Re: [PATCH v4 03/11] spi: spi-mem: Convert Aspeed SMC driver to spi-mem

[Pratyush Yadav <p.yadav@xxxxxx>]

On 04/04/22 09:06AM, Cédric Le Goater wrote:
> On 3/30/22 21:33, Pratyush Yadav wrote:
> > Hi Cedric,
> > 
> > Thanks for doing the conversion.
> > 
> > On 25/03/22 11:08AM, Cédric Le Goater wrote:
> > > This SPI driver adds support for the Aspeed static memory controllers
> > > of the AST2600, AST2500 and AST2400 SoCs using the spi-mem interface.
> > > 
> > >   * AST2600 Firmware SPI Memory Controller (FMC)
> > >     . BMC firmware
> > >     . 3 chip select pins (CE0 ~ CE2)
> > >     . Only supports SPI type flash memory
> > >     . different segment register interface
> > >     . single, dual and quad mode.
> > > 
> > >   * AST2600 SPI Flash Controller (SPI1 and SPI2)
> > >     . host firmware
> > >     . 2 chip select pins (CE0 ~ CE1)
> > >     . different segment register interface
> > >     . single, dual and quad mode.
> > > 
> > >   * AST2500 Firmware SPI Memory Controller (FMC)
> > >     . BMC firmware
> > >     . 3 chip select pins (CE0 ~ CE2)
> > >     . supports SPI type flash memory (CE0-CE1)
> > >     . CE2 can be of NOR type flash but this is not supported by the driver
> > >     . single, dual mode.
> > > 
> > >   * AST2500 SPI Flash Controller (SPI1 and SPI2)
> > >     . host firmware
> > >     . 2 chip select pins (CE0 ~ CE1)
> > >     . single, dual mode.
> > > 
> > >   * AST2400 New Static Memory Controller (also referred as FMC)
> > >     . BMC firmware
> > >     . New register set
> > >     . 5 chip select pins (CE0 ∼ CE4)
> > >     . supports NOR flash, NAND flash and SPI flash memory.
> > >     . single, dual and quad mode.
> > > 
> > > Each controller has a memory range on which flash devices contents are
> > > mapped. Each device is assigned a window that can be changed at bootime
> > > with the Segment Address Registers.
> > > 
> > > Each SPI flash device can then be accessed in two modes: Command and
> > > User. When in User mode, SPI transfers are initiated with accesses to
> > > the memory segment of a device. When in Command mode, memory
> > > operations on the memory segment of a device generate SPI commands
> > > automatically using a Control Register for the settings.
> > > 
> > > This initial patch adds support for User mode. Command mode needs a little
> > > more work to check that the memory window on the AHB bus fits the device
> > > size. It will come later when support for direct mapping is added.
> > > 
> > > Single and dual mode RX transfers are supported. Other types than SPI
> > > are not supported.
> > > 
> > > Reviewed-by: Joel Stanley <joel@xxxxxxxxx>
> > > Tested-by: Joel Stanley <joel@xxxxxxxxx>
> > > Tested-by: Tao Ren <rentao.bupt@xxxxxxxxx>
> > > Signed-off-by: Chin-Ting Kuo <chin-ting_kuo@xxxxxxxxxxxxxx>
> > > Signed-off-by: Cédric Le Goater <clg@xxxxxxxx>
> > > ---
> > >   drivers/mtd/spi-nor/controllers/aspeed-smc.c  | 910 ------------------
> > >   drivers/spi/spi-aspeed-smc.c                  | 709 ++++++++++++++
> > >   .../devicetree/bindings/mtd/aspeed-smc.txt    |  51 -
> > >   MAINTAINERS                                   |   1 +
> > >   drivers/mtd/spi-nor/controllers/Kconfig       |  10 -
> > >   drivers/mtd/spi-nor/controllers/Makefile      |   1 -
> > >   drivers/spi/Kconfig                           |  11 +
> > >   drivers/spi/Makefile                          |   1 +
> > >   8 files changed, 722 insertions(+), 972 deletions(-)
> > >   delete mode 100644 drivers/mtd/spi-nor/controllers/aspeed-smc.c
> > >   create mode 100644 drivers/spi/spi-aspeed-smc.c
> > >   delete mode 100644 Documentation/devicetree/bindings/mtd/aspeed-smc.txt
> > > 
> > [...]
> > > +static void aspeed_spi_send_cmd_addr(struct aspeed_spi_chip *chip, u8 addr_nbytes,
> > > +				     u64 offset, u32 opcode)
> > > +{
> > > +	struct aspeed_spi *aspi = chip->aspi;
> > > +	__be32 temp;
> > > +	u32 cmdaddr;
> > > +
> > > +	switch (addr_nbytes) {
> > > +	default:
> > > +		dev_warn_once(aspi->dev, "Unexpected address width %u, defaulting to 3",
> > > +			      addr_nbytes);
> > > +		fallthrough;
> > 
> > I think it is better if you reject ops where addr width is not 3 or 4.
> > This you can drop this. Or if you really want to keep it, you can change
> > it to a WARN_ON() and return an error.
> 
> OK. This is a left over from the initial driver.
> 
> I have added both at WARN_ONCE() and a 'return -EOPNOTSUPP'
> 
> > > +	case 3:
> > > +		cmdaddr = offset & 0xFFFFFF;
> > > +		cmdaddr |= opcode << 24;
> > > +
> > > +		temp = cpu_to_be32(cmdaddr);
> > > +		aspeed_spi_write_to_ahb(chip->ahb_base, &temp, 4);
> > > +		break;
> > > +	case 4:
> > > +		temp = cpu_to_be32(offset);
> > > +		aspeed_spi_write_to_ahb(chip->ahb_base, &opcode, 1);
> > > +		aspeed_spi_write_to_ahb(chip->ahb_base, &temp, 4);
> > > +		break;
> > > +	}
> > > +}
> > > +
> > [...]
> > > +/* support for 1-1-1, 1-1-2 or 1-1-4 */
> > > +static bool aspeed_spi_supports_op(struct spi_mem *mem, const struct spi_mem_op *op)
> > > +{
> > > +	if (op->cmd.buswidth > 1)
> > > +		return false;
> > > +
> > > +	if (op->addr.nbytes != 0) {
> > > +		if (op->addr.buswidth > 1 || op->addr.nbytes > 4)
> > 
> > As mentioned above, this should reject ops with addr width 1 and 2.
> 
> yes
> 
> > > +			return false;
> > > +	}
> > > +
> > > +	if (op->dummy.nbytes != 0) {
> > > +		if (op->dummy.buswidth > 1 || op->dummy.nbytes > 7)
> > > +			return false;
> > > +	}
> > > +
> > > +	if (op->data.nbytes != 0 && op->data.buswidth > 4)
> > > +		return false;
> > > +
> > > +	return spi_mem_default_supports_op(mem, op);
> > > +}
> > > +
> > > +static int do_aspeed_spi_exec_op(struct spi_mem *mem, const struct spi_mem_op *op)
> > > +{
> > > +	struct aspeed_spi *aspi = spi_controller_get_devdata(mem->spi->master);
> > > +	struct aspeed_spi_chip *chip = &aspi->chips[mem->spi->chip_select];
> > > +	u32 addr_mode, addr_mode_backup;
> > > +	u32 ctl_val;
> > > +	int ret = 0;
> > > +
> > > +	dev_dbg(aspi->dev,
> > > +		"CE%d %s OP %#x mode:%d.%d.%d.%d naddr:%#x ndummies:%#x len:%#x",
> > > +		chip->cs, op->data.dir == SPI_MEM_DATA_IN ? "read" : "write",
> > > +		op->cmd.opcode, op->cmd.buswidth, op->addr.buswidth,
> > > +		op->dummy.buswidth, op->data.buswidth,
> > > +		op->addr.nbytes, op->dummy.nbytes, op->data.nbytes);
> > > +
> > > +	addr_mode = readl(aspi->regs + CE_CTRL_REG);
> > > +	addr_mode_backup = addr_mode;
> > > +
> > > +	ctl_val = chip->ctl_val[ASPEED_SPI_BASE];
> > > +	ctl_val &= ~CTRL_IO_CMD_MASK;
> > > +
> > > +	ctl_val |= op->cmd.opcode << CTRL_COMMAND_SHIFT;
> > > +
> > > +	/* 4BYTE address mode */
> > > +	if (op->addr.nbytes) {
> > > +		if (op->addr.nbytes == 4)
> > > +			addr_mode |= (0x11 << chip->cs);
> > > +		else
> > > +			addr_mode &= ~(0x11 << chip->cs);
> > > +	}
> > > +
> > > +	if (op->dummy.buswidth && op->dummy.nbytes)
> > 
> > Nitpick: op->dummy.nbytes being set should imply op->dummy.buswidth > 0.
> > 
> > > +		ctl_val |= CTRL_IO_DUMMY_SET(op->dummy.nbytes / op->dummy.buswidth);
> > > +
> > > +	if (op->data.nbytes != 0) {
> > > +		if (op->data.buswidth)
> > 
> > Nitpick: op->data.nbytes != 0 should imply op->data.buswidth > 0.
> 
> Indeed. Fixed both.
> 
> > > +			ctl_val |= aspeed_spi_get_io_mode(op);
> > > +	}
> > > +
> > > +	if (op->data.dir == SPI_MEM_DATA_OUT)
> > > +		ctl_val |= CTRL_IO_MODE_WRITE;
> > > +	else
> > > +		ctl_val |= CTRL_IO_MODE_READ;
> > > +
> > > +	if (addr_mode != addr_mode_backup)
> > > +		writel(addr_mode, aspi->regs + CE_CTRL_REG);
> > > +	writel(ctl_val, chip->ctl);
> > > +
> > > +	if (op->data.dir == SPI_MEM_DATA_IN) {
> > > +		if (!op->addr.nbytes)
> > > +			ret = aspeed_spi_read_reg(chip, op);
> > > +		else
> > > +			ret = aspeed_spi_read_user(chip, op, op->addr.val,
> > > +						   op->data.nbytes, op->data.buf.in);
> > > +	} else {
> > > +		if (!op->addr.nbytes)
> > > +			ret = aspeed_spi_write_reg(chip, op);
> > > +		else
> > > +			ret = aspeed_spi_write_user(chip, op);
> > > +	}
> > > +
> > > +	/* Restore defaults */
> > > +	if (addr_mode != addr_mode_backup)
> > > +		writel(addr_mode_backup, aspi->regs + CE_CTRL_REG);
> > > +	writel(chip->ctl_val[ASPEED_SPI_READ], chip->ctl);
> > 
> > Why do you need to restore defaults here? Do you expect some other piece
> > of software to use it as well?
> 
> We expect the controller to be correctly set when dirmap_read() is called.
> But it is only required in the next patch.

Okay. This should be fine then.


-- 
Regards,
Pratyush Yadav
Texas Instruments Inc.