Re: [PATCH v1] sh: boards: fix CEU buffer size passed to dma_declare_coherent_memory()

On Mon, 24 Jul 2023 20:43:31 +0300
Laurent Pinchart <[email protected]> wrote:

> On Mon, Jul 24, 2023 at 07:29:27PM +0200, John Paul Adrian Glaubitz wrote:
> > On Mon, 2023-07-24 at 20:12 +0300, Laurent Pinchart wrote:  
> > > On Mon, Jul 24, 2023 at 02:07:42PM +0200, Petr Tesarik wrote:  
> > > > From: Petr Tesarik <[email protected]>
> > > > 
> > > > In all these cases, the last argument to dma_declare_coherent_memory() is
> > > > the buffer end address, but the expected value should be the size of the
> > > > reserved region.
> > > > 
> > > > Fixes: 39fb993038e1 ("media: arch: sh: ap325rxa: Use new renesas-ceu camera driver")
> > > > Fixes: c2f9b05fd5c1 ("media: arch: sh: ecovec: Use new renesas-ceu camera driver")
> > > > Fixes: f3590dc32974 ("media: arch: sh: kfr2r09: Use new renesas-ceu camera driver")
> > > > Fixes: 186c446f4b84 ("media: arch: sh: migor: Use new renesas-ceu camera driver")
> > > > Fixes: 1a3c230b4151 ("media: arch: sh: ms7724se: Use new renesas-ceu camera driver")
> > > > Signed-off-by: Petr Tesarik <[email protected]>  
> > > 
> > > Reviewed-by: Laurent Pinchart <[email protected]>
> > > 
> > > But given that nobody noticed for 5 years, maybe we should drop
> > > renesas-ceu support from those boards ? Or drop the boards completely ?  
> > 
> > arch/sh is being maintained again, so it's save to keep these boards. At some point, we're
> > going to convert the architecture to using Device Trees which should reduce the maintenance
> > burden anyways.  
> 
> Keeping the architecture is fine for newer systems, but is anyone really
> maintaining the Renesas SH board ?

FWIW I don't have any opinion here. I found these calls as a side
effect of making an overview of DMA-capable devices (in preparation for
moving all DMA-related members away from struct device). I don't have
any of those boards myself.

Petr T