Re: [PATCH v2] p54: Fix an error handling path in p54spi_probe()

[Christian Lamparter]

On 13/06/2022 22:57, Christophe JAILLET wrote:
> Le 13/06/2022 à 22:02, Christian Lamparter a écrit :
> > On Sun, Jun 12, 2022 at 11:12 PM Christophe JAILLET
> > <christophe.jaillet@xxxxxxxxxx> wrote:
> > > If an error occurs after a successful call to p54spi_request_firmware(), it
> > > must be undone by a corresponding release_firmware() as already done in
> > > the error handling path of p54spi_request_firmware() and in the .remove()
> > > function.
> > >
> > > Add the missing call in the error handling path and remove it from
> > > p54spi_request_firmware() now that it is the responsibility of the caller
> > > to release the firmawre
> >
> > that last word hast a typo:  firmware. (maybe Kalle can fix this in post).
>
> More or less the same typo twice in a row... _Embarrassed_
>
> > > Fixes: cd8d3d321285 ("p54spi: p54spi driver")
> > > Signed-off-by: Christophe JAILLET <christophe.jaillet@xxxxxxxxxx>
> > Acked-by: Christian Lamparter <chunkeey@xxxxxxxxx>
> > (Though, v1 was fine too.)
> > > ---
> > > v2: reduce diffstat and take advantage on the fact that release_firmware()
> > > checks for NULL
> >
> > Heh, ok ;) . Now that I see it,  the "ret = p54_parse_firmware(...); ... "
> > could have been replaced with "return p54_parse_firmware(dev, priv->firmware);"
> > so the p54spi.c could shrink another 5-6 lines.
> >
> > I think leaving p54spi_request_firmware() callee to deal with
> > releasing the firmware
> > in the error case as well is nicer because it gets rid of a "but in
> > this case" complexity.
>
>
> Take the one you consider being the best one.

well said!

> If it deserves a v3 to axe some lines of code, 
> I can do it but, as said previously,
> v1 is for me the cleaner and more future proof.

Gee, that last sentence about "future proof" is daring.
I don't know what's up on the horizon. For my part, I've been devresing
parts of carl9170 and now thinking about it. Because the various
request_firmware*() functions could be a target for devres too.
A driver usually loads the firmware in .probe(). It stays around because
of .suspend()+.resume() and gets freed by .release().
With devresing up request_firmware(), that release_firmware() would be
rendered obsolete in all of p54* cases.

There must be something that I have missed? right?

It's because there's already an extensive list of managed interfaces:
<https://www.kernel.org/doc/html/latest/driver-api/driver-model/devres.html>
But the firmware_class is not on it. Does somebody know the presumably
"very good" reason why not? I can't believe that this hasn't been done yet.

Regards,
Christian