Re: remove-page-list patch in -mm breaks m68k

Re: remove-page-list patch in -mm breaks m68k

[Geert Uytterhoeven]

On Tue, 23 Mar 2004, Matthias Urlichs wrote:
> The patch mentioned in $SUBJECT, which is included in Andrew's mm tree,
> does this to "struct page":
>
> diff -Nru a/include/linux/mm.h b/include/linux/mm.h
> --- a/include/linux/mm.h        Tue Mar 23 11:29:11 2004
> +++ b/include/linux/mm.h        Tue Mar 23 11:29:11 2004
> @@ -177,7 +177,6 @@
>         page_flags_t flags;             /* atomic flags, some possibly
>                                            updated asynchronously */
>         atomic_t count;                 /* Usage count, see below. */
> -       struct list_head list;          /* ->mapping has some page lists. */
>         struct address_space *mapping;  /* The inode (or ...) we belong to. */
>         pgoff_t index;                  /* Our offset within mapping. */
>         struct list_head lru;           /* Pageout list, eg. active_list;
>
> In principle, anything that makes "struct page" smaller is a good idea.
> Unfortunately, this change breaks m68k, as arch/m68k/mm/memory.c has,
> and actually uses,
>
> #define PD_PTABLE(page) ((ptable_desc *)&(virt_to_page(page)->list))
>
> Thus, apparently this change
>
> /* ++andreas: {get,free}_pointer_table rewritten to use unused fields from
>    struct page instead of separately kmalloced struct.  Stolen from
>    arch/sparc/mm/srmmu.c ... */

Apparently arch/sparc/mm/srmmu.c stopped using this a long time ago...

> needs to be reverted. It seems that a binary search through old kernel
> archives is in order.

I don't have a copy of the latest mm tree at hand, but in plain 2.6.5-rc2, we
have several users:

| tux$ find arch -type f | xargs grep 'page.*->list'
| arch/i386/mm/pageattr.c:                list_add(&kpte_page->list, &df_list);
| arch/i386/mm/hugetlbpage.c:     list_add(&page->list,
| arch/i386/mm/hugetlbpage.c:             list_del(&page->list);
| arch/i386/mm/hugetlbpage.c:     INIT_LIST_HEAD(&page->list);
| arch/m68k/mm/memory.c:#define PD_PTABLE(page) ((ptable_desc *)&(virt_to_page(page)->list))
| arch/sparc64/mm/hugetlbpage.c:  list_add(&page->list,
| arch/sparc64/mm/hugetlbpage.c:          list_del(&page->list);
| arch/sparc64/mm/hugetlbpage.c:  INIT_LIST_HEAD(&page->list);
| arch/ia64/mm/hugetlbpage.c:     list_add(&page->list,
| arch/ia64/mm/hugetlbpage.c:             list_del(&page->list);
| arch/ia64/mm/hugetlbpage.c:     INIT_LIST_HEAD(&page->list);
| arch/ppc64/mm/hugetlbpage.c:    list_add(&page->list,
| arch/ppc64/mm/hugetlbpage.c:            list_del(&page->list);
| arch/ppc64/mm/hugetlbpage.c:    INIT_LIST_HEAD(&page->list);
| arch/arm26/machine/small_page.c:                        list_del_init(&page->list);
| arch/arm26/machine/small_page.c:                        list_add(&page->list, &order->queue);
| arch/arm26/machine/small_page.c:                        list_add(&page->list, &order->queue);
| arch/arm26/machine/small_page.c:        list_del_init(&page->list);
| tux$

Are all of these (except for m68k) converted in the mm tree?

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

Re: remove-page-list patch in -mm breaks m68k

[Matthias Urlichs]

Hi,

Geert Uytterhoeven:
> Are all of these (except for m68k) converted in the mm tree?

The hugetlb stuff is covered; it basically replaces page->list with
page->lru.

> | arch/arm26/machine/small_page.c

That's missing from -mm. However, the code suggests that a patch for
arm26 would also just s/list/lru/g.

Something similar to the arm26 code might work for m68k..?

-- 
Matthias Urlichs

Re: remove-page-list patch in -mm breaks m68k

[Andrew Morton]

Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>
> On Tue, 23 Mar 2004, Matthias Urlichs wrote:
> > The patch mentioned in $SUBJECT, which is included in Andrew's mm tree,
> > does this to "struct page":
> >
> > diff -Nru a/include/linux/mm.h b/include/linux/mm.h
> > --- a/include/linux/mm.h        Tue Mar 23 11:29:11 2004
> > +++ b/include/linux/mm.h        Tue Mar 23 11:29:11 2004
> > @@ -177,7 +177,6 @@
> >         page_flags_t flags;             /* atomic flags, some possibly
> >                                            updated asynchronously */
> >         atomic_t count;                 /* Usage count, see below. */
> > -       struct list_head list;          /* ->mapping has some page lists. */
> >         struct address_space *mapping;  /* The inode (or ...) we belong to. */
> >         pgoff_t index;                  /* Our offset within mapping. */
> >         struct list_head lru;           /* Pageout list, eg. active_list;
> >
> > In principle, anything that makes "struct page" smaller is a good idea.
> > Unfortunately, this change breaks m68k, as arch/m68k/mm/memory.c has,
> > and actually uses,
> >
> > #define PD_PTABLE(page) ((ptable_desc *)&(virt_to_page(page)->list))
> >

ah-hah, thanks.

--- 25/arch/m68k/mm/memory.c~m68k-stop-using-page-list	2004-03-23 09:39:17.212548984 -0800
+++ 25-akpm/arch/m68k/mm/memory.c	2004-03-23 09:39:31.179425696 -0800
@@ -29,8 +29,8 @@
 typedef struct list_head ptable_desc;
 static LIST_HEAD(ptable_list);
 
-#define PD_PTABLE(page) ((ptable_desc *)&(virt_to_page(page)->list))
-#define PD_PAGE(ptable) (list_entry(ptable, struct page, list))
+#define PD_PTABLE(page) ((ptable_desc *)&(virt_to_page(page)->lru))
+#define PD_PAGE(ptable) (list_entry(ptable, struct page, lru))
 #define PD_MARKBITS(dp) (*(unsigned char *)&PD_PAGE(dp)->index)
 
 #define PTABLE_SIZE (PTRS_PER_PMD * sizeof(pmd_t))

_

> | tux$ find arch -type f | xargs grep 'page.*->list'
> | arch/i386/mm/pageattr.c:                list_add(&kpte_page->list, &df_list);
> | arch/i386/mm/hugetlbpage.c:     list_add(&page->list,
> | arch/i386/mm/hugetlbpage.c:             list_del(&page->list);
> | arch/i386/mm/hugetlbpage.c:     INIT_LIST_HEAD(&page->list);
> | arch/m68k/mm/memory.c:#define PD_PTABLE(page) ((ptable_desc *)&(virt_to_page(page)->list))
> | arch/sparc64/mm/hugetlbpage.c:  list_add(&page->list,
> | arch/sparc64/mm/hugetlbpage.c:          list_del(&page->list);
> | arch/sparc64/mm/hugetlbpage.c:  INIT_LIST_HEAD(&page->list);
> | arch/ia64/mm/hugetlbpage.c:     list_add(&page->list,
> | arch/ia64/mm/hugetlbpage.c:             list_del(&page->list);
> | arch/ia64/mm/hugetlbpage.c:     INIT_LIST_HEAD(&page->list);
> | arch/ppc64/mm/hugetlbpage.c:    list_add(&page->list,
> | arch/ppc64/mm/hugetlbpage.c:            list_del(&page->list);
> | arch/ppc64/mm/hugetlbpage.c:    INIT_LIST_HEAD(&page->list);
> | arch/arm26/machine/small_page.c:                        list_del_init(&page->list);
> | arch/arm26/machine/small_page.c:                        list_add(&page->list, &order->queue);
> | arch/arm26/machine/small_page.c:                        list_add(&page->list, &order->queue);
> | arch/arm26/machine/small_page.c:        list_del_init(&page->list);
> | tux$
> 
> Are all of these (except for m68k) converted in the mm tree?

All except for arm26:

--- 25/arch/arm26/machine/small_page.c~arm-stop-using-page-list	2004-03-23 09:35:33.994483288 -0800
+++ 25-akpm/arch/arm26/machine/small_page.c	2004-03-23 09:37:08.227157744 -0800
@@ -95,7 +95,7 @@ again:
 		offset = ffz(USED_MAP(page));
 		SET_USED(page, offset);
 		if (USED_MAP(page) == order->all_used)
-			list_del_init(&page->list);
+			list_del_init(&page->lru);
 		spin_unlock_irqrestore(&small_page_lock, flags);
 
 		return (unsigned long) page_address(page) + (offset << order->shift);
@@ -110,7 +110,7 @@ need_new_page:
 				goto no_page;
 			SetPageReserved(page);
 			USED_MAP(page) = 0;
-			list_add(&page->list, &order->queue);
+			list_add(&page->lru, &order->queue);
 			goto again;
 		}
 
@@ -151,7 +151,7 @@ static void __free_small_page(unsigned l
 		spin_lock_irqsave(&small_page_lock, flags);
 
 		if (USED_MAP(page) == order->all_used)
-			list_add(&page->list, &order->queue);
+			list_add(&page->lru, &order->queue);
 
 		if (!TEST_AND_CLEAR_USED(page, spage))
 			goto already_free;
@@ -167,7 +167,7 @@ free_page:
 	/*
 	 * unlink the page from the small page queue and free it
 	 */
-	list_del_init(&page->list);
+	list_del_init(&page->lru);
 	spin_unlock_irqrestore(&small_page_lock, flags);
 	ClearPageReserved(page);
 	__free_page(page);

_