On 2015-06-03 15:08, Sebastian Andrzej Siewior wrote:
> On 2015-03-25 17:28:24 [+0100], Stefan Agner wrote:
>> diff --git a/drivers/mtd/nand/vf610_nfc.c b/drivers/mtd/nand/vf610_nfc.c
>> new file mode 100644
>> index 0000000..23c1510
>> --- /dev/null
>> +++ b/drivers/mtd/nand/vf610_nfc.c
>> @@ -0,0 +1,686 @@
> …
>> +static inline u32 vf610_nfc_read(struct vf610_nfc *nfc, uint reg)
>> +{
>> + return readl(nfc->regs + reg);
>> +}
>> +
>> +static inline void vf610_nfc_write(struct vf610_nfc *nfc, uint reg, u32 val)
>> +{
>> + writel(val, nfc->regs + reg);
>> +}
> …
>
>> +static void vf610_nfc_send_command(struct vf610_nfc *nfc, u32 cmd_byte1,
>> + u32 cmd_code)
>> +{
>> + void __iomem *reg = nfc->regs + NFC_FLASH_CMD2;
>> + u32 tmp;
>> +
>> + vf610_nfc_clear_status(nfc);
>> +
>> + tmp = __raw_readl(reg);
>> + tmp &= ~(CMD_BYTE1_MASK | CMD_CODE_MASK | BUFNO_MASK);
>> + tmp |= cmd_byte1 << CMD_BYTE1_SHIFT;
>> + tmp |= cmd_code << CMD_CODE_SHIFT;
>> + __raw_writel(tmp, reg);
>> +}
>
> Why readl() vs __raw_readl() dito for write?
> vf610_nfc_{read|write} is good since for PPC we would need out_be32()
> here instead.
> It would be nice if you could abstract the __raw_ once as well. And I am
> not sure if you need those at all since the former functions should work
> here just fine.
As Boris guessed correctly, the reason I used the raw variant was due to
performance improvements due to the barrier. However, I will use
{read|write}l_relaxed instead, which should offer endian abstraction
while not having the performance penalty due to extensive barriers...
--
Stefan
--
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
