RE: [EXT] Re: [PATCH] spi: spi-fsl-qspi: Reduce devm_ioremap size to 4 times AHB buffer size

[Ashish Kumar]

Hi Frieder,

> -----Original Message-----
> From: Schrempf Frieder <frieder.schrempf@xxxxxxxxxx>
> Sent: Thursday, April 16, 2020 3:06 PM
> To: Ashish Kumar <ashish.kumar@xxxxxxx>; broonie@xxxxxxxxxx;
> boris.brezillon@xxxxxxxxxxxxx
> Cc: linux-spi@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Kuldeep Singh
> <kuldeep.singh@xxxxxxx>
> Subject: [EXT] Re: [PATCH] spi: spi-fsl-qspi: Reduce devm_ioremap size to 4
> times AHB buffer size
> 
> Caution: EXT Email
> 
> Hi Ashish,
> 
> On 16.04.20 10:33, Ashish Kumar wrote:
> > Reduce devm_ioremap size to (4 * AHB_BUFER_SIZE) rather than mapping
> > complete QSPI-Memmory as driver is now independent of flash size.
> > Flash of any size can be accessed.
> >
> > Issue was reported on platform where devm_ioremap failure is observed
> > with size > 256M.
> > Error log on LS1021ATWR :
> >   fsl-quadspi 1550000.spi: ioremap failed for resource [mem 0x40000000-
> 0x7fffffff]
> >   fsl-quadspi 1550000.spi: Freescale QuadSPI probe failed
> >   fsl-quadspi: probe of 1550000.spi failed with error -12
> >
> > This change was also suggested previously:
> > https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpatc
> >
> hwork.kernel.org%2Fpatch%2F10508753%2F%2322166385&amp;data=02%7C
> 01%7Ca
> >
> shish.kumar%40nxp.com%7C7e4a6913030f4f5a4a0408d7e1e9a342%7C686ea
> 1d3bc2
> >
> b4c6fa92cd99c5c301635%7C0%7C0%7C637226265903575933&amp;sdata=JDb
> k0rA2L
> > 5EwAah8S3srd5u26bOxaD1MVR3voUKFxyI%3D&amp;reserved=0
> 
> So if you want you could add a "Suggested-by" tag referring to Boris.
Sure, I will add in next version.

> 
> >
> > Signed-off-by: Kuldeep Singh <kuldeep.singh@xxxxxxx>
> > Signed-off-by: Ashish Kumar <Ashish.kumar@xxxxxxx>
> > ---
> >   drivers/spi/spi-fsl-qspi.c | 10 +++++++---
> >   1 file changed, 7 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/spi/spi-fsl-qspi.c b/drivers/spi/spi-fsl-qspi.c
> > index 63c9f7e..a41ce81 100644
> > --- a/drivers/spi/spi-fsl-qspi.c
> > +++ b/drivers/spi/spi-fsl-qspi.c
> > @@ -859,14 +859,15 @@ static int fsl_qspi_probe(struct platform_device
> > *pdev)
> >
> >       res = platform_get_resource_byname(pdev, IORESOURCE_MEM,
> >                                       "QuadSPI-memory");
> > -     q->ahb_addr = devm_ioremap_resource(dev, res);
> > +     q->memmap_phy = res->start;
> > +     /* Since there are 4 CS max MAP required is 4 time ahb_buf_size
> > + */
> 
> /* Since there are 4 cs, map size required is 4 times ahb_buf_size */
> 
> > +     q->ahb_addr = devm_ioremap(dev, q->memmap_phy,
> > +                                (q->devtype_data->ahb_buf_size * 4));
> >       if (IS_ERR(q->ahb_addr)) {
> >               ret = PTR_ERR(q->ahb_addr);
> >               goto err_put_ctrl;
> >       }
> >
> > -     q->memmap_phy = res->start;
> > -
> >       /* find the clocks */
> >       q->clk_en = devm_clk_get(dev, "qspi_en");
> >       if (IS_ERR(q->clk_en)) {
> > @@ -939,6 +940,9 @@ static int fsl_qspi_remove(struct platform_device
> > *pdev)
> >
> >       mutex_destroy(&q->lock);
> >
> > +     if (q->ahb_addr)
> > +             devm_iounmap(q->dev, q->ahb_addr);
> 
> You don't need to unmap explicitly. Using devm_*(), lifetime of the resource
> is already tied to the device.
Thanks. I will remove this.

Regards
Ashish 
> 
> Thanks,
> Frieder
> 
> > +
> >       return 0;
> >   }
> >
> >