Re: [PATCH] em28xx: input: fix oops on device removal

[Mauro Carvalho Chehab <mchehab@redhat.com>]

Em Fri, 21 Dec 2012 16:32:54 +0100
Frank Schäfer <fschaefer.oss@googlemail.com> escreveu:

> Am 13.12.2012 17:34, schrieb Frank Schäfer:
> > When em28xx_ir_init() fails du to an error in em28xx_ir_change_protocol(), it
> > frees the memory of struct em28xx_IR *ir, but doesn't set the corresponding
> > pointer in the device struct to NULL.
> > On device removal, em28xx_ir_fini() gets called, which then calls
> > rc_unregister_device() with a pointer to freed memory.
> >
> > Fixes bug 26572 (http://bugzilla.kernel.org/show_bug.cgi?id=26572)
> >
> > Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
> >
> > Cc: stable@kernel.org	# at least all kernels since 2.6.32 (incl.)
> > ---
> >  drivers/media/usb/em28xx/em28xx-input.c |    9 ++++-----
> >  1 Datei geändert, 4 Zeilen hinzugefügt(+), 5 Zeilen entfernt(-)
> >
> > diff --git a/drivers/media/usb/em28xx/em28xx-input.c b/drivers/media/usb/em28xx/em28xx-input.c
> > index 660bf80..5c7d768 100644
> > --- a/drivers/media/usb/em28xx/em28xx-input.c
> > +++ b/drivers/media/usb/em28xx/em28xx-input.c
> > @@ -538,7 +538,7 @@ static int em28xx_ir_init(struct em28xx *dev)
> >  	ir = kzalloc(sizeof(*ir), GFP_KERNEL);
> >  	rc = rc_allocate_device();
> >  	if (!ir || !rc)
> > -		goto err_out_free;
> > +		goto error;
> >  
> >  	/* record handles to ourself */
> >  	ir->dev = dev;
> > @@ -559,7 +559,7 @@ static int em28xx_ir_init(struct em28xx *dev)
> >  	rc_type = RC_BIT_UNKNOWN;
> >  	err = em28xx_ir_change_protocol(rc, &rc_type);
> >  	if (err)
> > -		goto err_out_free;
> > +		goto error;
> >  
> >  	/* This is how often we ask the chip for IR information */
> >  	ir->polling = 100; /* ms */
> > @@ -584,7 +584,7 @@ static int em28xx_ir_init(struct em28xx *dev)
> >  	/* all done */
> >  	err = rc_register_device(rc);
> >  	if (err)
> > -		goto err_out_stop;
> > +		goto error;
> >  
> >  	em28xx_register_i2c_ir(dev);
> >  
> > @@ -597,9 +597,8 @@ static int em28xx_ir_init(struct em28xx *dev)
> >  
> >  	return 0;
> >  
> > - err_out_stop:
> > +error:
> >  	dev->ir = NULL;
> > - err_out_free:
> >  	rc_free_device(rc);
> >  	kfree(ir);
> >  	return err;
> 
> Ping !?
> Mauro, this patch is really easy to review and it fixes a 2 years old bug...
> Isn't this one of those patches that should be applied immediately ?

This one is not on my queue... Patchwork doesn't seem to catch it:

http://patchwork.linuxtv.org/project/linux-media/list/?submitter=44&state=*

Hmm... perhaps it is due to the accent on your name. Weird that it got
other patches from you. You should likely thank python for discriminating
e-mails with accents.

Could you please try to re-submit it being sure that your email got
properly encoded with UTF-8?

Regards,
Mauro

PS.: my intention is to try to merge tomorrow all em28xx patches.