Re: [RESEND PATCH v7] mtd: spi-nor: add hisilicon spi-nor flash controller driver
From: Jiancheng Xue
Date: Tue Mar 08 2016 - 04:28:41 EST
Thank you very much for your comments.
On 2016/3/8 6:52, Brian Norris wrote:
>> diff --git a/Documentation/devicetree/bindings/spi/spi-hisi-sfc.txt b/Documentation/devicetree/bindings/spi/spi-hisi-sfc.txt
>> new file mode 100644
>> index 0000000..7407147
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/spi/spi-hisi-sfc.txt
> This is not a SPI controller (or is it? it looks like a SPI NOR
> controller that doesn't do generic SPI), so it shouldn't go in the spi/
OK. I'll move this file into Documentation/devicetree/bindings/mtd.
>> @@ -0,0 +1,25 @@
>> +HiSilicon SPI-NOR Flash Controller
>> +Required properties:
>> +- compatible : Should be "hisilicon,hisi-sfc" and one of the following strings:
>> + "hisilicon,hi3519-sfc"
>> +- address-cells : number of cells required to define a chip select
>> + address on the SPI bus. Should be set to 1. See spi-bus.txt.
> This is not exactly a typical SPI bus, though I suppose some of that
> documentation could apply...
OK. I'll modify the description.
>> +static int hisi_spi_nor_read_reg(struct spi_nor *nor, u8 opcode, u8 *buf,
>> + int len)
>> + struct hifmc_priv *priv = nor->priv;
>> + struct hifmc_host *host = priv->host;
>> + int ret;
>> + ret = hisi_spi_nor_send_cmd(nor, opcode, len);
>> + if (ret)
>> + return ret;
>> + memcpy(buf, host->iobase, len);
> sparse doesn't like any of these memcpy()'s, since that's __iomem. Are
> you sure you want a regular memcpy here, and elsewhere?
It's not a must to use memcpy though host->iobase represents a segment of inner memory
which stores data from flash device byte by byte. I will give up using memcpy here and
hisi_spi_nor_write_reg() in next version. Thank you!
>> + return 0;
>> +static void hisi_spi_nor_dma_transfer(struct spi_nor *nor, u32 start_off,
>> + u32 dma_buf, u32 len, u8 op_type)
> Does this controller support 64-bit addresses? What if you see LPAE, or
> an ARM64 CPU? It'd be nice if you don't truncate potentially 64-bit
> dma_buf down to u32.
No, this controller just supports 32-bit addresses. So I think I have to truncate 64-bit dma_buf. :(
Thank you very much!