Re: [PATCH v3 03/10] mtd: nand: add NAND driver for Broadcom STB NAND controller

[Ray Jui <rjui@broadcom.com>]

On 5/6/2015 2:05 PM, Brian Norris wrote:
> On Wed, May 06, 2015 at 09:17:36PM +0200, Arnd Bergmann wrote:
>> On Wednesday 06 May 2015 10:59:47 Brian Norris wrote:
>>> +
>>> +static inline u32 nand_readreg(struct brcmnand_controller *ctrl, u32 offs)
>>> +{
>>> +       return __raw_readl(ctrl->nand_base + offs);
>>> +}
>>> +
>>> +static inline void nand_writereg(struct brcmnand_controller *ctrl, u32 offs,
>>> +                                u32 val)
>>> +{
>>> +       __raw_writel(val, ctrl->nand_base + offs);
>>> +}
>>> +
>>>
>>
>> You had mentioned previously that there might be an endianess issue in this
>> driver.
> 
> Might. I have a patch already, but I failed to boot a BE kernel, so I
> kept it out for now. If you don't mind, I'd prefer patching something
> like this once it's testable on ARM BE. This *is*, however, extensively
> tested on MIPS (LE and BE) and ARM (LE).

Correct, extensive test and pass all MTD test cases. We should
eventually be able to test this on a working ARM BE platform, within the
next couple months.

> 
>> I think this won't work on big-endian architectures other than MIPS,
>> so it would be good to either list in the DT the endianess of the device
>> and use appropriate accessors here, or hardcode it based on the architecture
>> (using ioread32_be in big-endian mips, but readl elsewhere).
> 
> I suspect we wouldn't need a DT property but could just special-case
> MIPS BE, as you note.
> 
>> Using __raw_writel has another problem regarding the DMA capability of this
>> driver, as it will not flush any write buffers or synchronize caches before
>> sending data off to the device, so you risk data corruption.
> 
> We use mb() before kicking off DMA or other commands.
> 
>> Also, the
>> compiler can choose to split up the 32-bit word access into byte accesses,
>> which on most hardware does not do what you want.
> 
> Huh? Wouldn't that break just about every driver in existence? And how
> is writel() any different than __raw_writel() in that regard? From
> include/asm-generic/io.h:
> 
> static inline void writel(u32 value, volatile void __iomem *addr)
> {
>         __raw_writel(__cpu_to_le32(value), addr);
> }
> 
> And BTW, splitting isn't possible on ARM. From
> arch/arm/include/asm/io.h:
> 
> static inline void __raw_writel(u32 val, volatile void __iomem *addr)
> {
>         asm volatile("str %1, %0"
>                      : "+Qo" (*(volatile u32 __force *)addr)
>                      : "r" (val));
> }
> 
> Brian
>