On Tue, Mar 28, 2023 at 06:48:18PM +0300, Sakari Ailus wrote:
> On Tue, Mar 28, 2023 at 06:46:23PM +0300, Andy Shevchenko wrote:
> > On Tue, Mar 28, 2023 at 06:27:09PM +0300, Sakari Ailus wrote:
> > > On Tue, Mar 28, 2023 at 06:12:09PM +0300, Andy Shevchenko wrote:
> > > > On Tue, Mar 28, 2023 at 01:12:56PM +0300, Sakari Ailus wrote:
...
> > > > > + struct scan_check_crs_csi2_context inst_context = {
> > > > > + .handle = handle,
> > > > > + .res_head = LIST_HEAD_INIT(inst_context.res_head),
> > > > > + };
> > > > > + struct crs_csi2 *csi2;
> > > > > +
> > > > > + acpi_walk_resources(handle, METHOD_NAME__CRS,
> > > > > + scan_check_crs_csi2_instance, &inst_context);
> > > > > +
> > > > > + if (list_empty(&inst_context.res_head))
> > > > > + return;
> > > > > +
> > > > > + /*
> > > > > + * Found entry, so allocate memory for it, fill it and add it to the
> > > > > + * list.
> > > > > + */
> > > > > + csi2 = kmalloc(sizeof(*csi2), GFP_KERNEL);
> > > > > + if (!csi2)
> > > >
> > > > Who is going to release resources allocated in the callback above?
> > >
> > > This is done by crs_csi2_release(), called from acpi_bus_scan_crs_csi2().
> >
> > Isn't it a bit counter intuitive?
>
> Are you suggesting changing something?
>
> Basically we're gathering the information during an ACPI namespace walk.
> Once the information has been processed and used, the memory can be
> released.
Release allocated memory in the function which allocates it. Since it's called
via callback, the closest one is the one that calls acpi_walk_resources().
Can't we free them here?
--
With Best Regards,
Andy Shevchenko
