Re: [PATCH v6 03/16] dma-debug: refactor to use physical addresses for page mapping

[Leon Romanovsky <leon@kernel.org>]

On Tue, Sep 09, 2025 at 10:37:48PM +0300, Leon Romanovsky wrote:
> On Tue, Sep 09, 2025 at 04:27:31PM +0300, Leon Romanovsky wrote:
> > From: Leon Romanovsky <leonro@nvidia.com>
> 
> <...>
> 
> >  include/linux/page-flags.h         |  1 +
> 
> <...>
> 
> > --- a/include/linux/page-flags.h
> > +++ b/include/linux/page-flags.h
> > @@ -614,6 +614,7 @@ FOLIO_FLAG(dropbehind, FOLIO_HEAD_PAGE)
> >   * available at this point.
> >   */
> >  #define PageHighMem(__p) is_highmem_idx(page_zonenum(__p))
> > +#define PhysHighMem(__p) (PageHighMem(phys_to_page(__p)))
> 
> This was a not so great idea to add PhysHighMem() because of "else"
> below which unfolds to maze of macros and automatically generated
> functions with "static inline int Page##uname ..." signature.
> 
> >  #define folio_test_highmem(__f)	is_highmem_idx(folio_zonenum(__f))
> >  #else
> >  PAGEFLAG_FALSE(HighMem, highmem)

After sleeping over it, the following hunk will help:

diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index dfbc4ba86bba2..2a1f346178024 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -614,11 +614,11 @@ FOLIO_FLAG(dropbehind, FOLIO_HEAD_PAGE)
  * available at this point.
  */
 #define PageHighMem(__p) is_highmem_idx(page_zonenum(__p))
-#define PhysHighMem(__p) (PageHighMem(phys_to_page(__p)))
 #define folio_test_highmem(__f)        is_highmem_idx(folio_zonenum(__f))
 #else
 PAGEFLAG_FALSE(HighMem, highmem)
 #endif
+#define PhysHighMem(__p) (PageHighMem(phys_to_page(__p)))

 /* Does kmap_local_folio() only allow access to one page of the folio? */
 #ifdef CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP


> 
> Thanks
>