you could provide the output of ethtool -e eth0 that would be great. I will continue to play with my system and see if I can get a repro. Have you noticed any patterns - for example does it matter if the system is running on battery vs. power. Also you mentioned that you don't have cable connected - do you see the same spikes when cable is present? Thanks, Emil -----Original Message----- From: Aditya Rajgarhia [mailto:adity...@gmail.com] Sent: Tuesday, January 05, 2010 4:10 PM To: Tantilov, Emil S Cc: Brandeburg, Jesse; e1000-devel@lists.sourceforge.net; Allan, Bruce W Subject: Re: [E1000-devel] Problem with Linux 2.6.32 and e1000e Emil, that is strange. I wonder if mine being a T61p has some different hardware (although the main difference is supposed to be the video card). Please check the output of lspci that I had attached if needed. Since it is proving difficult to determine what in the kernel is causing me the problem, I decided to run the problematic 2.6.30.2 kernel (same configuration that I attached earlier) with different versions of the e1000e driver (from http://sourceforge.net/projects/e1000) and here is what I found (after booting into each a few times): Kernel-2.6.30.2 with e1000e-1.1.2 (the module shipped with the kernel) causes CPU spikes of ~50% as I reported earlier. Kernel-2.6.30.2 with e1000e-1.0.15 causes CPU spikes of ~50%. Kernel-2.6.30.2 with e1000e-1.0.2.5 causes CPU spikes of ~10%. Kernel-2.6.30.2 with e1000e-0.5.18.3 uses CPU spikes of ~5% but also causes bad irq messages while booting, so I am guessing it won't work (as I said, I don't have an ethernet connection available unless necessary to test something). With the e1000e-1.x drivers I see a message "0000:00:19.0: 0000:00:19.0: Hardware Error" while booting. The rest of the driver related messages are the same as with the kernel that works. I did use the ethernet even a few days ago with 2.6.30.4 and it worked, so I don't think the hardware is corrupted. Without the e1000e module loaded, there is no CPU spikes caused by the events thread. I am curious to know why there would be any, even negligible, CPU usage caused by an ethernet driver with no link connected. For the time being, I can use e1000e-1.0.2.5 if I need ethernet access, but given that this version may not compile with future kernels, hopefully the issue can be resolved. Perhaps the above numbers will help to locate the cause. Let me know if anything else is needed or if I should try something different Thanks, Aditya On 01/05/2010 02:05 PM, Tantilov, Emil S wrote: > Aditya, > > I installed vanilla 2.6.32.2 kernel using the config file from your first > email on my T61 laptop and at least off the bat did not see the issue you are > describing. > > I tested on the docking station and also with just a battery. Restarted the > OS couple of times, but the CPU usage is in line (9x% idle). > > Just an update on where we are atm with respect to this issue. > > Thanks, > Emil > > -----Original Message----- > From: Aditya Rajgarhia [mailto:adity...@gmail.com] > Sent: Monday, January 04, 2010 3:49 PM > To: Brandeburg, Jesse > Cc: e1000-devel@lists.sourceforge.net; Allan, Bruce W > Subject: Re: [E1000-devel] Problem with Linux 2.6.32 and e1000e > > It's indeed a ThinkPad T61p running Arch Linux 2009.08. > > I just tried with 2.6.32.2 from kernel.org and the problem is still > there. I used make oldconfig with the config file from my 2.6.30.4 > kernel (which does not have this problem) so it doesn't seem to be some > config option that the Arch Linux kernel enabled that is causing the > problem. > > I should have mentioned that this happens when there is no link/cable > present so there is no traffic at all. However, problem is not there > with 2.6.30.4 under the same circumstances. > > I have not tested it with an ethernet connection since I only have > wireless access at the moment, but could perhaps test with an ethernet > connection if needed (but I guess that will not help). Also, I do not > see any strange output or errors in the system log upon loading or > unloading the module. > > Thanks, > Aditya > > On 01/04/2010 02:39 PM, Brandeburg, Jesse wrote: > >> Also, do you have a continuous stream of receive traffic arriving? The >> events/0 thread freaking out is typically due to IPv6 setup, but under >> normal traffic conditions it only takes a few seconds to complete. >> >> -----Original Message----- >> From: Allan, Bruce W [mailto:bruce.w.al...@intel.com] >> Sent: Monday, January 04, 2010 12:09 PM >> To: Aditya Rajgarhia; e1000-devel@lists.sourceforge.net >> Subject: Re: [E1000-devel] Problem with Linux 2.6.32 and e1000e >> >> Apparently this is on a ThinkPad T61, correct? >> >> What version of Arch Linux are you running? 2009.08 or other? >> >> Have you tried a kernel.org version of the 2.6.32? >> >> >> >>> -----Original Message----- >>> From: Aditya Rajgarhia [mailto:adity...@gmail.com] >>> Sent: Saturday, January 02, 2010 1:09 PM >>> To: e1000-devel@lists.sourceforge.net >>> Subject: [E1000-devel] Problem with Linux 2.6.32 and e1000e >>> >>> Hi, >>> >>> I upgraded from 2.6.30.4 to 2.6.32.2 and am experiencing problems with >>> e1000e. The thread events/0 uses half the CPU (or all of one core I >>> guess) and the keyboard doesn't work properly (only some of the keys get >>> typed). >>> >>> I have isolated this problem to e1000e by disabling each hardware >>> independently (the issue appears if I load the e1000e module and the >>> eth0 interface, and disappears if I do the opposite). >>> >>> I used to compile my own kernel but this kernel which is giving problems >>> is the stock kernel from Arch Linux with tuxice, so I'm not sure if >>> there is something enabled in this kernel which causes the issue. >>> Anyway, it would be best if we can resolve it. >>> >>> Attached is the .config, and the outputs of lspci -vvv and >>> /proc/interrupts. >>> >>> Please let me know if you need any other information. >>> >>> Thanks, >>> >>> Aditya >>> >>> >> ------------------------------------------------------------------------------ >> This SF.Net email is sponsored by the Verizon Developer Community >> Take advantage of Verizon's best-in-class app development support >> A streamlined, 14 day to market process makes app distribution fast and easy >> Join now and get one step closer to millions of Verizon customers >> http://p.sf.net/sfu/verizon-dev2dev >> _______________________________________________ >> E1000-devel mailing list >> E1000-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/e1000-devel >> >> > ------------------------------------------------------------------------------ > This SF.Net email is sponsored by the Verizon Developer Community > Take advantage of Verizon's best-in-class app development support > A streamlined, 14 day to market process makes app distribution fast and easy > Join now and get one step closer to millions of Verizon customers > http://p.sf.net/sfu/verizon-dev2dev > _______________________________________________ > E1000-devel mailing list > E1000-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/e1000-devel > ------------------------------------------------------------------------------ This SF.Net email is sponsored by the Verizon Developer Community Take advantage of Verizon's best-in-class app development support A streamlined, 14 day to market process makes app distribution fast and easy Join now and get one step closer to millions of Verizon customers http://p.sf.net/sfu/verizon-dev2dev _______________________________________________ E1000-devel mailing list E1000-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/e1000-devel
