module: freebsd: fix aarch64 fpu handling

Just like x86, aarch64 needs to use the fpu_kern(9) API around FPU
usage, otherwise we panic promptly at boot as soon as ZFS attempts to
do checksum benchmarking.

Note that the build is broken both before and after this commit in the
openzfs repo... there's a non-trivial conflict of AT_UID/AT_GID in
sys/elf_common.h.  In FreeBSD, we solved it with a quick hack in
spl/sys/vnode.h to undef AT_UID/AT_GID before, but in the standalone
build the conflict is in the other direction.

It's not clear how to best handle that; presumably we don't want the
elf_common.h definition of AT_UID/AT_GID in ZFS code, but to solve it we
would need to undef these before including machine/elf.h in
spl/sys/simd_aarch64.h.  Punting on this for now.

Signed-off-by: Kyle Evans <kevans@FreeBSD.org>

include/os/freebsd/spl/sys/simd_aarch64.h

@@ -44,12 +44,23 @@
 #define	_FREEBSD_SIMD_AARCH64_H
 
 #include <sys/types.h>
+#include <sys/ucontext.h>
 #include <machine/elf.h>
+#include <machine/fpu.h>
+#include <machine/md_var.h>
+#include <machine/pcb.h>
 
 #define	kfpu_allowed()		1
 #define	kfpu_initialize(tsk)	do {} while (0)
-#define	kfpu_begin()		do {} while (0)
-#define	kfpu_end()		do {} while (0)
+#define	kfpu_begin() do {						\
+	if (__predict_false(!is_fpu_kern_thread(0)))			\
+		fpu_kern_enter(curthread, NULL, FPU_KERN_NOCTX);	\
+} while(0)
+
+#define	kfpu_end() do {							\
+	if (__predict_false(curthread->td_pcb->pcb_fpflags & PCB_FP_NOSAVE))	\
+		fpu_kern_leave(curthread, NULL);			\
+} while(0)
 #define	kfpu_init()		(0)
 #define	kfpu_fini()		do {} while (0)