xtensa: implement call0 ABI support in assembly

Replace hardcoded register and opcode names with ABI-agnostic macros.
Add register save/restore code where necessary. Conditionalize windowed
only or call0 only code. Add stack initialization matching _switch_to
epilogue to copy_thread.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>

Author: jcmvbkbc

arch/xtensa/boot/boot-redboot/bootstrap.S

@@ -3,6 +3,7 @@
 #include <asm/regs.h>
 #include <asm/asmmacro.h>
 #include <asm/cacheasm.h>
+#include <asm/processor.h>
 	/*
 	 * RB-Data: RedBoot data/bss
 	 * P:	    Boot-Parameters
@@ -36,7 +37,7 @@
 	.globl __start
 	/* this must be the first byte of the loader! */
 __start:
-	entry	sp, 32		# we do not intend to return
+	abi_entry(32)		# we do not intend to return
 	_call0	_start
 __start_a0:
 	.align 4
@@ -62,10 +63,12 @@ _start:
 	wsr	a4, windowstart
 	rsync
 
-	movi	a4, 0x00040000
+	movi	a4, KERNEL_PS_WOE_MASK
 	wsr	a4, ps
 	rsync
 
+KABI_C0	mov	abi_saved0, abi_arg0
+
 	/* copy the loader to its address
 	 * Note: The loader itself is a very small piece, so we assume we
 	 *       don't partially overlap. We also assume (even more important)
@@ -168,52 +171,52 @@ _reloc:
 
 	movi	a3, __image_load
 	sub	a4, a3, a4
-	add	a8, a0, a4
+	add	abi_arg2, a0, a4
 
 	# a1  Stack
 	# a8(a4)  Load address of the image
 
-	movi	a6, _image_start
-	movi	a10, _image_end
-	movi	a7, 0x1000000
-	sub	a11, a10, a6
-	movi	a9, complen
-	s32i	a11, a9, 0
+	movi	abi_arg0, _image_start
+	movi	abi_arg4, _image_end
+	movi	abi_arg1, 0x1000000
+	sub	abi_tmp0, abi_arg4, abi_arg0
+	movi	abi_arg3, complen
+	s32i	abi_tmp0, abi_arg3, 0
 
 	movi	a0, 0
 
-	# a6 destination
-	# a7 maximum size of destination
-	# a8 source
-	# a9 ptr to length
+	# abi_arg0 destination
+	# abi_arg1 maximum size of destination
+	# abi_arg2 source
+	# abi_arg3 ptr to length
 
 	.extern gunzip
-	movi	a4, gunzip
-	beqz	a4, 1f
+	movi	abi_tmp0, gunzip
+	beqz	abi_tmp0, 1f
 
-	callx4	a4
+	abi_callx	abi_tmp0
 
 	j	2f
 
 
-	# a6 destination start
-	# a7 maximum size of destination
-	# a8 source start
-	# a9 ptr to length
-	# a10 destination end
+	# abi_arg0 destination start
+	# abi_arg1 maximum size of destination
+	# abi_arg2 source start
+	# abi_arg3 ptr to length
+	# abi_arg4 destination end
 
 1:
-        l32i    a9, a8, 0
-        l32i    a11, a8, 4
-        s32i    a9, a6, 0
-        s32i    a11, a6, 4
-        l32i    a9, a8, 8
-        l32i    a11, a8, 12
-        s32i    a9, a6, 8
-        s32i    a11, a6, 12
-        addi    a6, a6, 16
-        addi    a8, a8, 16
-        blt     a6, a10, 1b
+        l32i    abi_tmp0, abi_arg2, 0
+        l32i    abi_tmp1, abi_arg2, 4
+        s32i    abi_tmp0, abi_arg0, 0
+        s32i    abi_tmp1, abi_arg0, 4
+        l32i    abi_tmp0, abi_arg2, 8
+        l32i    abi_tmp1, abi_arg2, 12
+        s32i    abi_tmp0, abi_arg0, 8
+        s32i    abi_tmp1, abi_arg0, 12
+        addi    abi_arg0, abi_arg0, 16
+        addi    abi_arg2, abi_arg2, 16
+        blt     abi_arg0, abi_arg4, 1b
 
 
 	/* jump to the kernel */
@@ -230,6 +233,7 @@ _reloc:
 
 	# a2  Boot parameter list
 
+KABI_C0	mov	abi_arg0, abi_saved0
 	movi	a0, _image_start
 	jx	a0