RE: [PATCH 3/3] phy: intel: Add Thunder Bay eMMC PHY support

From: A, Rashmi
Date: Tue Aug 10 2021 - 04:51:58 EST




> -----Original Message-----
> From: Vinod Koul <vkoul@xxxxxxxxxx>
> Sent: Friday, August 6, 2021 6:09 PM
> To: A, Rashmi <rashmi.a@xxxxxxxxx>
> Cc: linux-drivers-review-request@xxxxxxxxxxxxxxxxx; michal.simek@xxxxxxxxxx;
> ulf.hansson@xxxxxxxxxx; linux-mmc@xxxxxxxxxxxxxxx; linux-arm-
> kernel@xxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; kishon@xxxxxx;
> andriy.shevchenko@xxxxxxxxxxxxxxx; linux-phy@xxxxxxxxxxxxxxxxxxx;
> mgross@xxxxxxxxxxxxxxx; kris.pan@xxxxxxxxxxxxxxx; Zhou, Furong
> <furong.zhou@xxxxxxxxx>; Sangannavar, Mallikarjunappa
> <mallikarjunappa.sangannavar@xxxxxxxxx>; Hunter, Adrian
> <adrian.hunter@xxxxxxxxx>; Vaidya, Mahesh R
> <mahesh.r.vaidya@xxxxxxxxx>; Srikandan, Nandhini
> <nandhini.srikandan@xxxxxxxxx>; Demakkanavar, Kenchappa
> <kenchappa.demakkanavar@xxxxxxxxx>
> Subject: Re: [PATCH 3/3] phy: intel: Add Thunder Bay eMMC PHY support
>
> On 30-07-21, 12:03, rashmi.a@xxxxxxxxx wrote:
>
> > diff --git a/drivers/phy/intel/Kconfig b/drivers/phy/intel/Kconfig
> > index ac42bb2fb394..18a3cc5b98c0 100644
> > --- a/drivers/phy/intel/Kconfig
> > +++ b/drivers/phy/intel/Kconfig
> > @@ -46,3 +46,13 @@ config PHY_INTEL_LGM_EMMC
> > select GENERIC_PHY
> > help
> > Enable this to support the Intel EMMC PHY
> > +
> > +config PHY_INTEL_THUNDERBAY_EMMC
>
> Alphabetical sort please
>
> > + tristate "Intel Thunder Bay eMMC PHY driver"
> > + depends on OF && (ARCH_THUNDERBAY || COMPILE_TEST)
> > + select GENERIC_PHY
> > + help
> > + This option enables support for Intel Thunder Bay SoC eMMC PHY.
> > +
> > + To compile this driver as a module, choose M here: the module
> > + will be called phy-intel-thunderbay-emmc.ko.
> > diff --git a/drivers/phy/intel/Makefile b/drivers/phy/intel/Makefile
> > index 14550981a707..6a4db3ee7393 100644
> > --- a/drivers/phy/intel/Makefile
> > +++ b/drivers/phy/intel/Makefile
> > @@ -1,5 +1,6 @@
> > # SPDX-License-Identifier: GPL-2.0
> > obj-$(CONFIG_PHY_INTEL_KEEMBAY_EMMC) += phy-intel-
> keembay-emmc.o
> > +obj-$(CONFIG_PHY_INTEL_THUNDERBAY_EMMC) += phy-intel-
> thunderbay-emmc.o
>
> here as well
I will reorder the entry in Makefile. The Kconfig entries are in alphabetical order. Let me know if I am missing something.

-Rashmi
>
> > +static int thunderbay_emmc_phy_power(struct phy *phy, bool power_on)
> > +{
> > + struct thunderbay_emmc_phy *tbh_phy = phy_get_drvdata(phy);
> > + unsigned int freqsel = FREQSEL_200M_170M;
> > + unsigned long rate;
> > + static int lock;
> > + u32 val;
> > + int ret;
> > +
> > + /* Disable DLL */
> > + rate = clk_get_rate(tbh_phy->emmcclk);
> > + switch (rate) {
> > + case 200000000:
> > + /* lock dll only when it is used, i.e only if SEL_DLY_TXCLK/RXCLK are 0
> */
> > + update_reg(tbh_phy, PHY_CFG_0, DLL_EN_MASK,
> DLL_EN_SHIFT, 0x0);
>
> pls keep the same indent for comment and code!
Acknowledged

-Rashmi
>
> > + break;
> > + /* dll lock not required for other frequencies */
> > + case 50000000 ... 52000000:
> > + case 400000:
> > + default:
> > + break;
> > + }
> > +
> > + if (!power_on)
> > + return 0;
>
> should this not be the first thing you check...
During phy power on/off condition, dll should be disabled first. So power_on check is made after disabling dll.

-Rashmi
>
> > +
> > + rate = clk_get_rate(tbh_phy->emmcclk);
> > + switch (rate) {
> > + case 170000001 ... 200000000:
> > + freqsel = FREQSEL_200M_170M;
> > + break;
> > + case 140000001 ... 170000000:
> > + freqsel = FREQSEL_170M_140M;
> > + break;
> > + case 110000001 ... 140000000:
> > + freqsel = FREQSEL_140M_110M;
> > + break;
> > + case 80000001 ... 110000000:
> > + freqsel = FREQSEL_110M_80M;
> > + break;
> > + case 50000000 ... 80000000:
> > + freqsel = FREQSEL_80M_50M;
> > + break;
> > + case 250000001 ... 275000000:
> > + freqsel = FREQSEL_275M_250M;
> > + break;
> > + case 225000001 ... 250000000:
> > + freqsel = FREQSEL_250M_225M;
> > + break;
> > + case 200000001 ... 225000000:
> > + freqsel = FREQSEL_225M_200M;
> > + break;
> > + default:
> > + break;
> > + }
> > +
> > + if (rate > 200000000)
> > + /* only the upper limit is considered as the clock rate may fall low
> during init */
> > + dev_warn(&phy->dev, "Unsupported rate: %lu\n", rate);
>
> here as well! (checkpatch --strict should have told you so)
I will follow the proper alignment, though check-patch script didn't show any warnings.

-Rashmi
>
> > +static int thunderbay_emmc_phy_power_on(struct phy *phy) {
> > + struct thunderbay_emmc_phy *tbh_phy = phy_get_drvdata(phy);
> > + unsigned long rate;
> > +
> > + /* Overwrite capability bits configurable in bootloader */
> > + update_reg(tbh_phy, CTRL_CFG_0,
> > + SUPPORT_HS_MASK, SUPPORT_HS_SHIFT, 0x1);
> > + update_reg(tbh_phy, CTRL_CFG_0,
> > + SUPPORT_8B_MASK, SUPPORT_8B_SHIFT, 0x1);
> > + update_reg(tbh_phy, CTRL_CFG_1,
> > + SUPPORT_SDR50_MASK, SUPPORT_SDR50_SHIFT, 0x1);
> > + update_reg(tbh_phy, CTRL_CFG_1,
> > + SUPPORT_DDR50_MASK, SUPPORT_DDR50_SHIFT, 0x1);
> > + update_reg(tbh_phy, CTRL_CFG_1,
> > + SUPPORT_SDR104_MASK, SUPPORT_SDR104_SHIFT, 0x1);
> > + update_reg(tbh_phy, CTRL_CFG_1,
> > + SUPPORT_HS400_MASK, SUPPORT_HS400_SHIFT, 0x1);
> > + update_reg(tbh_phy, CTRL_CFG_1,
> > + SUPPORT_64B_MASK, SUPPORT_64B_SHIFT, 0x1);
> > +
> > + if (tbh_phy->phy_power_sts == PHY_UNINITIALIZED) {
> > + /* Indicates initialization, so settings to be done for init , same as
> 400KHZ setting */
> > + update_reg(tbh_phy, PHY_CFG_0, SEL_DLY_TXCLK_MASK,
> > +SEL_DLY_TXCLK_SHIFT, 0x1);
>
> inconsistent indent here too!
>
> > + update_reg(tbh_phy, PHY_CFG_0, SEL_DLY_RXCLK_MASK,
> SEL_DLY_RXCLK_SHIFT, 0x1);
> > + update_reg(tbh_phy, PHY_CFG_0, ITAP_DLY_ENA_MASK,
> ITAP_DLY_ENA_SHIFT, 0x0);
> > + update_reg(tbh_phy, PHY_CFG_0, ITAP_DLY_SEL_MASK,
> ITAP_DLY_SEL_SHIFT, 0x0);
> > + update_reg(tbh_phy, PHY_CFG_0, OTAP_DLY_ENA_MASK,
> OTAP_DLY_ENA_SHIFT, 0x0);
> > + update_reg(tbh_phy, PHY_CFG_0, OTAP_DLY_SEL_MASK,
> OTAP_DLY_SEL_SHIFT, 0);
> > + update_reg(tbh_phy, PHY_CFG_0, DLL_TRIM_ICP_MASK,
> DLL_TRIM_ICP_SHIFT, 0);
> > + update_reg(tbh_phy, PHY_CFG_0, DR_TY_MASK,
> DR_TY_SHIFT, 0x1);
> > +
> > + } else if (tbh_phy->phy_power_sts == PHY_INITIALIZED) {
> > + /* Indicates actual clock setting */
> > + rate = clk_get_rate(tbh_phy->emmcclk);
> > + switch (rate) {
> > + case 200000000:
> > + update_reg(tbh_phy, PHY_CFG_0,
> SEL_DLY_TXCLK_MASK,
> > + SEL_DLY_TXCLK_SHIFT, 0x0);
> > + update_reg(tbh_phy, PHY_CFG_0,
> SEL_DLY_RXCLK_MASK,
> > + SEL_DLY_RXCLK_SHIFT, 0x0);
> > + update_reg(tbh_phy, PHY_CFG_0,
> ITAP_DLY_ENA_MASK,
> > + ITAP_DLY_ENA_SHIFT, 0x0);
> > + update_reg(tbh_phy, PHY_CFG_0,
> ITAP_DLY_SEL_MASK,
> > + ITAP_DLY_SEL_SHIFT, 0x0);
> > + update_reg(tbh_phy, PHY_CFG_0,
> OTAP_DLY_ENA_MASK,
> > + OTAP_DLY_ENA_SHIFT, 0x1);
> > + update_reg(tbh_phy, PHY_CFG_0,
> OTAP_DLY_SEL_MASK,
> > + OTAP_DLY_SEL_SHIFT, 2);
> > + update_reg(tbh_phy, PHY_CFG_0,
> DLL_TRIM_ICP_MASK,
> > + DLL_TRIM_ICP_SHIFT, 0x8);
> > + update_reg(tbh_phy, PHY_CFG_0, DR_TY_MASK,
> > + DR_TY_SHIFT, 0x1);
> > + /* For HS400 only */
> > + update_reg(tbh_phy, PHY_CFG_2, SEL_STRB_MASK,
> > + SEL_STRB_SHIFT, STRB);
> > + break;
>
> pls give a empty line after break, helps to read the code
Acknowledged

-Rashmi