On Tue, Aug 25, 2015 at 12:22:10PM +0200, Marek Vasut wrote: > On Tuesday, August 25, 2015 at 12:17:37 PM, Cyrille Pitchen wrote: > > If you don't mind, I'd rather keep some of these inline functions. I have > > no strong justification, it's more a personal taste: it makes lines > > shorter as it avoids the need to add "->regs + ". > > Also it makes the code consistent with other Atmel drivers which already > > use such wrappers. > > > > However I'll fix the comment and remove the byte and word versions, which > > are not used. So only qspi_readl() and qspi_writel() are left. > > > > Does it sound good to you? > > In my mind, seeing explicit readl_relaxed() somewhere is much easier to > digest than seeing some wrapper, which I have to look up. But please do > wait for others to voice their concern too, I might not be the best person > to tell you what to do when it comes to wrapping IO accessors ;-) I could go either way, but there are times where local wrapper I/O accessors are useful. Case in point: it makes it really easy to make the choice between readl() and readl_relaxed() in one place (i.e., the discussion you had in another branch of this thread). That's been useful for me on brcmnand, where certain platforms (big-endian MIPS) have different assumptions about endianness than your average platform. Also, it helps with things like what Robert Jarzmik is trying to do on pxa3xx_nand -- add debug info to print every register read/write. Also as Cyrille mentioned, personal taste is a factor. Anyway, I'll go with whatever makes sense between y'all. I don't mind. Brian -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html