You are not logged in.
after upgrading to 3.2.12, everything seems ok. even though I checked the changelong, not sure why.
when I use 3.1.x kernel, seems everything is ok.
but after upgrade to 3.2.x, the system hangs when "Loading user-specified modules". even when I wait for several hours once, it is still busy.
the only solution I got is disable ACPI in /boot/grub/menu.lst "noapic nolapic acpi=off pci=noacpi apci=off irqpoll", with ACPI enable, I cannot even log in to system
but the drawback is if I disable the ACPI, I can only get 1 CPU, actually I have 8 cores( thinkpad w520)
seems everything is ok with linux-lts, except failed when Loading user-specified modules.
could anyone help me? I did many search on google, but get nothing. sorry for any inconvenience. Thanks ahead.
Last edited by cyberyoung (2012-03-25 13:05:25)
Offline
What is in your MODULES list in /etc/rc.conf?
"UNIX is simple and coherent" - Dennis Ritchie; "GNU's Not Unix" - Richard Stallman
Offline
What is in your MODULES list in /etc/rc.conf?
sorry, forgot to attach it:
MODULES=(fuse e1000e iwlagn rtl8192se brcmsmac thinkpad_acpi vboxdrv)
Offline
Maybe try disabling them one at a time and reboot... Painful but most probably effective. I'd personnally bet on vboxdrv :-)
V=RI sweet V=RI
Offline
You may want to search for threads on e1000e. I think there have been a few that have discussed this problem with that module - though I don't remember the details.
I'm also not sure why you're loading multiple wireless modules - though I doubt that is related to this issue.
"UNIX is simple and coherent" - Dennis Ritchie; "GNU's Not Unix" - Richard Stallman
Offline
One thing that changed between 3.1 and 3.2 is the name of the intel wireless module. iwlagn is now just an alias to the new name, iwlwifi, of the module. It's just a guess, but maybe it is worth a try to use the actual module name instead of the alias.
Offline
Thanks for all your help.
sorry, I used following modules this time
MODULES=(fuse e1000e iwlagn iwlwifi thinkpad_acpi)
iwlagn for 3.0 lts, iwlwifi for 3.2
everything is ok with lts, but still hangs with 3.2
I doubt if iwlwifi is the problem, it should complain about the error and show Loaidng user-specified modules [FAIL](which is actually when I log in to 3.0 lts), not hangs, I even cannot log in to system.
then I disable e1000e from MODULES, loading user-specified modules passed, but hangs in "Waiting for UDev uevents to be processed"....
Offline
Again, I don't recall the details of those e1000e threads, as I only skimmed them (I don't use that module). But this sounds just like the same problem.
I know the inital solution for many users on those threads was to stick with the older kernel or do without ethernet (ie remove e1000e). I suspect further progress has been made; you could find out by checking those threads.
EDIT: to make sure this is the right path to be sniffing on, you could temporarily remove e1000e and see if the new kernel fully boots.
Last edited by Trilby (2012-02-13 15:49:41)
"UNIX is simple and coherent" - Dennis Ritchie; "GNU's Not Unix" - Richard Stallman
Offline
Again, I don't recall the details of those e1000e threads, as I only skimmed them (I don't use that module). But this sounds just like the same problem.
I know the inital solution for many users on those threads was to stick with the older kernel or do without ethernet (ie remove e1000e). I suspect further progress has been made; you could find out by checking those threads.
EDIT: to make sure this is the right path to be sniffing on, you could temporarily remove e1000e and see if the new kernel fully boots.
Thanks, Trilby. As I said in last post, I removed e1000e, and "loading user-specified modules" passed, but hangs in "waiting for udev uevents to be processed" for ever.
would you please kindly tell me about the key word of e1000 thread? so I can search for it. how about "e1000e hangs boot"?
I also tried to add "nomodeset" to kernel boot like some thread said, but no help either.
now I tried to do following steps to log into my 3.2.x kernel:
1. remove /etc/modprobe.d/blacklist.conf, which has "mei" in it(I added it before).
2. I modified /etc/rc.d/functions:
change "udevadm settle" to "udevadm settle --timeout=0"
according to the manpage, "--timeout=seconds ...... A value of 0 will check if the queue is empty and always return immediately."
but not sure if this is harm to my system?
I think I should go on to dig the problem. from some thread, some one downgrade the kernel, saying like a bug in kernel, but I'm not sure about it. It's really funny that I have fight with it several weeks now, still get no perfect solution:)
Offline
I'm wrong..... it only succeed once, and now boot will hang randomly, once when "loading user-specified modules", once when "waiting for udev uevents to be processed", once "checking file systems"! I'm confused...who can help me? many thanks.
Offline
It's very simple. It's not any of those things causing the hang. You just assume that's the case, because it's the last thing printed on the screen before the hang. As to what exactly is wrong? No idea. Not possible to know without some serious debugging.
Offline
......I can say nothing now
Offline
Finally I got this problem resolve by following post:
https://mzanfardino.wordpress.com/2012/ … ring-udev/
so according to it, it's because the combition of nvidia and VT-D. we have several solutions:
1. disable VT
2. use integrated graphic card
3. use discrete graphic card and disable VT-d
because I have to use kvm, I tried 2&3, both are ok. many thanks to this guy!
Offline
It's been a year now. how time fly
Offline