Whenever the e1000 driver is enabled and Debug shared IRQ handlers I get a kernel oops when gentoo brings up the ethernet card from ifconfig Reproducible: Always Steps to Reproduce: 1.install latest ck-sources (2.6.21_p2) 2.enable e1000 3.enable Debug shared IRQ handlers Actual Results: Jun 27 08:47:11 itmanager-lnx BUG: unable to handle kernel NULL pointer dereference at virtual address 00000000 Jun 27 08:47:11 itmanager-lnx printing eip: Jun 27 08:47:11 itmanager-lnx 00000000 Jun 27 08:47:11 itmanager-lnx *pde = 00000000 Jun 27 08:47:11 itmanager-lnx Oops: 0000 [#1] Jun 27 08:47:11 itmanager-lnx PREEMPT SMP Jun 27 08:47:11 itmanager-lnx Modules linked in: snd_pcm_oss snd_mixer_oss cdc_acm ntfs cifs i2c_i801 i2c_core snd_intel8x0 s nd_ac97_codec ac97_bus snd_pcm snd_timer snd snd_page_alloc Jun 27 08:47:11 itmanager-lnx CPU: 1 Jun 27 08:47:11 itmanager-lnx EIP: 0060:[<00000000>] Not tainted VLI Jun 27 08:47:11 itmanager-lnx EFLAGS: 00010246 (2.6.21-ck2 #2) Jun 27 08:47:11 itmanager-lnx EIP is at 0x0 Jun 27 08:47:11 itmanager-lnx eax: f781c500 ebx: f8840000 ecx: 00000000 edx: f781de80 Jun 27 08:47:11 itmanager-lnx esi: f781c500 edi: 00000000 ebp: f781c728 esp: cb7e1e74 Jun 27 08:47:11 itmanager-lnx ds: 007b es: 007b fs: 00d8 gs: 0033 ss: 0068 Jun 27 08:47:11 itmanager-lnx Process ifconfig (pid: 5179, ti=cb7e0000 task=f7fed580 task.ti=cb7e0000) Jun 27 08:47:11 itmanager-lnx Stack: c028883a fffffff4 f76edbe0 00000000 c0288777 c0136081 00000010 01fffffe Jun 27 08:47:11 itmanager-lnx f781c500 f781c000 f781c000 00000000 c028820d f781c000 f781c000 f781c500 Jun 27 08:47:11 itmanager-lnx 00000000 f781c500 00000000 c028b042 f781c000 00001043 00001002 c0392bdf Jun 27 08:47:11 itmanager-lnx Call Trace: Jun 27 08:47:11 itmanager-lnx [<c028883a>] e1000_intr+0xc3/0x113 Jun 27 08:47:11 itmanager-lnx [<c0288777>] e1000_intr+0x0/0x113 Jun 27 08:47:11 itmanager-lnx [<c0136081>] request_irq+0xa7/0xce Jun 27 08:47:11 itmanager-lnx [<c028820d>] e1000_request_irq+0xad/0xe6 Jun 27 08:47:11 itmanager-lnx [<c028b042>] e1000_open+0x43/0xc0 Jun 27 08:47:11 itmanager-lnx [<c0392bdf>] dev_open+0x2b/0x62 Jun 27 08:47:11 itmanager-lnx [<c0391628>] dev_change_flags+0x47/0xe4 Jun 27 08:47:11 itmanager-lnx [<c03c487a>] devinet_ioctl+0x23e/0x55b Jun 27 08:47:11 itmanager-lnx [<c038a69e>] sock_ioctl+0x0/0x16a Jun 27 08:47:11 itmanager-lnx [<c038a7e9>] sock_ioctl+0x14b/0x16a Jun 27 08:47:11 itmanager-lnx [<c038a69e>] sock_ioctl+0x0/0x16a Jun 27 08:47:11 itmanager-lnx [<c015c967>] do_ioctl+0x1f/0x62 Jun 27 08:47:11 itmanager-lnx [<c015cbf0>] vfs_ioctl+0x246/0x258 Jun 27 08:47:11 itmanager-lnx [<c015cc35>] sys_ioctl+0x33/0x4a Jun 27 08:47:11 itmanager-lnx [<c01026b8>] sysenter_past_esp+0x5d/0x81 Jun 27 08:47:11 itmanager-lnx [<c03f0000>] _write_lock_irqsave+0x35/0x70 Jun 27 08:47:11 itmanager-lnx ======================= Jun 27 08:47:11 itmanager-lnx Code: Bad EIP value. Jun 27 08:47:11 itmanager-lnx EIP: [<00000000>] 0x0 SS:ESP 0068:cb7e1e74 Expected Results: I expect eth0 to come up and get an address from dhcpd. Kernel should not Oops! │ Symbol: E1000 [=y] │ │ Prompt: Intel(R) PRO/1000 Gigabit Ethernet support │ │ Defined at drivers/net/Kconfig:1943 │ │ Depends on: NET && NETDEVICES && !UML && PCI │ │ Location: │ │ -> Device Drivers │ │ -> Network device support │ │ -> Network device support (NETDEVICES [=y]) │ │ -> Ethernet (1000 Mbit) Symbol: DEBUG_SHIRQ [=y] │ │ Prompt: Debug shared IRQ handlers │ │ Defined at lib/Kconfig.debug:80 │ │ Depends on: DEBUG_KERNEL && GENERIC_HARDIRQS │ │ Location: │ │ -> Kernel hacking │ │ -> Kernel debugging (DEBUG_KERNEL [=y])
Does this still happen with the same config on either gentoo-sources or vanilla-sources or only with ck?
(In reply to comment #1) > Does this still happen with the same config on either gentoo-sources or > vanilla-sources or only with ck? > Yes, this same config worked fine on gentoo-sources I forgot to add that when I disable 'Debug shared IRQ handlers' in the kernel, it works fantastic. Actually, I turned off all kernel debugging. The reason I believe it's narrowed down to 'Debug shared IRQ handlers' is because of this: http://www.ussg.iu.edu/hypermail/linux/kernel/0702.1/2931.html Don't ask me why I even had kernel debugging on. It's just an option I have always had on in the kernel.
(In reply to comment #2) > (In reply to comment #1) > > Does this still happen with the same config on either gentoo-sources or > > vanilla-sources or only with ck? > > > > Yes, this same config worked fine on gentoo-sources > > I forgot to add that when I disable 'Debug shared IRQ handlers' in the kernel, > it works fantastic. Actually, I turned off all kernel debugging. The reason I > believe it's narrowed down to 'Debug shared IRQ handlers' is because of this: > http://www.ussg.iu.edu/hypermail/linux/kernel/0702.1/2931.html > > Don't ask me why I even had kernel debugging on. It's just an option I have > always had on in the kernel. > Ok, just to see if you are hitting the same bug you could try the patch that is proposed in that thread: http://www.ussg.iu.edu/hypermail/linux/kernel/0702.1/3023.html The patch was applied and reverted because the fix broke MSI, but I don't know for sure if an alternative fix was ever put in. http://www.ussg.iu.edu/hypermail/linux/kernel/0702.1/3023.html It may be related to this patch which is included in the latest gentoo sources but not ck-sources which would explain why gentoo-sources worked: http://lkml.org/lkml/2007/6/8/89 I bumped ck-sources to _p2-r1 to include genpatches-4 which has that patch so give that a try first once it hits the portage tree in an hour or so. If that doesn't fix it try the patch I suggested above, if it does we are good to go :-)
(In reply to comment #3) > I bumped ck-sources to _p2-r1 to include genpatches-4 which has that patch so > give that a try first once it hits the portage tree in an hour or so. If that > doesn't fix it try the patch I suggested above, if it does we are good to go > :-) > Okay, I will sync in an hour and try it again and post the results. I have some questions for you: Should a novice even have kernel debugging on? Does it have any noticeable negative effects on the performance of the system?
ck-sources were dropped from portage, maybe this bug should be closed
Oops, yeah, didn't clear this out. Closing.