[PATCH 1/2] pmdomain/rockchip: skip QoS operations for idle-only domains

[Shawn Lin]

在 2026/04/01 星期三 10:34, Daniel Bozeman 写道:
> The NanoPi Zero2 (RK3528) kernel panics during boot when a
> GPIO-controlled USB VBUS regulator is defined on GPIO4 (which
> is in PD_RKVENC). The goal of this series is to make USB host
> power work on boards that use GPIO4 for regulator control.
> 
> The root cause is a probe ordering issue. On RK3528, the power
> domain controller's first probe attempt fails because PD_GPU's
> clock lookup returns -EPROBE_DEFER (CRU hasn't probed yet).
> The driver then tears down all domains, including PD_RKVENC
> which would have registered successfully (it has no clock
> requirements). During this window, the USB regulator driver
> probes and requests GPIO4, which is in the now-unregistered
> PD_RKVENC. This triggers a synchronous external abort.
> 
> With patch 2 alone (skipping deferred domains), the idle-only
> domains register successfully. But the genpd framework then
> attempts to power them off via genpd_power_off_work_fn. This
> calls rockchip_pd_power(), which does QoS save and idle
> requests on domains with pwr_mask == 0 that cannot actually
> be powered off.
> 
> To your question about why QoS registers become inaccessible
> on idle-only domains: I have not root-caused that specifically.
> What I can confirm is the crash trace below, which occurs when
> patch 2 is applied without patch 1. The abort happens during


This sounds like a parent-child dependency which hasn't been sorted
out. My another question will be: with patch 1 applied, how to save-
restore qos registers during normal S2R usage?

> rockchip_pmu_set_idle_request on an idle-only domain:
> 
>    Internal error: synchronous external abort: 0000000096000010
>    CPU: 2 PID: 60 Comm: kworker/2:3
>    Workqueue: pm genpd_power_off_work_fn
>    pc : regmap_mmio_read32le+0x8/0x20
>    lr : regmap_mmio_read+0x44/0x70
>    Call trace:
>     regmap_mmio_read32le+0x8/0x20
>     _regmap_bus_reg_read+0x6c/0xac
>     _regmap_read+0x60/0xd8
>     regmap_read+0x4c/0x7c
>     rockchip_pmu_set_idle_request.isra.0+0x94/0x1b4
>     rockchip_pd_power+0x37c/0x608
>     rockchip_pd_power_off+0x14/0x38
>     genpd_power_off.isra.0+0x1f0/0x2f0
>     genpd_power_off_work_fn+0x34/0x54
> 
> The two patches work together: patch 1 prevents QoS access
> on idle-only domains, and patch 2 prevents the full probe
> teardown when a single domain defers.
> 
> Tested on NanoPi Zero2 (fixes panic) and Radxa E20C (no
> regression).
> 
> On Tue, Mar 31, 2026 at 6:17 PM Shawn Lin <shawn.lin at rock-chips.com 
> <mailto:shawn.lin at rock-chips.com>> wrote:
> 
>     Hi Daniel,
> 
>     在 2026/04/01 星期三 2:02, Daniel Bozeman 写道:
>      > Idle-only power domains (pwr_mask == 0) cannot actually be powered
>      > on or off. rockchip_do_pmu_set_power_domain() already returns early
>      > for these domains, but rockchip_pd_power() still attempts QoS save
>      > and idle requests before reaching that check.
>      >
>      > On RK3528, the idle-only domains (PD_RKVENC, PD_VO, PD_VPU) have
>      > QoS registers that may be inaccessible when the generic power domain
>      > framework attempts to power them off, leading to synchronous external
>      > aborts.
>      >
> 
>     Is it the real abort happened on your RK3528 board? I am trying to
>     understand the problem first. Even with idle-only powerdomain, the code
>     also save the QoS registers before set idle to the powerdomain, so
>     how the QoS registers become inaccessible?
> 
>      > Return early from rockchip_pd_power() when pwr_mask is zero, matching
>      > the existing guard in rockchip_do_pmu_set_power_domain().
>      >
>      > Fixes: 1fe767a56c32 ("soc: rockchip: power-domain: allow domains
>     only handling idle requests")
>      > Signed-off-by: Daniel Bozeman <daniel at orb.net
>     <mailto:daniel at orb.net>>
>      > ---
>      >   drivers/pmdomain/rockchip/pm-domains.c | 3 +++
>      >   1 file changed, 3 insertions(+)
>      >
>      > diff --git a/drivers/pmdomain/rockchip/pm-domains.c
>     b/drivers/pmdomain/rockchip/pm-domains.c
>      > index 490bbb1d1d..2eecae092a 100644
>      > --- a/drivers/pmdomain/rockchip/pm-domains.c
>      > +++ b/drivers/pmdomain/rockchip/pm-domains.c
>      > @@ -640,6 +640,9 @@ static int rockchip_pd_power(struct
>     rockchip_pm_domain *pd, bool power_on)
>      >       if (rockchip_pmu_domain_is_on(pd) == power_on)
>      >               return 0;
>      >
>      > +     if (pd->info->pwr_mask == 0)
>      > +             return 0;
>      > +
>      >       ret = clk_bulk_enable(pd->num_clks, pd->clks);
>      >       if (ret < 0) {
>      >               dev_err(pmu->dev, "failed to enable clocks\n");
>      >
>      > base-commit: bc330699801d3b4f99110365512caed5adcfaca3
>