[PATCH] fix pgd_lock deadlock

[Andrea Arcangeli <aarcange@redhat.com>]

Hello,

Without this patch we can deadlock in the page_table_lock with NR_CPUS
< 4 or THP on, with this patch we hopefully won't deadlock in the
pgd_lock (if taken from irq). I can't see anything taking it from irq
(maybe aio? to check I also tried the libaio testuite with no apparent
VM_BUG_ON triggering), so unless somebody sees it, I think we should
apply it. I've been running for a while with this patch applied
without apparent problems. Other archs may follow suit if it's proven
that there's nothing taking the pgd_lock from irq.

===
Subject: fix pgd_lock deadlock

From: Andrea Arcangeli <aarcange@redhat.com>

It's forbidden to take the page_table_lock with the irq disabled or if there's
contention the IPIs (for tlb flushes) sent with the page_table_lock held will
never run leading to a deadlock.

Apparently nobody takes the pgd_lock from irq so the _irqsave can be removed.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
---
 arch/x86/mm/fault.c    |    8 ++++----
 arch/x86/mm/init_64.c  |    7 ++++---
 arch/x86/mm/pageattr.c |   21 +++++++++++----------
 arch/x86/mm/pgtable.c  |   13 ++++++-------
 arch/x86/xen/mmu.c     |   10 ++++------
 5 files changed, 29 insertions(+), 30 deletions(-)

--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -229,15 +229,15 @@ void vmalloc_sync_all(void)
 	for (address = VMALLOC_START & PMD_MASK;
 	     address >= TASK_SIZE && address < FIXADDR_TOP;
 	     address += PMD_SIZE) {
-
-		unsigned long flags;
 		struct page *page;
 
-		spin_lock_irqsave(&pgd_lock, flags);
+		VM_BUG_ON(in_interrupt());
+		spin_lock(&pgd_lock);
 		list_for_each_entry(page, &pgd_list, lru) {
 			spinlock_t *pgt_lock;
 			pmd_t *ret;
 
+			/* the pgt_lock only for Xen */
 			pgt_lock = &pgd_page_get_mm(page)->page_table_lock;
 
 			spin_lock(pgt_lock);
@@ -247,7 +247,7 @@ void vmalloc_sync_all(void)
 			if (!ret)
 				break;
 		}
-		spin_unlock_irqrestore(&pgd_lock, flags);
+		spin_unlock(&pgd_lock);
 	}
 }
 
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -105,18 +105,19 @@ void sync_global_pgds(unsigned long star
 
 	for (address = start; address <= end; address += PGDIR_SIZE) {
 		const pgd_t *pgd_ref = pgd_offset_k(address);
-		unsigned long flags;
 		struct page *page;
 
 		if (pgd_none(*pgd_ref))
 			continue;
 
-		spin_lock_irqsave(&pgd_lock, flags);
+		VM_BUG_ON(in_interrupt());
+		spin_lock(&pgd_lock);
 		list_for_each_entry(page, &pgd_list, lru) {
 			pgd_t *pgd;
 			spinlock_t *pgt_lock;
 
 			pgd = (pgd_t *)page_address(page) + pgd_index(address);
+			/* the pgt_lock only for Xen */
 			pgt_lock = &pgd_page_get_mm(page)->page_table_lock;
 			spin_lock(pgt_lock);
 
@@ -128,7 +129,7 @@ void sync_global_pgds(unsigned long star
 
 			spin_unlock(pgt_lock);
 		}
-		spin_unlock_irqrestore(&pgd_lock, flags);
+		spin_unlock(&pgd_lock);
 	}
 }
 
--- a/arch/x86/mm/pageattr.c
+++ b/arch/x86/mm/pageattr.c
@@ -57,12 +57,11 @@ static unsigned long direct_pages_count[
 
 void update_page_count(int level, unsigned long pages)
 {
-	unsigned long flags;
-
 	/* Protect against CPA */
-	spin_lock_irqsave(&pgd_lock, flags);
+	VM_BUG_ON(in_interrupt());
+	spin_lock(&pgd_lock);
 	direct_pages_count[level] += pages;
-	spin_unlock_irqrestore(&pgd_lock, flags);
+	spin_unlock(&pgd_lock);
 }
 
 static void split_page_count(int level)
@@ -394,7 +393,7 @@ static int
 try_preserve_large_page(pte_t *kpte, unsigned long address,
 			struct cpa_data *cpa)
 {
-	unsigned long nextpage_addr, numpages, pmask, psize, flags, addr, pfn;
+	unsigned long nextpage_addr, numpages, pmask, psize, addr, pfn;
 	pte_t new_pte, old_pte, *tmp;
 	pgprot_t old_prot, new_prot, req_prot;
 	int i, do_split = 1;
@@ -403,7 +402,8 @@ try_preserve_large_page(pte_t *kpte, uns
 	if (cpa->force_split)
 		return 1;
 
-	spin_lock_irqsave(&pgd_lock, flags);
+	VM_BUG_ON(in_interrupt());
+	spin_lock(&pgd_lock);
 	/*
 	 * Check for races, another CPU might have split this page
 	 * up already:
@@ -498,14 +498,14 @@ try_preserve_large_page(pte_t *kpte, uns
 	}
 
 out_unlock:
-	spin_unlock_irqrestore(&pgd_lock, flags);
+	spin_unlock(&pgd_lock);
 
 	return do_split;
 }
 
 static int split_large_page(pte_t *kpte, unsigned long address)
 {
-	unsigned long flags, pfn, pfninc = 1;
+	unsigned long pfn, pfninc = 1;
 	unsigned int i, level;
 	pte_t *pbase, *tmp;
 	pgprot_t ref_prot;
@@ -519,7 +519,8 @@ static int split_large_page(pte_t *kpte,
 	if (!base)
 		return -ENOMEM;
 
-	spin_lock_irqsave(&pgd_lock, flags);
+	VM_BUG_ON(in_interrupt());
+	spin_lock(&pgd_lock);
 	/*
 	 * Check for races, another CPU might have split this page
 	 * up for us already:
@@ -591,7 +592,7 @@ out_unlock:
 	 */
 	if (base)
 		__free_page(base);
-	spin_unlock_irqrestore(&pgd_lock, flags);
+	spin_unlock(&pgd_lock);
 
 	return 0;
 }
--- a/arch/x86/mm/pgtable.c
+++ b/arch/x86/mm/pgtable.c
@@ -121,14 +121,13 @@ static void pgd_ctor(struct mm_struct *m
 
 static void pgd_dtor(pgd_t *pgd)
 {
-	unsigned long flags; /* can be called from interrupt context */
-
 	if (SHARED_KERNEL_PMD)
 		return;
 
-	spin_lock_irqsave(&pgd_lock, flags);
+	VM_BUG_ON(in_interrupt());
+	spin_lock(&pgd_lock);
 	pgd_list_del(pgd);
-	spin_unlock_irqrestore(&pgd_lock, flags);
+	spin_unlock(&pgd_lock);
 }
 
 /*
@@ -260,7 +259,6 @@ pgd_t *pgd_alloc(struct mm_struct *mm)
 {
 	pgd_t *pgd;
 	pmd_t *pmds[PREALLOCATED_PMDS];
-	unsigned long flags;
 
 	pgd = (pgd_t *)__get_free_page(PGALLOC_GFP);
 
@@ -280,12 +278,13 @@ pgd_t *pgd_alloc(struct mm_struct *mm)
 	 * respect to anything walking the pgd_list, so that they
 	 * never see a partially populated pgd.
 	 */
-	spin_lock_irqsave(&pgd_lock, flags);
+	VM_BUG_ON(in_interrupt());
+	spin_lock(&pgd_lock);
 
 	pgd_ctor(mm, pgd);
 	pgd_prepopulate_pmd(mm, pgd, pmds);
 
-	spin_unlock_irqrestore(&pgd_lock, flags);
+	spin_unlock(&pgd_lock);
 
 	return pgd;
 
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -986,10 +986,9 @@ static void xen_pgd_pin(struct mm_struct
  */
 void xen_mm_pin_all(void)
 {
-	unsigned long flags;
 	struct page *page;
 
-	spin_lock_irqsave(&pgd_lock, flags);
+	spin_lock(&pgd_lock);
 
 	list_for_each_entry(page, &pgd_list, lru) {
 		if (!PagePinned(page)) {
@@ -998,7 +997,7 @@ void xen_mm_pin_all(void)
 		}
 	}
 
-	spin_unlock_irqrestore(&pgd_lock, flags);
+	spin_unlock(&pgd_lock);
 }
 
 /*
@@ -1099,10 +1098,9 @@ static void xen_pgd_unpin(struct mm_stru
  */
 void xen_mm_unpin_all(void)
 {
-	unsigned long flags;
 	struct page *page;
 
-	spin_lock_irqsave(&pgd_lock, flags);
+	spin_lock(&pgd_lock);
 
 	list_for_each_entry(page, &pgd_list, lru) {
 		if (PageSavePinned(page)) {
@@ -1112,7 +1110,7 @@ void xen_mm_unpin_all(void)
 		}
 	}
 
-	spin_unlock_irqrestore(&pgd_lock, flags);
+	spin_unlock(&pgd_lock);
 }
 
 void xen_activate_mm(struct mm_struct *prev, struct mm_struct *next)