s390/uaccess: rework uaccess code - fix locking issues

The current uaccess code uses a page table walk in some circumstances,
e.g. in case of the in atomic futex operations or if running on old
hardware which doesn't support the mvcos instruction.

However it turned out that the page table walk code does not correctly
lock page tables when accessing page table entries.
In other words: a different cpu may invalidate a page table entry while
the current cpu inspects the pte. This may lead to random data corruption.

Adding correct locking however isn't trivial for all uaccess operations.
Especially copy_in_user() is problematic since that requires to hold at
least two locks, but must be protected against ABBA deadlock when a
different cpu also performs a copy_in_user() operation.

So the solution is a different approach where we change address spaces:

User space runs in primary address mode, or access register mode within
vdso code, like it currently already does.

The kernel usually also runs in home space mode, however when accessing
user space the kernel switches to primary or secondary address mode if
the mvcos instruction is not available or if a compare-and-swap (futex)
instruction on a user space address is performed.
KVM however is special, since that requires the kernel to run in home
address space while implicitly accessing user space with the sie
instruction.

So we end up with:

User space:
- runs in primary or access register mode
- cr1 contains the user asce
- cr7 contains the user asce
- cr13 contains the kernel asce

Kernel space:
- runs in home space mode
- cr1 contains the user or kernel asce
  -> the kernel asce is loaded when a uaccess requires primary or
     secondary address mode
- cr7 contains the user or kernel asce, (changed with set_fs())
- cr13 contains the kernel asce

In case of uaccess the kernel changes to:
- primary space mode in case of a uaccess (copy_to_user) and uses
  e.g. the mvcp instruction to access user space. However the kernel
  will stay in home space mode if the mvcos instruction is available
- secondary space mode in case of futex atomic operations, so that the
  instructions come from primary address space and data from secondary
  space

In case of kvm the kernel runs in home space mode, but cr1 gets switched
to contain the gmap asce before the sie instruction gets executed. When
the sie instruction is finished cr1 will be switched back to contain the
user asce.

A context switch between two processes will always load the kernel asce
for the next process in cr1. So the first exit to user space is a bit
more expensive (one extra load control register instruction) than before,
however keeps the code rather simple.

In sum this means there is no need to perform any error prone page table
walks anymore when accessing user space.

The patch seems to be rather large, however it mainly removes the
the page table walk code and restores the previously deleted "standard"
uaccess code, with a couple of changes.

The uaccess without mvcos mode can be enforced with the "uaccess_primary"
kernel parameter.

Reported-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

1 files changed, 0 insertions, 471 deletions

diff --git a/arch/s390/lib/uaccess_pt.c b/arch/s390/lib/uaccess_pt.c
deleted file mode 100644
index 8d39760bae68..000000000000
--- a/arch/s390/lib/uaccess_pt.c
+++ /dev/null
@@ -1,471 +0,0 @@
-/*
- *  User access functions based on page table walks for enhanced
- *  system layout without hardware support.
- *
- *    Copyright IBM Corp. 2006, 2012
- *    Author(s): Gerald Schaefer (gerald.schaefer@de.ibm.com)
- */
-
-#include <linux/errno.h>
-#include <linux/hardirq.h>
-#include <linux/mm.h>
-#include <linux/hugetlb.h>
-#include <asm/uaccess.h>
-#include <asm/futex.h>
-#include "uaccess.h"
-
-#ifndef CONFIG_64BIT
-#define AHI	"ahi"
-#define SLR	"slr"
-#else
-#define AHI	"aghi"
-#define SLR	"slgr"
-#endif
-
-static unsigned long strnlen_kernel(const char __user *src, unsigned long count)
-{
-	register unsigned long reg0 asm("0") = 0UL;
-	unsigned long tmp1, tmp2;
-
-	asm volatile(
-		"   la	  %2,0(%1)\n"
-		"   la	  %3,0(%0,%1)\n"
-		"  "SLR"  %0,%0\n"
-		"0: srst  %3,%2\n"
-		"   jo	  0b\n"
-		"   la	  %0,1(%3)\n"	/* strnlen_kernel results includes \0 */
-		"  "SLR"  %0,%1\n"
-		"1:\n"
-		EX_TABLE(0b,1b)
-		: "+a" (count), "+a" (src), "=a" (tmp1), "=a" (tmp2)
-		: "d" (reg0) : "cc", "memory");
-	return count;
-}
-
-static unsigned long copy_in_kernel(void __user *to, const void __user *from,
-				    unsigned long count)
-{
-	unsigned long tmp1;
-
-	asm volatile(
-		"  "AHI"  %0,-1\n"
-		"   jo	  5f\n"
-		"   bras  %3,3f\n"
-		"0:"AHI"  %0,257\n"
-		"1: mvc	  0(1,%1),0(%2)\n"
-		"   la	  %1,1(%1)\n"
-		"   la	  %2,1(%2)\n"
-		"  "AHI"  %0,-1\n"
-		"   jnz	  1b\n"
-		"   j	  5f\n"
-		"2: mvc	  0(256,%1),0(%2)\n"
-		"   la	  %1,256(%1)\n"
-		"   la	  %2,256(%2)\n"
-		"3:"AHI"  %0,-256\n"
-		"   jnm	  2b\n"
-		"4: ex	  %0,1b-0b(%3)\n"
-		"5:"SLR"  %0,%0\n"
-		"6:\n"
-		EX_TABLE(1b,6b) EX_TABLE(2b,0b) EX_TABLE(4b,0b)
-		: "+a" (count), "+a" (to), "+a" (from), "=a" (tmp1)
-		: : "cc", "memory");
-	return count;
-}
-
-/*
- * Returns kernel address for user virtual address. If the returned address is
- * >= -4095 (IS_ERR_VALUE(x) returns true), a fault has occurred and the
- * address contains the (negative) exception code.
- */
-#ifdef CONFIG_64BIT
-
-static unsigned long follow_table(struct mm_struct *mm,
-				  unsigned long address, int write)
-{
-	unsigned long *table = (unsigned long *)__pa(mm->pgd);
-
-	if (unlikely(address > mm->context.asce_limit - 1))
-		return -0x38UL;
-	switch (mm->context.asce_bits & _ASCE_TYPE_MASK) {
-	case _ASCE_TYPE_REGION1:
-		table = table + ((address >> 53) & 0x7ff);
-		if (unlikely(*table & _REGION_ENTRY_INVALID))
-			return -0x39UL;
-		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
-		/* fallthrough */
-	case _ASCE_TYPE_REGION2:
-		table = table + ((address >> 42) & 0x7ff);
-		if (unlikely(*table & _REGION_ENTRY_INVALID))
-			return -0x3aUL;
-		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
-		/* fallthrough */
-	case _ASCE_TYPE_REGION3:
-		table = table + ((address >> 31) & 0x7ff);
-		if (unlikely(*table & _REGION_ENTRY_INVALID))
-			return -0x3bUL;
-		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
-		/* fallthrough */
-	case _ASCE_TYPE_SEGMENT:
-		table = table + ((address >> 20) & 0x7ff);
-		if (unlikely(*table & _SEGMENT_ENTRY_INVALID))
-			return -0x10UL;
-		if (unlikely(*table & _SEGMENT_ENTRY_LARGE)) {
-			if (write && (*table & _SEGMENT_ENTRY_PROTECT))
-				return -0x04UL;
-			return (*table & _SEGMENT_ENTRY_ORIGIN_LARGE) +
-				(address & ~_SEGMENT_ENTRY_ORIGIN_LARGE);
-		}
-		table = (unsigned long *)(*table & _SEGMENT_ENTRY_ORIGIN);
-	}
-	table = table + ((address >> 12) & 0xff);
-	if (unlikely(*table & _PAGE_INVALID))
-		return -0x11UL;
-	if (write && (*table & _PAGE_PROTECT))
-		return -0x04UL;
-	return (*table & PAGE_MASK) + (address & ~PAGE_MASK);
-}
-
-#else /* CONFIG_64BIT */
-
-static unsigned long follow_table(struct mm_struct *mm,
-				  unsigned long address, int write)
-{
-	unsigned long *table = (unsigned long *)__pa(mm->pgd);
-
-	table = table + ((address >> 20) & 0x7ff);
-	if (unlikely(*table & _SEGMENT_ENTRY_INVALID))
-		return -0x10UL;
-	table = (unsigned long *)(*table & _SEGMENT_ENTRY_ORIGIN);
-	table = table + ((address >> 12) & 0xff);
-	if (unlikely(*table & _PAGE_INVALID))
-		return -0x11UL;
-	if (write && (*table & _PAGE_PROTECT))
-		return -0x04UL;
-	return (*table & PAGE_MASK) + (address & ~PAGE_MASK);
-}
-
-#endif /* CONFIG_64BIT */
-
-static inline unsigned long __user_copy_pt(unsigned long uaddr, void *kptr,
-					   unsigned long n, int write_user)
-{
-	struct mm_struct *mm = current->mm;
-	unsigned long offset, done, size, kaddr;
-	void *from, *to;
-
-	if (!mm)
-		return n;
-	done = 0;
-retry:
-	spin_lock(&mm->page_table_lock);
-	do {
-		kaddr = follow_table(mm, uaddr, write_user);
-		if (IS_ERR_VALUE(kaddr))
-			goto fault;
-
-		offset = uaddr & ~PAGE_MASK;
-		size = min(n - done, PAGE_SIZE - offset);
-		if (write_user) {
-			to = (void *) kaddr;
-			from = kptr + done;
-		} else {
-			from = (void *) kaddr;
-			to = kptr + done;
-		}
-		memcpy(to, from, size);
-		done += size;
-		uaddr += size;
-	} while (done < n);
-	spin_unlock(&mm->page_table_lock);
-	return n - done;
-fault:
-	spin_unlock(&mm->page_table_lock);
-	if (__handle_fault(uaddr, -kaddr, write_user))
-		return n - done;
-	goto retry;
-}
-
-/*
- * Do DAT for user address by page table walk, return kernel address.
- * This function needs to be called with current->mm->page_table_lock held.
- */
-static inline unsigned long __dat_user_addr(unsigned long uaddr, int write)
-{
-	struct mm_struct *mm = current->mm;
-	unsigned long kaddr;
-	int rc;
-
-retry:
-	kaddr = follow_table(mm, uaddr, write);
-	if (IS_ERR_VALUE(kaddr))
-		goto fault;
-
-	return kaddr;
-fault:
-	spin_unlock(&mm->page_table_lock);
-	rc = __handle_fault(uaddr, -kaddr, write);
-	spin_lock(&mm->page_table_lock);
-	if (!rc)
-		goto retry;
-	return 0;
-}
-
-unsigned long copy_from_user_pt(void *to, const void __user *from, unsigned long n)
-{
-	unsigned long rc;
-
-	if (segment_eq(get_fs(), KERNEL_DS))
-		return copy_in_kernel((void __user *) to, from, n);
-	rc = __user_copy_pt((unsigned long) from, to, n, 0);
-	if (unlikely(rc))
-		memset(to + n - rc, 0, rc);
-	return rc;
-}
-
-unsigned long copy_to_user_pt(void __user *to, const void *from, unsigned long n)
-{
-	if (segment_eq(get_fs(), KERNEL_DS))
-		return copy_in_kernel(to, (void __user *) from, n);
-	return __user_copy_pt((unsigned long) to, (void *) from, n, 1);
-}
-
-unsigned long clear_user_pt(void __user *to, unsigned long n)
-{
-	void *zpage = (void *) empty_zero_page;
-	unsigned long done, size, ret;
-
-	done = 0;
-	do {
-		if (n - done > PAGE_SIZE)
-			size = PAGE_SIZE;
-		else
-			size = n - done;
-		if (segment_eq(get_fs(), KERNEL_DS))
-			ret = copy_in_kernel(to, (void __user *) zpage, n);
-		else
-			ret = __user_copy_pt((unsigned long) to, zpage, size, 1);
-		done += size;
-		to += size;
-		if (ret)
-			return ret + n - done;
-	} while (done < n);
-	return 0;
-}
-
-unsigned long strnlen_user_pt(const char __user *src, unsigned long count)
-{
-	unsigned long uaddr = (unsigned long) src;
-	struct mm_struct *mm = current->mm;
-	unsigned long offset, done, len, kaddr;
-	unsigned long len_str;
-
-	if (unlikely(!count))
-		return 0;
-	if (segment_eq(get_fs(), KERNEL_DS))
-		return strnlen_kernel(src, count);
-	if (!mm)
-		return 0;
-	done = 0;
-retry:
-	spin_lock(&mm->page_table_lock);
-	do {
-		kaddr = follow_table(mm, uaddr, 0);
-		if (IS_ERR_VALUE(kaddr))
-			goto fault;
-
-		offset = uaddr & ~PAGE_MASK;
-		len = min(count - done, PAGE_SIZE - offset);
-		len_str = strnlen((char *) kaddr, len);
-		done += len_str;
-		uaddr += len_str;
-	} while ((len_str == len) && (done < count));
-	spin_unlock(&mm->page_table_lock);
-	return done + 1;
-fault:
-	spin_unlock(&mm->page_table_lock);
-	if (__handle_fault(uaddr, -kaddr, 0))
-		return 0;
-	goto retry;
-}
-
-long strncpy_from_user_pt(char *dst, const char __user *src, long count)
-{
-	unsigned long done, len, offset, len_str;
-
-	if (unlikely(count <= 0))
-		return 0;
-	done = 0;
-	do {
-		offset = (unsigned long)src & ~PAGE_MASK;
-		len = min(count - done, PAGE_SIZE - offset);
-		if (segment_eq(get_fs(), KERNEL_DS)) {
-			if (copy_in_kernel((void __user *) dst, src, len))
-				return -EFAULT;
-		} else {
-			if (__user_copy_pt((unsigned long) src, dst, len, 0))
-				return -EFAULT;
-		}
-		len_str = strnlen(dst, len);
-		done += len_str;
-		src += len_str;
-		dst += len_str;
-	} while ((len_str == len) && (done < count));
-	return done;
-}
-
-unsigned long copy_in_user_pt(void __user *to, const void __user *from,
-			      unsigned long n)
-{
-	struct mm_struct *mm = current->mm;
-	unsigned long offset_max, uaddr, done, size, error_code;
-	unsigned long uaddr_from = (unsigned long) from;
-	unsigned long uaddr_to = (unsigned long) to;
-	unsigned long kaddr_to, kaddr_from;
-	int write_user;
-
-	if (segment_eq(get_fs(), KERNEL_DS))
-		return copy_in_kernel(to, from, n);
-	if (!mm)
-		return n;
-	done = 0;
-retry:
-	spin_lock(&mm->page_table_lock);
-	do {
-		write_user = 0;
-		uaddr = uaddr_from;
-		kaddr_from = follow_table(mm, uaddr_from, 0);
-		error_code = kaddr_from;
-		if (IS_ERR_VALUE(error_code))
-			goto fault;
-
-		write_user = 1;
-		uaddr = uaddr_to;
-		kaddr_to = follow_table(mm, uaddr_to, 1);
-		error_code = (unsigned long) kaddr_to;
-		if (IS_ERR_VALUE(error_code))
-			goto fault;
-
-		offset_max = max(uaddr_from & ~PAGE_MASK,
-				 uaddr_to & ~PAGE_MASK);
-		size = min(n - done, PAGE_SIZE - offset_max);
-
-		memcpy((void *) kaddr_to, (void *) kaddr_from, size);
-		done += size;
-		uaddr_from += size;
-		uaddr_to += size;
-	} while (done < n);
-	spin_unlock(&mm->page_table_lock);
-	return n - done;
-fault:
-	spin_unlock(&mm->page_table_lock);
-	if (__handle_fault(uaddr, -error_code, write_user))
-		return n - done;
-	goto retry;
-}
-
-#define __futex_atomic_op(insn, ret, oldval, newval, uaddr, oparg)	\
-	asm volatile("0: l   %1,0(%6)\n"				\
-		     "1: " insn						\
-		     "2: cs  %1,%2,0(%6)\n"				\
-		     "3: jl  1b\n"					\
-		     "   lhi %0,0\n"					\
-		     "4:\n"						\
-		     EX_TABLE(0b,4b) EX_TABLE(2b,4b) EX_TABLE(3b,4b)	\
-		     : "=d" (ret), "=&d" (oldval), "=&d" (newval),	\
-		       "=m" (*uaddr)					\
-		     : "0" (-EFAULT), "d" (oparg), "a" (uaddr),		\
-		       "m" (*uaddr) : "cc" );
-
-static int __futex_atomic_op_pt(int op, u32 __user *uaddr, int oparg, int *old)
-{
-	int oldval = 0, newval, ret;
-
-	switch (op) {
-	case FUTEX_OP_SET:
-		__futex_atomic_op("lr %2,%5\n",
-				  ret, oldval, newval, uaddr, oparg);
-		break;
-	case FUTEX_OP_ADD:
-		__futex_atomic_op("lr %2,%1\nar %2,%5\n",
-				  ret, oldval, newval, uaddr, oparg);
-		break;
-	case FUTEX_OP_OR:
-		__futex_atomic_op("lr %2,%1\nor %2,%5\n",
-				  ret, oldval, newval, uaddr, oparg);
-		break;
-	case FUTEX_OP_ANDN:
-		__futex_atomic_op("lr %2,%1\nnr %2,%5\n",
-				  ret, oldval, newval, uaddr, oparg);
-		break;
-	case FUTEX_OP_XOR:
-		__futex_atomic_op("lr %2,%1\nxr %2,%5\n",
-				  ret, oldval, newval, uaddr, oparg);
-		break;
-	default:
-		ret = -ENOSYS;
-	}
-	if (ret == 0)
-		*old = oldval;
-	return ret;
-}
-
-int __futex_atomic_op_inuser(int op, u32 __user *uaddr, int oparg, int *old)
-{
-	int ret;
-
-	if (segment_eq(get_fs(), KERNEL_DS))
-		return __futex_atomic_op_pt(op, uaddr, oparg, old);
-	if (unlikely(!current->mm))
-		return -EFAULT;
-	spin_lock(&current->mm->page_table_lock);
-	uaddr = (u32 __force __user *)
-		__dat_user_addr((__force unsigned long) uaddr, 1);
-	if (!uaddr) {
-		spin_unlock(&current->mm->page_table_lock);
-		return -EFAULT;
-	}
-	get_page(virt_to_page(uaddr));
-	spin_unlock(&current->mm->page_table_lock);
-	ret = __futex_atomic_op_pt(op, uaddr, oparg, old);
-	put_page(virt_to_page(uaddr));
-	return ret;
-}
-
-static int __futex_atomic_cmpxchg_pt(u32 *uval, u32 __user *uaddr,
-				     u32 oldval, u32 newval)
-{
-	int ret;
-
-	asm volatile("0: cs   %1,%4,0(%5)\n"
-		     "1: la   %0,0\n"
-		     "2:\n"
-		     EX_TABLE(0b,2b) EX_TABLE(1b,2b)
-		     : "=d" (ret), "+d" (oldval), "=m" (*uaddr)
-		     : "0" (-EFAULT), "d" (newval), "a" (uaddr), "m" (*uaddr)
-		     : "cc", "memory" );
-	*uval = oldval;
-	return ret;
-}
-
-int futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr,
-				  u32 oldval, u32 newval)
-{
-	int ret;
-
-	if (segment_eq(get_fs(), KERNEL_DS))
-		return __futex_atomic_cmpxchg_pt(uval, uaddr, oldval, newval);
-	if (unlikely(!current->mm))
-		return -EFAULT;
-	spin_lock(&current->mm->page_table_lock);
-	uaddr = (u32 __force __user *)
-		__dat_user_addr((__force unsigned long) uaddr, 1);
-	if (!uaddr) {
-		spin_unlock(&current->mm->page_table_lock);
-		return -EFAULT;
-	}
-	get_page(virt_to_page(uaddr));
-	spin_unlock(&current->mm->page_table_lock);
-	ret = __futex_atomic_cmpxchg_pt(uval, uaddr, oldval, newval);
-	put_page(virt_to_page(uaddr));
-	return ret;
-}