Re: [PATCH v4 2/5] soc: mediatek: pwrap: add arbiter capability

From: Matthias Brugger
Date: Sun Jan 31 2021 - 09:20:30 EST

Next message: Hariprasad Kelam: "[Patch v3 net-next 2/7] octeontx2-af: Add new CGX_CMD to get PHY FEC statistics"
Previous message: Hariprasad Kelam: "[Patch v3 net-next 7/7] octeontx2-pf: ethtool physical link configuration"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On 21/12/2020 03:33, Nicolas Boichat wrote:
> On Wed, Nov 18, 2020 at 8:08 PM Hsin-Hsiung Wang
> <hsin-hsiung.wang@xxxxxxxxxxxx> wrote:
>>
>> Add arbiter capability for pwrap driver.
>> The arbiter capability uses new design to judge the priority and latency
>> for multi-channel.
>> This patch is preparing for adding mt6873/8192 pwrap support.
>>
>> Signed-off-by: Hsin-Hsiung Wang <hsin-hsiung.wang@xxxxxxxxxxxx>
>> ---
>> drivers/soc/mediatek/mtk-pmic-wrap.c | 57 ++++++++++++++++++++++++++++++------
>> 1 file changed, 48 insertions(+), 9 deletions(-)
>>
>> diff --git a/drivers/soc/mediatek/mtk-pmic-wrap.c b/drivers/soc/mediatek/mtk-pmic-wrap.c
>> index c897205..5678f46 100644
>> --- a/drivers/soc/mediatek/mtk-pmic-wrap.c
>> +++ b/drivers/soc/mediatek/mtk-pmic-wrap.c
>> @@ -25,10 +25,12 @@
>>
>> /* macro for wrapper status */
>> #define PWRAP_GET_WACS_RDATA(x) (((x) >> 0) & 0x0000ffff)
>> +#define PWRAP_GET_WACS_ARB_FSM(x) (((x) >> 1) & 0x00000007)
>> #define PWRAP_GET_WACS_FSM(x) (((x) >> 16) & 0x00000007)
>> #define PWRAP_GET_WACS_REQ(x) (((x) >> 19) & 0x00000001)
>> #define PWRAP_STATE_SYNC_IDLE0 BIT(20)
>> #define PWRAP_STATE_INIT_DONE0 BIT(21)
>> +#define PWRAP_STATE_INIT_DONE1 BIT(15)
>>
>> /* macro for WACS FSM */
>> #define PWRAP_WACS_FSM_IDLE 0x00
>> @@ -74,6 +76,7 @@
>> #define PWRAP_CAP_DCM BIT(2)
>> #define PWRAP_CAP_INT1_EN BIT(3)
>> #define PWRAP_CAP_WDT_SRC1 BIT(4)
>> +#define PWRAP_CAP_ARB BIT(5)
>>
>> /* defines for slave device wrapper registers */
>> enum dew_regs {
>> @@ -340,6 +343,8 @@ enum pwrap_regs {
>> PWRAP_DCM_DBC_PRD,
>> PWRAP_EINT_STA0_ADR,
>> PWRAP_EINT_STA1_ADR,
>> + PWRAP_SWINF_2_WDATA_31_0,
>> + PWRAP_SWINF_2_RDATA_31_0,
>>
>> /* MT2701 only regs */
>> PWRAP_ADC_CMD_ADDR,
>> @@ -1108,14 +1113,22 @@ static void pwrap_writel(struct pmic_wrapper *wrp, u32 val, enum pwrap_regs reg)
>>
>> static bool pwrap_is_fsm_idle(struct pmic_wrapper *wrp)
>> {
>> - u32 val = pwrap_readl(wrp, PWRAP_WACS2_RDATA);
>> + u32 val;
>> +
>> + val = pwrap_readl(wrp, PWRAP_WACS2_RDATA);
>> + if (HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB))
>> + return PWRAP_GET_WACS_ARB_FSM(val) == PWRAP_WACS_FSM_IDLE;
>>
>> return PWRAP_GET_WACS_FSM(val) == PWRAP_WACS_FSM_IDLE;
>> }
>>
>> static bool pwrap_is_fsm_vldclr(struct pmic_wrapper *wrp)
>> {
>> - u32 val = pwrap_readl(wrp, PWRAP_WACS2_RDATA);
>> + u32 val;
>> +
>> + val = pwrap_readl(wrp, PWRAP_WACS2_RDATA);
>> + if (HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB))
>> + return PWRAP_GET_WACS_ARB_FSM(val) == PWRAP_WACS_FSM_WFVLDCLR;
>
> This code is now copied twice. Do you think it'd be better to create a
> new function?
>
> static u32 pwrap_get_fsm_state(struct pmic_wrapper *wrp) {
> if (HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB))
> return PWRAP_GET_WACS_ARB_FSM(val);
> else
> return PWRAP_GET_WACS_FSM(val);
> }
>
>>
>> return PWRAP_GET_WACS_FSM(val) == PWRAP_WACS_FSM_WFVLDCLR;
>> }
>> @@ -1165,6 +1178,7 @@ static int pwrap_wait_for_state(struct pmic_wrapper *wrp,
>> static int pwrap_read16(struct pmic_wrapper *wrp, u32 adr, u32 *rdata)
>> {
>> int ret;
>> + u32 val;
>>
>> ret = pwrap_wait_for_state(wrp, pwrap_is_fsm_idle);
>> if (ret) {
>> @@ -1172,13 +1186,21 @@ static int pwrap_read16(struct pmic_wrapper *wrp, u32 adr, u32 *rdata)
>> return ret;
>> }
>>
>> - pwrap_writel(wrp, (adr >> 1) << 16, PWRAP_WACS2_CMD);
>> + if (HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB))
>> + val = adr;
>> + else
>> + val = (adr >> 1) << 16;
>> + pwrap_writel(wrp, val, PWRAP_WACS2_CMD);
>>
>> ret = pwrap_wait_for_state(wrp, pwrap_is_fsm_vldclr);
>> if (ret)
>> return ret;
>>
>> - *rdata = PWRAP_GET_WACS_RDATA(pwrap_readl(wrp, PWRAP_WACS2_RDATA));
>> + if (HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB))
>> + val = pwrap_readl(wrp, PWRAP_SWINF_2_RDATA_31_0);
>> + else
>> + val = pwrap_readl(wrp, PWRAP_WACS2_RDATA);
>> + *rdata = PWRAP_GET_WACS_RDATA(val);
>>
>> pwrap_writel(wrp, 1, PWRAP_WACS2_VLDCLR);
>>
>> @@ -1228,8 +1250,13 @@ static int pwrap_write16(struct pmic_wrapper *wrp, u32 adr, u32 wdata)
>> return ret;
>> }
>>
>> - pwrap_writel(wrp, (1 << 31) | ((adr >> 1) << 16) | wdata,
>> - PWRAP_WACS2_CMD);
>> + if (HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB)) {
>> + pwrap_writel(wrp, wdata, PWRAP_SWINF_2_WDATA_31_0);
>> + pwrap_writel(wrp, BIT(29) | adr, PWRAP_WACS2_CMD);
>> + } else {
>> + pwrap_writel(wrp, BIT(31) | ((adr >> 1) << 16) | wdata,
>> + PWRAP_WACS2_CMD);
>> + }
>>
>> return 0;
>> }
>> @@ -2022,6 +2049,7 @@ MODULE_DEVICE_TABLE(of, of_pwrap_match_tbl);
>> static int pwrap_probe(struct platform_device *pdev)
>> {
>> int ret, irq;
>> + u32 mask_done;
>> struct pmic_wrapper *wrp;
>> struct device_node *np = pdev->dev.of_node;
>> const struct of_device_id *of_slave_id = NULL;
>> @@ -2116,14 +2144,21 @@ static int pwrap_probe(struct platform_device *pdev)
>> }
>> }
>>
>> - if (!(pwrap_readl(wrp, PWRAP_WACS2_RDATA) & PWRAP_STATE_INIT_DONE0)) {
>> + if (HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB))
>> + mask_done = PWRAP_STATE_INIT_DONE1;
>> + else
>> + mask_done = PWRAP_STATE_INIT_DONE0;
>> +
>> + if (!(pwrap_readl(wrp, PWRAP_WACS2_RDATA) & mask_done)) {
>> dev_dbg(wrp->dev, "initialization isn't finished\n");
>> ret = -ENODEV;
>> goto err_out2;
>> }
>>
>> /* Initialize watchdog, may not be done by the bootloader */
>> - pwrap_writel(wrp, 0xf, PWRAP_WDT_UNIT);
>> + if (!HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB))
>> + pwrap_writel(wrp, 0xf, PWRAP_WDT_UNIT);
>> +
>
> To expand on Matthias' question on v3
> (https://patchwork.kernel.org/project/linux-mediatek/patch/1600686235-27979-3-git-send-email-hsin-hsiung.wang@xxxxxxxxxxxx/):
> is there any PWRAP implementation where a design with an arbiter is
> still able to control the watchdog?
>
> If not, at the very least, it'd be good to expand the comment above
> (e.g. "designs with arbiter support cannot change the watchdog
> timer").
>
>> /*
>> * Since STAUPD was not used on mt8173 platform,
>> * so STAUPD of WDT_SRC which should be turned off
>> @@ -2132,7 +2167,11 @@ static int pwrap_probe(struct platform_device *pdev)
>> if (HAS_CAP(wrp->master->caps, PWRAP_CAP_WDT_SRC1))
>> pwrap_writel(wrp, wrp->master->wdt_src, PWRAP_WDT_SRC_EN_1);
>>
>> - pwrap_writel(wrp, 0x1, PWRAP_TIMER_EN);
>> + if (!HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB))
>
> Please invert this if test:
>
> if (HAS_CAP(wrp->master->caps, PWRAP_CAP_ARB))
> ... 0x3 ...
> else
> ... 0x1 ...
>
>> + pwrap_writel(wrp, 0x1, PWRAP_TIMER_EN);
>> + else
>> + pwrap_writel(wrp, 0x3, PWRAP_TIMER_EN);
>> +

Thanks for the review Nicolas, I think with these two comments addressed the
patch is ready for mainline. Given of course that 4/5 isn't a hack with all the
registers being not defined.

Regards,
Matthias

>> pwrap_writel(wrp, wrp->master->int_en_all, PWRAP_INT_EN);
>> /*
>> * We add INT1 interrupt to handle starvation and request exception
>> --
>> 2.6.4
>> _______________________________________________
>> Linux-mediatek mailing list
>> Linux-mediatek@xxxxxxxxxxxxxxxxxxx
>> http://lists.infradead.org/mailman/listinfo/linux-mediatek

Next message: Hariprasad Kelam: "[Patch v3 net-next 2/7] octeontx2-af: Add new CGX_CMD to get PHY FEC statistics"
Previous message: Hariprasad Kelam: "[Patch v3 net-next 7/7] octeontx2-pf: ethtool physical link configuration"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]