[<prev] [next>] [day] [month] [year] [list]
Message-ID: <20251003221030.GA1162775@ax162>
Date: Fri, 3 Oct 2025 15:10:30 -0700
From: Nathan Chancellor <nathan@...nel.org>
To: Ingo Molnar <mingo@...nel.org>, x86@...nel.org
Cc: Kees Cook <kees@...nel.org>,
Linus Torvalds <torvalds@...ux-foundation.org>,
linux-kernel@...r.kernel.org
Subject: Semantic conflict around CONFIG_CFI_CLANG between Linus's tree and
x86/core
Hi Ingo and other x86 folks,
Just a heads up, there will be a semantic conflict (for somewhat obvious
reasons) between
23ef9d439769 ("kcfi: Rename CONFIG_CFI_CLANG to CONFIG_CFI")
that went into Linus's tree via the hardening tree and
038c7dc66e27 ("compiler_types.h: Move __nocfi out of compiler-specific header")
which is currently in x86/core. I noticed this while kexec'ing under a
CFI kernel, which needs __nocfi as a result of
2114796ca041 ("x86/kexec: Mark machine_kexec() with __nocfi")
otherwise there is a nice big splat and the machine dies :)
Cheers,
Nathan
diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index 41c16fb8eb40..59288a2c1ad2 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h
@@ -455,7 +455,7 @@ struct ftrace_likely_data {
# define __noscs
#endif
-#if defined(CONFIG_CFI_CLANG)
+#if defined(CONFIG_CFI)
# define __nocfi __attribute__((__no_sanitize__("kcfi")))
#else
# define __nocfi
Powered by blists - more mailing lists