ftrace: Synchronize the initialization of ftrace_opsgrafted

There's some internal state that ftrace_ops needs to have set, but since
it can be declared outside of the ftrace.c code, it calls
ftrace_ops_init() on the ops in every global function. The issue is that
if two tasks call it on the same ops at the same time it is possible to
have the initialization of one corrupt the initialization of the other
call.

Create a ops_mutex to use to synchronize every initialization of the
ftrace_ops. The mutex is taken within checking the ftrace_ops flag that
states it was initializied but the flag is checked again after the mutex
has been taken. Checking first outside the mutex allows it to shortcut
having to take the mutex. But then the check needs to be done again after
the mute is taken in case of races.

Cc: stable@vger.kernel.org
Link: https://patch.msgid.link/20260902095501.6b59af20@gandalf.local.home
Fixes: f04f24fb7e48d ("ftrace, kprobes: Fix a deadlock on ftrace_regex_lock")
Reported-by: sashiko-bot@kernel.org
Close: https://lore.kernel.org/all/20260829025528.49A831F000E9@smtp.kernel.org/
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

1 files changed, 95 insertions, 0 deletions

diff --git a/arch/arm/lib/putuser.S b/arch/arm/lib/putuser.S
new file mode 100644
index 000000000..bdd8836dc
--- /dev/null
+++ b/arch/arm/lib/putuser.S
@@ -0,0 +1,95 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ *  linux/arch/arm/lib/putuser.S
+ *
+ *  Copyright (C) 2001 Russell King
+ *
+ *  Idea from x86 version, (C) Copyright 1998 Linus Torvalds
+ *
+ * These functions have a non-standard call interface to make
+ * them more efficient, especially as they return an error
+ * value in addition to the "real" return value.
+ *
+ * __put_user_X
+ *
+ * Inputs:	r0 contains the address
+ *		r1 contains the address limit, which must be preserved
+ *		r2, r3 contains the value
+ * Outputs:	r0 is the error code
+ *		lr corrupted
+ *
+ * No other registers must be altered.  (see <asm/uaccess.h>
+ * for specific ASM register usage).
+ *
+ * Note that ADDR_LIMIT is either 0 or 0xc0000000
+ * Note also that it is intended that __put_user_bad is not global.
+ */
+#include <linux/linkage.h>
+#include <asm/assembler.h>
+#include <asm/errno.h>
+#include <asm/domain.h>
+
+ENTRY(__put_user_1)
+	check_uaccess r0, 1, r1, ip, __put_user_bad
+1: TUSER(strb)	r2, [r0]
+	mov	r0, #0
+	ret	lr
+ENDPROC(__put_user_1)
+
+ENTRY(__put_user_2)
+	check_uaccess r0, 2, r1, ip, __put_user_bad
+#if __LINUX_ARM_ARCH__ >= 6
+
+2: TUSER(strh)	r2, [r0]
+
+#else
+
+	mov	ip, r2, lsr #8
+#ifndef __ARMEB__
+2: TUSER(strb)	r2, [r0], #1
+3: TUSER(strb)	ip, [r0]
+#else
+2: TUSER(strb)	ip, [r0], #1
+3: TUSER(strb)	r2, [r0]
+#endif
+
+#endif /* __LINUX_ARM_ARCH__ >= 6 */
+	mov	r0, #0
+	ret	lr
+ENDPROC(__put_user_2)
+
+ENTRY(__put_user_4)
+	check_uaccess r0, 4, r1, ip, __put_user_bad
+4: TUSER(str)	r2, [r0]
+	mov	r0, #0
+	ret	lr
+ENDPROC(__put_user_4)
+
+ENTRY(__put_user_8)
+	check_uaccess r0, 8, r1, ip, __put_user_bad
+#ifdef CONFIG_THUMB2_KERNEL
+5: TUSER(str)	r2, [r0]
+6: TUSER(str)	r3, [r0, #4]
+#else
+5: TUSER(str)	r2, [r0], #4
+6: TUSER(str)	r3, [r0]
+#endif
+	mov	r0, #0
+	ret	lr
+ENDPROC(__put_user_8)
+
+__put_user_bad:
+	mov	r0, #-EFAULT
+	ret	lr
+ENDPROC(__put_user_bad)
+
+.pushsection __ex_table, "a"
+	.long	1b, __put_user_bad
+	.long	2b, __put_user_bad
+#if __LINUX_ARM_ARCH__ < 6
+	.long	3b, __put_user_bad
+#endif
+	.long	4b, __put_user_bad
+	.long	5b, __put_user_bad
+	.long	6b, __put_user_bad
+.popsection