arm64: Preserve FPU registers in setjmp/longjmp and arch_context_swap.

Change-Id: If0ca2ecbfa45b663dab39d1e9cb2562f071c3b27
Reviewed-on: https://review.haiku-os.org/c/haiku/+/5268
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: David Karoly <karolyd577@gmail.com>
Reviewed-by: waddlesplash <waddlesplash@gmail.com>

headers/private/kernel/arch/arm64/arch_thread_types.h

@@ -10,7 +10,8 @@
 
 
 struct arch_thread {
-	uint64			regs[13]; // x19-x30, sp
+	uint64 regs[13]; // x19-x30, sp
+	uint64 fp_regs[8]; // d8-d15
 };
 
 struct arch_team {

src/system/kernel/arch/arm64/arch_asm.S

@@ -241,6 +241,7 @@ FUNCTION(_fp_restore):
 FUNCTION_END(_fp_restore)
 
 FUNCTION(_arch_context_swap):
+	// save
 	stp x19, x20, [x0], #16
 	stp x21, x22, [x0], #16
 	stp x23, x24, [x0], #16
@@ -249,8 +250,14 @@ FUNCTION(_arch_context_swap):
 	stp x29, x30, [x0], #16
 
 	mov x2, sp
-	str x2, [x0]
+	str x2, [x0], #8
 
+	stp  d8,  d9, [x0], #16
+	stp d10, d11, [x0], #16
+	stp d12, d13, [x0], #16
+	stp d14, d15, [x0], #16
+
+	// restore
 	ldp x19, x20, [x1], #16
 	ldp x21, x22, [x1], #16
 	ldp x23, x24, [x1], #16
@@ -258,9 +265,15 @@ FUNCTION(_arch_context_swap):
 	ldp x27, x28, [x1], #16
 	ldp x29, x30, [x1], #16
 
-	ldr x2, [x1]
+	ldr x2, [x1], #8
 	mov sp, x2
 
+	ldp  d8,  d9, [x1], #16
+	ldp d10, d11, [x1], #16
+	ldp d12, d13, [x1], #16
+	ldp d14, d15, [x1], #16
+
+	// pass x29 as argument to thread entry function
 	mov x0, x29
 	ret
 FUNCTION_END(_arch_context_swap)

src/system/libroot/posix/arch/arm64/siglongjmp.S

@@ -18,9 +18,14 @@ FUNCTION(_longjmp):
 	ldp x27, x28, [x0], #16
 	ldp x29, x30, [x0], #16
 
-	ldr x2, [x0]
+	ldr x2, [x0], #8
 	mov sp, x2
 
+	ldp  d8,  d9, [x0], #16
+	ldp d10, d11, [x0], #16
+	ldp d12, d13, [x0], #16
+	ldp d14, d15, [x0], #16
+
 	b __longjmp_return
 FUNCTION_END(siglongjmp)
 FUNCTION_END(longjmp)

src/system/libroot/posix/arch/arm64/sigsetjmp.S

@@ -16,7 +16,12 @@ FUNCTION(sigsetjmp):
 	stp x29, x30, [x0], #16
 
 	mov x2, sp
-	str x2, [x0]
+	str x2, [x0], #8
+
+	stp  d8,  d9, [x0], #16
+	stp d10, d11, [x0], #16
+	stp d12, d13, [x0], #16
+	stp d14, d15, [x0], #16
 
 	b __setjmp_save_sigs
 FUNCTION_END(sigsetjmp)