* Stephen Warren wrote:
> Thierry Reding wrote at Tuesday, November 22, 2011 11:58 PM:
> > * Stephen Warren wrote:
> > > module_platform_drive saves some boiler-plate code.
> > >
> > > The devm_ APIs remove the need to manually clean up allocations,
> > > thus removing some code.
> ...
> > > static int __devexit tegra_das_remove(struct platform_device *pdev)
> > > {
> > > - struct resource *res;
> > > -
> > > if (!das)
> > > return -ENODEV;
> > >
> > > - platform_set_drvdata(pdev, NULL);
> > > -
> > [...]
> >
> > Setting the driver data to NULL may still be a good idea.
>
> When Mark Brown reviewed the TrimSlice machine driver, he mentioned that
> clearing the drvdata was pointless; nothing should be using it when the
> device is not created.
>
> As background, soon after that, I modified the tegra_wm8903.c machine
> driver along the same lines, but evidently didn't update tegra_das.c
> in a similar way, but would have if I'd been paying attention...
Thinking about this some more I have to agree. The only one using the driver
data would be the driver, which in turn shouldn't be doing anything with it
after remove is called.
I was just commenting on it because I've seen it done in a lot of drivers and
thought it was common (best) practice.
Thierry
Attachment:
pgpVvJimPO59S.pgp
Description: PGP signature
