xtensa: userspace: simplify syscall trampoline a bit

There is no need to do a call4 and jx. Simply do a callx4
is enough.

Also amended the now incorrect comment about how syscall
trampoline is set up. It is now a straight call4 instead of
the old 2x call4.

Signed-off-by: Daniel Leung <daniel.leung@intel.com>
diff --git a/arch/xtensa/core/userspace.S b/arch/xtensa/core/userspace.S
index c798e12..cd446c4 100644
--- a/arch/xtensa/core/userspace.S
+++ b/arch/xtensa/core/userspace.S
@@ -122,9 +122,14 @@
 	wur.THREADPTR a0
 #endif
 
-	/* Set syscall parameters. We have an initial call4 to set up the
-	 * the stack and then a new call4 for the syscall function itself.
-	 * So parameters should be put as if it was a call8.
+	/* Set syscall parameters by moving them into place before we do
+	 * a call4 for the syscall function itself.
+	 * arg1 = a6
+	 * arg2 = a3 (clobbered above, so we need to reload it)
+	 * arg3 = a4
+	 * arg4 = a5
+	 * arg5 = a8
+	 * arg6 = a9
 	 */
 	mov a10, a8
 	mov a11, a9
@@ -141,7 +146,7 @@
 	wsr.ps a0
 	rsync
 
-	call4 _syscall_call0
+	callx4 a2
 
 	/* copy return value. Lets put it in the top of stack
 	 * because registers will be clobbered in
@@ -152,12 +157,6 @@
 
 	j _syscall_returned
 
-.align 4
-_syscall_call0:
-	/* We want an ENTRY to set a bit in windowstart */
-	jx a2
-
-
 _syscall_returned:
 	call0 xtensa_restore_high_regs