Opened 18 years ago

Closed 17 years ago

#1714 closed defect (invalid)

LFS-6.1.1: USB segfaults on "legacy" system

Reported by: Paul Rogers Owned by: Matthew Burgess
Priority: normal Milestone:
Component: Book Version: 6.1.1
Severity: normal Keywords: USB segfault


This is a Socket-7 Pentium system with an unused header block for USB, support disabled in the BIOS. Building module-inits-3.1 with book instructions (--prefix="") put programs in /usr/local/..., --prefix=/ got them into /sbin, et al. Booting produces the following errors: ./hotplug.functions: line 127: 1565 Segmentation fault $MODPROBE -n $MODULE >.dev.null 2>&1 /etc/hotplug/usb.rc: line 122: "X" Segmentation fault modprobe -q "Y" >dev/null 2>&1

X Y 1569 usbcore 1570 ehci-hcd 1571 ohci-hcd 1572 uhci-hcd 1573 usb-ohci 1574 uhci 1575 usb-uhci

Kernel is built with USB=off.

It looks like init.d/hotplug looks for /etc/hotplug/usb.rc and then runs it with "maybe-start" which tries to modprobe usbcore assuming it's there. Is that right? But shouldn't it check for USB presence/support first? Did somebody forget "legacy" systems don't necessarily have/need USB? /proc/bus/usb doesn't exist here--well, the BIOS has some support and the ports exist, I suppose.

Change History (5)

comment:1 by Matthew Burgess, 18 years ago

Owner: changed from lfs-book@… to Matthew Burgess
Status: newassigned

Thanks for the report Paul. We're heading towards being able to finally drop the entire hotplug package and related bootscripts, so I'd like to mark this bug as WONTFIX (or even FIXED, I guess). However, it's important that our new udev-only setup works on your configuration. Are you in a position to be able to build from the udev_update branch ( and test that your kernel configuration works as you'd expect?

comment:2 by alexander@…, 18 years ago

This is either upstream module-init-tools (or kernel) bug, or your local issue caused by i686 code in one of the packages. It is not an error to attempt brute-force loading of modules, as the hotplug script currently does for USB host controllers. Loading a module for hardware that doesn't exist should result in the module being loaded but attached to no hardware.

Please look if anything interesting is in dmesg output, because modprobe segfault usually accompanies some kernel oops.

This is neither "wontfix" nor "fixed". After udev update that would become "worksforme", because this change (not loading the USB host controller driver) only hides the segfault.

comment:3 by Matthew Burgess, 18 years ago

Thanks for your input Alexander. Paul, could you therefore try upgrading to module-init-tools-3.2.2 and see if that fixes your issue, rather than potentially hiding the problem by migrating to the new udev-only setup.

comment:4 by archaic@…, 18 years ago

Has anyone heard back from Paul regarding this issue? If not, I suggest closing this bug since no one else has claimed to have the same problem.

comment:5 by archaic@…, 17 years ago

Resolution: invalid
Status: assignedclosed

This a non-issue per:

Closing ticket as invalid.

Note: See TracTickets for help on using tickets.