Re: [PATCH 3.4 088/125] ser_gigaset: fix deallocation of platform device structure

From: Paul Bolle
Date: Thu Oct 13 2016 - 04:14:41 EST


On Thu, 2016-10-13 at 10:52 +0800, Zefan Li wrote:
> On 2016/10/12 20:52, Paul Bolle wrote:
> > There has been a follow up for this fix. I'll have to dive into my
> > archive to see why that was needed.
> >
> > It was complicated, because there has been a short period in which this
> > fix was correct. Something like that, I'm speaking from memory.
> > (Perhaps Tilman's memory is less imperfect.)
> >
> > I'll try get back to this shortly (in a day or so).
> >
>
> Thanks for looking into this.

So what I think you also need _on top of_ this patch:
- commit 8aeb3c3d655e ("ser_gigaset: remove unnecessary kfree() calls
from release method"), for context changes; and
- commit 8d2c3ab44456 ("ser_gigaset: use container_of() instead of
detour"), the proper fix.

I could not get v3.4 to build _at all_ on my current Fedora 24 machine.
(v3.4 was probably released when Fedora 16 was still shiny and new.)
Lack of coffee? So I've only visually inspected these three commits on
top of v3.4.112. Is that acceptable to you?


Paul Bolle