Re: [PATCH] Cleanups for: line length, printk KERN_ argument, stack frame size > 2048 (added a kmalloc/kfree), style/formatting errors, incorrect include files

[Greg KH <gregkh@suse.de>]

On Fri, Oct 09, 2009 at 02:41:41PM -0700, Bruce B wrote:
> 
> On Oct 9, 2009, at 2:29 PM, Greg KH wrote:
> 
> > On Fri, Oct 09, 2009 at 02:27:42PM -0700, Randy Dunlap wrote:
> >> On Fri, 9 Oct 2009 14:16:10 -0700 Greg KH wrote:
> >>
> >>>
> >>> Try changing one thing at a time per patch please.
> >>>
> >>> There's some bugs in here:
> >>>
> >>>> -		struct config_t dig_in_config[32];
> >>>> -		struct config_t dig_out_config[32];
> >>>> -		struct config_t chan_in_config[32];
> >>>> -		struct config_t chan_out_config[32];
> >>>> 		int i;
> >>>> +		config_data = kmalloc(sizeof(struct config_data_t), GFP_KERNEL);
> >>>
> >>> Shouldn't that be:
> >>> 		config_data = kmalloc(sizeof(struct config_data_t)*32,  
> >>> GFP_KERNEL);
> >>
> >> The new struct already includes the [32]s.
> >
> > Ah, missed that.
> >
> > But your other comment about checking the kmalloc is valid.  We should
> > not add new calls to kmalloc that doesn't check, let's not _add_ new
> > errors to the code :)
> >
> > Bruce, care to split this up into individual patches, each doing only
> > one thing, and check for the return value of this call?
> >
> > thanks,
> >
> > greg k-h
> 
> Sure. Glad to.  A little head scratching will be required for the  
> general cleanups in the .open routing for the kmalloc calls.
> There are several different types of formatting errors that I have  
> addressed. Those can be bundled as one patch?

Sure, just try to do one logical thing per patch to make it easy to
review and say "of course that is obvious, why would I even think of
rejecting that patch?"

thanks,

greg k-h