Re: [PATCH v6 2/3] mtd: mtk-nor: mtk serial flash controller driver

From: Brian Norris
Date: Mon Nov 09 2015 - 22:01:25 EST


Hi Bayi,

A few small comments.

On Fri, Nov 06, 2015 at 11:48:08PM +0800, Bayi Cheng wrote:
> add spi nor flash driver for mediatek controller
>
> Signed-off-by: Bayi Cheng <bayi.cheng@xxxxxxxxxxxx>
> ---
> drivers/mtd/spi-nor/Kconfig | 7 +
> drivers/mtd/spi-nor/Makefile | 1 +
> drivers/mtd/spi-nor/mtk-quadspi.c | 475 ++++++++++++++++++++++++++++++++++++++
> 3 files changed, 483 insertions(+)
> create mode 100644 drivers/mtd/spi-nor/mtk-quadspi.c


> diff --git a/drivers/mtd/spi-nor/Makefile b/drivers/mtd/spi-nor/Makefile
> index e53333e..0bf3a7f8 100644
> --- a/drivers/mtd/spi-nor/Makefile
> +++ b/drivers/mtd/spi-nor/Makefile
> @@ -1,3 +1,4 @@
> obj-$(CONFIG_MTD_SPI_NOR) += spi-nor.o
> obj-$(CONFIG_SPI_FSL_QUADSPI) += fsl-quadspi.o
> +obj-$(CONFIG_MTD_MT81xx_NOR) += mtk-quadspi.o
> obj-$(CONFIG_SPI_NXP_SPIFI) += nxp-spifi.o
> diff --git a/drivers/mtd/spi-nor/mtk-quadspi.c b/drivers/mtd/spi-nor/mtk-quadspi.c
> new file mode 100644
> index 0000000..6582811
> --- /dev/null
> +++ b/drivers/mtd/spi-nor/mtk-quadspi.c
> @@ -0,0 +1,475 @@

...

> +/* Can shift up to 48 bits (6 bytes) of TX/RX */
> +#define MTK_NOR_MAX_SHIFT 6

...

> +static int mt8173_nor_do_tx_rx(struct mt8173_nor *mt8173_nor, u8 op,
> + u8 *tx, int txlen, u8 *rx, int rxlen)
> +{
> + int len = 1 + txlen + rxlen;
> + int i, ret, idx;
> +
> + if (len > MTK_NOR_MAX_SHIFT + 1)
> + return -EINVAL;

So I see you adjusted these bounds to add 1, which inspired one of my
questions on the cover letter.

Why do you allow len=7, which means you'd program 7*8 to the COUNT
register, when the SoC manual says it has a max of 48? Is the manual
wrong?

I notice you added the '+ 1' to your driver, so it allows:

do_tx_rx( txlen = 0 , rxlen = 6) /* e.g., for READ ID */

but that means your driver also allows:

do_tx_rx( txlen = 6, rxlen = 0) /* ERROR: this will allow out of
bounds write on PRGDATA
register -1 */

If I understand correctly, the following constraints are more correct:

/* Can only transmit 1 opcode and 5 other bytes */
if (1 + txlen > MTK_NOR_MAX_SHIFT)
return -EINVAL;

/* Can only receive 6 bytes after the opcode */
if (rxlen > MTK_NOR_MAX_SHIFT)
return -EINVAL;

/* Can only handle XXX bytes total */
/* How many bytes is the max for register MTK_NOR_CNT_REG ? */
if (len > XXX)
return -EINVAL;

> +
> + writeb(len * 8, mt8173_nor->base + MTK_NOR_CNT_REG);
> +
> + /* start at PRGDATA5, go down to PRGDATA0 */
> + idx = MTK_NOR_MAX_SHIFT - 1;
> +
> + /* opcode */
> + writeb(op, mt8173_nor->base + MTK_NOR_PRG_REG(idx));
> + idx--;
> +
> + /* program TX data */
> + for (i = 0; i < txlen; i++, idx--)
> + writeb(tx[i], mt8173_nor->base + MTK_NOR_PRG_REG(idx));
> +
> + /* clear out rest of TX registers */
> + while (idx >= 0) {
> + writeb(0, mt8173_nor->base + MTK_NOR_PRG_REG(idx));
> + idx--;
> + }
> +
> + ret = mt8173_nor_execute_cmd(mt8173_nor, MTK_NOR_PRG_CMD);
> + if (ret)
> + return ret;
> +
> + /* restart at first RX byte */
> + idx = rxlen - 1;
> +
> + /* read out RX data */
> + for (i = 0; i < rxlen; i++, idx--)
> + rx[i] = readb(mt8173_nor->base + MTK_NOR_SHREG(idx));
> +
> + return 0;
> +}
> +

...

> +static int __init mtk_nor_init(struct mt8173_nor *mt8173_nor,
> + struct device_node *flash_node)
> +{
> + struct mtd_part_parser_data ppdata = {
> + .of_node = flash_node,
> + };
> + int ret;
> + struct spi_nor *nor;

Normally we'd have a blank line here.

> + /* initialize controller to accept commands */
> + writel(MTK_NOR_ENABLE_SF_CMD, mt8173_nor->base + MTK_NOR_WRPROT_REG);
> +
> + nor = &mt8173_nor->nor;
> + nor->dev = mt8173_nor->dev;
> + nor->priv = mt8173_nor;
> + nor->flash_node = flash_node;
> +
> + /* fill the hooks to spi nor */
> + nor->read = mt8173_nor_read;
> + nor->read_reg = mt8173_nor_read_reg;
> + nor->write = mt8173_nor_write;
> + nor->write_reg = mt8173_nor_write_reg;
> + nor->mtd.owner = THIS_MODULE;
> + nor->mtd.name = "mtk_nor";
> + /* initialized with NULL */
> + ret = spi_nor_scan(nor, NULL, SPI_NOR_DUAL);
> + if (ret)
> + return ret;
> +
> + return mtd_device_parse_register(&nor->mtd, NULL, &ppdata, NULL, 0);
> +}

...

Brian
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/