Re: [PATCH v3 4/6] mm: ioremap: Add arch_ioremap/iounmap()

[Arnd Bergmann <arnd@arndb.de>]

On Tue, May 24, 2022 at 4:32 PM Kefeng Wang <wangkefeng.wang@huawei.com> wrote:
> On 2022/5/24 20:37, Arnd Bergmann wrote:
> > On Thu, May 19, 2022 at 10:25 AM Kefeng Wang <wangkefeng.wang@huawei.com> wrote:
> >> Add special hook for architecture to verify or setup addr, size
> >> or prot when ioremap() or iounmap(), which will make the generic
> >> ioremap more useful.
> >>
> >>    arch_ioremap() return a pointer,
> >>      - IS_ERR means return an error
> >>      - NULL means continue to remap
> >>      - a non-NULL, non-IS_ERR pointer is directly returned
> >>    arch_iounmap() return a int value,
> >>      - 0 means continue to vunmap
> >>      - error code means skip vunmap and return directly
> >>
> >> Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
> > I don't really like interfaces that mix error pointers and NULL pointer
> > returns.
> >
> > Would it be possible to have a special error code other than NULL
> > for the fallback case?
>
> I don't find a good error code, maybe  ENOTSUPP, any better suggestion?

I had another look at the resulting arm64 function, and it appears that
you never actually return a non-error pointer here. If I didn't miss anything,
I think the best way would be to change the return type to just indicate
success or an error code, and drop the case of returning the actual result,
and changing the function name accordingly.

Would that work, or do you actually require returning an __iomem
token from somewhere else?

       Arnd