On Mon, Sep 27, 2010 at 09:06:57AM +0800, Eric Miao wrote: > On Mon, Sep 27, 2010 at 12:09 AM, Wolfram Sang <w.sang@xxxxxxxxxxxxxx> wrote: > > On Sun, Sep 26, 2010 at 04:52:06AM -0400, zhangfei gao wrote: > >> Add several call back to sdhci-pltfm.c, help give suggestion > > > > Just formal things, without looking at the code yet. > > > > Make seperate patches out of these, everyone with a proper description. > > > >> 1. struct sdhci_host *(*alloc_host)(struct device *dev), since > >> specific driver need some private variable to allocate and free, > >> including clk. > >> 2. unsigned int (*get_quirk)(struct sdhci_host *host); add this > >> because one driver may serve several device, each one may require > >> different quirk, and specific driver is right one to know. > >> 3. void (*set_max_speed)(struct sdhci_host *host); this should be done > >> after add_host, which impact f_max. > >> 4. int (*init)(struct sdhci_host *host, struct sdhci_pltfm_data > >> *pdata, void* priv_pdata); copy from Wolfram Sang's patch to transfer > >> platform data, copy here for test. > > > > Just a rough idea, considering the potential differences (I believe > there will be > more in the future) between the SDHCI of different SoCs, would it make more > sense to make sdhci-pltfm.c as a common function library for sdhci-<soc>.c? Yeah, I think Alan Cox mentioned this idea, too. My guess is that it will be well received, if somebody actually does it ;) Up to that point, it probably makes sense to keep redundancy low by the means we have today, i.e. pltfm. That should help a later migration process. Regards, Wolfram -- Pengutronix e.K. | Wolfram Sang | Industrial Linux Solutions | http://www.pengutronix.de/ |
Attachment:
signature.asc
Description: Digital signature
