[PATCH v2 1/2] mtd: sh_flctl: Setup and release DMA channels

[Guennadi Liakhovetski]

On Fri, 28 Sep 2012, Bastian Hecht wrote:

[snip]

> >> diff --git a/include/linux/mtd/sh_flctl.h b/include/linux/mtd/sh_flctl.h
> >> index 01e4b15..20d3f48 100644
> >> --- a/include/linux/mtd/sh_flctl.h
> >> +++ b/include/linux/mtd/sh_flctl.h
> >> @@ -20,10 +20,12 @@
> >>  #ifndef __SH_FLCTL_H__
> >>  #define __SH_FLCTL_H__
> >>
> >> +#include <linux/dmaengine.h>
> >>  #include <linux/mtd/mtd.h>
> >>  #include <linux/mtd/nand.h>
> >>  #include <linux/mtd/partitions.h>
> >>  #include <linux/pm_qos.h>
> >> +#include <linux/sh_dma.h>
> >
> > Here headers aren't actually needed - you can just forward-declare "struct
> > dma_chan." OTOH, what _is_ indeed needed is <linux/completion.h>.
> 
> Oh yes, I'll add <linux/completion.h>.
> About the forward declration: Do I really need it? We use a struct
> pointer and I am unsure if the compiler just needs to know the name
> and nothing more. I've tested it and included sh_flctl.h (without any
> DMA headers) in some driver that has no idea about DMA.
> gcc didn't complain.

I'm not sure about this, I think, sometimes there might be cases, when gcc 
doesn't complain. Or (more likely) the header is pulled in via another 
one, or some other header has this forward-declaration. But in any case I 
wouldn't rely on either of those and declare what's needed explicitly.

> Adding a non-pointer member
> + struct sh_dma test;
> lead to an error, so struct sh_dma is really unknown.

Of course, the compiler has to know the size of the struct to allocate 
space for it:-)

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/