> On 03/26/2014 12:41 PM, Jason Jin wrote: > > This board sepecific initialization mechanism is not feasible i for > > corenet platform as the corenet platform file is a abstraction of > > serveral platforms. > > You can't make it 100% abstract. The DIU driver requires some sort of > board support. I think you can make a generic platform file for the DIU. > [Jason Jin-R64188] Provide the generic diu platform file is reasonable, but it is will be just the board specific initialization file collection, if we really need to reinitialize something in kernel.
> > However, the DIU is already initialized in u-boot and we can rely on > > the settings in u-boot for corenet platform, > > I don't like that at all. [Jason Jin-R64188] What's the potential issue of this? Most of the IPs need the platform initialization in u-boot. For example, the hwconfig in u-boot used for board specific settings for most platform. The diu_ops structure is still open for any board specific initialization if the platform needed. We at least can let the platform to select to use the diu_ops or not. This is the purse for this patch. > > > the only > > issue is that when DIU wake up from the deepsleep, some of the board > > specific initialization will lost, such as the pixel clock setting. > > That is a BIG issue. This is why we have a platform file -- to handle > the platform issues. [Jason Jin-R64188] Yes, this is an issue, Actually, this is a SOC level issue, We can try to fix it in the driver by saving the pixel clock in suspend function and restored it in resume function. Thanks. _______________________________________________ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
