[PATCH] mm: pagefault_{disable,enable}()

Introduce pagefault_{disable,enable}() and use these where previously we did
manual preempt increments/decrements to make the pagefault handler do the
atomic thing.

Currently they still rely on the increased preempt count, but do not rely on
the disabled preemption, this might go away in the future.

(NOTE: the extra barrier() in pagefault_disable might fix some holes on
       machines which have too many registers for their own good)

[heiko.carstens@de.ibm.com: s390 fix]
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Nick Piggin <npiggin@suse.de>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

17 files changed, 88 insertions, 62 deletions

diff --git a/arch/frv/kernel/futex.c b/arch/frv/kernel/futex.c
index eae874a970c..53dc5ed1ebd 100644
--- a/arch/frv/kernel/futex.c
+++ b/arch/frv/kernel/futex.c
@@ -200,7 +200,7 @@ int futex_atomic_op_inuser(int encoded_op, int __user *uaddr)
 	if (!access_ok(VERIFY_WRITE, uaddr, sizeof(int)))
 		return -EFAULT;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	switch (op) {
 	case FUTEX_OP_SET:
@@ -223,7 +223,7 @@ int futex_atomic_op_inuser(int encoded_op, int __user *uaddr)
 		break;
 	}
 
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (!ret) {
 		switch (cmp) {
diff --git a/arch/i386/mm/highmem.c b/arch/i386/mm/highmem.c
index f9f647cdbc7..178bbfe6cba 100644
--- a/arch/i386/mm/highmem.c
+++ b/arch/i386/mm/highmem.c
@@ -32,7 +32,7 @@ void *kmap_atomic(struct page *page, enum km_type type)
 	unsigned long vaddr;
 
 	/* even !CONFIG_PREEMPT needs this, for in_atomic in do_page_fault */
-	inc_preempt_count();
+	pagefault_disable();
 	if (!PageHighMem(page))
 		return page_address(page);
 
@@ -52,8 +52,7 @@ void kunmap_atomic(void *kvaddr, enum km_type type)
 
 #ifdef CONFIG_DEBUG_HIGHMEM
 	if (vaddr >= PAGE_OFFSET && vaddr < (unsigned long)high_memory) {
-		dec_preempt_count();
-		preempt_check_resched();
+		pagefault_enable();
 		return;
 	}
 
@@ -68,8 +67,7 @@ void kunmap_atomic(void *kvaddr, enum km_type type)
 	 */
 	kpte_clear_flush(kmap_pte-idx, vaddr);
 
-	dec_preempt_count();
-	preempt_check_resched();
+	pagefault_enable();
 }
 
 /* This is the same as kmap_atomic() but can map memory that doesn't
@@ -80,7 +78,7 @@ void *kmap_atomic_pfn(unsigned long pfn, enum km_type type)
 	enum fixed_addresses idx;
 	unsigned long vaddr;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	idx = type + KM_TYPE_NR*smp_processor_id();
 	vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
diff --git a/arch/mips/mm/highmem.c b/arch/mips/mm/highmem.c
index 99ebf3ccc22..675502ada5a 100644
--- a/arch/mips/mm/highmem.c
+++ b/arch/mips/mm/highmem.c
@@ -39,7 +39,7 @@ void *__kmap_atomic(struct page *page, enum km_type type)
 	unsigned long vaddr;
 
 	/* even !CONFIG_PREEMPT needs this, for in_atomic in do_page_fault */
-	inc_preempt_count();
+	pagefault_disable();
 	if (!PageHighMem(page))
 		return page_address(page);
 
@@ -62,8 +62,7 @@ void __kunmap_atomic(void *kvaddr, enum km_type type)
 	enum fixed_addresses idx = type + KM_TYPE_NR*smp_processor_id();
 
 	if (vaddr < FIXADDR_START) { // FIXME
-		dec_preempt_count();
-		preempt_check_resched();
+		pagefault_enable();
 		return;
 	}
 
@@ -78,8 +77,7 @@ void __kunmap_atomic(void *kvaddr, enum km_type type)
 	local_flush_tlb_one(vaddr);
 #endif
 
-	dec_preempt_count();
-	preempt_check_resched();
+	pagefault_enable();
 }
 
 #ifndef CONFIG_LIMITED_DMA
@@ -92,7 +90,7 @@ void *kmap_atomic_pfn(unsigned long pfn, enum km_type type)
 	enum fixed_addresses idx;
 	unsigned long vaddr;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	idx = type + KM_TYPE_NR*smp_processor_id();
 	vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
diff --git a/arch/s390/lib/uaccess_std.c b/arch/s390/lib/uaccess_std.c
index 2d549ed2e11..bbaca66fa29 100644
--- a/arch/s390/lib/uaccess_std.c
+++ b/arch/s390/lib/uaccess_std.c
@@ -11,7 +11,7 @@
 
 #include <linux/errno.h>
 #include <linux/mm.h>
-#include <asm/uaccess.h>
+#include <linux/uaccess.h>
 #include <asm/futex.h>
 
 #ifndef __s390x__
@@ -258,7 +258,7 @@ int futex_atomic_op(int op, int __user *uaddr, int oparg, int *old)
 {
 	int oldval = 0, newval, ret;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	switch (op) {
 	case FUTEX_OP_SET:
@@ -284,7 +284,7 @@ int futex_atomic_op(int op, int __user *uaddr, int oparg, int *old)
 	default:
 		ret = -ENOSYS;
 	}
-	dec_preempt_count();
+	pagefault_enable();
 	*old = oldval;
 	return ret;
 }
diff --git a/arch/sparc/mm/highmem.c b/arch/sparc/mm/highmem.c
index 4d8ed9c6518..01fc6c25429 100644
--- a/arch/sparc/mm/highmem.c
+++ b/arch/sparc/mm/highmem.c
@@ -35,7 +35,7 @@ void *kmap_atomic(struct page *page, enum km_type type)
 	unsigned long vaddr;
 
 	/* even !CONFIG_PREEMPT needs this, for in_atomic in do_page_fault */
-	inc_preempt_count();
+	pagefault_disable();
 	if (!PageHighMem(page))
 		return page_address(page);
 
@@ -70,8 +70,7 @@ void kunmap_atomic(void *kvaddr, enum km_type type)
 	unsigned long idx = type + KM_TYPE_NR*smp_processor_id();
 
 	if (vaddr < FIXADDR_START) { // FIXME
-		dec_preempt_count();
-		preempt_check_resched();
+		pagefault_enable();
 		return;
 	}
 
@@ -97,8 +96,7 @@ void kunmap_atomic(void *kvaddr, enum km_type type)
 #endif
 #endif
 
-	dec_preempt_count();
-	preempt_check_resched();
+	pagefault_enable();
 }
 
 /* We may be fed a pagetable here by ptep_to_xxx and others. */
diff --git a/include/asm-frv/highmem.h b/include/asm-frv/highmem.h
index 0f390f41f81..ff4d6cdeb15 100644
--- a/include/asm-frv/highmem.h
+++ b/include/asm-frv/highmem.h
@@ -115,7 +115,7 @@ static inline void *kmap_atomic(struct page *page, enum km_type type)
 {
 	unsigned long paddr;
 
-	inc_preempt_count();
+	pagefault_disable();
 	paddr = page_to_phys(page);
 
 	switch (type) {
@@ -170,8 +170,7 @@ static inline void kunmap_atomic(void *kvaddr, enum km_type type)
 	default:
 		BUG();
 	}
-	dec_preempt_count();
-	preempt_check_resched();
+	pagefault_enable();
 }
 
 #endif /* !__ASSEMBLY__ */
diff --git a/include/asm-generic/futex.h b/include/asm-generic/futex.h
index df893c16031..f422df0956a 100644
--- a/include/asm-generic/futex.h
+++ b/include/asm-generic/futex.h
@@ -21,7 +21,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 	if (! access_ok (VERIFY_WRITE, uaddr, sizeof(int)))
 		return -EFAULT;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	switch (op) {
 	case FUTEX_OP_SET:
@@ -33,7 +33,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 		ret = -ENOSYS;
 	}
 
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (!ret) {
 		switch (cmp) {
diff --git a/include/asm-i386/futex.h b/include/asm-i386/futex.h
index 946d97cfea2..438ef0ec710 100644
--- a/include/asm-i386/futex.h
+++ b/include/asm-i386/futex.h
@@ -56,7 +56,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 	if (! access_ok (VERIFY_WRITE, uaddr, sizeof(int)))
 		return -EFAULT;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	if (op == FUTEX_OP_SET)
 		__futex_atomic_op1("xchgl %0, %2", ret, oldval, uaddr, oparg);
@@ -88,7 +88,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 		}
 	}
 
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (!ret) {
 		switch (cmp) {
diff --git a/include/asm-ia64/futex.h b/include/asm-ia64/futex.h
index 07d77f3a8cb..8a98a265413 100644
--- a/include/asm-ia64/futex.h
+++ b/include/asm-ia64/futex.h
@@ -59,7 +59,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 	if (! access_ok (VERIFY_WRITE, uaddr, sizeof(int)))
 		return -EFAULT;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	switch (op) {
 	case FUTEX_OP_SET:
@@ -83,7 +83,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 		ret = -ENOSYS;
 	}
 
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (!ret) {
 		switch (cmp) {
diff --git a/include/asm-mips/futex.h b/include/asm-mips/futex.h
index 927a216bd53..47e5679c235 100644
--- a/include/asm-mips/futex.h
+++ b/include/asm-mips/futex.h
@@ -88,7 +88,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 	if (! access_ok (VERIFY_WRITE, uaddr, sizeof(int)))
 		return -EFAULT;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	switch (op) {
 	case FUTEX_OP_SET:
@@ -115,7 +115,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 		ret = -ENOSYS;
 	}
 
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (!ret) {
 		switch (cmp) {
diff --git a/include/asm-parisc/futex.h b/include/asm-parisc/futex.h
index d84bbb283fd..dbee6e60aa8 100644
--- a/include/asm-parisc/futex.h
+++ b/include/asm-parisc/futex.h
@@ -21,7 +21,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 	if (! access_ok (VERIFY_WRITE, uaddr, sizeof(int)))
 		return -EFAULT;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	switch (op) {
 	case FUTEX_OP_SET:
@@ -33,7 +33,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 		ret = -ENOSYS;
 	}
 
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (!ret) {
 		switch (cmp) {
diff --git a/include/asm-powerpc/futex.h b/include/asm-powerpc/futex.h
index 936422e5489..3f3673fd3ff 100644
--- a/include/asm-powerpc/futex.h
+++ b/include/asm-powerpc/futex.h
@@ -43,7 +43,7 @@ static inline int futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 	if (! access_ok (VERIFY_WRITE, uaddr, sizeof(int)))
 		return -EFAULT;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	switch (op) {
 	case FUTEX_OP_SET:
@@ -65,7 +65,7 @@ static inline int futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 		ret = -ENOSYS;
 	}
 
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (!ret) {
 		switch (cmp) {
diff --git a/include/asm-ppc/highmem.h b/include/asm-ppc/highmem.h
index 1d2c4ef81c2..f7b21ee302b 100644
--- a/include/asm-ppc/highmem.h
+++ b/include/asm-ppc/highmem.h
@@ -79,7 +79,7 @@ static inline void *kmap_atomic(struct page *page, enum km_type type)
 	unsigned long vaddr;
 
 	/* even !CONFIG_PREEMPT needs this, for in_atomic in do_page_fault */
-	inc_preempt_count();
+	pagefault_disable();
 	if (!PageHighMem(page))
 		return page_address(page);
 
@@ -101,8 +101,7 @@ static inline void kunmap_atomic(void *kvaddr, enum km_type type)
 	unsigned int idx = type + KM_TYPE_NR*smp_processor_id();
 
 	if (vaddr < KMAP_FIX_BEGIN) { // FIXME
-		dec_preempt_count();
-		preempt_check_resched();
+		pagefault_enable();
 		return;
 	}
 
@@ -115,8 +114,7 @@ static inline void kunmap_atomic(void *kvaddr, enum km_type type)
 	pte_clear(&init_mm, vaddr, kmap_pte+idx);
 	flush_tlb_page(NULL, vaddr);
 #endif
-	dec_preempt_count();
-	preempt_check_resched();
+	pagefault_enable();
 }
 
 static inline struct page *kmap_atomic_to_page(void *ptr)
diff --git a/include/asm-sparc64/futex.h b/include/asm-sparc64/futex.h
index 7392fc4a954..876312fe82c 100644
--- a/include/asm-sparc64/futex.h
+++ b/include/asm-sparc64/futex.h
@@ -45,7 +45,7 @@ static inline int futex_atomic_op_inuser(int encoded_op, int __user *uaddr)
 	if (encoded_op & (FUTEX_OP_OPARG_SHIFT << 28))
 		oparg = 1 << oparg;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	switch (op) {
 	case FUTEX_OP_SET:
@@ -67,7 +67,7 @@ static inline int futex_atomic_op_inuser(int encoded_op, int __user *uaddr)
 		ret = -ENOSYS;
 	}
 
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (!ret) {
 		switch (cmp) {
diff --git a/include/asm-x86_64/futex.h b/include/asm-x86_64/futex.h
index 9804bf07b09..5cdfb08013c 100644
--- a/include/asm-x86_64/futex.h
+++ b/include/asm-x86_64/futex.h
@@ -55,7 +55,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 	if (! access_ok (VERIFY_WRITE, uaddr, sizeof(int)))
 		return -EFAULT;
 
-	inc_preempt_count();
+	pagefault_disable();
 
 	switch (op) {
 	case FUTEX_OP_SET:
@@ -78,7 +78,7 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 		ret = -ENOSYS;
 	}
 
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (!ret) {
 		switch (cmp) {
diff --git a/include/linux/uaccess.h b/include/linux/uaccess.h
index a48d7f11c7b..67918c22339 100644
--- a/include/linux/uaccess.h
+++ b/include/linux/uaccess.h
@@ -1,8 +1,43 @@
 #ifndef __LINUX_UACCESS_H__
 #define __LINUX_UACCESS_H__
 
+#include <linux/preempt.h>
 #include <asm/uaccess.h>
 
+/*
+ * These routines enable/disable the pagefault handler in that
+ * it will not take any locks and go straight to the fixup table.
+ *
+ * They have great resemblance to the preempt_disable/enable calls
+ * and in fact they are identical; this is because currently there is
+ * no other way to make the pagefault handlers do this. So we do
+ * disable preemption but we don't necessarily care about that.
+ */
+static inline void pagefault_disable(void)
+{
+	inc_preempt_count();
+	/*
+	 * make sure to have issued the store before a pagefault
+	 * can hit.
+	 */
+	barrier();
+}
+
+static inline void pagefault_enable(void)
+{
+	/*
+	 * make sure to issue those last loads/stores before enabling
+	 * the pagefault handler again.
+	 */
+	barrier();
+	dec_preempt_count();
+	/*
+	 * make sure we do..
+	 */
+	barrier();
+	preempt_check_resched();
+}
+
 #ifndef ARCH_HAS_NOCACHE_UACCESS
 
 static inline unsigned long __copy_from_user_inatomic_nocache(void *to,
@@ -35,9 +70,9 @@ static inline unsigned long __copy_from_user_nocache(void *to,
 	({						\
 		long ret;				\
 							\
-		inc_preempt_count();			\
+		pagefault_disable();			\
 		ret = __get_user(retval, addr);		\
-		dec_preempt_count();			\
+		pagefault_enable();			\
 		ret;					\
 	})
 
diff --git a/kernel/futex.c b/kernel/futex.c
index 93ef30ba209..af7b81cbde3 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -282,9 +282,9 @@ static inline int get_futex_value_locked(u32 *dest, u32 __user *from)
 {
 	int ret;
 
-	inc_preempt_count();
+	pagefault_disable();
 	ret = __copy_from_user_inatomic(dest, from, sizeof(u32));
-	dec_preempt_count();
+	pagefault_enable();
 
 	return ret ? -EFAULT : 0;
 }
@@ -585,9 +585,9 @@ static int wake_futex_pi(u32 __user *uaddr, u32 uval, struct futex_q *this)
 	if (!(uval & FUTEX_OWNER_DIED)) {
 		newval = FUTEX_WAITERS | new_owner->pid;
 
-		inc_preempt_count();
+		pagefault_disable();
 		curval = futex_atomic_cmpxchg_inatomic(uaddr, uval, newval);
-		dec_preempt_count();
+		pagefault_enable();
 		if (curval == -EFAULT)
 			return -EFAULT;
 		if (curval != uval)
@@ -618,9 +618,9 @@ static int unlock_futex_pi(u32 __user *uaddr, u32 uval)
 	 * There is no waiter, so we unlock the futex. The owner died
 	 * bit has not to be preserved here. We are the owner:
 	 */
-	inc_preempt_count();
+	pagefault_disable();
 	oldval = futex_atomic_cmpxchg_inatomic(uaddr, uval, 0);
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (oldval == -EFAULT)
 		return oldval;
@@ -1158,9 +1158,9 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
 	 */
 	newval = current->pid;
 
-	inc_preempt_count();
+	pagefault_disable();
 	curval = futex_atomic_cmpxchg_inatomic(uaddr, 0, newval);
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (unlikely(curval == -EFAULT))
 		goto uaddr_faulted;
@@ -1183,9 +1183,9 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
 	uval = curval;
 	newval = uval | FUTEX_WAITERS;
 
-	inc_preempt_count();
+	pagefault_disable();
 	curval = futex_atomic_cmpxchg_inatomic(uaddr, uval, newval);
-	dec_preempt_count();
+	pagefault_enable();
 
 	if (unlikely(curval == -EFAULT))
 		goto uaddr_faulted;
@@ -1215,10 +1215,10 @@ static int futex_lock_pi(u32 __user *uaddr, int detect, unsigned long sec,
 			newval = current->pid |
 				FUTEX_OWNER_DIED | FUTEX_WAITERS;
 
-			inc_preempt_count();
+			pagefault_disable();
 			curval = futex_atomic_cmpxchg_inatomic(uaddr,
 							       uval, newval);
-			dec_preempt_count();
+			pagefault_enable();
 
 			if (unlikely(curval == -EFAULT))
 				goto uaddr_faulted;
@@ -1390,9 +1390,9 @@ retry_locked:
 	 * anyone else up:
 	 */
 	if (!(uval & FUTEX_OWNER_DIED)) {
-		inc_preempt_count();
+		pagefault_disable();
 		uval = futex_atomic_cmpxchg_inatomic(uaddr, current->pid, 0);
-		dec_preempt_count();
+		pagefault_enable();
 	}
 
 	if (unlikely(uval == -EFAULT))