[PATCH v4 0/4] StarFive's SDIO/eMMC driver support

[Ulf Hansson]

On Mon, 27 Feb 2023 at 08:47, William Qiu <william.qiu at starfivetech.com> wrote:
>
>
>
> On 2023/2/15 20:37, Ulf Hansson wrote:
> > On Wed, 15 Feb 2023 at 12:32, William Qiu <william.qiu at starfivetech.com> wrote:
> >>
> >> Hi,
> >>
> >> This patchset adds initial rudimentary support for the StarFive
> >> designware mobile storage host controller driver. And this driver will
> >> be used in StarFive's VisionFive 2 board. The main purpose of adding
> >> this driver is to accommodate the ultra-high speed mode of eMMC.
> >>
> >> The last patch should be applied after the patchset [1]:
> >> [1] https://lore.kernel.org/all/20221220011247.35560-1-hal.feng@starfivetech.com/
> >>
> >> Changes v3->v4:
> >> - Added documentation to describe StarFive System Controller Registers.
> >> - Added aon_syscon and stg_syscon node.
> >> - Fixed some checkpatch errors/warnings.
> >>
> >> Changes v2->v3:
> >> - Wraped commit message according to Linux coding style.
> >> - Rephrased the description of the patches.
> >> - Changed the description of syscon regsiter.
> >> - Dropped redundant properties.
> >>
> >> Changes v1->v2:
> >> - Renamed the dt-binding 'starfive,jh7110-sdio.yaml' to 'starfive,jh7110-mmc.yaml'.
> >> - Changed the type of 'starfive,syscon' and modify its description.
> >> - Deleted unused head files like '#include <linux/gpio.h>'.
> >> - Added comment for the 'rise_point' and 'fall_point'.
> >> - Changed the API 'num_caps' to 'common_caps'.
> >> - Changed the node name 'sys_syscon' to 'syscon'.
> >> - Changed the node name 'sdio' to 'mmc'.
> >>
> >> The patch series is based on v6.1.
> >>
> >> William Qiu (4):
> >>   dt-bindings: mmc: Add StarFive MMC module
> >>   mmc: starfive: Add sdio/emmc driver support
> >>   riscv: dts: starfive: Add mmc node
> >>   dt-bindings: syscon: Add StarFive syscon doc
> >>
> >>  .../bindings/mmc/starfive,jh7110-mmc.yaml     |  77 ++++++++
> >>  .../bindings/soc/starfive/jh7110-syscon.yaml  |  51 +++++
> >>  MAINTAINERS                                   |  11 ++
> >>  .../jh7110-starfive-visionfive-2.dtsi         |  23 +++
> >>  arch/riscv/boot/dts/starfive/jh7110.dtsi      |  47 +++++
> >>  drivers/mmc/host/Kconfig                      |  10 +
> >>  drivers/mmc/host/Makefile                     |   1 +
> >>  drivers/mmc/host/dw_mmc-starfive.c            | 186 ++++++++++++++++++
> >>  8 files changed, 406 insertions(+)
> >>  create mode 100644 Documentation/devicetree/bindings/mmc/starfive,jh7110-mmc.yaml
> >>  create mode 100644 Documentation/devicetree/bindings/soc/starfive/jh7110-syscon.yaml
> >>  create mode 100644 drivers/mmc/host/dw_mmc-starfive.c
> >>
> >
> > I have dropped the v3 patches and applied patch1 and patch2 from the
> > v4 series instead, for my next branch, thanks!
> >
> > Kind regards
> > Uffe
>
> Hi Uffe,
>
> Sorry to bother you.But I found a bug that in drivers/mmc/host/dw_mmc-starfive.c:
>
>     47 static int dw_mci_starfive_execute_tuning(struct dw_mci_slot *slot,
>     48                                              u32 opcode)
>     49 {
>     50         static const int grade  = MAX_DELAY_CHAIN;
>     51         struct dw_mci *host = slot->host;
>     52         struct starfive_priv *priv = host->priv;
>     53         int rise_point = -1, fall_point = -1;
>     54         int err, prev_err;
>     55         int i;
>     56         bool found = 0;
>     57         u32 regval;
>     58
>     59         /*
>     60          * Use grade as the max delay chain, and use the rise_point and
>     61          * fall_point to ensure the best sampling point of a data input
>     62          * signals.
>     63          */
>     64         for (i = 0; i < grade; i++) {
>     65                 regval = i << priv->syscon_shift;
>     66                 err = regmap_update_bits(priv->reg_syscon, priv->syscon_offset,
>     67                                                 priv->syscon_mask, regval);
>     68                 if (err)
>     69                         return err;
>     70                 mci_writel(host, RINTSTS, ALL_INT_CLR);
>     71
>     72                 err = mmc_send_tuning(slot->mmc, opcode, NULL);
>     73                 if (!err)
>     74                         found = 1;
>     75
>     76                 if (i > 0) {
> --> 77                         if (err && !prev_err)
>
> prev_err was never initialized to zero.
>
> So I'm here to ask for your suggestion, should I send a new version
> to fix it or send you a patch with a fixes tag?

Please send a new incremental patch on top. I will queue it up as a
fix for v6.3-rc[n].

Kind regards
Uffe