[PATCH 06/12] x86_64/coreibwl/mullo_basecase.asm: Add X86_ENDBR

[H.J. Lu]

Add X86_ENDBR to indirect branch targets.  Since adding X86_ENDBR makes
one byte displacement of "jrcxz L(end)" out of range, move L(end) closer
to "jrcxz L(end)".

	* mpn/x86_64/coreibwl/mullo_basecase.asm: Add X86_ENDBR to
	indirect branch targets.  Move L(end) closer to "jrcxz L(end)".
---
 mpn/x86_64/coreibwl/mullo_basecase.asm | 58 ++++++++++++++++----------
 1 file changed, 37 insertions(+), 21 deletions(-)

diff --git a/mpn/x86_64/coreibwl/mullo_basecase.asm b/mpn/x86_64/coreibwl/mullo_basecase.asm
index b3e435b35..d7f3bd55e 100644
--- a/mpn/x86_64/coreibwl/mullo_basecase.asm
+++ b/mpn/x86_64/coreibwl/mullo_basecase.asm
@@ -134,13 +134,15 @@ ifdef(`PIC',
 	jmp	*(%r10,%rax,8)
 ')
 
-L(mf0):	mulx(	(up), %r10, %r8)
+L(mf0):	X86_ENDBR
+	mulx(	(up), %r10, %r8)
 	lea	56(up), up
 	lea	-8(rp), rp
 	lea	L(f7)(%rip), jmpreg
 	jmp	L(mb0)
 
-L(mf3):	mulx(	(up), %r9, %rax)
+L(mf3):	X86_ENDBR
+	mulx(	(up), %r9, %rax)
 	lea	16(up), up
 	lea	16(rp), rp
 	jrcxz	L(mc)
@@ -157,38 +159,44 @@ L(mc):	mulx(	-8,(up), %r10, %r8)
 	mov	%r9, (rp)
 	jmp	L(c2)
 
-L(mf4):	mulx(	(up), %r10, %r8)
+L(mf4):	X86_ENDBR
+	mulx(	(up), %r10, %r8)
 	lea	24(up), up
 	lea	24(rp), rp
 	inc	R32(n)
 	lea	L(f3)(%rip), jmpreg
 	jmp	L(mb4)
 
-L(mf5):	mulx(	(up), %r9, %rax)
+L(mf5):	X86_ENDBR
+	mulx(	(up), %r9, %rax)
 	lea	32(up), up
 	lea	32(rp), rp
 	inc	R32(n)
 	lea	L(f4)(%rip), jmpreg
 	jmp	L(mb5)
 
-L(mf6):	mulx(	(up), %r10, %r8)
+L(mf6):	X86_ENDBR
+	mulx(	(up), %r10, %r8)
 	lea	40(up), up
 	lea	40(rp), rp
 	inc	R32(n)
 	lea	L(f5)(%rip), jmpreg
 	jmp	L(mb6)
 
-L(mf7):	mulx(	(up), %r9, %rax)
+L(mf7):	X86_ENDBR
+	mulx(	(up), %r9, %rax)
 	lea	48(up), up
 	lea	48(rp), rp
 	lea	L(f6)(%rip), jmpreg
 	jmp	L(mb7)
 
-L(mf1):	mulx(	(up), %r9, %rax)
+L(mf1):	X86_ENDBR
+	mulx(	(up), %r9, %rax)
 	lea	L(f0)(%rip), jmpreg
 	jmp	L(mb1)
 
-L(mf2):	mulx(	(up), %r10, %r8)
+L(mf2):	X86_ENDBR
+	mulx(	(up), %r10, %r8)
 	lea	8(up), up
 	lea	8(rp), rp
 	lea	L(f1)(%rip), jmpreg
@@ -235,17 +243,26 @@ L(mend):mov	%r10, -8(rp)
 	shr	$3, R32(nn)
 	jmp	L(ent)
 
-L(f0):	mulx(	(up), %r10, %r8)
+L(f0):	X86_ENDBR
+	mulx(	(up), %r10, %r8)
 	lea	-8(up), up
 	lea	-8(rp), rp
 	lea	L(f7)(%rip), jmpreg
 	jmp	L(b0)
 
-L(f1):	mulx(	(up), %r9, %rax)
+L(f1):	X86_ENDBR
+	mulx(	(up), %r9, %rax)
 	lea	-1(nn), R32(nn)
 	lea	L(f0)(%rip), jmpreg
 	jmp	L(b1)
 
+L(f7):	X86_ENDBR
+	mulx(	(up), %r9, %rax)
+	lea	-16(up), up
+	lea	-16(rp), rp
+	lea	L(f6)(%rip), jmpreg
+	jmp	L(b7)
+
 L(end):	adox(	(rp), %r9)
 	mov	%r9, (rp)
 	adox(	%rcx, %rax)		C relies on rcx = 0
@@ -261,13 +278,8 @@ L(ent):	mulx(	8,(up), %r10, %r8)	C r8 unused (use imul?)
 	or	R32(nn), R32(n)		C copy count, clear CF,OF (n = 0 prior)
 	jmp	*jmpreg
 
-L(f7):	mulx(	(up), %r9, %rax)
-	lea	-16(up), up
-	lea	-16(rp), rp
-	lea	L(f6)(%rip), jmpreg
-	jmp	L(b7)
-
-L(f2):	mulx(	(up), %r10, %r8)
+L(f2):	X86_ENDBR
+	mulx(	(up), %r10, %r8)
 	lea	8(up), up
 	lea	8(rp), rp
 	mulx(	(up), %r9, %rax)
@@ -313,25 +325,29 @@ L(b3):	adox(	48,(rp), %r9)
 	mulx(	(up), %r9, %rax)
 	jmp	L(top)
 
-L(f6):	mulx(	(up), %r10, %r8)
+L(f6):	X86_ENDBR
+	mulx(	(up), %r10, %r8)
 	lea	40(up), up
 	lea	-24(rp), rp
 	lea	L(f5)(%rip), jmpreg
 	jmp	L(b6)
 
-L(f5):	mulx(	(up), %r9, %rax)
+L(f5):	X86_ENDBR
+	mulx(	(up), %r9, %rax)
 	lea	32(up), up
 	lea	-32(rp), rp
 	lea	L(f4)(%rip), jmpreg
 	jmp	L(b5)
 
-L(f4):	mulx(	(up), %r10, %r8)
+L(f4):	X86_ENDBR
+	mulx(	(up), %r10, %r8)
 	lea	24(up), up
 	lea	-40(rp), rp
 	lea	L(f3)(%rip), jmpreg
 	jmp	L(b4)
 
-L(f3):	mulx(	(up), %r9, %rax)
+L(f3):	X86_ENDBR
+	mulx(	(up), %r9, %rax)
 	lea	16(up), up
 	lea	-48(rp), rp
 	jrcxz	L(cor)
-- 
2.24.1