Re: [PATCH v16 3/7] soc: mediatek: SVS: introduce MTK SVS engine

[Guenter Roeck]

On 5/13/21 8:10 PM, Roger Lu wrote:
> Hi Guenter,
>
> Sorry for the late reply and thanks for the notice.
>
> On Wed, 2021-05-05 at 21:51 -0700, Guenter Roeck wrote:
> > On Wed, Apr 28, 2021 at 02:54:36PM +0800, Roger Lu wrote:
> > > The Smart Voltage Scaling(SVS) engine is a piece of hardware
> > > which calculates suitable SVS bank voltages to OPP voltage table.
> > > Then, DVFS driver could apply those SVS bank voltages to PMIC/Buck
> > > when receiving OPP_EVENT_ADJUST_VOLTAGE.
> > >
> > > Signed-off-by: Roger Lu <roger.lu@xxxxxxxxxxxx>
> > > ---
> > >   drivers/soc/mediatek/Kconfig   |   10 +
> > >   drivers/soc/mediatek/Makefile  |    1 +
> > >   drivers/soc/mediatek/mtk-svs.c | 1723
> > > ++++++++++++++++++++++++++++++++
> > >   3 files changed, 1734 insertions(+)
> > >   create mode 100644 drivers/soc/mediatek/mtk-svs.c
> >
> > [ ... ]
> >
> > > +
> > > +	svsp_irq = irq_of_parse_and_map(svsp->dev->of_node, 0);
> > > +	ret = devm_request_threaded_irq(svsp->dev, svsp_irq, NULL,
> > > svs_isr,
> > > +					svsp->irqflags, svsp->name,
> > > svsp);
> >
> > 0-day reports:
> >
> > drivers/soc/mediatek/mtk-svs.c:1663:7-32: ERROR:
> > 	Threaded IRQ with no primary handler requested without
> > IRQF_ONESHOT
> >
> > I would be a bit concerned about this. There is no primary (hard)
> > interrupt handler, meaning the hard interrupt may be re-enabled after
> > the default hard interrupt handler runs. This might result in endless
> > interrupts.
>
> Oh, we add IRQF_ONESHOT in "svs_get_svs_mt8183_platform_data()" for
> threaded irq. So, please kindly let us know if we need to set more
> flags or any other potential risks we should be aware. Thanks in
> advance.

After reviewing the code, I think this was actually a false alarm,
at least if svsp->irqflags always includes IRQF_ONESHOT.
The code is kind of unusual, though. Unless I am missing something,
svsp->irqflags is only set in one place and it is always set
to IRQF_TRIGGER_LOW | IRQF_ONESHOT. If there is a remote chance
that the flag is ever different, it would have been better (and less
confusing) to specify IRQF_ONESHOT directly when requesting the
interrupt (because it is always needed, no matter which SOC).
If the flags are always the same, there is no reason for having
the svsp->irqflags variable in the first place.

Thanks,
Guenter