Description of problem: While diff'ing the boot logs of a -43.el5 and a -44.el5, the following diff showed up on v40z-01.rhts.boston.redhat.com +k8_edac: Unknown symbol edac_mc_del_mc +k8_edac: Unknown symbol edac_mc_find +k8_edac: Unknown symbol edac_mc_add_mc +k8_edac: Unknown symbol edac_mc_alloc +k8_edac: Unknown symbol edac_mc_handle_ue_no_info +k8_edac: Unknown symbol edac_mc_free +k8_edac: Unknown symbol edac_mc_handle_ce +k8_edac: Unknown symbol edac_mc_handle_ue +k8_edac: Unknown symbol edac_mc_handle_ce_no_info edac_mc seems to be loaded, so these warnings are unusual. further info http://rhts.lab.boston.redhat.com/cgi-bin/rhts/test_log.cgi?id=588961 (click on tmp.uz3842 for boot log) Version-Release number of selected component (if applicable): using distro RHEL5.1-Server-20070822.1 How reproducible: not sure Steps to Reproduce: 1.boot -44.el5 2. 3.
adding Aris as he already started to look at it
EDAC MC: Ver: 2.0.1 Aug 27 2007 /* edac_mc gets loaded */ FDC 0 is a post-1991 82077 hdc: ATAPI 24X DVD-ROM drive, 256kB Cache, UDMA(33) Uniform CD-ROM driver Revision: 3.20 k8_edac: Unknown symbol edac_mc_del_mc k8_edac: Unknown symbol edac_mc_find k8_edac: Unknown symbol edac_mc_add_mc k8_edac: Unknown symbol edac_mc_alloc k8_edac: Unknown symbol edac_mc_handle_ue_no_info k8_edac: Unknown symbol edac_mc_free k8_edac: Unknown symbol edac_mc_handle_ce k8_edac: Unknown symbol edac_mc_handle_ue k8_edac: Unknown symbol edac_mc_handle_ce_no_info /* k8_edac fails to load missing symbols present in edac_mc !? */ EDAC MC0: Giving out device to k8_edac Athlon64/Opteron: DEV 0000:00:18.2 EDAC MC1: Giving out device to k8_edac Athlon64/Opteron: DEV 0000:00:19.2 EDAC MC2: Giving out device to k8_edac Athlon64/Opteron: DEV 0000:00:1a.2 EDAC MC3: Giving out device to k8_edac Athlon64/Opteron: DEV 0000:00:1b.2 /* k8_edac loads, edac_mc is responsible by those messages */
Just to make an update on this bug: Still no clue. Checked the module loading process and the module init function (the one that will print "EDAC MC: Ver: 2.0.1 Aug 27 2007") only will be called after making all the symbols available. I also ran a test (in the same machine and the same version used by jburke) by 72h (by changing /etc/rc.local to keep rebooting until find the messages or mail me if the problem is detected), couldn't reproduce the problem. Will try to reproduce again.
Oct 29 10:13:21 manus kernel: Serial: 8250/16550 driver $Revision: 1.90 $ 4 ports, IRQ sharing enabled Oct 29 10:13:21 manus kernel: 8250_pnp: Unknown symbol serial8250_unregister_port Oct 29 10:13:21 manus kernel: 8250_pnp: Unknown symbol serial8250_register_port Oct 29 10:13:21 manus kernel: 8250_pnp: Unknown symbol serial8250_unregister_port Oct 29 10:13:21 manus kernel: 8250_pnp: Unknown symbol serial8250_register_port Oct 29 10:13:21 manus kernel: ACPI: PCI Interrupt 0000:00:1f.5[B] -> GSI 17 (level, low) -> IRQ 17 it appears to be the same bug, only different modules
Adding Jon to the BZ. Perhaps he may have some ideas.
There was a change upstream to the module loading code such that we might end up calling modprobe before the required symbols have been made available - it's being discussed on LKML at the moment, and Jan Glauber is debating this with other folks. He's seeing funky stuff on s390x. I am going to look into this and try to give an informed opinion - how urgent is it this week? Is this for 5.2 now? Jon.
Aris, Jeff B. suggested that you might want to look into this. There's an upstream patch from Rusty, which will dtrt., but it's currently broken (Jan has confirmed), and Rusty is away for the next week. I am planning to look at this at the weekend if there is some time to do so, but let me know if you want to own this issue. Jon.
Upstream thread has the subject: "[PATCH] module loader should not complain about unknown symbol" Jon.
Jon, I don't think I'll have time to work on this issue until next week, so feel free to work on it on weekend. If you don't I'll resume the work next week. Anyway I'll follow this thread.
Still triggers on 2.6.18-90.el5. Attaching reproducer.
Created attachment 302808 [details] reproducer edit the script replacing the edac module used on the current machine
the reproducer doesn't work. working on a new one.
Updating PM score.
I'm pretty sure we fixed this a long time ago. Not sure why this bug is still open. Don, you still see this happening?
I haven't really looked, but feel free to close as someone else should have noticed it by now if it was still a problem. Cheers, Don