Re: [RFC][PATCH] OMAP3: add support for 2 SDRAM chip selects (was: Re: Beagleboard rev C memory timings & suspend/resume)

[Jean Pihet <jpihet@mvista.com>]

Hi Paul,

OK I will rework the code and send a patch when done.

Regards,
Jean

On Wednesday 03 June 2009 01:40:13 Paul Walmsley wrote:
> Hi Jean,
>
> a minor request: it is easier to comment on these patches if they are
> included inline in the E-mail message, rather than attached.  That way
> code comments can be inlined in the reply.
>
> On Tue, 26 May 2009, Jean Pihet wrote:
> > Here is a patch for the SDRC 2nd CS support. It applies on top of the
> > current pm branch.
>
> Thanks for doing this work.
>
> > I have some questions:
> > - Is it OK to copy the micron sdram params file to a new file with the 2
> > CSes params? One could use a unique file with #ifdef SDRC_SUPPORT_2_CSES.
>
> Is it possible for the SDRAM parameter files to remain unchanged, and to
> simply pass two struct omap_sdrc_params * to omap2_init_common_hw() and
> then to omap2_sdrc_init()?  Boards with only CS0 in use should pass NULL
> for the second omap_sdrc_params *.
>
> So something like this (I realize the PM branch has additional parameters
> here also):
>
> void __init omap2_init_common_hw(struct omap_sdrc_params *sdrc_cs0,
>                                  struct omap_sdrc_params *sdrc_cs1)
>
> Then:
>
> void __init omap2_sdrc_init(struct omap_sdrc_params *sdrc_cs0,
>                             struct omap_sdrc_params *sdrc_cs1)
>
> I would prefer that approach.
>
> It would also be good to avoid changing the SDRC CS1 parameters in the
> SRAM code if the board does not use CS1.  Maybe pass in a flag that
> indicates whether CS1 is in use, and if not, avoid programming those
> registers?  The (admittedly minor) overhead of loading the CS1 registers
> off the stack would be nice to avoid also.
>
> > - Does the RX51 board have 2 sdram parts? If so I need to update the
> > board file as well.
>
> Probably best if someone from Nokia handles this.
>
>
> - Paul