|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-devel] [PATCH] linux/i386: relax highpte pinning
>>> Keir Fraser <keir@xxxxxxxxxxxxx> 17.01.07 15:51 >>>
>I like the cleanups in this patch, including making PageForeign a generic
>page flag (which it does indeed have to become).
>
>However I'm fundamentally confused about why you believe highptes have to be
>explicitly pinned. They will be pinned automatically by mm_pin() when it
>pins the pgdir. The act of populating a pmd entry of a pinned mm will cause
>the pte page to be pinned also -- you don't need to do it explicitly.
>
>So I'm not sure if you're working around an issue I haven't foreseen or if
>there's simply some confusion around this issue.
Here's the updated patch, without any addition of explicit pinning.
Signed-off-by: Jan Beulich <jbeulich@xxxxxxxxxx>
Index: sle10sp1-2007-01-22/arch/i386/mm/highmem-xen.c
===================================================================
--- sle10sp1-2007-01-22.orig/arch/i386/mm/highmem-xen.c 2007-01-22
12:20:20.000000000 +0100
+++ sle10sp1-2007-01-22/arch/i386/mm/highmem-xen.c 2007-01-22
08:53:57.000000000 +0100
@@ -55,7 +55,9 @@ void *kmap_atomic(struct page *page, enu
/* Same as kmap_atomic but with PAGE_KERNEL_RO page protection. */
void *kmap_atomic_pte(struct page *page, enum km_type type)
{
- return __kmap_atomic(page, type, PAGE_KERNEL_RO);
+ return __kmap_atomic(page, type,
+ test_bit(PG_pinned, &page->flags)
+ ? PAGE_KERNEL_RO : kmap_prot);
}
void kunmap_atomic(void *kvaddr, enum km_type type)
Index: sle10sp1-2007-01-22/arch/i386/mm/pgtable-xen.c
===================================================================
--- sle10sp1-2007-01-22.orig/arch/i386/mm/pgtable-xen.c 2007-01-22
12:20:20.000000000 +0100
+++ sle10sp1-2007-01-22/arch/i386/mm/pgtable-xen.c 2007-01-22
11:16:51.000000000 +0100
@@ -25,7 +25,6 @@
#include <asm/mmu_context.h>
#include <xen/features.h>
-#include <xen/foreign_page.h>
#include <asm/hypervisor.h>
static void pgd_test_and_unpin(pgd_t *pgd);
@@ -239,14 +238,6 @@ struct page *pte_alloc_one(struct mm_str
#ifdef CONFIG_HIGHPTE
pte = alloc_pages(GFP_KERNEL|__GFP_HIGHMEM|__GFP_REPEAT|__GFP_ZERO, 0);
- if (pte && PageHighMem(pte)) {
- struct mmuext_op op;
-
- kmap_flush_unused();
- op.cmd = MMUEXT_PIN_L1_TABLE;
- op.arg1.mfn = pfn_to_mfn(page_to_pfn(pte));
- BUG_ON(HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF) < 0);
- }
#else
pte = alloc_pages(GFP_KERNEL|__GFP_REPEAT|__GFP_ZERO, 0);
#endif
@@ -267,13 +258,8 @@ struct page *pte_alloc_one(struct mm_str
if (!pte_write(*virt_to_ptep(va)))
BUG_ON(HYPERVISOR_update_va_mapping(
va, pfn_pte(pfn, PAGE_KERNEL), 0));
- } else {
- struct mmuext_op op;
-
- op.cmd = MMUEXT_UNPIN_TABLE;
- op.arg1.mfn = pfn_to_mfn(pfn);
- BUG_ON(HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF) < 0);
- }
+ } else
+ clear_bit(PG_pinned, &pte->flags);
ClearPageForeign(pte);
init_page_count(pte);
@@ -587,46 +601,48 @@ void make_pages_writable(void *va, unsig
}
}
-static inline void pgd_walk_set_prot(void *pt, pgprot_t flags)
+static inline int pgd_walk_set_prot(struct page *page, pgprot_t flags)
{
- struct page *page = virt_to_page(pt);
unsigned long pfn = page_to_pfn(page);
if (PageHighMem(page))
- return;
+ return pgprot_val(flags) & _PAGE_RW
+ ? test_and_clear_bit(PG_pinned, &page->flags)
+ : !test_and_set_bit(PG_pinned, &page->flags);
+
BUG_ON(HYPERVISOR_update_va_mapping(
(unsigned long)__va(pfn << PAGE_SHIFT),
pfn_pte(pfn, flags), 0));
+
+ return 0;
}
-static void pgd_walk(pgd_t *pgd_base, pgprot_t flags)
+static int pgd_walk(pgd_t *pgd_base, pgprot_t flags)
{
pgd_t *pgd = pgd_base;
pud_t *pud;
pmd_t *pmd;
- pte_t *pte;
- int g, u, m;
+ int g, u, m, flush;
if (xen_feature(XENFEAT_auto_translated_physmap))
- return;
+ return 0;
- for (g = 0; g < USER_PTRS_PER_PGD; g++, pgd++) {
+ for (g = 0, flush = 0; g < USER_PTRS_PER_PGD; g++, pgd++) {
if (pgd_none(*pgd))
continue;
pud = pud_offset(pgd, 0);
if (PTRS_PER_PUD > 1) /* not folded */
- pgd_walk_set_prot(pud,flags);
+ flush |= pgd_walk_set_prot(virt_to_page(pud),flags);
for (u = 0; u < PTRS_PER_PUD; u++, pud++) {
if (pud_none(*pud))
continue;
pmd = pmd_offset(pud, 0);
if (PTRS_PER_PMD > 1) /* not folded */
- pgd_walk_set_prot(pmd,flags);
+ flush |=
pgd_walk_set_prot(virt_to_page(pmd),flags);
for (m = 0; m < PTRS_PER_PMD; m++, pmd++) {
if (pmd_none(*pmd))
continue;
- pte = pte_offset_kernel(pmd,0);
- pgd_walk_set_prot(pte,flags);
+ flush |=
pgd_walk_set_prot(pmd_page(*pmd),flags);
}
}
}
@@ -635,11 +653,14 @@ static void pgd_walk(pgd_t *pgd_base, pg
(unsigned long)pgd_base,
pfn_pte(virt_to_phys(pgd_base)>>PAGE_SHIFT, flags),
UVMF_TLB_FLUSH));
+
+ return flush;
}
static void __pgd_pin(pgd_t *pgd)
{
- pgd_walk(pgd, PAGE_KERNEL_RO);
+ if (pgd_walk(pgd, PAGE_KERNEL_RO))
+ kmap_flush_unused();
xen_pgd_pin(__pa(pgd));
set_bit(PG_pinned, &virt_to_page(pgd)->flags);
}
@@ -647,7 +651,8 @@ static void __pgd_pin(pgd_t *pgd)
static void __pgd_unpin(pgd_t *pgd)
{
xen_pgd_unpin(__pa(pgd));
- pgd_walk(pgd, PAGE_KERNEL);
+ if (pgd_walk(pgd, PAGE_KERNEL))
+ kmap_flush_unused();
clear_bit(PG_pinned, &virt_to_page(pgd)->flags);
}
Index: sle10sp1-2007-01-22/include/asm-i386/mach-xen/asm/page.h
===================================================================
--- sle10sp1-2007-01-22.orig/include/asm-i386/mach-xen/asm/page.h
2007-01-22 12:20:20.000000000 +0100
+++ sle10sp1-2007-01-22/include/asm-i386/mach-xen/asm/page.h 2007-01-22
08:53:57.000000000 +0100
@@ -28,13 +28,12 @@
#include <asm/bug.h>
#include <xen/interface/xen.h>
#include <xen/features.h>
-#include <xen/foreign_page.h>
-#define arch_free_page(_page,_order) \
-({ int foreign = PageForeign(_page); \
- if (foreign) \
- (PageForeignDestructor(_page))(_page); \
- foreign; \
+#define arch_free_page(_page,_order) \
+({ int foreign = PageForeign(_page); \
+ if (foreign) \
+ PageForeignDestructor(_page); \
+ foreign; \
})
#define HAVE_ARCH_FREE_PAGE
Index: sle10sp1-2007-01-22/include/asm-i386/mach-xen/asm/pgalloc.h
===================================================================
--- sle10sp1-2007-01-22.orig/include/asm-i386/mach-xen/asm/pgalloc.h
2007-01-22 12:20:20.000000000 +0100
+++ sle10sp1-2007-01-22/include/asm-i386/mach-xen/asm/pgalloc.h 2007-01-22
12:20:39.000000000 +0100
@@ -6,27 +6,23 @@
#include <linux/mm.h> /* for struct page */
#include <asm/io.h> /* for phys_to_virt and page_to_pseudophys */
-/* Is this pagetable pinned? */
-#define PG_pinned PG_arch_1
-
#define pmd_populate_kernel(mm, pmd, pte) \
set_pmd(pmd, __pmd(_PAGE_TABLE + __pa(pte)))
#define pmd_populate(mm, pmd, pte) \
do { \
+ unsigned long pfn = page_to_pfn(pte); \
if (test_bit(PG_pinned, &virt_to_page((mm)->pgd)->flags)) { \
if (!PageHighMem(pte)) \
BUG_ON(HYPERVISOR_update_va_mapping( \
- (unsigned long)__va(page_to_pfn(pte)<<PAGE_SHIFT),\
- pfn_pte(page_to_pfn(pte), PAGE_KERNEL_RO), 0));\
- set_pmd(pmd, __pmd(_PAGE_TABLE + \
- ((unsigned long long)page_to_pfn(pte) << \
- (unsigned long long) PAGE_SHIFT))); \
- } else { \
- *(pmd) = __pmd(_PAGE_TABLE + \
- ((unsigned long long)page_to_pfn(pte) << \
- (unsigned long long) PAGE_SHIFT)); \
- } \
+ (unsigned long)__va(pfn << PAGE_SHIFT), \
+ pfn_pte(pfn, PAGE_KERNEL_RO), 0)); \
+ else if (!test_and_set_bit(PG_pinned, &pte->flags)) \
+ kmap_flush_unused(); \
+ set_pmd(pmd, \
+ __pmd(_PAGE_TABLE + ((paddr_t)pfn << PAGE_SHIFT))); \
+ } else \
+ *(pmd) = __pmd(_PAGE_TABLE + ((paddr_t)pfn << PAGE_SHIFT)); \
} while (0)
/*
Index: sle10sp1-2007-01-22/include/asm-i386/mach-xen/asm/pgtable.h
===================================================================
--- sle10sp1-2007-01-22.orig/include/asm-i386/mach-xen/asm/pgtable.h
2007-01-22 12:20:20.000000000 +0100
+++ sle10sp1-2007-01-22/include/asm-i386/mach-xen/asm/pgtable.h 2007-01-22
08:53:57.000000000 +0100
@@ -25,6 +25,9 @@
#include <linux/list.h>
#include <linux/spinlock.h>
+/* Is this pagetable pinned? */
+#define PG_pinned PG_arch_1
+
struct mm_struct;
struct vm_area_struct;
Index: sle10sp1-2007-01-22/include/asm-x86_64/mach-xen/asm/page.h
===================================================================
--- sle10sp1-2007-01-22.orig/include/asm-x86_64/mach-xen/asm/page.h
2007-01-22 12:20:20.000000000 +0100
+++ sle10sp1-2007-01-22/include/asm-x86_64/mach-xen/asm/page.h 2007-01-22
08:53:57.000000000 +0100
@@ -8,13 +8,12 @@
#include <asm/bug.h>
#endif
#include <xen/interface/xen.h>
-#include <xen/foreign_page.h>
-#define arch_free_page(_page,_order) \
-({ int foreign = PageForeign(_page); \
- if (foreign) \
- (PageForeignDestructor(_page))(_page); \
- foreign; \
+#define arch_free_page(_page,_order) \
+({ int foreign = PageForeign(_page); \
+ if (foreign) \
+ PageForeignDestructor(_page); \
+ foreign; \
})
#define HAVE_ARCH_FREE_PAGE
Index: sle10-sp1-2007-01-12/include/linux/page-flags.h
===================================================================
--- sle10-sp1-2007-01-12.orig/include/linux/page-flags.h 2007-01-15
17:59:57.000000000 +0100
+++ sle10-sp1-2007-01-12/include/linux/page-flags.h 2007-01-15
14:17:18.000000000 +0100
@@ -76,6 +76,8 @@
#define PG_nosave_free 18 /* Free, should not be written */
#define PG_uncached 19 /* Page has been mapped as uncached */
+#define PG_foreign 20 /* Page is owned by foreign allocator.
*/
+
/*
* Global page accounting. One instance per CPU. Only unsigned longs are
* allowed.
@@ -344,6 +346,18 @@ extern void __mod_page_state_offset(unsi
#define SetPageUncached(page) set_bit(PG_uncached, &(page)->flags)
#define ClearPageUncached(page) clear_bit(PG_uncached, &(page)->flags)
+#define PageForeign(page) test_bit(PG_foreign, &(page)->flags)
+#define SetPageForeign(page, dtor) do { \
+ set_bit(PG_foreign, &(page)->flags); \
+ (page)->mapping = (void *)dtor; \
+} while (0)
+#define ClearPageForeign(page) do { \
+ clear_bit(PG_foreign, &(page)->flags); \
+ (page)->mapping = NULL; \
+} while (0)
+#define PageForeignDestructor(page) \
+ ( (void (*) (struct page *)) (page)->mapping )(page)
+
struct page; /* forward declaration */
int test_clear_page_dirty(struct page *page);
Index: sle10-sp1-2007-01-12/include/xen/foreign_page.h
===================================================================
--- sle10-sp1-2007-01-12.orig/include/xen/foreign_page.h 2007-01-15
17:59:57.000000000 +0100
+++ /dev/null 1970-01-01 00:00:00.000000000 +0000
@@ -1,30 +0,0 @@
-/******************************************************************************
- * foreign_page.h
- *
- * Provide a "foreign" page type, that is owned by a foreign allocator and
- * not the normal buddy allocator in page_alloc.c
- *
- * Copyright (c) 2004, K A Fraser
- */
-
-#ifndef __ASM_XEN_FOREIGN_PAGE_H__
-#define __ASM_XEN_FOREIGN_PAGE_H__
-
-#define PG_foreign PG_arch_1
-
-#define PageForeign(page) test_bit(PG_foreign, &(page)->flags)
-
-#define SetPageForeign(page, dtor) do { \
- set_bit(PG_foreign, &(page)->flags); \
- (page)->mapping = (void *)dtor; \
-} while (0)
-
-#define ClearPageForeign(page) do { \
- clear_bit(PG_foreign, &(page)->flags); \
- (page)->mapping = NULL; \
-} while (0)
-
-#define PageForeignDestructor(page) \
- ( (void (*) (struct page *)) (page)->mapping )
-
-#endif /* __ASM_XEN_FOREIGN_PAGE_H__ */
Index: sle10-sp1-2007-01-12/mm/page_alloc.c
===================================================================
--- sle10-sp1-2007-01-12.orig/mm/page_alloc.c 2007-01-15 17:59:57.000000000
+0100
+++ sle10-sp1-2007-01-12/mm/page_alloc.c 2007-01-16 12:41:24.000000000
+0100
@@ -154,7 +154,11 @@ static void bad_page(struct page *page)
1 << PG_slab |
1 << PG_swapcache |
1 << PG_writeback |
- 1 << PG_buddy );
+ 1 << PG_buddy |
+#ifdef CONFIG_X86_XEN
+ 1 << PG_pinned |
+#endif
+ 1 << PG_foreign );
set_page_count(page, 0);
reset_page_mapcount(page);
page->mapping = NULL;
@@ -389,7 +393,11 @@ static inline int free_pages_check(struc
1 << PG_swapcache |
1 << PG_writeback |
1 << PG_reserved |
- 1 << PG_buddy ))))
+ 1 << PG_buddy |
+#ifdef CONFIG_X86_XEN
+ 1 << PG_pinned |
+#endif
+ 1 << PG_foreign ))))
bad_page(page);
if (PageDirty(page))
__ClearPageDirty(page);
@@ -539,7 +547,11 @@ static int prep_new_page(struct page *pa
1 << PG_swapcache |
1 << PG_writeback |
1 << PG_reserved |
- 1 << PG_buddy ))))
+ 1 << PG_buddy |
+#ifdef CONFIG_X86_XEN
+ 1 << PG_pinned |
+#endif
+ 1 << PG_foreign ))))
bad_page(page);
/*
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
 |
Lists.xenproject.org is hosted with RackSpace, monitoring our |