Bug 72611

Created attachment 130201 [details]
lsusb -v

On Fri, Mar 21, 2014 at 06:55:44PM +0000, bugzilla-daemon@bugzilla.kernel.org wrote:
> Merge tag 'driver-core-3.13-rc1' of
> git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
> Pull driver core / sysfs patches from Greg KH: "Here's the big driver core /
> sysfs update for 3.13-rc1.
> 
>
> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=0324e74534241f3f00910ec04ef67de1fe1542f4

That's a merge, not a "real" commit.  Can you find the commit in that
merge that is causing you problems?

(In reply to Greg Kroah-Hartman from comment #2)

> That's a merge, not a "real" commit.  Can you find the commit in that
> merge that is causing you problems?

Hi Greg. Ok, so I did a bisect between 1071ec7 (good) and 0324e74 (bad) and I got the following:

64c862a839a8db2c02bbaa88b923d13e1208919d is the first bad commit
commit 64c862a839a8db2c02bbaa88b923d13e1208919d
Author: Joe Perches <joe@perches.com>

    devres: add kernel standard devm_k.alloc functions

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=64c862a839a8db2c02bbaa88b923d13e1208919d

********************************************************

The problem is: this commit causes a kernel panic. In fact several commits caused a kernel panic, so I couldn't test the bug itself:

git bisect start
# good: [1071ec7bc2dabd0a9d12a1ae5570f4fd3ba944ca] Merge tag 'char-misc-3.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc
git bisect good 1071ec7bc2dabd0a9d12a1ae5570f4fd3ba944ca
# bad: [0324e74534241f3f00910ec04ef67de1fe1542f4] Merge tag 'driver-core-3.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
git bisect bad 0324e74534241f3f00910ec04ef67de1fe1542f4
# good: [f9b9a6217cf10fd5d3002627cc13c4789a777213] sysfs: prepare path write for unified regular / bin file handling
git bisect good f9b9a6217cf10fd5d3002627cc13c4789a777213
# bad: [fb3fed7926545e44ce36574e1b1c5cdeb018db5c] ide: convert bus code to use dev_groups
git bisect bad fb3fed7926545e44ce36574e1b1c5cdeb018db5c
# bad: [64c862a839a8db2c02bbaa88b923d13e1208919d] devres: add kernel standard devm_k.alloc functions
git bisect bad 64c862a839a8db2c02bbaa88b923d13e1208919d
# good: [bcc8edb52f05c1a9e75118d6b3bc04996a750593] driver core: remove dev_attrs from struct class
git bisect good bcc8edb52f05c1a9e75118d6b3bc04996a750593
# good: [1461c5be7becc6e65dba5cadb31fb5f4339609f5] kobject: show debug info on delayed kobject release
git bisect good 1461c5be7becc6e65dba5cadb31fb5f4339609f5
# good: [d723a92dd465d549bf79dd481c09d59f0be02936] sysfs/bin: Fix size handling overflow for bin_attribute
git bisect good d723a92dd465d549bf79dd481c09d59f0be02936
# first bad commit: [64c862a839a8db2c02bbaa88b923d13e1208919d] devres: add kernel standard devm_k.alloc functions

But the "bad" commits are "bad" because they causes a kernel panic (I couldn't test for the bug itself).

************************************************

Do you have any suggestions since the kernel panic is preventing me to test correctly? Thanks.

Created attachment 130281 [details]
64c862a commit kernel panic screenshot

Maybe it's useless, but here's the kernel panic. It seems the 64c862a commit prevents the kernel from mounting the root partition.

Ok, to properly bisect I had to apply 6fffcfa commit on all stages of bisect (6fffcfa corrects the defective 64c862a commit).

Now I found the real bad commit:

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=f1125f81feca956288ef4ae4fcf47961865ab2e2

commit f1125f81feca956288ef4ae4fcf47961865ab2e2
Author: Deng-Cheng Zhu <dengcheng.zhu@imgtec.com>
Date:   Fri Oct 4 15:14:34 2013 -0700

    USB/host: Use existing macros instead of hard-coded values in uhci-debug.c
    
    Now that UHCI IO registers have been defined in uhci-hcd.h, use them.
    
    Reviewed-by: James Hogan <james.hogan@imgtec.com>
    Signed-off-by: Deng-Cheng Zhu <dengcheng.zhu@imgtec.com>
    Acked-by: Alan Stern <stern@rowland.harvard.edu>
    Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

***

Now we know the bad commit how can I help you to solve this? If you need more information, just ask! Thank you.

On Sat, Mar 22, 2014 at 02:28:15PM +0000, bugzilla-daemon@bugzilla.kernel.org wrote:
> Now I found the real bad commit:
> 
>
> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=f1125f81feca956288ef4ae4fcf47961865ab2e2
> 
> commit f1125f81feca956288ef4ae4fcf47961865ab2e2
> Author: Deng-Cheng Zhu <dengcheng.zhu@imgtec.com>
> Date:   Fri Oct 4 15:14:34 2013 -0700
> 
>     USB/host: Use existing macros instead of hard-coded values in
>     uhci-debug.c
> 
>     Now that UHCI IO registers have been defined in uhci-hcd.h, use them.
> 
>     Reviewed-by: James Hogan <james.hogan@imgtec.com>
>     Signed-off-by: Deng-Cheng Zhu <dengcheng.zhu@imgtec.com>
>     Acked-by: Alan Stern <stern@rowland.harvard.edu>
>     Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> 
> ***
> 
> Now we know the bad commit how can I help you to solve this? If you need more
> information, just ask! Thank you.

Great, can you send this information to the linux-usb@vger.kernel.org
mailing list, and cc: the above people as well, and we can take it from
there.

(In reply to Greg Kroah-Hartman from comment #6)
> Great, can you send this information to the linux-usb@vger.kernel.org
> mailing list, and cc: the above people as well, and we can take it from
> there.

Ok, I'll do that. Before I'll attach 2 logs Alan Stern asked me:

1) USB debug log

2) usbmon trace

Created attachment 130301 [details]
USB debug log

Full log with CONFIG_USB_DEBUG enabled.

Created attachment 130311 [details]
usbmon trace

usbmon trace with mouse pointer stuck in the screen

Created attachment 130321 [details]
dmesg requested by Alan Stern

This is a full dmesg as requested by Alan Stern (with printk).

Created attachment 130331 [details]
usbmon trace 2

Created attachment 130511 [details]
dmesg without commit f1125f81fe

Created attachment 130591 [details]
3.12.14 dmesg (working)

I'll mark it as invalid since it doesn't seem related to USB.