x86/mm: Validate kernel_physical_mapping_init() PTE population

The usage of __flush_tlb_all() in the kernel_physical_mapping_init()
path is not necessary. In general flushing the TLB is not required when
updating an entry from the !present state. However, to give confidence
in the future removal of TLB flushing in this path, use the new
set_pte_safe() family of helpers to assert that the !present assumption
is true in this path.

[ mingo: Minor readability edits. ]

Suggested-by: Peter Zijlstra <peterz@infradead.org>
Suggested-by: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/154395944177.32119.8524957429632012270.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>

2 files changed, 39 insertions, 12 deletions

diff --git a/arch/x86/include/asm/pgalloc.h b/arch/x86/include/asm/pgalloc.h
index ec7f43327033..1ea41aaef68b 100644
--- a/arch/x86/include/asm/pgalloc.h
+++ b/arch/x86/include/asm/pgalloc.h
@@ -80,6 +80,13 @@ static inline void pmd_populate_kernel(struct mm_struct *mm,
 	set_pmd(pmd, __pmd(__pa(pte) | _PAGE_TABLE));
 }
 
+static inline void pmd_populate_kernel_safe(struct mm_struct *mm,
+				       pmd_t *pmd, pte_t *pte)
+{
+	paravirt_alloc_pte(mm, __pa(pte) >> PAGE_SHIFT);
+	set_pmd_safe(pmd, __pmd(__pa(pte) | _PAGE_TABLE));
+}
+
 static inline void pmd_populate(struct mm_struct *mm, pmd_t *pmd,
 				struct page *pte)
 {
@@ -132,6 +139,12 @@ static inline void pud_populate(struct mm_struct *mm, pud_t *pud, pmd_t *pmd)
 	paravirt_alloc_pmd(mm, __pa(pmd) >> PAGE_SHIFT);
 	set_pud(pud, __pud(_PAGE_TABLE | __pa(pmd)));
 }
+
+static inline void pud_populate_safe(struct mm_struct *mm, pud_t *pud, pmd_t *pmd)
+{
+	paravirt_alloc_pmd(mm, __pa(pmd) >> PAGE_SHIFT);
+	set_pud_safe(pud, __pud(_PAGE_TABLE | __pa(pmd)));
+}
 #endif	/* CONFIG_X86_PAE */
 
 #if CONFIG_PGTABLE_LEVELS > 3
@@ -141,6 +154,12 @@ static inline void p4d_populate(struct mm_struct *mm, p4d_t *p4d, pud_t *pud)
 	set_p4d(p4d, __p4d(_PAGE_TABLE | __pa(pud)));
 }
 
+static inline void p4d_populate_safe(struct mm_struct *mm, p4d_t *p4d, pud_t *pud)
+{
+	paravirt_alloc_pud(mm, __pa(pud) >> PAGE_SHIFT);
+	set_p4d_safe(p4d, __p4d(_PAGE_TABLE | __pa(pud)));
+}
+
 static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long addr)
 {
 	gfp_t gfp = GFP_KERNEL_ACCOUNT;
@@ -173,6 +192,14 @@ static inline void pgd_populate(struct mm_struct *mm, pgd_t *pgd, p4d_t *p4d)
 	set_pgd(pgd, __pgd(_PAGE_TABLE | __pa(p4d)));
 }
 
+static inline void pgd_populate_safe(struct mm_struct *mm, pgd_t *pgd, p4d_t *p4d)
+{
+	if (!pgtable_l5_enabled())
+		return;
+	paravirt_alloc_p4d(mm, __pa(p4d) >> PAGE_SHIFT);
+	set_pgd_safe(pgd, __pgd(_PAGE_TABLE | __pa(p4d)));
+}
+
 static inline p4d_t *p4d_alloc_one(struct mm_struct *mm, unsigned long addr)
 {
 	gfp_t gfp = GFP_KERNEL_ACCOUNT;
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index 5fab264948c2..3e25ac2793ef 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -432,7 +432,7 @@ phys_pte_init(pte_t *pte_page, unsigned long paddr, unsigned long paddr_end,
 					     E820_TYPE_RAM) &&
 			    !e820__mapped_any(paddr & PAGE_MASK, paddr_next,
 					     E820_TYPE_RESERVED_KERN))
-				set_pte(pte, __pte(0));
+				set_pte_safe(pte, __pte(0));
 			continue;
 		}
 
@@ -452,7 +452,7 @@ phys_pte_init(pte_t *pte_page, unsigned long paddr, unsigned long paddr_end,
 			pr_info("   pte=%p addr=%lx pte=%016lx\n", pte, paddr,
 				pfn_pte(paddr >> PAGE_SHIFT, PAGE_KERNEL).pte);
 		pages++;
-		set_pte(pte, pfn_pte(paddr >> PAGE_SHIFT, prot));
+		set_pte_safe(pte, pfn_pte(paddr >> PAGE_SHIFT, prot));
 		paddr_last = (paddr & PAGE_MASK) + PAGE_SIZE;
 	}
 
@@ -487,7 +487,7 @@ phys_pmd_init(pmd_t *pmd_page, unsigned long paddr, unsigned long paddr_end,
 					     E820_TYPE_RAM) &&
 			    !e820__mapped_any(paddr & PMD_MASK, paddr_next,
 					     E820_TYPE_RESERVED_KERN))
-				set_pmd(pmd, __pmd(0));
+				set_pmd_safe(pmd, __pmd(0));
 			continue;
 		}
 
@@ -524,7 +524,7 @@ phys_pmd_init(pmd_t *pmd_page, unsigned long paddr, unsigned long paddr_end,
 		if (page_size_mask & (1<<PG_LEVEL_2M)) {
 			pages++;
 			spin_lock(&init_mm.page_table_lock);
-			set_pte((pte_t *)pmd,
+			set_pte_safe((pte_t *)pmd,
 				pfn_pte((paddr & PMD_MASK) >> PAGE_SHIFT,
 					__pgprot(pgprot_val(prot) | _PAGE_PSE)));
 			spin_unlock(&init_mm.page_table_lock);
@@ -536,7 +536,7 @@ phys_pmd_init(pmd_t *pmd_page, unsigned long paddr, unsigned long paddr_end,
 		paddr_last = phys_pte_init(pte, paddr, paddr_end, new_prot);
 
 		spin_lock(&init_mm.page_table_lock);
-		pmd_populate_kernel(&init_mm, pmd, pte);
+		pmd_populate_kernel_safe(&init_mm, pmd, pte);
 		spin_unlock(&init_mm.page_table_lock);
 	}
 	update_page_count(PG_LEVEL_2M, pages);
@@ -573,7 +573,7 @@ phys_pud_init(pud_t *pud_page, unsigned long paddr, unsigned long paddr_end,
 					     E820_TYPE_RAM) &&
 			    !e820__mapped_any(paddr & PUD_MASK, paddr_next,
 					     E820_TYPE_RESERVED_KERN))
-				set_pud(pud, __pud(0));
+				set_pud_safe(pud, __pud(0));
 			continue;
 		}
 
@@ -611,7 +611,7 @@ phys_pud_init(pud_t *pud_page, unsigned long paddr, unsigned long paddr_end,
 		if (page_size_mask & (1<<PG_LEVEL_1G)) {
 			pages++;
 			spin_lock(&init_mm.page_table_lock);
-			set_pte((pte_t *)pud,
+			set_pte_safe((pte_t *)pud,
 				pfn_pte((paddr & PUD_MASK) >> PAGE_SHIFT,
 					PAGE_KERNEL_LARGE));
 			spin_unlock(&init_mm.page_table_lock);
@@ -624,7 +624,7 @@ phys_pud_init(pud_t *pud_page, unsigned long paddr, unsigned long paddr_end,
 					   page_size_mask, prot);
 
 		spin_lock(&init_mm.page_table_lock);
-		pud_populate(&init_mm, pud, pmd);
+		pud_populate_safe(&init_mm, pud, pmd);
 		spin_unlock(&init_mm.page_table_lock);
 	}
 	__flush_tlb_all();
@@ -659,7 +659,7 @@ phys_p4d_init(p4d_t *p4d_page, unsigned long paddr, unsigned long paddr_end,
 					     E820_TYPE_RAM) &&
 			    !e820__mapped_any(paddr & P4D_MASK, paddr_next,
 					     E820_TYPE_RESERVED_KERN))
-				set_p4d(p4d, __p4d(0));
+				set_p4d_safe(p4d, __p4d(0));
 			continue;
 		}
 
@@ -677,7 +677,7 @@ phys_p4d_init(p4d_t *p4d_page, unsigned long paddr, unsigned long paddr_end,
 					   page_size_mask);
 
 		spin_lock(&init_mm.page_table_lock);
-		p4d_populate(&init_mm, p4d, pud);
+		p4d_populate_safe(&init_mm, p4d, pud);
 		spin_unlock(&init_mm.page_table_lock);
 	}
 	__flush_tlb_all();
@@ -723,9 +723,9 @@ kernel_physical_mapping_init(unsigned long paddr_start,
 
 		spin_lock(&init_mm.page_table_lock);
 		if (pgtable_l5_enabled())
-			pgd_populate(&init_mm, pgd, p4d);
+			pgd_populate_safe(&init_mm, pgd, p4d);
 		else
-			p4d_populate(&init_mm, p4d_offset(pgd, vaddr), (pud_t *) p4d);
+			p4d_populate_safe(&init_mm, p4d_offset(pgd, vaddr), (pud_t *) p4d);
 		spin_unlock(&init_mm.page_table_lock);
 		pgd_changed = true;
 	}