x86/static-call: provide a way to do very early static-call updates

Add static_call_update_early() for updating static-call targets in
very early boot.

This will be needed for support of Xen guest type specific hypercall
functions.

This is part of XSA-466 / CVE-2024-53241.

Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Juergen Gross <jgross@suse.com>
Co-developed-by: Peter Zijlstra <peterz@infradead.org>
Co-developed-by: Josh Poimboeuf <jpoimboe@redhat.com>

Author: jgross1

arch/x86/include/asm/static_call.h

@@ -65,4 +65,19 @@
 
 extern bool __static_call_fixup(void *tramp, u8 op, void *dest);
 
+extern void __static_call_update_early(void *tramp, void *func);
+
+#define static_call_update_early(name, _func)				\
+({									\
+	typeof(&STATIC_CALL_TRAMP(name)) __F = (_func);			\
+	if (static_call_initialized) {					\
+		__static_call_update(&STATIC_CALL_KEY(name),		\
+				     STATIC_CALL_TRAMP_ADDR(name), __F);\
+	} else {							\
+		WRITE_ONCE(STATIC_CALL_KEY(name).func, _func);		\
+		__static_call_update_early(STATIC_CALL_TRAMP_ADDR(name),\
+					   __F);			\
+	}								\
+})
+
 #endif /* _ASM_STATIC_CALL_H */

arch/x86/include/asm/sync_core.h

@@ -8,7 +8,7 @@
 #include <asm/special_insns.h>
 
 #ifdef CONFIG_X86_32
-static inline void iret_to_self(void)
+static __always_inline void iret_to_self(void)
 {
 	asm volatile (
 		"pushfl\n\t"
@@ -19,7 +19,7 @@ static inline void iret_to_self(void)
 		: ASM_CALL_CONSTRAINT : : "memory");
 }
 #else
-static inline void iret_to_self(void)
+static __always_inline void iret_to_self(void)
 {
 	unsigned int tmp;
 
@@ -55,7 +55,7 @@ static inline void iret_to_self(void)
  * Like all of Linux's memory ordering operations, this is a
  * compiler barrier as well.
  */
-static inline void sync_core(void)
+static __always_inline void sync_core(void)
 {
 	/*
 	 * The SERIALIZE instruction is the most straightforward way to

arch/x86/kernel/static_call.c

@@ -172,6 +172,15 @@ void arch_static_call_transform(void *site, void *tramp, void *func, bool tail)
 }
 EXPORT_SYMBOL_GPL(arch_static_call_transform);
 
+noinstr void __static_call_update_early(void *tramp, void *func)
+{
+	BUG_ON(system_state != SYSTEM_BOOTING);
+	BUG_ON(!early_boot_irqs_disabled);
+	BUG_ON(static_call_initialized);
+	__text_gen_insn(tramp, JMP32_INSN_OPCODE, tramp, func, JMP32_INSN_SIZE);
+	sync_core();
+}
+
 #ifdef CONFIG_MITIGATION_RETHUNK
 /*
  * This is called by apply_returns() to fix up static call trampolines,

include/linux/compiler.h

@@ -216,28 +216,43 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
 
 #endif /* __KERNEL__ */
 
+/**
+ * offset_to_ptr - convert a relative memory offset to an absolute pointer
+ * @off:	the address of the 32-bit offset value
+ */
+static inline void *offset_to_ptr(const int *off)
+{
+	return (void *)((unsigned long)off + *off);
+}
+
+#endif /* __ASSEMBLY__ */
+
+#ifdef CONFIG_64BIT
+#define ARCH_SEL(a,b) a
+#else
+#define ARCH_SEL(a,b) b
+#endif
+
 /*
  * Force the compiler to emit 'sym' as a symbol, so that we can reference
  * it from inline assembler. Necessary in case 'sym' could be inlined
  * otherwise, or eliminated entirely due to lack of references that are
  * visible to the compiler.
  */
-#define ___ADDRESSABLE(sym, __attrs) \
-	static void * __used __attrs \
+#define ___ADDRESSABLE(sym, __attrs)						\
+	static void * __used __attrs						\
 	__UNIQUE_ID(__PASTE(__addressable_,sym)) = (void *)(uintptr_t)&sym;
+
 #define __ADDRESSABLE(sym) \
 	___ADDRESSABLE(sym, __section(".discard.addressable"))
 
-/**
- * offset_to_ptr - convert a relative memory offset to an absolute pointer
- * @off:	the address of the 32-bit offset value
- */
-static inline void *offset_to_ptr(const int *off)
-{
-	return (void *)((unsigned long)off + *off);
-}
+#define __ADDRESSABLE_ASM(sym)						\
+	.pushsection .discard.addressable,"aw";				\
+	.align ARCH_SEL(8,4);						\
+	ARCH_SEL(.quad, .long) __stringify(sym);			\
+	.popsection;
 
-#endif /* __ASSEMBLY__ */
+#define __ADDRESSABLE_ASM_STR(sym) __stringify(__ADDRESSABLE_ASM(sym))
 
 #ifdef __CHECKER__
 #define __BUILD_BUG_ON_ZERO_MSG(e, msg) (0)

include/linux/static_call.h

@@ -138,6 +138,7 @@
 #ifdef CONFIG_HAVE_STATIC_CALL
 #include <asm/static_call.h>
 
+extern int static_call_initialized;
 /*
  * Either @site or @tramp can be NULL.
  */

kernel/static_call_inline.c

@@ -15,7 +15,7 @@ extern struct static_call_site __start_static_call_sites[],
 extern struct static_call_tramp_key __start_static_call_tramp_key[],
 				    __stop_static_call_tramp_key[];
 
-static int static_call_initialized;
+int static_call_initialized;
 
 /*
  * Must be called before early_initcall() to be effective.