Re: [PATCH 01/11] mm/ioremap: change the return value of io[re|un]map_allowed

[Alexander Gordeev <agordeev@linux.ibm.com>]

On Sat, Aug 06, 2022 at 10:29:03AM +0800, Kefeng Wang wrote:
> 
> On 2022/8/4 23:42, Alexander Gordeev wrote:
> > On Mon, Aug 01, 2022 at 10:40:19PM +0800, Baoquan He wrote:
> > 
> > Hi Baoquan,
> > 
> > > --- a/arch/arm64/mm/ioremap.c
> > > +++ b/arch/arm64/mm/ioremap.c
> > > @@ -3,19 +3,20 @@
> > >   #include <linux/mm.h>
> > >   #include <linux/io.h>
> > > -bool ioremap_allowed(phys_addr_t phys_addr, size_t size, unsigned long prot)
> > > +void __iomem *ioremap_allowed(phys_addr_t phys_addr, size_t size, unsigned long prot)
> > >   {
> > >   	unsigned long last_addr = phys_addr + size - 1;
> > > +	int ret = -EINVAL;
> > If ret variable is really needed?
> > 
> > > diff --git a/include/asm-generic/io.h b/include/asm-generic/io.h
> > > index 72974cb81343..d72eb310fb3c 100644
> > > --- a/include/asm-generic/io.h
> > > +++ b/include/asm-generic/io.h
> > > @@ -967,26 +967,27 @@ static inline void iounmap(volatile void __iomem *addr)
> > >   /*
> > >    * Arch code can implement the following two hooks when using GENERIC_IOREMAP
> > >    * ioremap_allowed() return a bool,
> > > - *   - true means continue to remap
> > > - *   - false means skip remap and return directly
> > > + *   - IS_ERR means return an error
> > > + *   - NULL means continue to remap
> > > + *   - a non-NULL, non-IS_ERR pointer is returned directly
> > If ioremap_allowed() returns a valid pointer, then the function name
> > is not as precise anymore.
> 
> Maybe use arch_ioremap/unmap as before, or some better name.
> 
> > 
> > > @@ -28,8 +29,11 @@ void __iomem *ioremap_prot(phys_addr_t phys_addr, size_t size,
> > >   	phys_addr -= offset;
> > >   	size = PAGE_ALIGN(size + offset);
> > > -	if (!ioremap_allowed(phys_addr, size, prot))
> > > +	base = ioremap_allowed(phys_addr, size, prot);
> > > +	if (IS_ERR(base))
> > >   		return NULL;
> > > +	else if (base)
> > > +		return base;
> > It is probably just me, but the base name bit misleading here.
> We could reuse vaddr, not add new base.

vaddr name is wrong AFAICT. ioremap_allowed() returns __iomem address,
not the virtual one.

> > 
> > > @@ -50,9 +54,9 @@ EXPORT_SYMBOL(ioremap_prot);
> > >   void iounmap(volatile void __iomem *addr)
> > >   {
> > > -	void *vaddr = (void *)((unsigned long)addr & PAGE_MASK);
> > > +	void __iomem *vaddr = (void __iomem *)((unsigned long)addr & PAGE_MASK);

Same here.

> > > -	if (!iounmap_allowed(vaddr))
> > > +	if (iounmap_allowed(vaddr))
> > I guess, iounmap_allowed() should accept void __iomem *, not void *.
> > Then addr needs to be passed to iounmap_allowed() not vaddr.
> 
> The following is_vmalloc_addr()  and vunmap() in iounmap() use void *,
> 
> so we could simply use void* for iounmap_allowed().

iounmap_allowed() accepts void __iomem * and I that looks correct to me.

Passing void * on the other hand means you pass a pointer that
in theory differs from what architecture previously returned
with ioremap_allowed() and "knows" nothing about.

I think you need to pass addr to iounmap_allowed() as is.