[PATCH 09/12] x86/aors_n.asm: Add X86_ENDBR to indirect jump targets

[H.J. Lu]

x86/aors_n.asm uses a trick to implment jump tables with LEA.  We can't
use conditional branches nor normal jump tables since jump table entries
use EFLAGS set by jump table index.  This patch adds X86_ENDBR to indirect
jump targets and adjust destination for X86_ENDBR.

	* mpn/x86/aors_n.asm: Save and restore %ebx if X86_ENDBR is
	endbr32.  Add X86_ENDBR to indirect jump targets and adjust
	jump destination for X86_ENDBR.
---
 mpn/x86/aors_n.asm | 32 ++++++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/mpn/x86/aors_n.asm b/mpn/x86/aors_n.asm
index 95d006a5a..94fa16ed8 100644
--- a/mpn/x86/aors_n.asm
+++ b/mpn/x86/aors_n.asm
@@ -80,6 +80,9 @@ deflit(`FRAME',0)
 	movl	PARAM_SRC2,%edx
 	movl	PARAM_SIZE,%ecx
 
+ifelse(X86_ENDBR,`endbr32',
+`	pushl	%ebx		FRAME_pushl()')
+
 	movl	%ecx,%eax
 	shrl	$3,%ecx			C compute count for unrolled loop
 	negl	%eax
@@ -92,6 +95,10 @@ deflit(`FRAME',0)
 	subl	%eax,%edx		C ... enter the loop
 	shrl	$2,%eax			C restore previous value
 
+	C Count for 4-byte endbr32.
+ifelse(X86_ENDBR,`endbr32',
+`	leal	-4(,%eax,4),%ebx')
+
 ifdef(`PIC',`
 	C Calculate start address in loop for PIC.  Due to limitations in
 	C old gas, LF(M4_function_n,oop)-L(0a)-3 cannot be put into the leal
@@ -105,6 +112,10 @@ L(0a):	leal	(%eax,%eax,8),%eax
 	leal	L(oop)-3(%eax,%eax,8),%eax
 ')
 
+	C Adjust for endbr32
+ifelse(X86_ENDBR,`endbr32',
+`	addl	%ebx,%eax')
+
 	C These lines initialize carry from the 5th parameter.  Should be
 	C possible to simplify.
 	pushl	%ebp		FRAME_pushl()
@@ -129,6 +140,9 @@ deflit(`FRAME',0)
 	movl	PARAM_SRC2,%edx
 	movl	PARAM_SIZE,%ecx
 
+ifelse(X86_ENDBR,`endbr32',
+`	pushl	%ebx		FRAME_pushl()')
+
 	movl	%ecx,%eax
 	shrl	$3,%ecx			C compute count for unrolled loop
 	negl	%eax
@@ -141,6 +155,10 @@ deflit(`FRAME',0)
 	subl	%eax,%edx		C ... enter the loop
 	shrl	$2,%eax			C restore previous value
 
+	C Count for 4-byte endbr32.
+ifelse(X86_ENDBR,`endbr32',
+`	leal	-4(,%eax,4),%ebx')
+
 ifdef(`PIC',`
 	C Calculate start address in loop for PIC.  Due to limitations in
 	C some assemblers, L(oop)-L(0b)-3 cannot be put into the leal
@@ -153,6 +171,10 @@ L(0b):	leal	(%eax,%eax,8),%eax
 	C Calculate start address in loop for non-PIC.
 	leal	L(oop)-3(%eax,%eax,8),%eax
 ')
+	C Adjust for endbr32
+ifelse(X86_ENDBR,`endbr32',
+`	addl	%ebx,%eax')
+
 	jmp	*%eax			C jump into loop
 
 L(oopgo):
@@ -165,24 +187,31 @@ L(oopgo):
 L(oop):	movl	(%esi),%eax
 	M4_inst	(%edx),%eax
 	movl	%eax,(%edi)
+	X86_ENDBR
 	movl	4(%esi),%eax
 	M4_inst	4(%edx),%eax
 	movl	%eax,4(%edi)
+	X86_ENDBR
 	movl	8(%esi),%eax
 	M4_inst	8(%edx),%eax
 	movl	%eax,8(%edi)
+	X86_ENDBR
 	movl	12(%esi),%eax
 	M4_inst	12(%edx),%eax
 	movl	%eax,12(%edi)
+	X86_ENDBR
 	movl	16(%esi),%eax
 	M4_inst	16(%edx),%eax
 	movl	%eax,16(%edi)
+	X86_ENDBR
 	movl	20(%esi),%eax
 	M4_inst	20(%edx),%eax
 	movl	%eax,20(%edi)
+	X86_ENDBR
 	movl	24(%esi),%eax
 	M4_inst	24(%edx),%eax
 	movl	%eax,24(%edi)
+	X86_ENDBR
 	movl	28(%esi),%eax
 	M4_inst	28(%edx),%eax
 	movl	%eax,28(%edi)
@@ -195,6 +224,9 @@ L(oop):	movl	(%esi),%eax
 	sbbl	%eax,%eax
 	negl	%eax
 
+ifelse(X86_ENDBR,`endbr32',
+`	popl	%ebx')
+
 	popl	%esi
 	popl	%edi
 	ret
-- 
2.24.1