[RFC PATCH 03/10] devfreq: rk3399_dmc: Pass ODT and auto power down parameters to TF-A.
Enric Balletbo Serra
eballetbo at gmail.com
Sat Jun 16 10:15:34 UTC 2018
Hi Chanwoo,
I'll send a new version soon, just wanted to ask some questions here. See below.
Missatge de Chanwoo Choi <cw00.choi at samsung.com> del dia dt., 15 de
maig 2018 a les 0:21:
>
> Hi,
>
> On 2018년 05월 15일 06:16, Enric Balletbo i Serra wrote:
> > Trusted Firmware-A (TF-A) for rk3399 implements a SiP call to get the
> > on-die termination (ODT) and auto power down parameters from kernel,
> > this patch adds the functionality to do this. Also, if DDR clock
> > frequency is lower than the on-die termination (ODT) disable frequency
> > this driver should disable the DDR ODT.
>
> I have a question.
> 'disable frequency' is the same meaning of 'disable the DDR ODT'?
>
Yes, the DT defines an odt_dis_freq parameter, when the DDR frequency
is less than the value in this parameter we disable the ODT on the
DRAM.
> >
> > Signed-off-by: Enric Balletbo i Serra <enric.balletbo at collabora.com>
> > ---
> >
> > drivers/devfreq/rk3399_dmc.c | 50 ++++++++++++++++++++++++++++-
> > include/soc/rockchip/rockchip_sip.h | 1 +
> > 2 files changed, 50 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/devfreq/rk3399_dmc.c b/drivers/devfreq/rk3399_dmc.c
> > index d5c03e5abe13..cc1bbca3fb15 100644
> > --- a/drivers/devfreq/rk3399_dmc.c
> > +++ b/drivers/devfreq/rk3399_dmc.c
> > @@ -18,14 +18,17 @@
> > #include <linux/devfreq.h>
> > #include <linux/devfreq-event.h>
> > #include <linux/interrupt.h>
> > +#include <linux/mfd/syscon.h>
> > #include <linux/module.h>
> > #include <linux/of.h>
> > #include <linux/platform_device.h>
> > #include <linux/pm_opp.h>
> > +#include <linux/regmap.h>
> > #include <linux/regulator/consumer.h>
> > #include <linux/rwsem.h>
> > #include <linux/suspend.h>
> >
> > +#include <soc/rockchip/rk3399_grf.h>
> > #include <soc/rockchip/rockchip_sip.h>
> >
> > struct dram_timing {
> > @@ -69,8 +72,11 @@ struct rk3399_dmcfreq {
> > struct mutex lock;
> > struct dram_timing timing;
> > struct regulator *vdd_center;
> > + struct regmap *regmap_pmu;
> > unsigned long rate, target_rate;
> > unsigned long volt, target_volt;
> > + unsigned int odt_dis_freq;
> > + int odt_pd_arg0, odt_pd_arg1;
> > };
> >
> > static int rk3399_dmcfreq_target(struct device *dev, unsigned long *freq,
> > @@ -80,6 +86,8 @@ static int rk3399_dmcfreq_target(struct device *dev, unsigned long *freq,
> > struct dev_pm_opp *opp;
> > unsigned long old_clk_rate = dmcfreq->rate;
> > unsigned long target_volt, target_rate;
> > + struct arm_smccc_res res;
> > + int dram_flag;
> > int err;
> >
> > opp = devfreq_recommended_opp(dev, freq, flags);
> > @@ -95,6 +103,15 @@ static int rk3399_dmcfreq_target(struct device *dev, unsigned long *freq,
> >
> > mutex_lock(&dmcfreq->lock);
> >
> > + dram_flag = 0;
>
> Also, if dram_flag is 0, it mean that disable ODT frequency?
Yes, not a good name, maybe I should just rename it to odt_enable to
be more clear.
> If it's right, you better to define the precise variables as following
> instead of just integer(0 or 1).
> For example,
> - ROCKCHIP_SIP_DRAM_FREQ_ENABLE
> - ROCKCHIP_SIP_DRAM_FREQ_DISABLE
>
> > + if (target_rate >= dmcfreq->odt_dis_freq)
> > + dram_flag = 1;
> > +
> > + arm_smccc_smc(ROCKCHIP_SIP_DRAM_FREQ, dmcfreq->odt_pd_arg0,
> > + dmcfreq->odt_pd_arg1,
> > + ROCKCHIP_SIP_CONFIG_DRAM_SET_ODT_PD,
> > + dram_flag, 0, 0, 0, &res);
> > +
>
> This operation is special for only rk3399_dmc. It is difficult
> to understand what to do. I recommend you better to add the detailed comment
> with code.
Will do.
>
> > /*
> > * If frequency scaling from low to high, adjust voltage first.
> > * If frequency scaling from high to low, adjust frequency first.
> > @@ -294,11 +311,13 @@ static int rk3399_dmcfreq_probe(struct platform_device *pdev)
> > {
> > struct arm_smccc_res res;
> > struct device *dev = &pdev->dev;
> > - struct device_node *np = pdev->dev.of_node;
> > + struct device_node *np = pdev->dev.of_node, *node;
> > struct rk3399_dmcfreq *data;
> > int ret, index, size;
> > uint32_t *timing;
> > struct dev_pm_opp *opp;
> > + u32 ddr_type;
> > + u32 val;
> >
> > data = devm_kzalloc(dev, sizeof(struct rk3399_dmcfreq), GFP_KERNEL);
> > if (!data)
> > @@ -334,6 +353,29 @@ static int rk3399_dmcfreq_probe(struct platform_device *pdev)
> > return ret;
> > }
> >
> > + /* Try to find the optional reference to the pmu syscon */
> > + node = of_parse_phandle(np, "rockchip,pmu", 0);
> > + if (node) {
> > + data->regmap_pmu = syscon_node_to_regmap(node);
> > + if (IS_ERR(data->regmap_pmu))
> > + return PTR_ERR(data->regmap_pmu);
> > + }
> > +
> > + /* Get DDR type */
> > + regmap_read(data->regmap_pmu, RK3399_PMUGRF_OS_REG2, &val);
> > + ddr_type = (val >> RK3399_PMUGRF_DDRTYPE_SHIFT) &
> > + RK3399_PMUGRF_DDRTYPE_MASK;
> > +
> > + /* Get the odt_dis_freq parameter in function of the DDR type */
> > + if (ddr_type == RK3399_PMUGRF_DDRTYPE_DDR3)
> > + data->odt_dis_freq = data->timing.ddr3_odt_dis_freq;
> > + else if (ddr_type == RK3399_PMUGRF_DDRTYPE_LPDDR3)
> > + data->odt_dis_freq = data->timing.lpddr3_odt_dis_freq;
> > + else if (ddr_type == RK3399_PMUGRF_DDRTYPE_LPDDR4)
> > + data->odt_dis_freq = data->timing.lpddr4_odt_dis_freq;
> > + else
> > + return -EINVAL;
> > +
>
> how about using 'switch' statement?
>
Ok
> > /*
> > * Get dram timing and pass it to arm trust firmware,
> > * the dram drvier in arm trust firmware will get these
> > @@ -358,6 +400,12 @@ static int rk3399_dmcfreq_probe(struct platform_device *pdev)
> > ROCKCHIP_SIP_CONFIG_DRAM_INIT,
> > 0, 0, 0, 0, &res);
> >
> > + data->odt_pd_arg0 = (data->timing.sr_idle & 0xff) |
> > + ((data->timing.sr_mc_gate_idle & 0xff) << 8) |
> > + ((data->timing.standby_idle & 0xffff) << 16);
> > + data->odt_pd_arg1 = (data->timing.pd_idle & 0xfff) |
> > + ((data->timing.srpd_lite_idle & 0xfff) << 16);
> > +
>
> odt_pd_arg0 and odt_pd_arg1 might be used for disabling/enabling the ODT frequency.
> As I commented, it depend on only rk3399_dmc. You better to add detailed comment.
>
Ok
> And I prefer to define the XXX_SHIFT/XXX_MASK definition instead of
> using 8/16/0xff/0xffff for the readability.
>
I tried to add the XXX_SHIFT/XXX_MASK definitions and IMHO the
readability is worst if I use a maximum line length of 80 characters.
These masks are only used here, let me try to convince you by adding a
good doc in the next version and if you still prefer I add the
definition I'll do.
> > /*
> > * We add a devfreq driver to our parent since it has a device tree node
> > * with operating points.
> > diff --git a/include/soc/rockchip/rockchip_sip.h b/include/soc/rockchip/rockchip_sip.h
> > index 7e28092c4d3d..ad9482c56797 100644
> > --- a/include/soc/rockchip/rockchip_sip.h
> > +++ b/include/soc/rockchip/rockchip_sip.h
> > @@ -23,5 +23,6 @@
> > #define ROCKCHIP_SIP_CONFIG_DRAM_GET_RATE 0x05
> > #define ROCKCHIP_SIP_CONFIG_DRAM_CLR_IRQ 0x06
> > #define ROCKCHIP_SIP_CONFIG_DRAM_SET_PARAM 0x07
> > +#define ROCKCHIP_SIP_CONFIG_DRAM_SET_ODT_PD 0x08
> >
> > #endif
> >
>
>
> --
> Best Regards,
> Chanwoo Choi
> Samsung Electronics
>
> _______________________________________________
> Linux-rockchip mailing list
> Linux-rockchip at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-rockchip
Cheers,
Enric
More information about the dri-devel
mailing list