[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAL_JsqJmcTPqopCcETeFUsgG=nDSUMVvVBAMMuCF_j-SzfoAEw@mail.gmail.com>
Date: Wed, 29 Sep 2021 09:58:11 -0500
From: Rob Herring <robh+dt@...nel.org>
To: trix@...hat.com
Cc: Frank Rowand <frowand.list@...il.com>, devicetree@...r.kernel.org,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] of: remove duplicate declaration of of_iomap()
On Wed, Sep 29, 2021 at 9:26 AM Rob Herring <robh+dt@...nel.org> wrote:
>
> On Tue, Sep 28, 2021 at 3:12 PM <trix@...hat.com> wrote:
> >
> > From: Tom Rix <trix@...hat.com>
> >
> > A ranconfig produces this linker error
> > irq-al-fic.c:252: undefined reference to `of_iomap'
> >
> > The declaration of of_iomap() is dependent on OF
> > The definition of of_iomap() is dependent on OF_ADDRESS
> > These should match. There are duplicate declarations
> > of of_iomap(), remove of_iomap() and the
> > of_address_to_resource() duplicate.
> >
> > Signed-off-by: Tom Rix <trix@...hat.com>
> > ---
> > include/linux/of_address.h | 8 +-------
> > 1 file changed, 1 insertion(+), 7 deletions(-)
> >
> > diff --git a/include/linux/of_address.h b/include/linux/of_address.h
> > index 45598dbec269..a190996b4b0b 100644
> > --- a/include/linux/of_address.h
> > +++ b/include/linux/of_address.h
> > @@ -122,13 +122,7 @@ static inline bool of_dma_is_coherent(struct device_node *np)
> > {
> > return false;
> > }
> > -#endif /* CONFIG_OF_ADDRESS */
> >
> > -#ifdef CONFIG_OF
> > -extern int of_address_to_resource(struct device_node *dev, int index,
> > - struct resource *r);
> > -void __iomem *of_iomap(struct device_node *node, int index);
>
> This is going to break sparc which has !OF_ADDRESS and its own
> of_iomap and of_address_to_resource implementations. I don't want to
> add CONFIG_SPARC in here, so I think we should solve this in kconfig.
> OF and !OF_ADDRESS is supposed to mean the arch provides these
> functions.
>
> I'd really like to do away with HAS_IOMEM. It doesn't serve much
> purpose other than disabling a bunch of drivers.
Following UML (the only other arch with configurable HAS_IOMEM), I
think dummy ioremap() implementations is the way to go here. UML added
its own dummy functions, but it would be better if we can just do
this:
diff --git a/include/asm-generic/io.h b/include/asm-generic/io.h
index e93375c710b9..a9d02403b2fe 100644
--- a/include/asm-generic/io.h
+++ b/include/asm-generic/io.h
@@ -946,7 +946,7 @@ static inline void *phys_to_virt(unsigned long address)
* can provide stricter non-posted write semantics if the architecture
* implements them.
*/
-#ifndef CONFIG_MMU
+#if !defined(CONFIG_GENERIC_IOREMAP)
#ifndef ioremap
#define ioremap ioremap
static inline void __iomem *ioremap(phys_addr_t offset, size_t size)
@@ -961,7 +961,7 @@ static inline void iounmap(void __iomem *addr)
{
}
#endif
-#elif defined(CONFIG_GENERIC_IOREMAP)
+#else /* CONFIG_GENERIC_IOREMAP */
#include <linux/pgtable.h>
void __iomem *ioremap_prot(phys_addr_t addr, size_t size, unsigned long prot);
@@ -972,7 +972,7 @@ static inline void __iomem *ioremap(phys_addr_t
addr, size_t size)
/* _PAGE_IOREMAP needs to be supplied by the architecture */
return ioremap_prot(addr, size, _PAGE_IOREMAP);
}
-#endif /* !CONFIG_MMU || CONFIG_GENERIC_IOREMAP */
+#endif /* CONFIG_GENERIC_IOREMAP */
#ifndef ioremap_wc
#define ioremap_wc ioremap
Then we can apply the following patch:
diff --git a/drivers/of/Kconfig b/drivers/of/Kconfig
index 3dfeae8912df..4a2453e4d4f3 100644
--- a/drivers/of/Kconfig
+++ b/drivers/of/Kconfig
@@ -64,7 +64,7 @@ config OF_DYNAMIC
config OF_ADDRESS
def_bool y
- depends on !SPARC && (HAS_IOMEM || UML)
+ depends on !SPARC
config OF_IRQ
def_bool y
Powered by blists - more mailing lists