s390: fix ptrace of user area if the inferior uses vector registers

The floating point registers of a process that uses vector instruction are
not store into task->thread.fp_regs anymore but in the upper halves of the
first 16 vector registers.
The ptrace interface for the peeks and pokes to the user area fails to take
this into account. Fix __peek_user[_compat] and __poke_user[_compat]
to use the vector array for the floating pointer register if the process
has one.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

arch/s390/kernel/ptrace.c

@@ -248,14 +248,27 @@ static unsigned long __peek_user(struct task_struct *child, addr_t addr)
 		 */
 		tmp = 0;
 
+	} else if (addr == (addr_t) &dummy->regs.fp_regs.fpc) {
+		/*
+		 * floating point control reg. is in the thread structure
+		 */
+		tmp = child->thread.fp_regs.fpc;
+		tmp <<= BITS_PER_LONG - 32;
+
 	} else if (addr < (addr_t) (&dummy->regs.fp_regs + 1)) {
-		/* 
-		 * floating point regs. are stored in the thread structure
+		/*
+		 * floating point regs. are either in child->thread.fp_regs
+		 * or the child->thread.vxrs array
 		 */
-		offset = addr - (addr_t) &dummy->regs.fp_regs;
-		tmp = *(addr_t *)((addr_t) &child->thread.fp_regs + offset);
-		if (addr == (addr_t) &dummy->regs.fp_regs.fpc)
-			tmp <<= BITS_PER_LONG - 32;
+		offset = addr - (addr_t) &dummy->regs.fp_regs.fprs;
+#ifdef CONFIG_64BIT
+		if (child->thread.vxrs)
+			tmp = *(addr_t *)
+			       ((addr_t) child->thread.vxrs + 2*offset);
+		else
+#endif
+			tmp = *(addr_t *)
+			       ((addr_t) &child->thread.fp_regs.fprs + offset);
 
 	} else if (addr < (addr_t) (&dummy->regs.per_info + 1)) {
 		/*
@@ -383,16 +396,29 @@ static int __poke_user(struct task_struct *child, addr_t addr, addr_t data)
 		 */
 		return 0;
 
+	} else if (addr == (addr_t) &dummy->regs.fp_regs.fpc) {
+		/*
+		 * floating point control reg. is in the thread structure
+		 */
+		if ((unsigned int) data != 0 ||
+		    test_fp_ctl(data >> (BITS_PER_LONG - 32)))
+			return -EINVAL;
+		child->thread.fp_regs.fpc = data >> (BITS_PER_LONG - 32);
+
 	} else if (addr < (addr_t) (&dummy->regs.fp_regs + 1)) {
 		/*
-		 * floating point regs. are stored in the thread structure
+		 * floating point regs. are either in child->thread.fp_regs
+		 * or the child->thread.vxrs array
 		 */
-		if (addr == (addr_t) &dummy->regs.fp_regs.fpc)
-			if ((unsigned int) data != 0 ||
-			    test_fp_ctl(data >> (BITS_PER_LONG - 32)))
-				return -EINVAL;
-		offset = addr - (addr_t) &dummy->regs.fp_regs;
-		*(addr_t *)((addr_t) &child->thread.fp_regs + offset) = data;
+		offset = addr - (addr_t) &dummy->regs.fp_regs.fprs;
+#ifdef CONFIG_64BIT
+		if (child->thread.vxrs)
+			*(addr_t *)((addr_t)
+				child->thread.vxrs + 2*offset) = data;
+		else
+#endif
+			*(addr_t *)((addr_t)
+				&child->thread.fp_regs.fprs + offset) = data;
 
 	} else if (addr < (addr_t) (&dummy->regs.per_info + 1)) {
 		/*
@@ -611,12 +637,26 @@ static u32 __peek_user_compat(struct task_struct *child, addr_t addr)
 		 */
 		tmp = 0;
 
+	} else if (addr == (addr_t) &dummy32->regs.fp_regs.fpc) {
+		/*
+		 * floating point control reg. is in the thread structure
+		 */
+		tmp = child->thread.fp_regs.fpc;
+
 	} else if (addr < (addr_t) (&dummy32->regs.fp_regs + 1)) {
 		/*
-		 * floating point regs. are stored in the thread structure 
+		 * floating point regs. are either in child->thread.fp_regs
+		 * or the child->thread.vxrs array
 		 */
-	        offset = addr - (addr_t) &dummy32->regs.fp_regs;
-		tmp = *(__u32 *)((addr_t) &child->thread.fp_regs + offset);
+		offset = addr - (addr_t) &dummy32->regs.fp_regs.fprs;
+#ifdef CONFIG_64BIT
+		if (child->thread.vxrs)
+			tmp = *(__u32 *)
+			       ((addr_t) child->thread.vxrs + 2*offset);
+		else
+#endif
+			tmp = *(__u32 *)
+			       ((addr_t) &child->thread.fp_regs.fprs + offset);
 
 	} else if (addr < (addr_t) (&dummy32->regs.per_info + 1)) {
 		/*
@@ -722,15 +762,28 @@ static int __poke_user_compat(struct task_struct *child,
 		 */
 		return 0;
 
-	} else if (addr < (addr_t) (&dummy32->regs.fp_regs + 1)) {
+	} else if (addr == (addr_t) &dummy32->regs.fp_regs.fpc) {
 		/*
-		 * floating point regs. are stored in the thread structure 
+		 * floating point control reg. is in the thread structure
 		 */
-		if (addr == (addr_t) &dummy32->regs.fp_regs.fpc &&
-		    test_fp_ctl(tmp))
+		if (test_fp_ctl(tmp))
 			return -EINVAL;
-	        offset = addr - (addr_t) &dummy32->regs.fp_regs;
-		*(__u32 *)((addr_t) &child->thread.fp_regs + offset) = tmp;
+		child->thread.fp_regs.fpc = data;
+
+	} else if (addr < (addr_t) (&dummy32->regs.fp_regs + 1)) {
+		/*
+		 * floating point regs. are either in child->thread.fp_regs
+		 * or the child->thread.vxrs array
+		 */
+		offset = addr - (addr_t) &dummy32->regs.fp_regs.fprs;
+#ifdef CONFIG_64BIT
+		if (child->thread.vxrs)
+			*(__u32 *)((addr_t)
+				child->thread.vxrs + 2*offset) = tmp;
+		else
+#endif
+			*(__u32 *)((addr_t)
+				&child->thread.fp_regs.fprs + offset) = tmp;
 
 	} else if (addr < (addr_t) (&dummy32->regs.per_info + 1)) {
 		/*