Re: [PATCH v2] USB: host: Use zeroing memory allocator rather than allocator/memset

[Alan Stern]

On Mon, 1 Jan 2018, Himanshu Jha wrote:

> On Sun, Dec 31, 2017 at 04:20:45PM -0500, Alan Stern wrote:
> > On Sun, 31 Dec 2017, Himanshu Jha wrote:
> > 
> > > Use dma_zalloc_coherent for allocating zeroed
> > > memory and remove unnecessary memset function.
> > > 
> > > Done using Coccinelle.
> > > Generated-by: scripts/coccinelle/api/alloc/kzalloc-simple.cocci
> > > 0-day tested with no failures.
> > > 
> > > Suggested-by: Luis R. Rodriguez <mcgrof@xxxxxxxxxx>
> > > Signed-off-by: Himanshu Jha <himanshujha199640@xxxxxxxxx>
> > > ---
> > > v2:
> > >    -align argumenst as they were before applying the SmPL rule.
> > 
> > For the UHCI portion:
> > 
> > Acked-by: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>
> > 
> > But there is something pecular about the patch...
> > 
> > >  drivers/usb/host/uhci-hcd.c | 7 +++----
> > >  drivers/usb/host/xhci-mem.c | 7 ++-----
> > >  2 files changed, 5 insertions(+), 9 deletions(-)
> > > 
> > > diff --git a/drivers/usb/host/uhci-hcd.c b/drivers/usb/host/uhci-hcd.c
> > > index f5c9021..ac53398 100644
> > > --- a/drivers/usb/host/uhci-hcd.c
> > > +++ b/drivers/usb/host/uhci-hcd.c
> > > @@ -600,15 +600,14 @@ static int uhci_start(struct usb_hcd *hcd)
> > >  	uhci->dentry = dentry;
> > >  #endif
> > >  
> > > -	uhci->frame = dma_alloc_coherent(uhci_dev(uhci),
> > > -			UHCI_NUMFRAMES * sizeof(*uhci->frame),
> > > -			&uhci->frame_dma_handle, GFP_KERNEL);
> > > +	uhci->frame = dma_zalloc_coherent(uhci_dev(uhci),
> > > +			UHCI_NUMFRAMES * sizeof(*uhci->frame),
> > > +			&uhci->frame_dma_handle, GFP_KERNEL);
> > 
> > The second and third "changed" lines here actually are identical.  What 
> > program would produce a diff file showing that they were changed?
> 
> I'm sorry, I can't understand the problem.

There is no problem.  I merely asked a question.

> The patch was generated by cocci script specified above and then you
> told me to change the whitespace scheme as it was before.

Yes, I know.  But did you change the cocci script to make it produce
this new patch, or did you make the patch by manually editing the old
version?  In other words, how was this patch generated?

>  It is also
> 0-day tested as specified above. It's also free from warnings reported by
> checkpatch.

I never said there was anything wrong with the patch.  It's just
peculiar.  For example, you could never get diff (or git diff) to
produce that patch.

Alan Stern

> If are still confused please take a look at the rule :
> https://lkml.org/lkml/2017/12/26/205
> which will be soon merged to the linux-next tree.
> 
> Thanks
> Himanshu Jha
>