Skip to contentSkip to conversation

x86/bugs: Only log missing retpoline when it's actually the missing mitigation

PATCH In review Kim Phillips 2 participants 3 messages Aug 26 – Aug 31

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86/bugs: Only log missing retpoline when it's actually the missing mitigation

spectre_v2_select_retpoline() unconditionally emits

"Kernel not compiled with retpoline; no mitigation available!"

when the kernel lacks retpoline support, before returning SPECTRE_V2_NONE to its callers. SPECTRE_V2_NONE is not the final answer, though:
spectre_v2_update_mitigation() runs afterwards and can still upgrade the mitigation.

On an Intel CPU affected by RETBleed, built with CONFIG_MITIGATION_IBRS_ENTRY=y and CONFIG_MITIGATION_RETPOLINE=n, and booted with spectre_v2=auto, retbleed_select_mitigation() picks RETBLEED_MITIGATION_IBRS, and spectre_v2_update_mitigation() then promotes SPECTRE_V2_NONE to SPECTRE_V2_IBRS. The boot log claims no mitigation is available immediately before reporting "Spectre V2 : Mitigation: IBRS", and sysfs reports the machine as mitigated. The message is alarming and simply wrong.

Drop the pr_err from the helper and emit it once from spectre_v2_update_mitigation(), after the final mitigation is known, so it only fires when the system really is left unmitigated. Guard it on !IS_ENABLED(CONFIG_MITIGATION_RETPOLINE) so it fires only when retpoline truly cannot be built in, and restrict it to the cases where retpoline was the implied choice: SPECTRE_V2_CMD_FORCE, or SPECTRE_V2_CMD_AUTO when should_mitigate_vuln(X86_BUG_SPECTRE_V2) indicates a mitigation was actually intended.

Fixes 480e803dacf8 ("x86/bugs: Restructure spectre_v2 mitigation")
Assisted-by ClaudeCode:claude-opus-4-7
---
 arch/x86/kernel/cpu/bugs.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
arch/x86/kernel/cpu/bugs.c +8 -3
index 82436b3534fa..46c68feef339 100644
@@ -1956,10 +1956,8 @@ early_param("spectre_v2", spectre_v2_parse_cmdline);
 
 static enum spectre_v2_mitigation __init spectre_v2_select_retpoline(void)
 {
-	if (!IS_ENABLED(CONFIG_MITIGATION_RETPOLINE)) {
-		pr_err("Kernel not compiled with retpoline; no mitigation available!");
+	if (!IS_ENABLED(CONFIG_MITIGATION_RETPOLINE))
 		return SPECTRE_V2_NONE;
-	}
 
 	return SPECTRE_V2_RETPOLINE;
 }
@@ -2247,6 +2245,13 @@ static void __init spectre_v2_update_mitigation(void)
 		}
 	}
 
+	if (!IS_ENABLED(CONFIG_MITIGATION_RETPOLINE) &&
+	    spectre_v2_enabled == SPECTRE_V2_NONE &&
+	    (spectre_v2_cmd == SPECTRE_V2_CMD_FORCE ||
+	     (spectre_v2_cmd == SPECTRE_V2_CMD_AUTO &&
+	      should_mitigate_vuln(X86_BUG_SPECTRE_V2))))
+		pr_err("Kernel not compiled with retpoline; no mitigation available!");
+
 	if (boot_cpu_has_bug(X86_BUG_SPECTRE_V2))
 		pr_info("%s\n", spectre_v2_strings[spectre_v2_enabled]);
 }

base-commit: 8d3ae59288f1e7d58d76558a6ee96d533bc5019f

Signature · 1 line
2.43.0
reviewer
#1 lore raw
Quoted 40 lines · Kim Phillips

spectre_v2_select_retpoline() unconditionally emits

"Kernel not compiled with retpoline; no mitigation available!"

when the kernel lacks retpoline support, before returning SPECTRE_V2_NONE to its callers. SPECTRE_V2_NONE is not the final answer, though:
spectre_v2_update_mitigation() runs afterwards and can still upgrade the mitigation.

On an Intel CPU affected by RETBleed, built with CONFIG_MITIGATION_IBRS_ENTRY=y and CONFIG_MITIGATION_RETPOLINE=n, and booted with spectre_v2=auto, retbleed_select_mitigation() picks RETBLEED_MITIGATION_IBRS, and spectre_v2_update_mitigation() then promotes SPECTRE_V2_NONE to SPECTRE_V2_IBRS. The boot log claims no mitigation is available immediately before reporting "Spectre V2 : Mitigation: IBRS", and sysfs reports the machine as mitigated. The message is alarming and simply wrong.

Drop the pr_err from the helper and emit it once from spectre_v2_update_mitigation(), after the final mitigation is known, so it only fires when the system really is left unmitigated. Guard it on !IS_ENABLED(CONFIG_MITIGATION_RETPOLINE) so it fires only when retpoline truly cannot be built in, and restrict it to the cases where retpoline was the implied choice: SPECTRE_V2_CMD_FORCE, or SPECTRE_V2_CMD_AUTO when should_mitigate_vuln(X86_BUG_SPECTRE_V2) indicates a mitigation was actually intended.

Fixes 480e803dacf8 ("x86/bugs: Restructure spectre_v2 mitigation")
Assisted-by ClaudeCode:claude-opus-4-7
---
 arch/x86/kernel/cpu/bugs.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
arch/x86/kernel/cpu/bugs.c +0 -0
index 82436b3534fa..46c68feef339 100644

arch/x86/kernel/cpu/bugs.c early_param("spectre_v2", spectre_v2_parse_cmdline); · line 1960

Quoted 5 lines
arch/x86/kernel/cpu/bugs.c +0 -2
@@ -1956,10 +1956,8 @@ early_param("spectre_v2", spectre_v2_parse_cmdline);
 
 static enum spectre_v2_mitigation __init spectre_v2_select_retpoline(void)
 {
-	if (!IS_ENABLED(CONFIG_MITIGATION_RETPOLINE)) {
-		pr_err("Kernel not compiled with retpoline; no mitigation available!");

The claim "No mitigation available" is really the problem. And, it is redundant here because the final mitigation state is anyways printed in spectre_v2_update_mitigation().

How about just change the message to:

	pr_err("Kernel not compiled with retpoline");

It stays accurate here, telling retpoline mitigation was attempted.

Quoted 7 lines
arch/x86/kernel/cpu/bugs.c +1 -1
+	if (!IS_ENABLED(CONFIG_MITIGATION_RETPOLINE))
 		return SPECTRE_V2_NONE;
-	}
 
 	return SPECTRE_V2_RETPOLINE;
 }

arch/x86/kernel/cpu/bugs.c static void __init spectre_v2_update_mitigation(void) · line 2254

Quoted 10 lines
arch/x86/kernel/cpu/bugs.c +7 -0
@@ -2247,6 +2245,13 @@ static void __init spectre_v2_update_mitigation(void)
 		}
 	}
 
+	if (!IS_ENABLED(CONFIG_MITIGATION_RETPOLINE) &&
+	    spectre_v2_enabled == SPECTRE_V2_NONE &&
+	    (spectre_v2_cmd == SPECTRE_V2_CMD_FORCE ||
+	     (spectre_v2_cmd == SPECTRE_V2_CMD_AUTO &&
+	      should_mitigate_vuln(X86_BUG_SPECTRE_V2))))
+		pr_err("Kernel not compiled with retpoline; no mitigation available!");
+

With the above message tweak, this check can be dropped.

Quoted 3 lines
arch/x86/kernel/cpu/bugs.c +0 -0
 	if (boot_cpu_has_bug(X86_BUG_SPECTRE_V2))
 		pr_info("%s\n", spectre_v2_strings[spectre_v2_enabled]);
 }
#2 lore raw
Quoted 13 lines · Pawan Gupta

arch/x86/kernel/cpu/bugs.c early_param("spectre_v2", spectre_v2_parse_cmdline); · line 1960

Quoted 7 lines · Kim Phillips

+++ b/arch/x86/kernel/cpu/bugs.c

arch/x86/kernel/cpu/bugs.c +0 -2
@@ -1956,10 +1956,8 @@ early_param("spectre_v2", spectre_v2_parse_cmdline);
 
  static enum spectre_v2_mitigation __init spectre_v2_select_retpoline(void)
  {
-	if (!IS_ENABLED(CONFIG_MITIGATION_RETPOLINE)) {
-		pr_err("Kernel not compiled with retpoline; no mitigation available!");

The claim "No mitigation available" is really the problem. And, it is redundant here because the final mitigation state is anyways printed in spectre_v2_update_mitigation().

Quoted 5 lines

How about just change the message to:

	pr_err("Kernel not compiled with retpoline");

It stays accurate here, telling retpoline mitigation was attempted.

Thanks for your review. I've made these changes and reworded the commit text and submitted a v2 here:

https://lore.kernel.org/lkml/[email protected]/

Thanks,

Kim