Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Subject: Re: [RFC PATCH 1/2] soc: rockchip: power-domain: Manage resource
 conflicts with firmware
To:     Brian Norris <briannorris@chromium.org>,
        MyungJoo Ham <myungjoo.ham@samsung.com>,
        Kyungmin Park <kyungmin.park@samsung.com>,
        Heiko Stuebner <heiko@sntech.de>
Cc:     linux-kernel@vger.kernel.org,
        Elaine Zhang <zhangqing@rock-chips.com>,
        linux-pm@vger.kernel.org, Doug Anderson <dianders@chromium.org>,
        linux-arm-kernel@lists.infradead.org,
        linux-rockchip@lists.infradead.org
From:   Chanwoo Choi <cw00.choi@samsung.com>
Organization: Samsung Electronics
Message-ID: <c8664eae-4a10-bd1a-8898-01b96c05331e@samsung.com>
Date:   Thu, 7 Apr 2022 14:04:09 +0900
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:59.0) Gecko/20100101
        Thunderbird/59.0
MIME-Version: 1.0
In-Reply-To: <20220405184816.RFC.1.Ib865f199d15221eab4ff77f70bd7e9e2eb04d32f@changeid>
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset="utf-8"
CMS-TYPE: 101P
DLP-Filter: Pass
References: <20220406014842.2771799-1-briannorris@chromium.org>
        <CGME20220406094558epcas1p4fa0c77a5acd6b73c192f6b19136cd5f9@epcas1p4.samsung.com>
        <20220405184816.RFC.1.Ib865f199d15221eab4ff77f70bd7e9e2eb04d32f@changeid>
Precedence: bulk

Hi Brian,

Thanks for very detailed description.
I understand this issue.

Instead of adding the specific function for only rockchip,
how about adding new function pointer (like block/unblock or start/stop and others)
into 'struct generic_pm_domain'? And add new pm_genpd_* function
to control the power domain. Because it is better to use subsystem interface.

Best Regards,
Chanwoo Choi


On 4/6/22 10:48 AM, Brian Norris wrote:
> On RK3399 platforms, power domains are managed mostly by the kernel
> (drivers/soc/rockchip/pm_domains.c), but there are a few exceptions
> where ARM Trusted Firmware has to be involved:
> 
> (1) system suspend/resume
> (2) DRAM DVFS (a.k.a., "ddrfreq")
> 
> Exception (1) does not cause much conflict, since the kernel has
> quiesced itself by the time we make the relevant PSCI call.
> 
> Exception (2) can cause conflict, because of two actions:
> 
> (a) ARM Trusted Firmware needs to read/modify/write the PMU_BUS_IDLE_REQ
>     register to idle the memory controller domain; the kernel driver
>     also has to touch this register for other domains.
> (b) ARM Trusted Firmware needs to manage the clocks associated with
>     these domains.
> 
> To elaborate on (b): idling a power domain has always required ungating
> an array of clocks; see this old explanation from Rockchip:
> https://lore.kernel.org/linux-arm-kernel/54503C19.9060607@rock-chips.com/
> 
> Historically, ARM Trusted Firmware has avoided this issue by using a
> special PMU_CRU_GATEDIS_CON0 register -- this register ungates all the
> necessary clocks -- when idling the memory controller. Unfortunately,
> we've found that this register is not 100% sufficient; it does not turn
> the relevant PLLs on [0].
> 
> So it's possible to trigger issues with something like the following:
> 
> 1. enable a power domain (e.g., RK3399_PD_VDU) -- kernel will
>    temporarily enable relevant clocks/PLLs, then turn them back off
>    2. a PLL (e.g., PLL_NPLL) is part of the clock tree for
>       RK3399_PD_VDU's clocks but otherwise unused; NPLL is disabled
> 3. perform a ddrfreq transition (rk3399_dmcfreq_target() -> ...
>    drivers/clk/rockchip/clk-ddr.c / ROCKCHIP_SIP_DRAM_FREQ)
>    4. ARM Trusted Firmware unagates VDU clocks (via PMU_CRU_GATEDIS_CON0)
>    5. ARM Trusted firmware idles the memory controller domain
>    6. Step 5 waits on the VDU domain/clocks, but NPLL is still off
> 
> i.e., we hang the system.
> 
> So for (b), we need to at a minimum manage the relevant PLLs on behalf
> of firmware. It's easier to simply manage the whole clock tree, in a
> similar way we do in rockchip_pd_power().
> 
> For (a), we need to provide mutual exclusion betwen rockchip_pd_power()
> and firmware. To resolve that, we simply grab the PMU mutex and release
> it when ddrfreq is done.
> 
> The Chromium OS kernel has been carrying versions of part of this hack
> for a while, based on some new custom notifiers [1]. I've rewritten as a
> simple function call between the drivers, which is OK because:
> 
>  * the PMU driver isn't enabled, and we don't have this problem at all
>    (the firmware should have left us in an OK state, and there are no
>    runtime conflicts); or
>  * the PMU driver is present, and is a single instance.
> 
> And the power-domain driver cannot be removed, so there's no lifetime
> management to worry about.
> 
> For completeness, there's a 'dmc_pmu_mutex' to guard (likely
> theoretical?) probe()-time races. It's OK for the memory controller
> driver to start running before the PMU, because the PMU will avoid any
> critical actions during the block() sequence.
> 
> [0] The RK3399 TRM for PMU_CRU_GATEDIS_CON0 only talks about ungating
>     clocks. Based on experimentation, we've found that it does not power
>     up the necessary PLLs.
> 
> [1] CHROMIUM: soc: rockchip: power-domain: Add notifier to dmc driver
>     https://chromium-review.googlesource.com/q/I242dbd706d352f74ff706f5cbf42ebb92f9bcc60
>     Notably, the Chromium solution only handled conflict (a), not (b).
>     In practice, item (b) wasn't a problem in many cases because we
>     never managed to fully power off PLLs. Now that the (upstream) video
>     decoder driver performs runtime clock management, we often power off
>     NPLL.
> 
> Signed-off-by: Brian Norris <briannorris@chromium.org>
> ---
> 
>  drivers/soc/rockchip/pm_domains.c | 118 ++++++++++++++++++++++++++++++
>  include/soc/rockchip/pm_domains.h |  25 +++++++
>  2 files changed, 143 insertions(+)
>  create mode 100644 include/soc/rockchip/pm_domains.h
> 
> diff --git a/drivers/soc/rockchip/pm_domains.c b/drivers/soc/rockchip/pm_domains.c
> index 1b029e494274..bc0afc52299b 100644
> --- a/drivers/soc/rockchip/pm_domains.c
> +++ b/drivers/soc/rockchip/pm_domains.c
> @@ -8,6 +8,7 @@
>  #include <linux/io.h>
>  #include <linux/iopoll.h>
>  #include <linux/err.h>
> +#include <linux/mutex.h>
>  #include <linux/pm_clock.h>
>  #include <linux/pm_domain.h>
>  #include <linux/of_address.h>
> @@ -16,6 +17,7 @@
>  #include <linux/clk.h>
>  #include <linux/regmap.h>
>  #include <linux/mfd/syscon.h>
> +#include <soc/rockchip/pm_domains.h>
>  #include <dt-bindings/power/px30-power.h>
>  #include <dt-bindings/power/rk3036-power.h>
>  #include <dt-bindings/power/rk3066-power.h>
> @@ -139,6 +141,109 @@ struct rockchip_pmu {
>  #define DOMAIN_RK3568(name, pwr, req, wakeup)		\
>  	DOMAIN_M(name, pwr, pwr, req, req, req, wakeup)
>  
> +/*
> + * Dynamic Memory Controller may need to coordinate with us -- see
> + * rockchip_pmu_block().
> + *
> + * dmc_pmu_mutex protects registration-time races, so DMC driver doesn't try to
> + * block() while we're initializing the PMU.
> + */
> +static DEFINE_MUTEX(dmc_pmu_mutex);
> +static struct rockchip_pmu *dmc_pmu;
> +
> +/*
> + * Block PMU transitions and make sure they don't interfere with ARM Trusted
> + * Firmware operations. There are two conflicts, noted in the comments below.
> + *
> + * Caller must unblock PMU transitions via rockchip_pmu_unblock().
> + */
> +int rockchip_pmu_block(void)
> +{
> +	struct rockchip_pmu *pmu;
> +	struct generic_pm_domain *genpd;
> +	struct rockchip_pm_domain *pd;
> +	int i, ret;
> +
> +	mutex_lock(&dmc_pmu_mutex);
> +
> +	/* No PMU (yet)? Then we just block rockchip_pmu_probe(). */
> +	if (!dmc_pmu)
> +		return 0;
> +	pmu = dmc_pmu;
> +
> +	/*
> +	 * mutex blocks all idle transitions: we can't touch the
> +	 * PMU_BUS_IDLE_REQ (our ".idle_offset") register while ARM Trusted
> +	 * Firmware might be using it.
> +	 */
> +	mutex_lock(&pmu->mutex);
> +
> +	/*
> +	 * Power domain clocks: Per Rockchip, we *must* keep certain clocks
> +	 * enabled for the duration of power-domain transitions. Most
> +	 * transitions are handled by this driver, but some cases (in
> +	 * particular, DRAM DVFS / memory-controller idle) must be handled by
> +	 * firmware. Firmware can handle most clock management via a special
> +	 * "ungate" register (PMU_CRU_GATEDIS_CON0), but unfortunately, this
> +	 * doesn't handle PLLs. We can assist this transition by doing the
> +	 * clock management on behalf of firmware.
> +	 */
> +	for (i = 0; i < pmu->genpd_data.num_domains; i++) {
> +		genpd = pmu->genpd_data.domains[i];
> +		if (genpd) {
> +			pd = to_rockchip_pd(genpd);
> +			ret = clk_bulk_enable(pd->num_clks, pd->clks);
> +			if (ret < 0) {
> +				dev_err(pmu->dev,
> +					"failed to enable clks for domain '%s': %d\n",
> +					genpd->name, ret);
> +				goto err;
> +			}
> +		}
> +	}
> +
> +	return 0;
> +
> +err:
> +	for (i = i - 1; i >= 0; i--) {
> +		genpd = pmu->genpd_data.domains[i];
> +		if (genpd) {
> +			pd = to_rockchip_pd(genpd);
> +			clk_bulk_disable(pd->num_clks, pd->clks);
> +		}
> +	}
> +	mutex_unlock(&pmu->mutex);
> +	mutex_unlock(&dmc_pmu_mutex);
> +
> +	return ret;
> +}
> +EXPORT_SYMBOL_GPL(rockchip_pmu_block);
> +
> +/* Unblock PMU transitions. */
> +void rockchip_pmu_unblock(void)
> +{
> +	struct rockchip_pmu *pmu;
> +	struct generic_pm_domain *genpd;
> +	struct rockchip_pm_domain *pd;
> +	int i;
> +
> +	if (dmc_pmu) {
> +		pmu = dmc_pmu;
> +		for (i = 0; i < pmu->genpd_data.num_domains; i++) {
> +			genpd = pmu->genpd_data.domains[i];
> +			if (genpd) {
> +				pd = to_rockchip_pd(genpd);
> +				clk_bulk_disable(pd->num_clks, pd->clks);
> +			}
> +		}
> +
> +		mutex_unlock(&pmu->mutex);
> +	}
> +
> +	mutex_unlock(&dmc_pmu_mutex);
> +}
> +EXPORT_SYMBOL_GPL(rockchip_pmu_unblock);
> +
>  static bool rockchip_pmu_domain_is_idle(struct rockchip_pm_domain *pd)
>  {
>  	struct rockchip_pmu *pmu = pd->pmu;
> @@ -690,6 +795,12 @@ static int rockchip_pm_domain_probe(struct platform_device *pdev)
>  
>  	error = -ENODEV;
>  
> +	/*
> +	 * Prevent any rockchip_pmu_block() from racing with the remainder of
> +	 * setup (clocks, register initialization).
> +	 */
> +	mutex_lock(&dmc_pmu_mutex);
> +
>  	for_each_available_child_of_node(np, node) {
>  		error = rockchip_pm_add_one_domain(pmu, node);
>  		if (error) {
> @@ -719,10 +830,17 @@ static int rockchip_pm_domain_probe(struct platform_device *pdev)
>  		goto err_out;
>  	}
>  
> +	/* We only expect one PMU. */
> +	if (!WARN_ON_ONCE(dmc_pmu))
> +		dmc_pmu = pmu;
> +
> +	mutex_unlock(&dmc_pmu_mutex);
> +
>  	return 0;
>  
>  err_out:
>  	rockchip_pm_domain_cleanup(pmu);
> +	mutex_unlock(&dmc_pmu_mutex);
>  	return error;
>  }
>  
> diff --git a/include/soc/rockchip/pm_domains.h b/include/soc/rockchip/pm_domains.h
> new file mode 100644
> index 000000000000..7dbd941fc937
> --- /dev/null
> +++ b/include/soc/rockchip/pm_domains.h
> @@ -0,0 +1,25 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Copyright 2022, The Chromium OS Authors. All rights reserved.
> + */
> +
> +#ifndef __SOC_ROCKCHIP_PM_DOMAINS_H__
> +#define __SOC_ROCKCHIP_PM_DOMAINS_H__
> +
> +#ifdef CONFIG_ROCKCHIP_PM_DOMAINS
> +
> +int rockchip_pmu_block(void);
> +void rockchip_pmu_unblock(void);
> +
> +#else /* CONFIG_ROCKCHIP_PM_DOMAINS */
> +
> +static inline int rockchip_pmu_block(void)
> +{
> +	return 0;
> +}
> +
> +static inline void rockchip_pmu_unblock(void) { }
> +
> +#endif /* CONFIG_ROCKCHIP_PM_DOMAINS */
> +
> +#endif /* __SOC_ROCKCHIP_PM_DOMAINS_H__ */
> 


-- 
Best Regards,
Chanwoo Choi
Samsung Electronics