[S390] fix io_return critical section cleanup

If a machine check interrupts the io interrupt handler on one of the
instructions between io_return and io_leave the critical section
cleanup code will move the return psw to io_work_loop. By doing that
the switch from the asynchronous interrupt stack to the process stack
is skipped. If e.g. TIF_NEED_RESCHED is set things break because
the scheduler is called with the asynchronous interrupts stack.
Moving the psw back to io_return instead fixes the problem.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

arch/s390/kernel/entry.S

@@ -964,7 +964,7 @@ cleanup_critical:
 	clc	4(4,%r12),BASED(cleanup_table_io_work_loop)
 	bl	BASED(0f)
 	clc	4(4,%r12),BASED(cleanup_table_io_work_loop+4)
-	bl	BASED(cleanup_io_return)
+	bl	BASED(cleanup_io_work_loop)
 0:
 	br	%r14
 
@@ -1038,6 +1038,12 @@ cleanup_sysc_leave_insn:
 	.long	sysc_done - 8 + 0x80000000
 
 cleanup_io_return:
+	mvc	__LC_RETURN_PSW(4),0(%r12)
+	mvc	__LC_RETURN_PSW+4(4),BASED(cleanup_table_io_return)
+	la	%r12,__LC_RETURN_PSW
+	br	%r14
+
+cleanup_io_work_loop:
 	mvc	__LC_RETURN_PSW(4),0(%r12)
 	mvc	__LC_RETURN_PSW+4(4),BASED(cleanup_table_io_work_loop)
 	la	%r12,__LC_RETURN_PSW

arch/s390/kernel/entry64.S

@@ -946,7 +946,7 @@ cleanup_critical:
 	clc	8(8,%r12),BASED(cleanup_table_io_work_loop)
 	jl	0f
 	clc	8(8,%r12),BASED(cleanup_table_io_work_loop+8)
-	jl	cleanup_io_return
+	jl	cleanup_io_work_loop
 0:
 	br	%r14
 
@@ -1020,6 +1020,12 @@ cleanup_sysc_leave_insn:
 	.quad	sysc_done - 16
 
 cleanup_io_return:
+	mvc	__LC_RETURN_PSW(8),0(%r12)
+	mvc	__LC_RETURN_PSW+8(8),BASED(cleanup_table_io_return)
+	la	%r12,__LC_RETURN_PSW
+	br	%r14
+
+cleanup_io_work_loop:
 	mvc	__LC_RETURN_PSW(8),0(%r12)
 	mvc	__LC_RETURN_PSW+8(8),BASED(cleanup_table_io_work_loop)
 	la	%r12,__LC_RETURN_PSW