[PATCH V5 2/2] mtd: spi-nor: Add driver for Cadence Quad SPI Flash Controller.

[Vikas MANOCHA]

Hi Graham,

> -----Original Message-----
> From: Graham Moore [mailto:grmoore at opensource.altera.com]
> Sent: Monday, July 27, 2015 10:40 AM
> To: Vikas MANOCHA
> Cc: Marek Vasut; linux-mtd at lists.infradead.org; David Woodhouse; Brian
> Norris; linux-kernel at vger.kernel.org; Alan Tull; Dinh Nguyen; Yves
> Vandervennet
> Subject: Re: [PATCH V5 2/2] mtd: spi-nor: Add driver for Cadence Quad SPI
> Flash Controller.
> 
> Hi Vikas,
> 
> On 07/24/2015 07:02 PM, vikasm wrote:
> > Hi Graham,
> >
> > On 07/24/2015 10:17 AM, Graham Moore wrote:
> >> Signed-off-by: Graham Moore <grmoore at opensource.altera.com>
> >> ---
> >> V2: use NULL instead of modalias in spi_nor_scan call
> >> V3: Use existing property is-decoded-cs instead of creating duplicate.
> >> V4: Support Micron quad mode by snooping command stream for EVCR
> >> command and subsequently configuring Cadence controller for quad
> mode.
> >> V5: Clean up sparse and smatch complaints.  Remove snooping of Micron
> >> quad mode.  Add comment on XIP mode bit and dummy clock cycles.  Set
> >> up SRAM partition at 1:1 during init.
> >> ---
> >>   arch/arm/boot/dts/socfpga.dtsi               |    1 +
> >>   arch/arm/boot/dts/socfpga_cyclone5_socdk.dts |    1 -
> >>   drivers/mtd/spi-nor/Kconfig                  |    6 +
> >>   drivers/mtd/spi-nor/Makefile                 |    1 +
> >>   drivers/mtd/spi-nor/cadence-quadspi.c        | 1261
> ++++++++++++++++++++++++++
> >>   5 files changed, 1269 insertions(+), 1 deletion(-)
> >>   create mode 100644 drivers/mtd/spi-nor/cadence-quadspi.c
> >>
> >> diff --git a/arch/arm/boot/dts/socfpga.dtsi
> >> b/arch/arm/boot/dts/socfpga.dtsi index c71a705..e9ecdce 100644
> >> --- a/arch/arm/boot/dts/socfpga.dtsi
> >> +++ b/arch/arm/boot/dts/socfpga.dtsi
> >> @@ -695,6 +695,7 @@
> >>                          is-decoded-cs = <1>;
> >>                          fifo-depth = <128>;
> >>                          status = "disabled";
> >> +                       m25p,fast-read;
> >
> > This patch is not applicable to l2-mtd master or linus master repo, might
> need to rebase it.
> >
> 
> Argh, my mistake, these dts files are not supposed to be in the patch.
> 
> ...
> 
> >> +#include <linux/timer.h>
> >> +
> >> +#define CQSPI_NAME                     "cadence-qspi"
> >
> > replace space with tabs.
> >
> 
> They *are* tabs in my original patch.  Not sure how they got changed to
> spaces...
> 
> ...
> >> +
> >> +#define CQSPI_FIFO_WIDTH                       4
> >
> > FIFO width could be 4 or 8 etc depending on the SOC, it would be better to
> get it from device.
> > You can refer to u-boot code for the same.
> >
> 
> OK
> ...
> >> +
> >> +#define CQSPI_INDIRECTTRIGGER_ADDR_MASK                0xFFFFF
> >
> > Mask value of 0xFFFFF is specific to socfpga platform.
> > Please refer to u-boot patchset for same discussion.
> >
> 
> OK
> ...
> >> +static void cqspi_fifo_read(void *dest, const void __iomem
> *src_ahb_addr,
> >> +                           unsigned int bytes) {
> >> +       unsigned int temp;
> >> +       int remaining = bytes;
> >> +       unsigned int *dest_ptr = (unsigned int *)dest;
> >> +
> >> +       while (remaining >= CQSPI_FIFO_WIDTH) {
> >> +               *dest_ptr = readl(src_ahb_addr);
> >> +               dest_ptr++;
> >> +               remaining -= CQSPI_FIFO_WIDTH;
> >
> > this logic only works when fifo width is 4 with "unsigned int" data of 4
> bytes.
> > It has been corrected in mainline u-boot or in the u-boot patches.
> >
> 
> OK
> ...
> >> +static int cqspi_indirect_read_setup(struct spi_nor *nor,
> >> +                                    unsigned int from_addr) {
> >> +       unsigned int reg;
> >> +       unsigned int dummy_clk = 0;
> >> +       struct cqspi_st *cqspi = nor->priv;
> >> +       void __iomem *reg_base = cqspi->iobase;
> >> +       unsigned int ahb_phy_addr = cqspi->ahb_phy_addr;
> >> +
> >> +       writel((ahb_phy_addr & CQSPI_INDIRECTTRIGGER_ADDR_MASK),
> >> +              reg_base + CQSPI_REG_INDIRECTTRIGGER);
> >> +       writel(from_addr, reg_base + CQSPI_REG_INDIRECTRDSTARTADDR);
> >
> > Base trigger register address (0x1c register) corresponds to the
> > address which should be put on AHB bus to handle indirect transfer
> triggered before.
> >
> > To handle indirect transfer we need to issue addresses from (value of
> > 0x1c) to (value of 0x1c) + 15*4 ("4" corresponds to size of SRAM location).
> > There are no obstacles in issuing const address just equal to 0x1c.
> > Important thing to note is that indirect trigger address has nothing
> > in common with your physical or mapped NOR Flash address.
> >
> > Transfer read/write start addresses should be programmed with the
> > absolute flash address to be read/written.
> >
> 
> OK, I'll add the trigger address to the device tree, etc.  You seem to be
> concerned about the fact that on Altera SoC, the trigger address and the ahb
> address are different.  I'm not seeing any problem with that.

No problem with different or same trigger address & ahb address. Trigger address should be passed through device tree as it could
be different for different SoCs. Masking it with value like 0xf_ffff to make it suitable for one SOC in driver does not seems right approach.

> The CPU reads/writes the FIFO using an address, but that is not the same
> address that goes on the bus way down in the interconnect.  What's wrong
> with that?  It's the way our SoC works.  If we use the trigger address for the
> trigger address register, and the ahb address for reading/writing the FIFO,
> then it works fine.

This implementation is your Soc specific & not generic for the qspi peripheral. In this way, it won't be possible to use this driver for Socs other than altera soc.

Rgds,
Vikas

> 
> ...
> 
> >> +       watermark = cqspi->fifo_depth * CQSPI_FIFO_WIDTH / 2;
> >> +       writel(watermark, reg_base +
> CQSPI_REG_INDIRECTRDWATERMARK);
> >> +       writel(remaining, reg_base + CQSPI_REG_INDIRECTRDBYTES);
> >> +       writel(cqspi->fifo_depth - CQSPI_REG_SRAM_RESV_WORDS,
> >> +              reg_base + CQSPI_REG_SRAMPARTITION);
> >
> > sram partioning is not needed to be changed at every read/write, it
> > should be ok to configure it in the init like divide half for read & half for
> write.
> >
> 
> OK
> ...
> >
> > Rgds,
> > Vikas
> >
> 
> Best regards,
> Graham