On Tue, Mar 22, 2022 at 07:32:16PM +0530, Srinivasa Rao Mandadapu wrote:
>
> On 3/22/2022 7:24 PM, Dan Carpenter wrote:
> Thanks for Your Time Dan!!!
> > On Tue, Mar 22, 2022 at 07:03:23PM +0530, Srinivasa Rao Mandadapu wrote:
> > > > 9e3d83c52844f95 Srinivasa Rao Mandadapu 2022-02-26 14 struct lpass_macro *lpass_macro_pds_init(struct device *dev)
> > > > 9e3d83c52844f95 Srinivasa Rao Mandadapu 2022-02-26 15 {
> > > > 9e3d83c52844f95 Srinivasa Rao Mandadapu 2022-02-26 16 struct lpass_macro *l_pds;
> > > > 9e3d83c52844f95 Srinivasa Rao Mandadapu 2022-02-26 17 int ret;
> > > > 9e3d83c52844f95 Srinivasa Rao Mandadapu 2022-02-26 18
> > > > 9e3d83c52844f95 Srinivasa Rao Mandadapu 2022-02-26 19 if (!of_find_property(dev->of_node, "power-domains", NULL))
> > > > 9e3d83c52844f95 Srinivasa Rao Mandadapu 2022-02-26 20 return NULL;
> > > >
> > > > Returning NULL here will lead to a crash in tx_macro_runtime_resume()
> > > >
> > > > When a function returns a mix of NULL and error pointers, then NULL
> > > > means the feature is deliberately disabled. It's not an error, it's a
> > > > deliberate choice by the distro or sys admin. The caller has to
> > > > be written to allow the feature to be disabled.
> > > >
> > > > An example of this might be LEDs. Maybe people don't want LEDs so code
> > > > has to asume that the led->ops pointer might be NULL and check for that
> > > > before dereferencing it.
> > > Actually, it's optional here. For some targets, with lpass ADSP enabled,
> > > power domains are not required.
> > >
> > > So is the reason, returning NULL Here.
> > >
> > Unfortunately, the caller is not written to handle NULLs so it will
> > crash.
>
> Yes agree. will add error check with in lpass_macro_pds_exit().
>
> Will that make sense?
Yep. Thanks!
regards,
dan carpenter
