RE: [PATCH v5 3/9] spi: Add a driver for the Freescale/NXP QuadSPI controller

[Yogesh Narayan Gaur <yogeshnarayan.gaur@xxxxxxx>]

Hi Frieder,

> -----Original Message-----
> From: Schrempf Frieder [mailto:frieder.schrempf@xxxxxxxxxx]
> Sent: Friday, November 16, 2018 3:12 PM
> To: Yogesh Narayan Gaur <yogeshnarayan.gaur@xxxxxxx>
> Cc: Boris Brezillon <boris.brezillon@xxxxxxxxxxx>; linux-mtd@xxxxxxxxxxxxxxxxxxx;
> linux-spi@xxxxxxxxxxxxxxx; Marek Vasut <marek.vasut@xxxxxxxxx>; Mark
> Brown <broonie@xxxxxxxxxx>; Han Xu <han.xu@xxxxxxx>;
> dwmw2@xxxxxxxxxxxxx; computersforpeace@xxxxxxxxx; richard@xxxxxx;
> miquel.raynal@xxxxxxxxxxx; David Wolfe <david.wolfe@xxxxxxx>; Fabio
> Estevam <fabio.estevam@xxxxxxx>; Prabhakar Kushwaha
> <prabhakar.kushwaha@xxxxxxx>; shawnguo@xxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH v5 3/9] spi: Add a driver for the Freescale/NXP QuadSPI
> controller
> 
> Hi Yogesh,
> 
> On 16.11.18 06:41, Yogesh Narayan Gaur wrote:
> > Hi Frieder,
> >
> >> -----Original Message-----
> >> From: Schrempf Frieder [mailto:frieder.schrempf@xxxxxxxxxx]
> >> Sent: Thursday, November 15, 2018 7:32 PM
> >> To: Yogesh Narayan Gaur <yogeshnarayan.gaur@xxxxxxx>
> >> Cc: Boris Brezillon <boris.brezillon@xxxxxxxxxxx>;
> >> linux-mtd@xxxxxxxxxxxxxxxxxxx; linux-spi@xxxxxxxxxxxxxxx; Marek Vasut
> >> <marek.vasut@xxxxxxxxx>; Mark Brown <broonie@xxxxxxxxxx>; Han Xu
> >> <han.xu@xxxxxxx>; dwmw2@xxxxxxxxxxxxx;
> computersforpeace@xxxxxxxxx;
> >> richard@xxxxxx; miquel.raynal@xxxxxxxxxxx; David Wolfe
> >> <david.wolfe@xxxxxxx>; Fabio Estevam <fabio.estevam@xxxxxxx>;
> >> Prabhakar Kushwaha <prabhakar.kushwaha@xxxxxxx>;
> shawnguo@xxxxxxxxxx;
> >> linux- kernel@xxxxxxxxxxxxxxx
> >> Subject: Re: [PATCH v5 3/9] spi: Add a driver for the Freescale/NXP
> >> QuadSPI controller
> >>
> >> Hi Yogesh,
> >>
> >> On 15.11.18 14:12, Boris Brezillon wrote:
> >>> On Thu, 15 Nov 2018 11:43:05 +0000
> >>> Schrempf Frieder <frieder.schrempf@xxxxxxxxxx> wrote:
> >>>
> >>>> On 15.11.18 07:22, Yogesh Narayan Gaur wrote:
> >>>>> Hi Frieder,
> >>>>>
> >>>>> With below patch on top of your v5, Read/Write/Erase on CS1 is
> >>>>> working
> >> fine for me.
> >>>>
> >>>> Ok, are you sure, that AHB read is working too with this patch?
> >>>> You are removing the memmap_phy offset from SFAR and the SFXXAD
> >>>> register values.
> >>>>
> >>>> I can understand that selection of the CS and IP commands will work
> >>>> like this, but I can't understand how AHB read should work without
> >>>> the base address of the mapped memory.
> >>>>
> >>>> I'm afraid I still don't fully understand the background of these
> >>>> things,
> >>>
> >>> Same here. Yogesh, can you give us more detail on why you decided to
> >>> drop the memmap_phy offset?
> >>
> >> Your changes do not work on my setup (i.MX6UL). It looks like your
> >> hardware is different.
> >>
> >> I found this patch for LS2080A: [1]. This would explain why you need
> >> to remove the offset to make it work.
> >>
> >> To verify this, could you please test your setup with the current
> >> spi-nor driver (fsl_quadspi.c). If our assumptions are right, it
> >> should only work on CS0 and CS1 with [1] applied.
> >>
> >
> > Yes, I need to remove the offset to make it work and this is required for the
> NXP Layerscape-2.x SoCs like LS208x/Ls108x etc.
> >
> > I have modified the patch and have introduced entry in quirks for ls2080a. With
> this Read/Write/Erase are working for me for both CS.
> 
> Ok, what I was asking for is a test with the original, unmodified SPI NOR driver in
> mtd/spi-nor/fsl-quadspi.c. We need this to confirm that the problem is really
> what we think, or to find out if we missed something.
> 
> Can you please do a quick test? If it confirms our assumptions, I will send a new
> version with the quirk and hopefully we can then move on.
> 

Yes, problem exist with original un-modified upstread SPI-NOR driver also.
Actually, internally we are maintaining driver with some local change and one of the change is related to same i.e. making having map_addr as 0 for layerscape chips.

I have tested, that by removing that CS1 access shows error.

Please integrate these changes in your next version.

--
Regards
Yogesh Gaur

> Thanks,
> Frieder
> 
> >
> > diff --git a/drivers/spi/spi-fsl-qspi.c b/drivers/spi/spi-fsl-qspi.c
> > index ce45e8e..5d26f73 100644
> > --- a/drivers/spi/spi-fsl-qspi.c
> > +++ b/drivers/spi/spi-fsl-qspi.c
> > @@ -175,6 +175,9 @@
> >   /* TKT245618, the controller cannot wake up from wait mode */
> >   #define QUADSPI_QUIRK_TKT245618                BIT(3)
> >
> > +/* QSPI_AMBA_BASE is internally added by SOC design for LS-2.x architecture
> */
> > +#define QUADSPI_AMBA_BASE_INTERNAL     BIT(4)
> > +
> >   struct fsl_qspi_devtype_data {
> >          unsigned int rxfifo;
> >          unsigned int txfifo;
> > @@ -227,7 +230,7 @@ static const struct fsl_qspi_devtype_data
> ls2080a_data = {
> >          .rxfifo = SZ_128,
> >          .txfifo = SZ_64,
> >          .ahb_buf_size = SZ_1K,
> > -       .quirks = QUADSPI_QUIRK_TKT253890,
> > +       .quirks = QUADSPI_QUIRK_TKT253890 |
> > + QUADSPI_AMBA_BASE_INTERNAL,
> >          .little_endian = true,
> >   };
> >
> > @@ -235,6 +238,7 @@ struct fsl_qspi {
> >          void __iomem *iobase;
> >          void __iomem *ahb_addr;
> >          u32 memmap_phy;
> > +       u32 amba_base_addr;
> >          struct clk *clk, *clk_en;
> >          struct device *dev;
> >          struct completion c;
> > @@ -264,6 +268,11 @@ static inline int needs_wakeup_wait_mode(struct
> fsl_qspi *q)
> >          return q->devtype_data->quirks & QUADSPI_QUIRK_TKT245618;
> >   }
> >
> > +static inline int has_added_amba_base_internal(struct fsl_qspi *q) {
> > +       return q->devtype_data->quirks & QUADSPI_AMBA_BASE_INTERNAL; }
> > +
> >   /*
> >    * An IC bug makes it necessary to rearrange the 32-bit data.
> >    * Later chips, such as IMX6SLX, have fixed this bug.
> > @@ -489,29 +498,11 @@ static void fsl_qspi_invalidate(struct fsl_qspi *q)
> >   static void fsl_qspi_select_mem(struct fsl_qspi *q, struct spi_device *spi)
> >   {
> >          unsigned long rate = spi->max_speed_hz;
> > -       int ret, i;
> > -       u32 map_addr;
> > +       int ret;
> >
> >          if (q->selected == spi->chip_select)
> >                  return;
> >
> > -       /*
> > -        * In HW there can be a maximum of four chips on two buses with
> > -        * two chip selects on each bus. We use four chip selects in SW
> > -        * to differentiate between the four chips.
> > -        * We use the SFA1AD, SFA2AD, SFB1AD, SFB2AD registers to select
> > -        * the chip we want to access.
> > -        */
> > -       for (i = 0; i < 4; i++) {
> > -               if (i < spi->chip_select)
> > -                       map_addr = q->memmap_phy;
> > -               else
> > -                       map_addr = q->memmap_phy +
> > -                                  2 * q->devtype_data->ahb_buf_size;
> > -
> > -               qspi_writel(q, map_addr, q->iobase + QUADSPI_SFA1AD + (i * 4));
> > -       }
> > -
> >          if (needs_4x_clock(q))
> >                  rate *= 4;
> >
> > @@ -534,7 +525,9 @@ static void fsl_qspi_select_mem(struct fsl_qspi
> > *q, struct spi_device *spi)
> >
> >   static void fsl_qspi_read_ahb(struct fsl_qspi *q, const struct spi_mem_op *op)
> >   {
> > -       memcpy_fromio(op->data.buf.in, q->ahb_addr, op->data.nbytes);
> > +       memcpy_fromio(op->data.buf.in,
> > +                     q->ahb_addr + q->selected * q->devtype_data->ahb_buf_size,
> > +                     op->data.nbytes);
> >   }
> >
> >   static void fsl_qspi_fill_txfifo(struct fsl_qspi *q, @@ -634,7
> > +627,9 @@ static int fsl_qspi_exec_op(struct spi_mem *mem, const
> > struct spi_mem_op *op)
> >
> >          fsl_qspi_select_mem(q, mem->spi);
> >
> > -       qspi_writel(q, q->memmap_phy, base + QUADSPI_SFAR);
> > +       qspi_writel(q, q->amba_base_addr +
> > +                   q->selected * q->devtype_data->ahb_buf_size,
> > +                   base + QUADSPI_SFAR);
> >
> >          qspi_writel(q, qspi_readl(q, base + QUADSPI_MCR) |
> >                      QUADSPI_MCR_CLR_RXF_MASK |
> > QUADSPI_MCR_CLR_TXF_MASK, @@ -733,6 +728,23 @@ static int
> fsl_qspi_default_setup(struct fsl_qspi *q)
> >                      QUADSPI_BUF3CR_ADATSZ(q->devtype_data->ahb_buf_size / 8),
> >                      base + QUADSPI_BUF3CR);
> >
> > +       /*
> > +        * In HW there can be a maximum of four chips on two buses with
> > +        * two chip selects on each bus. We use four chip selects in SW
> > +        * to differentiate between the four chips.
> > +        * We use the SFA1AD, SFA2AD, SFB1AD, SFB2AD registers to select
> > +        * the chip we want to access.
> > +        */
> > +
> > +       qspi_writel(q, q->amba_base_addr + q->devtype_data->ahb_buf_size,
> > +                   base + QUADSPI_SFA1AD);
> > +       qspi_writel(q, q->amba_base_addr + q->devtype_data->ahb_buf_size * 2,
> > +                   base + QUADSPI_SFA2AD);
> > +       qspi_writel(q, q->amba_base_addr + q->devtype_data->ahb_buf_size * 3,
> > +                   base + QUADSPI_SFB1AD);
> > +       qspi_writel(q, q->amba_base_addr + q->devtype_data->ahb_buf_size * 4,
> > +                   base + QUADSPI_SFB2AD);
> > +
> >          q->selected = -1;
> >
> >          /* Enable the module */
> > @@ -825,6 +837,11 @@ static int fsl_qspi_probe(struct platform_device
> > *pdev)
> >
> >          q->memmap_phy = res->start;
> >
> > +       if (has_added_amba_base_internal(q))
> > +               q->amba_base_addr = 0;
> > +       else
> > +               q->amba_base_addr = q->memmap_phy;
> > +
> >          /* find the clocks */
> >          q->clk_en = devm_clk_get(dev, "qspi_en");
> >          if (IS_ERR(q->clk_en)) {
> >
> > --
> > Regards
> > Yogesh Gaur
> > [...]
> >