On 11/5/19 7:03 PM, Anshuman Khandual wrote:

But should not pfn_pmd() be encapsulated inside HAVE_ARCH_TRANSPARENT_HUGEPAGE
at the minimum (but I would say it should be available always, nonetheless) when
the platform subscribes to THP irrespective of whether THP is enabled or not.

For ARC it was only introduced/needed when I added THP support so it is dependent
in some way.

I could see in the file (arch/arc/include/asm/pgtable.h) that fetching pfn_pmd()
and all other basic PMD definitions is conditional on CONFIG_TRANSPARENT_HUGEPAGE.

#ifdef CONFIG_TRANSPARENT_HUGEPAGE
#include <asm/hugepage.h>
#endif

IIUC, CONFIG_TRANSPARENT_HUGEPAGE should only encapsulate PMD page table helpers
which are expected from generic THP code (pmd_trans_huge, pmdp_set_access_flags
etc) but not the basic PMD helpers like pmd_pfn, pmd_mkyoung, pmd_mkdirty,
pmd_mkclean etc. 

ARC only has 2 levels of paging, so these don't make any sense in general and
needed only for THP case.
I case of arch/arm you see it is only defined in pgtable-3level.h

Hence wondering will it be possible to accommodate following
code change on arc platform (not even compiled) in order to fix the problem ?

I'm open to making changes in ARC code but lets do the right thing.

  */
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE
 #include <asm/hugepage.h>
 #endif

This in wrong.  CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE is a just a glue toggle,
used only in Kconfig files (and not in any "C" code).  It enables generic Kconfig
code to allow visibility of CONFIG_TRANSPARENT_HUGEPAGE w/o every arch needing to
do a me too.

I think you need to use CONFIG_TRANSPARENT_HUGEPAGE to guard appropriate tests. I
understand that it only

-Vineet