Re: [RFC PATCH 2/3] mmc: sdhci-esdhc-imx: add NXP S32G2 support

[Chester Lin <clin@suse.com>]

Hi Haibo,

On Thu, Oct 21, 2021 at 08:00:43AM +0000, Bough Chen wrote:
> > -----Original Message-----
> > From: Chester Lin [mailto:clin@suse.com]
> > Sent: 2021年10月21日 15:31
> > To: Ulf Hansson <ulf.hansson@linaro.org>; dl-S32 <S32@nxp.com>; dl-linux-imx
> > <linux-imx@nxp.com>; Bough Chen <haibo.chen@nxp.com>; Aisheng Dong
> > <aisheng.dong@nxp.com>; linux-mmc@vger.kernel.org
> > Cc: Rob Herring <robh+dt@kernel.org>; Shawn Guo <shawnguo@kernel.org>;
> > Sascha Hauer <s.hauer@pengutronix.de>; Pengutronix Kernel Team
> > <kernel@pengutronix.de>; Fabio Estevam <festevam@gmail.com>;
> > devicetree@vger.kernel.org; linux-arm-kernel@lists.infradead.org;
> > linux-kernel@vger.kernel.org; Radu-nicolae Pirea (OSS)
> > <radu-nicolae.pirea@oss.nxp.com>; Andreas Färber <afaerber@suse.de>;
> > Matthias Brugger <mbrugger@suse.com>; Ivan T . Ivanov <iivanov@suse.de>;
> > Lee, Chun-Yi <jlee@suse.com>
> > Subject: Re: [RFC PATCH 2/3] mmc: sdhci-esdhc-imx: add NXP S32G2 support
> > 
> > Hi NXP S32 and i.MX Linux teams,
> > 
> > On Thu, Oct 21, 2021 at 03:13:32PM +0800, Chester Lin wrote:
> > > Support the SDHCI controller found on NXP S32G2 platform. The new flag
> > > ESDHC_FLAG_SKIP_ERR004536 is used because the hardware erratum bit is
> > > not applicable for S32G2.
> > >
> > > Signed-off-by: Chester Lin <clin@suse.com>
> > > ---
> > >  drivers/mmc/host/sdhci-esdhc-imx.c | 17 +++++++++++++++--
> > >  1 file changed, 15 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/drivers/mmc/host/sdhci-esdhc-imx.c
> > > b/drivers/mmc/host/sdhci-esdhc-imx.c
> > > index f18d169bc8ff..d0f7d46a0354 100644
> > > --- a/drivers/mmc/host/sdhci-esdhc-imx.c
> > > +++ b/drivers/mmc/host/sdhci-esdhc-imx.c
> > > @@ -196,6 +196,9 @@
> > >   */
> > >  #define ESDHC_FLAG_BROKEN_AUTO_CMD23	BIT(16)
> > >
> > > +/* ERR004536 is not applicable for the IP  */
> > > +#define ESDHC_FLAG_SKIP_ERR004536	BIT(17)
> > > +
> > >  enum wp_types {
> > >  	ESDHC_WP_NONE,		/* no WP, neither controller nor gpio */
> > >  	ESDHC_WP_CONTROLLER,	/* mmc controller internal WP */
> > > @@ -289,6 +292,13 @@ static const struct esdhc_soc_data
> > usdhc_imx7d_data = {
> > >  			| ESDHC_FLAG_BROKEN_AUTO_CMD23,
> > >  };
> > >
> > > +static struct esdhc_soc_data usdhc_s32g2_data = {
> > > +	.flags = ESDHC_FLAG_USDHC | ESDHC_FLAG_MAN_TUNING
> > > +			| ESDHC_FLAG_HAVE_CAP1 | ESDHC_FLAG_HS200
> > > +			| ESDHC_FLAG_HS400 | ESDHC_FLAG_HS400_ES
> > > +			| ESDHC_FLAG_SKIP_ERR004536,
> > > +};
> > > +
> > >  static struct esdhc_soc_data usdhc_imx7ulp_data = {
> > >  	.flags = ESDHC_FLAG_USDHC | ESDHC_FLAG_STD_TUNING
> > >  			| ESDHC_FLAG_HAVE_CAP1 | ESDHC_FLAG_HS200 @@ -347,6
> > +357,7 @@
> > > static const struct of_device_id imx_esdhc_dt_ids[] = {
> > >  	{ .compatible = "fsl,imx7ulp-usdhc", .data = &usdhc_imx7ulp_data, },
> > >  	{ .compatible = "fsl,imx8qxp-usdhc", .data = &usdhc_imx8qxp_data, },
> > >  	{ .compatible = "fsl,imx8mm-usdhc", .data = &usdhc_imx8mm_data, },
> > > +	{ .compatible = "nxp,s32g2-usdhc", .data = &usdhc_s32g2_data, },
> > >  	{ /* sentinel */ }
> > >  };
> > >  MODULE_DEVICE_TABLE(of, imx_esdhc_dt_ids); @@ -1359,8 +1370,10
> > @@
> > > static void sdhci_esdhc_imx_hwinit(struct sdhci_host *host)
> > >  		 * erratum ESDHC_FLAG_ERR004536 fix for MX6Q TO1.2 and MX6DL
> > >  		 * TO1.1, it's harmless for MX6SL
> > >  		 */
> > > -		writel(readl(host->ioaddr + 0x6c) & ~BIT(7),
> > > -			host->ioaddr + 0x6c);
> > > +		if (!(imx_data->socdata->flags & ESDHC_FLAG_SKIP_ERR004536)) {
> > > +			writel(readl(host->ioaddr + 0x6c) & ~BIT(7),
> > > +				host->ioaddr + 0x6c);
> > > +		}
> > 
> > Hope you don't might that I raise this question here. Is it really necessary to
> > unconditionally apply the erratum bit even if some SoCs might not need this
> > workaround? From the S32 implementation in CodeAurora[1], I noticed that
> > this bit is not required by S32V/S32G so I wonder if there's any better way to
> > refine this part?
> > 
> 
> I confirmed with IP owner before, for SoC contain this errata fixup, clear this bit 7
> will enable the fixup, set the bit 7 will disable the fixup.
> For SoC which do not contain this errata fixup, this bit 7 has no definition.
> So it's okay to clear this bit 7 unconditionally.
> 

Thanks for your reply. If I understand correctly, this bit should be almost the
same even if the IP can be used in different SoCs. Actually I haven't found an
issue even if I have tried to clear the bit-7 on S32G274A although this patch
just gets the driver working on S32G, which still has limited functions [e.g.
pins_100mhz and pins_200mhz are missing]. I just wonder if any case should
avoid touching this bit since the s32 downstream kernel has a specific handling
for this part.

@NXP S32 team:
Please let us know if any concern about this bit.

Thanks,
Chester
> 
> 
> > Thanks,
> > Chester
> > 
> > [1]
> > https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fsource.
> > codeaurora.org%2Fexternal%2Fautobsps32%2Flinux%2Ftree%2Fdrivers%2Fm
> > mc%2Fhost%2Fsdhci-esdhc-imx.c%3Fh%3Drelease%2Fbsp30.0-5.4-rt%23n1268
> > &amp;data=04%7C01%7Chaibo.chen%40nxp.com%7Cec36a273354b4b66c45b
> > 08d99464c449%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63770
> > 3982782606697%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJ
> > QIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=hvzsy
> > 3W%2FpKXBmXxS8%2F73huzb157a%2FuHa5G4lFWj5ABQ%3D&amp;reserved=
> > 0
> > 
> > >
> > >  		/* disable DLL_CTRL delay line settings */
> > >  		writel(0x0, host->ioaddr + ESDHC_DLL_CTRL);
> > > --
> > > 2.30.0
> > >
>