Well, I don't know if this can be somehow related to a bug in the driver, but on
my system, after I've unplugged the USB trasnsitter, ksoftirqd/0 begin to use
89% of my CPU time (top). I suspect that there are some issue that should be
fixed somewhere.
Actually, it's probably me who broke the driver. I think Greg merged
most of my patches after 2.6.28 came out. Sandro, if you have some
time, please consider doing git bisect to find out which changeset
causes the problem.
Well, actually my 2.6.29 copy seems to be broken:
arch/x86/kernel/amd_iommu.o: In function `debugfs_create_size_t':
amd_iommu.c:(.text+0x0): multiple definition of `debugfs_create_size_t'
arch/x86/kernel/kdebugfs.o:kdebugfs.c:(.text+0x0): first defined here
make[1]: *** [arch/x86/kernel/built-in.o] Error 1
make: *** [arch/x86/kernel] Error 2
# git pull
Already up-to-date.
# git status
# On branch master
# Changed but not updated:
# (use "git add <file>..." to update what will be committed)
# (use "git checkout -- <file>..." to discard changes in working directory)
#
# modified: drivers/staging/winbond/wbusb.c
#
no changes added to commit (use "git add" and/or "git commit -a")
I don't know too much how to use git to do what you're asking. I've some years
of experience with svn and cvs but git is quite a new tool for me.
I'll wait a git kernel that compile while I'll try to find out how to do what
you're asking.
I have not so much free time to spend in a checkout / build / test if the test
phase need a kernel reinstall / system reboot. If it's only the driver module I
think I can try to test it for each changeset in a couple of weeks /starting
from the day I know how to bisect :-) ).