x86/paravirt: Detect over-sized patching bugs in paravirt_patch_call()

paravirt_patch_call() currently handles patching failures inconsistently:
we generate a warning in the retpoline case, but don't in other cases where
we might end up with a non-working kernel as well.

So just convert it all to a BUG_ON(), these patching calls are *not* supposed
to fail, and if they do we want to know it immediately.

This also makes the kernel smaller and removes an #ifdef ugly.

I tried it with a richly paravirt-enabled kernel and no patching bugs
were detected.

Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20190425095039.GC115378@gmail.comSigned-off-by: NIngo Molnar <mingo@kernel.org>

arch/x86/kernel/paravirt.c

@@ -73,21 +73,21 @@ struct branch {
 static unsigned paravirt_patch_call(void *insnbuf, const void *target,
 				    unsigned long addr, unsigned len)
 {
+	const int call_len = 5;
 	struct branch *b = insnbuf;
-	unsigned long delta = (unsigned long)target - (addr+5);
+	unsigned long delta = (unsigned long)target - (addr+call_len);
 
-	if (len < 5) {
-#ifdef CONFIG_RETPOLINE
-		WARN_ONCE(1, "Failing to patch indirect CALL in %ps\n", (void *)addr);
-#endif
-		return len;	/* call too long for patch site */
+	if (len < call_len) {
+		pr_warn("paravirt: Failed to patch indirect CALL at %ps\n", (void *)addr);
+		/* Kernel might not be viable if patching fails, bail out: */
+		BUG_ON(1);
 	}
 
 	b->opcode = 0xe8; /* call */
 	b->delta = delta;
-	BUILD_BUG_ON(sizeof(*b) != 5);
+	BUILD_BUG_ON(sizeof(*b) != call_len);
 
-	return 5;
+	return call_len;
 }
 
 #ifdef CONFIG_PARAVIRT_XXL