Hi Sandro,
On Mon, Jan 26, 2009 at 9:31 PM, Sandro Bonazzola
So the driver doesn't work in 2.6.29-rc1 but does in 2.6.28.1? I
wonder what changed in the networking/wireless stack that broke the
winbond driver.... I'm cc'ing netdev even though this is a driver in
staging.Pekka
--
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.Pekka
--
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1Well, 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 beWell, 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 :-) ).- --
Sandro
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.orgiEYEARECAAYFAkl+GWIACgkQV871CybFezlRXACfQTBrMm2hL9b+o5oU1/6/xZ+f
gaYAnjyyawOFPBPIe/scq5pYupNrRLa3
=d1SV
-----END PGP SIGNATURE-----
--
Here's a tutorial on git bisect.
Assuming 2.6.29-rc1 is broken, you initialize the bisection with:
git bisect start -- drivers/staging/winbond/
git bisect good v2.6.28
git bisect bad v2.6.29-rc1Then you continue doing this:
<apply my patch, compile, test>
and
git bisect good
or
git bisect bad
depending on whether the kernel worked or not. Repeat until git bisect
tells you which changeset broke the kernel. The process shouldn't take
more than 5 compiles/reboots.Pekka
--
| Linus Torvalds | Linux 2.6.21-rc4 |
| Greg Kroah-Hartman | [PATCH 008/196] Chinese: add translation of volatile-considered-harmful.txt |
| Andrew Morton | -mm merge plans for 2.6.23 |
| Stephane Eranian | Re: [PATCH] fix up perfmon to build on -mm |
git: | |
| David Miller | [GIT]: Networking |
| David Miller | Re: [PATCH] pkt_sched: Destroy gen estimators under rtnl_lock(). |
| Gerrit Renker | [PATCH 27/37] dccp: Integration of dynamic feature activation - part 2 (server side) |
| Natalie Protasevich | [BUG] New Kernel Bugs |
| Linus Torvalds | Re: silent semantic changes with reiser4 |
| jim owens | Re: ext4 - getting at birth time (file create time) and getting/setting nanosecond... |
| Alan Cox | Re: impact of 4k sector size on the IO & FS stack |
| Peter Zijlstra | Re: + mm-balance_dirty_pages-reduce-calls-to-global_page_state-to-reduce-c ache-re... |
