Re: 2.6.31-rc5 regression: Oops when USB Serial disconnected while in use

[Greg KH <gregkh@suse.de>]

On Tue, Aug 18, 2009 at 07:00:16PM +0200, Bruno Prémont wrote:
> Hi Greg,
> 
> Were you able to reproduce this with a FTDI usbserial device?

No, I can't find one in my box.  I'll go dig in the attic next to try to
find on, I knew I had one before...

> I just had a short look with 2.6.31-rc6, it crashes harder... once
> kernel did panic with general protection fault after reporting
>   tty_port_close_start: count=-1
> The second time id started output of an Oops but only half of the first
> line ever showed up on screen (Intel KMS).
> 
> Note: this was with just usbserial.ko and ftdi_sio.ko rebuilt with
> CONFIG_DEBUG_INFO=y in the hope to find out what did access a bad
> pointer. (make CONFIG_DEBUG_INFO=y drivers/usb/serial/$module.ko)
> 
> Will retry now with a fresh compile with CONFIG_DEBUG_INFO enabled for
> whole kernel and see if I can get better information, eventually
> switching to vesa framebuffer in case the partial output happens again.
> 
> This pretty much looks like unbalanced ref-count...

Yeah, it sounds like it.  More info would be great to have if you can
get it.

thanks,

greg k-h