On Sat, May 6, 2023 at 4:35 AM John Paul Adrian Glaubitz
[off-list ref] wrote:

Hi Vishal!

On Mon, 2023-05-01 at 12:28 -0700, Vishal Moola (Oracle) wrote:

quoted

Part of the conversions to replace pgtable constructor/destructors with
ptdesc equivalents. Also cleans up some spacing issues.

Signed-off-by: Vishal Moola (Oracle) <redacted>
---
 arch/sh/include/asm/pgalloc.h | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/arch/sh/include/asm/pgalloc.h b/arch/sh/include/asm/pgalloc.h
index a9e98233c4d4..ce2ba99dbd84 100644
--- a/arch/sh/include/asm/pgalloc.h
+++ b/arch/sh/include/asm/pgalloc.h

@@ -2,6 +2,7 @@
 #ifndef __ASM_SH_PGALLOC_H
 #define __ASM_SH_PGALLOC_H

+#include <linux/mm.h>
 #include <asm/page.h>

 #define __HAVE_ARCH_PMD_ALLOC_ONE

@@ -31,10 +32,10 @@ static inline void pmd_populate(struct mm_struct *mm, pmd_t *pmd,
      set_pmd(pmd, __pmd((unsigned long)page_address(pte)));
 }

-#define __pte_free_tlb(tlb,pte,addr)                 \
-do {                                                 \
-     pgtable_pte_page_dtor(pte);                     \
-     tlb_remove_page((tlb), (pte));                  \
+#define __pte_free_tlb(tlb, pte, addr)                               \
+do {                                                         \
+     ptdesc_pte_dtor(page_ptdesc(pte));                      \
+     tlb_remove_page_ptdesc((tlb), (page_ptdesc(pte)));      \
 } while (0)

 #endif /* __ASM_SH_PGALLOC_H */

Looking at the patch which introduces tlb_remove_page_ptdesc() [1], it seems that
tlb_remove_page_ptdesc() already calls tlb_remove_page() with ptdesc_page(pt), so
I'm not sure whether the above tlb_remove_page_ptdesc((tlb), (page_ptdesc(pte)))
is correct.

Shouldn't it just be tlb_remove_page_ptdesc((tlb), (pte))?

As of this patchset all implementations of __pte_free_tlb() take in a
struct page. Eventually we'll want it to be tlb_remove_page_ptdesc(tlb, pte),
but for now the cast is necessary here.