Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Takashi Iwai <tiwai@suse.de>]

At Sat, 16 Apr 2005 02:31:06 +0200 (CEST),
karsten wiese wrote:
> 
> [1  <text/plain; iso-8859-1 (8bit)>]
> 
> --- Clemens Ladisch <clemens@ladisch.de> wrote:
> > Takashi Iwai wrote:
> > > karsten wiese wrote:
> > > > (-: You apply my patch now?
> > >
> > > No, not enough :)  This is exactly why
> > snd_card_free_in_thread() was
> > > introduced.
> > >
> > > As I asked, please make sure that the files are all
> > released before
> > > snd_card_free().  Otherwise the disconnect callback
> > hangs at this
> > > point because of this check.
> > 
> > Most programs close the device file when an error occurs,
> > so you have
> > to test it with something like this to keep the file
> > open:
> > 
> > #include <stdio.h>
> > int main()
> > {
> >   fopen("/dev/snd/hwC1D0", "rb");
> >   getchar();
> > }
> > 
> Tested with it, see attached shell log (copied+pasted).
> Any more tests needed?
> I bet a kasten bier that it oopses reliably on first try,
> if I revert to snd_card_free_in_thread() nothing else
> changed.
> And I'll inform you , if any oops should show when i shut
> down (pc is still in the state like @ the end of the log).
> 
> Regards,
> Karsten
> 
> P.S.: how about adding a kernelversion depending macro for
> backward compatibility,... when was
> snd_card_free_in_thread() introduced?

Agreed.  The problem is that how to know which version has still this
problem..?  At least, 2.4 kernel must have a problem without
snd_card_free_in_thread().


Takashi


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728