powerpc: use switch frame for ret_from_kernel_thread parameters

The kernel thread path in copy_thread creates a user interrupt frame on
stack and stores the function and arg parameters there, and
ret_from_kernel_thread loads them. This is a slightly confusing way to
overload that frame. Non-volatile registers are loaded from the switch
frame, so the parameters can be stored there. The user interrupt frame
is now only used by user threads when they return to user.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20230325122904.2375060-4-npiggin@gmail.com

arch/powerpc/kernel/entry_32.S

@@ -188,7 +188,6 @@ ret_from_fork:
 
 	.globl	ret_from_kernel_thread
 ret_from_kernel_thread:
-	REST_NVGPRS(r1)
 	bl	schedule_tail
 	mtctr	r14
 	mr	r3,r15

arch/powerpc/kernel/interrupt_64.S

@@ -741,7 +741,6 @@ _GLOBAL(ret_from_fork)
 
 _GLOBAL(ret_from_kernel_thread)
 	bl	schedule_tail
-	REST_NVGPRS(r1)
 	mtctr	r14
 	mr	r3,r15
 #ifdef CONFIG_PPC64_ELF_ABI_V2

arch/powerpc/kernel/process.c

@@ -1765,14 +1765,10 @@ int copy_thread(struct task_struct *p, const struct kernel_clone_args *args)
 		((unsigned long *)sp)[0] = 0;
 		memset(childregs, 0, sizeof(struct pt_regs));
 		childregs->gpr[1] = sp + STACK_USER_INT_FRAME_SIZE;
-		/* function */
-		if (args->fn)
-			childregs->gpr[14] = ppc_function_entry((void *)args->fn);
 #ifdef CONFIG_PPC64
 		clear_tsk_thread_flag(p, TIF_32BIT);
 		childregs->softe = IRQS_ENABLED;
 #endif
-		childregs->gpr[15] = (unsigned long)args->fn_arg;
 		p->thread.regs = NULL;	/* no user register state */
 		ti->flags |= _TIF_RESTOREALL;
 		f = ret_from_kernel_thread;
@@ -1811,6 +1807,15 @@ int copy_thread(struct task_struct *p, const struct kernel_clone_args *args)
 	((unsigned long *)sp)[0] = sp + STACK_SWITCH_FRAME_SIZE;
 	kregs = (struct pt_regs *)(sp + STACK_SWITCH_FRAME_REGS);
 	kregs->nip = ppc_function_entry(f);
+	if (unlikely(args->fn)) {
+		/*
+		 * Put kthread fn, arg parameters in non-volatile GPRs in the
+		 * switch frame so they are loaded by _switch before it returns
+		 * to ret_from_kernel_thread.
+		 */
+		kregs->gpr[14] = ppc_function_entry((void *)args->fn);
+		kregs->gpr[15] = (unsigned long)args->fn_arg;
+	}
 	p->thread.ksp = sp;
 
 #ifdef CONFIG_HAVE_HW_BREAKPOINT