[PATCH] x86/mce: Report unavailable MCE support at informational level

[PATCH] x86/mce: Report unavailable MCE support at informational level

[Dexuan Cui]

mcheck_init_device() reports an error and returns -EIO when
mce_available() is false. However, MCE support being unavailable does
not necessarily indicate an initialization failure.

For example, a TDX guest with TD_CTLS.REDUCE_VE set and
FEATURE_PARAVIRT_CTLS.MCA clear does not enumerate the MCE and MCA
CPUID capabilities. This is an expected configuration, but it causes
the kernel to print:

  mce: Unable to init MCE device (rc: -5)

MCE can also be unavailable because it was explicitly disabled with
mce=off or nomce. When it was disabled due to an earlier initialization
problem, the relevant path already reports the reason.

Report MCE unavailability at informational level and return -ENODEV
directly instead of routing it through the common error path.

Signed-off-by: Dexuan Cui <decui@microsoft.com>
---
 arch/x86/kernel/cpu/mce/core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
index f3a793e3a6c8..24ac10075263 100644
--- a/arch/x86/kernel/cpu/mce/core.c
+++ b/arch/x86/kernel/cpu/mce/core.c
@@ -2825,8 +2825,8 @@ static __init int mcheck_init_device(void)
 	MAYBE_BUILD_BUG_ON(__VIRTUAL_MASK_SHIFT >= 63);
 
 	if (!mce_available(&boot_cpu_data)) {
-		err = -EIO;
-		goto err_out;
+		pr_info("MCE is unavailable\n");
+		return -ENODEV;
 	}
 
 	if (!zalloc_cpumask_var(&mce_device_initialized, GFP_KERNEL)) {
-- 
2.52.0

Re: [PATCH] x86/mce: Report unavailable MCE support at informational level

[Naman Jain]

On Wed, Sep 23, 2026 at 10:30:33AM -0700, Dexuan Cui wrote:
> mcheck_init_device() reports an error and returns -EIO when
> mce_available() is false. However, MCE support being unavailable does
> not necessarily indicate an initialization failure.
> For example, a TDX guest with TD_CTLS.REDUCE_VE set and
> FEATURE_PARAVIRT_CTLS.MCA clear does not enumerate the MCE and MCA
> CPUID capabilities. This is an expected configuration, but it causes
> the kernel to print:
>
>   mce: Unable to init MCE device (rc: -5)
>
> MCE can also be unavailable because it was explicitly disabled with
> mce=off or nomce. When it was disabled due to an earlier initialization
> problem, the relevant path already reports the reason.
> Report MCE unavailability at informational level and return -ENODEV
> directly instead of routing it through the common error path.
>
> Signed-off-by: Dexuan Cui <decui@microsoft.com>
> ---
>  arch/x86/kernel/cpu/mce/core.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/kernel/cpu/mce/core.c b/arch/x86/kernel/cpu/mce/core.c
> index f3a793e3a6c8..24ac10075263 100644
> --- a/arch/x86/kernel/cpu/mce/core.c
> +++ b/arch/x86/kernel/cpu/mce/core.c
> @@ -2825,8 +2825,8 @@ static __init int mcheck_init_device(void)
>  	MAYBE_BUILD_BUG_ON(__VIRTUAL_MASK_SHIFT >= 63);
>
>  	if (!mce_available(&boot_cpu_data)) {
> -		err = -EIO;
> -		goto err_out;
> +		pr_info("MCE is unavailable\n");
> +		return -ENODEV;
>  	}
>
>  	if (!zalloc_cpumask_var(&mce_device_initialized, GFP_KERNEL)) {

LGTM.

Reviewed-by: Naman Jain <namjain@linux.microsoft.com>

Regards,
Naman Jain