[RFC PATCH v3 10/15] x86/alternatives: Record text_poke's of JUMP_TYPE_FORCEFUL labels

[Valentin Schneider]

Forceful static keys are used in early entry code where it is unsafe to
defer the sync_core() IPIs, and flagged as such via their ->type field.

Record that information when creating a text_poke_loc. The
text_poke_loc.old field is written to when first iterating a text_poke()
entry, and as such can be (ab)used to store this information at the start
of text_poke_bp_batch().

Signed-off-by: Valentin Schneider <vschneid@xxxxxxxxxx>
---
 arch/x86/include/asm/text-patching.h | 12 ++++++++++--
 arch/x86/kernel/alternative.c        | 16 ++++++++++------
 arch/x86/kernel/jump_label.c         |  7 ++++---
 3 files changed, 24 insertions(+), 11 deletions(-)

diff --git a/arch/x86/include/asm/text-patching.h b/arch/x86/include/asm/text-patching.h
index 6259f1937fe77..e34de36cab61e 100644
--- a/arch/x86/include/asm/text-patching.h
+++ b/arch/x86/include/asm/text-patching.h
@@ -38,9 +38,17 @@ extern void *text_poke_copy(void *addr, const void *opcode, size_t len);
 extern void *text_poke_copy_locked(void *addr, const void *opcode, size_t len, bool core_ok);
 extern void *text_poke_set(void *addr, int c, size_t len);
 extern int poke_int3_handler(struct pt_regs *regs);
-extern void text_poke_bp(void *addr, const void *opcode, size_t len, const void *emulate);
+extern void __text_poke_bp(void *addr, const void *opcode, size_t len, const void *emulate, bool force_ipi);
+static inline void text_poke_bp(void *addr, const void *opcode, size_t len, const void *emulate)
+{
+	__text_poke_bp(addr, opcode, len, emulate, false);
+}
 
-extern void text_poke_queue(void *addr, const void *opcode, size_t len, const void *emulate);
+extern void __text_poke_queue(void *addr, const void *opcode, size_t len, const void *emulate, bool force_ipi);
+static inline void text_poke_queue(void *addr, const void *opcode, size_t len, const void *emulate)
+{
+	__text_poke_queue(addr, opcode, len, emulate, false);
+}
 extern void text_poke_finish(void);
 
 #define INT3_INSN_SIZE		1
diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c
index d17518ca19b8b..954c4c0f7fc58 100644
--- a/arch/x86/kernel/alternative.c
+++ b/arch/x86/kernel/alternative.c
@@ -2098,7 +2098,10 @@ struct text_poke_loc {
 	u8 opcode;
 	const u8 text[POKE_MAX_OPCODE_SIZE];
 	/* see text_poke_bp_batch() */
-	u8 old;
+	union {
+		u8 old;
+		u8 force_ipi;
+	};
 };
 
 struct bp_patching_desc {
@@ -2385,7 +2388,7 @@ static void text_poke_bp_batch(struct text_poke_loc *tp, unsigned int nr_entries
 }
 
 static void text_poke_loc_init(struct text_poke_loc *tp, void *addr,
-			       const void *opcode, size_t len, const void *emulate)
+			       const void *opcode, size_t len, const void *emulate, bool force_ipi)
 {
 	struct insn insn;
 	int ret, i = 0;
@@ -2402,6 +2405,7 @@ static void text_poke_loc_init(struct text_poke_loc *tp, void *addr,
 	tp->rel_addr = addr - (void *)_stext;
 	tp->len = len;
 	tp->opcode = insn.opcode.bytes[0];
+	tp->force_ipi = force_ipi;
 
 	if (is_jcc32(&insn)) {
 		/*
@@ -2493,14 +2497,14 @@ void text_poke_finish(void)
 	text_poke_flush(NULL);
 }
 
-void __ref text_poke_queue(void *addr, const void *opcode, size_t len, const void *emulate)
+void __ref __text_poke_queue(void *addr, const void *opcode, size_t len, const void *emulate, bool force_ipi)
 {
 	struct text_poke_loc *tp;
 
 	text_poke_flush(addr);
 
 	tp = &tp_vec[tp_vec_nr++];
-	text_poke_loc_init(tp, addr, opcode, len, emulate);
+	text_poke_loc_init(tp, addr, opcode, len, emulate, force_ipi);
 }
 
 /**
@@ -2514,10 +2518,10 @@ void __ref text_poke_queue(void *addr, const void *opcode, size_t len, const voi
  * dynamically allocated memory. This function should be used when it is
  * not possible to allocate memory.
  */
-void __ref text_poke_bp(void *addr, const void *opcode, size_t len, const void *emulate)
+void __ref __text_poke_bp(void *addr, const void *opcode, size_t len, const void *emulate, bool force_ipi)
 {
 	struct text_poke_loc tp;
 
-	text_poke_loc_init(&tp, addr, opcode, len, emulate);
+	text_poke_loc_init(&tp, addr, opcode, len, emulate, force_ipi);
 	text_poke_bp_batch(&tp, 1);
 }
diff --git a/arch/x86/kernel/jump_label.c b/arch/x86/kernel/jump_label.c
index f5b8ef02d172c..e03a4f56b30fd 100644
--- a/arch/x86/kernel/jump_label.c
+++ b/arch/x86/kernel/jump_label.c
@@ -101,8 +101,8 @@ __jump_label_transform(struct jump_entry *entry,
 		text_poke_early((void *)jump_entry_code(entry), jlp.code, jlp.size);
 		return;
 	}
-
-	text_poke_bp((void *)jump_entry_code(entry), jlp.code, jlp.size, NULL);
+	__text_poke_bp((void *)jump_entry_code(entry), jlp.code, jlp.size, NULL,
+		     jump_entry_key(entry)->type & JUMP_TYPE_FORCEFUL);
 }
 
 static void __ref jump_label_transform(struct jump_entry *entry,
@@ -135,7 +135,8 @@ bool arch_jump_label_transform_queue(struct jump_entry *entry,
 
 	mutex_lock(&text_mutex);
 	jlp = __jump_label_patch(entry, type);
-	text_poke_queue((void *)jump_entry_code(entry), jlp.code, jlp.size, NULL);
+	__text_poke_queue((void *)jump_entry_code(entry), jlp.code, jlp.size, NULL,
+			jump_entry_key(entry)->type & JUMP_TYPE_FORCEFUL);
 	mutex_unlock(&text_mutex);
 	return true;
 }
-- 
2.43.0