[PATCH 1/8] x86/microcode/intel: Reject problematic loading on GNR systems

[PATCH 1/8] x86/microcode/intel: Reject problematic loading on GNR systems

[Chang S. Bae]

Revision 0x1000405 contains internal microcode changes that are required
by subsequent revisions to avoid #MC during loading. This dependency
logically fits the minimum revision requirement.

The minimum revision check, however, currently applies only to the late
loading path, since the dependency was primarily intended for OS-visible
changes. The early loading path is therefore still vulnerable to this
issue.

Furthermore, one of the subsequent revisions does not correctly specify
the minimum revision, so unfortunately the late loading cannot rely on
that check either in this case.

Prevent loading 0x1000405 or later when the system has not yet been
updated to 0x1000405 or later. Apply this blocking to both early- and
late-loading paths.

Rename is_blacklisted() to is_late_loading_denied() so the new function
that covers both loading paths is not confused with the late-load only
one.

Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Cc: <stable@vger.kernel.org>
---
The GNR errata page:
https://edc.intel.com/content/www/jp/ja/design/products-and-solutions/processors-and-chipsets/birch-stream/xeon-6900-6700-6500-series-processors-with-p-cores-specification-update/016US/errata-summary-table/

Thanks to Sohil, I noticed the naming guideline in
Documentation/process/coding-style.rst:
    For symbol names and documentation, avoid introducing new usage of
    'master / slave' (or 'slave' independent of 'master') and 'blacklist /
    whitelist'.
---
 arch/x86/kernel/cpu/microcode/intel.c | 42 +++++++++++++++++++++++++--
 1 file changed, 40 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
index 1142183c950c..c502138fd8e9 100644
--- a/arch/x86/kernel/cpu/microcode/intel.c
+++ b/arch/x86/kernel/cpu/microcode/intel.c
@@ -309,6 +309,38 @@ static void save_microcode_patch(struct microcode_intel *patch)
 		pr_err("Unable to allocate microcode memory size: %u\n", size);
 }
 
+static bool is_loading_denied(struct cpu_signature *sig, u32 rev)
+{
+	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
+
+	/*
+	 * Revision 0x1000405 contains prerequisite changes for subsequent
+	 * microcode updates on Granite Rapids systems. Updates directly from
+	 * an older revision to this or a newer one can result in #MC (GNR98).
+	 *
+	 * This dependency can be indicated from the minimum revision field.
+	 * However, revision 0x1000423 has an incorrect minimum revision in its
+	 * header (GNR101).
+	 *
+	 * Prevent loading 0x1000405 or later unless the CPU has already been
+	 * updated to 0x1000405 or later.
+	 */
+	if (vfm == INTEL_GRANITERAPIDS_X &&
+	    x86_stepping(sig->sig) == 1 &&
+	    sig->pf & 0x95 &&
+	    sig->rev < 0x1000405 &&
+	    rev >= 0x1000405) {
+		if (rev == 0x1000405)
+			pr_err_once("Erratum GNR98: 0x1000405 is not loadable.\n");
+		else
+			pr_err_once("Erratum GNR98: 0x1000405 is required before 0x%x.\n", rev);
+		pr_err_once("Please update the system BIOS or firmware.\n");
+		return true;
+	}
+
+	return false;
+}
+
 /* Scan blob for microcode matching the boot CPUs family, model, stepping */
 static __init struct microcode_intel *scan_microcode(void *data, size_t size,
 						     struct ucode_cpu_info *uci,
@@ -330,6 +362,9 @@ static __init struct microcode_intel *scan_microcode(void *data, size_t size,
 		if (!intel_find_matching_signature(data, &uci->cpu_sig))
 			continue;
 
+		if (is_loading_denied(&uci->cpu_sig, mc_header->rev))
+			continue;
+
 		/*
 		 * For saving the early microcode, find the matching revision which
 		 * was loaded on the BSP.
@@ -878,6 +913,9 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
 		if (!intel_find_matching_signature(mc, &uci->cpu_sig))
 			continue;
 
+		if (is_loading_denied(&uci->cpu_sig, mc_header.rev))
+			continue;
+
 		is_safe = ucode_validate_minrev(&mc_header);
 		if (force_minrev && !is_safe)
 			continue;
@@ -905,7 +943,7 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
 	return UCODE_ERROR;
 }
 
-static bool is_blacklisted(unsigned int cpu)
+static bool is_late_loading_denied(unsigned int cpu)
 {
 	struct cpuinfo_x86 *c = &cpu_data(cpu);
 
@@ -936,7 +974,7 @@ static enum ucode_state request_microcode_fw(int cpu, struct device *device)
 	struct kvec kvec;
 	char name[30];
 
-	if (is_blacklisted(cpu))
+	if (is_late_loading_denied(cpu))
 		return UCODE_NFOUND;
 
 	sprintf(name, "intel-ucode/%02x-%02x-%02x",
-- 
2.53.0

Re: [PATCH 1/8] x86/microcode/intel: Reject problematic loading on GNR systems

[Sohil Mehta]

On 9/1/2026 4:16 PM, Chang S. Bae wrote:
> Revision 0x1000405 contains internal microcode changes that are required
> by subsequent revisions to avoid #MC during loading. 

Before going to the solution space, can we add some more
context/background here?

For example:

GNR revision 0x1000405 introduces a breaking change that causes a #MC
when the OS updates the microcode from any version older than 0x1000405
to a newer version. This is applicable to late load as well as early
loading during boot.

A BIOS or firmware update is required to update the revision to
0x1000405 or newer before any OS updates can be safely loaded.
This dependency ...

> This dependency logically fits the minimum revision requirement.
> 
> The minimum revision check, however, currently applies only to the late
> loading path, since the dependency was primarily intended for OS-visible
> changes. The early loading path is therefore still vulnerable to this
> issue.
> 
> Furthermore, one of the subsequent revisions does not correctly specify
> the minimum revision, so unfortunately the late loading cannot rely on
> that check either in this case.
> 
> Prevent loading 0x1000405 or later when the system has not yet been
> updated to 0x1000405 or later. 

This is a bit confusing. Should it say jumping from any anything older
than 0x1000405 to anything newer?

> Apply this blocking to both early- and late-loading paths.
> 

> diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
> index 1142183c950c..c502138fd8e9 100644
> --- a/arch/x86/kernel/cpu/microcode/intel.c
> +++ b/arch/x86/kernel/cpu/microcode/intel.c
> @@ -309,6 +309,38 @@ static void save_microcode_patch(struct microcode_intel *patch)
>  		pr_err("Unable to allocate microcode memory size: %u\n", size);
>  }
>  
> +static bool is_loading_denied(struct cpu_signature *sig, u32 rev)
> +{
> +	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
> +
> +	/*
> +	 * Revision 0x1000405 contains prerequisite changes for subsequent
> +	 * microcode updates on Granite Rapids systems. Updates directly from
> +	 * an older revision to this or a newer one can result in #MC (GNR98).
> +	 *
> +	 * This dependency can be indicated from the minimum revision field.
> +	 * However, revision 0x1000423 has an incorrect minimum revision in its
> +	 * header (GNR101).
> +	 *
> +	 * Prevent loading 0x1000405 or later unless the CPU has already been
> +	 * updated to 0x1000405 or later.
> +	 */
> +	if (vfm == INTEL_GRANITERAPIDS_X &&
> +	    x86_stepping(sig->sig) == 1 &&
> +	    sig->pf & 0x95 &&
> +	    sig->rev < 0x1000405 &&
> +	    rev >= 0x1000405) {

I don't think we have a helper that can be used here directly. Also,
this is tagged for stable so adding a new one probably doesn't make sense.

But, 0x1000405 is repeated way too many times in this function :)
At a minimum, can we add something like this?

#define GNR98_UCODE_MIN_REV 	0x1000405

Maybe add defines for the PF as well?


> +		if (rev == 0x1000405)
> +			pr_err_once("Erratum GNR98: 0x1000405 is not loadable.\n");

						   ^ revision

With early load, his is probably one of the first messages that folks
will see but it will only be displayed on platforms that have the issue
until they update their microcode. Should we be more verbose here?


> +		else
> +			pr_err_once("Erratum GNR98: 0x1000405 is required before 0x%x.\n", rev);

"Erratum GNR98: revision 0x1000405 is required before revision 0x%x can
be loaded. \n"

Should this be 0x1000405 *or later* ? We don't want common users to
specifically try to find the revision 0x1000405 and load it, right?

I am wondering what is the use of the if-else? Is that intended to guide
the late loading users? I think for other users the second message would
be confusing.


> +		pr_err_once("Please update the system BIOS or firmware.\n");
> +		return true;
> +	}
> +
> +	return false;
> +}
> +

RE: [PATCH 1/8] x86/microcode/intel: Reject problematic loading on GNR systems

[Van De Ven, Arjan]

> On 9/1/2026 4:16 PM, Chang S. Bae wrote:
> > Revision 0x1000405 contains internal microcode changes that are required
> > by subsequent revisions to avoid #MC during loading.
> 
> Before going to the solution space, can we add some more
> context/background here?

I think, taking as step back, the world is sort of simple and more examples maybe helping or maybe not.

At the request of "linux folks" (many on the CC), Intel added a "minrev" field for the microcode files, to express a "don't load this unless you are at least at version XYZ", for various compatibility reasons.

At the time, we added this as a check to "runtime" loading, but not during early loading. In hindsight this was a mistake, compatibility issues are not limited to late runtime loading, but can also happen during earlier-in-the-OS loading.

The example mentioned is a case where this became no longer theoretical.... 

Now, these kind of compat breaks are somewhat rare (we can debate what rare means I suppose), but clearly they happen (the bug mentioned is clear existence proof).

The more detailed nicer solution is in later patches -- this patch just makes the kernel honor the minrev field more consistently, which gets the kernel into a "safe" space again. The reason this patch is first is one of -stable backports. One can imagine that -stable folks like to take the most simple solution back first before even considering a more complex solution.... 

Re: [PATCH 1/8] x86/microcode/intel: Reject problematic loading on GNR systems

[Chang S. Bae]

On 9/2/2026 6:11 AM, Sohil Mehta wrote:
> On 9/1/2026 4:16 PM, Chang S. Bae wrote:
>> Revision 0x1000405 contains internal microcode changes that are required
>> by subsequent revisions to avoid #MC during loading.
> 
> Before going to the solution space, can we add some more
> context/background here?
> 
> For example:
> 
> GNR revision 0x1000405 introduces a breaking change that causes a #MC
> when the OS updates the microcode from any version older than 0x1000405
> to a newer version. This is applicable to late load as well as early
> loading during boot.
> 
> A BIOS or firmware update is required to update the revision to
> 0x1000405 or newer before any OS updates can be safely loaded.
> This dependency ...

Thanks. I ended up taking Dave's version for V2.


>> Prevent loading 0x1000405 or later when the system has not yet been
>> updated to 0x1000405 or later.
> 
> This is a bit confusing. Should it say jumping from any anything older
> than 0x1000405 to anything newer?

I tweaked it in V2 to:
  "Prevent loading 0x1000405 or later unless the running revision is
   already at least 0x1000405."


>> +	if (vfm == INTEL_GRANITERAPIDS_X &&
>> +	    x86_stepping(sig->sig) == 1 &&
>> +	    sig->pf & 0x95 &&
>> +	    sig->rev < 0x1000405 &&
>> +	    rev >= 0x1000405) {
> 
> I don't think we have a helper that can be used here directly. Also,
> this is tagged for stable so adding a new one probably doesn't make sense.
> 
> But, 0x1000405 is repeated way too many times in this function :)
> At a minimum, can we add something like this?
> 
> #define GNR98_UCODE_MIN_REV 	0x1000405
> 
> Maybe add defines for the PF as well?


I saw exactly the same suggestion from AI, but I was not sure much value 
beyond more cosmetic. The revision is a one-off specific to the blocking 
check, so putting the revision value directly instead of a new define.


> 
> 
>> +		if (rev == 0x1000405)
>> +			pr_err_once("Erratum GNR98: 0x1000405 is not loadable.\n");
> 
> 						   ^ revision

Let me add `revision` in the message. Thanks.

> 
> With early load, his is probably one of the first messages that folks
> will see but it will only be displayed on platforms that have the issue
> until they update their microcode. Should we be more verbose here?
> 
> 
>> +		else
>> +			pr_err_once("Erratum GNR98: 0x1000405 is required before 0x%x.\n", rev);
> 
> "Erratum GNR98: revision 0x1000405 is required before revision 0x%x can
> be loaded. \n"
> 
> Should this be 0x1000405 *or later* ? We don't want common users to
> specifically try to find the revision 0x1000405 and load it, right?


For the error messages, `GNR98` needs to be decoded anyway, since the 
full context of the erratum is needed to understand for the next step. 
The message then is intended to point the erratum with a brief 
explanation. So I don't think it is too terse.

> 
> I am wondering what is the use of the if-else? Is that intended to guide
> the late loading users? I think for other users the second message would
> be confusing.

When the incoming revision is exactly 0x1000405, saying "revision 
0x1000405 is required" doesn't make any sense. For the later revisions, 
the second message can explain more directly too. So I split the two cases.

Thanks,
Chang

Re: [PATCH 1/8] x86/microcode/intel: Reject problematic loading on GNR systems

[Dave Hansen]

On 9/1/26 16:16, Chang S. Bae wrote:
> Revision 0x1000405 contains internal microcode changes that are required
> by subsequent revisions to avoid #MC during loading. This dependency
> logically fits the minimum revision requirement.
> 
> The minimum revision check, however, currently applies only to the late
> loading path, since the dependency was primarily intended for OS-visible
> changes. The early loading path is therefore still vulnerable to this
> issue.

Honestly, I think minrevs are a distraction here. Let's keep this short
and sweet:

Microcode updates can usually jump revisions. However, there is an
erratum on Granite Rapids systems. If they "jump over" revision
0x1000405, they machine check <or whatever, I forget how they die>.

> Furthermore, one of the subsequent revisions does not correctly specify
> the minimum revision, so unfortunately the late loading cannot rely on
> that check either in this case.
> 
> Prevent loading 0x1000405 or later when the system has not yet been
> updated to 0x1000405 or later. Apply this blocking to both early- and
> late-loading paths.
> 
> Rename is_blacklisted() to is_late_loading_denied() so the new function
> that covers both loading paths is not confused with the late-load only
> one.

Whee, and get rid of a chunk of old terminology while we're at it.

Re: [PATCH 1/8] x86/microcode/intel: Reject problematic loading on GNR systems

[Borislav Petkov]

On Tue, Sep 01, 2026 at 11:16:26PM +0000, Chang S. Bae wrote:
> +static bool is_loading_denied(struct cpu_signature *sig, u32 rev)

This naming is not better, sorry.

Is loading denied means the loading in general is denied because <raisin> or
are you trying to check whether this particular revision should not be loaded?

I think it is latter.

So you wanna say

	revision_blacklisted()

or so.

> +{
> +	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
> +
> +	/*
> +	 * Revision 0x1000405 contains prerequisite changes for subsequent
> +	 * microcode updates on Granite Rapids systems. Updates directly from
> +	 * an older revision to this or a newer one can result in #MC (GNR98).
> +	 *
> +	 * This dependency can be indicated from the minimum revision field.
> +	 * However, revision 0x1000423 has an incorrect minimum revision in its

So you lost me here: 0x1000423 is not tested anywhere - just mentioned here.

So what I understand is: usually, dependencies like that can be expressed with
minrev but *in addition* to the current issue, patch 0x1000423 has minrev
wrong so that dependency cannot be upheld there either.

But then why even mention it if you're not going to test it?

Why do we care about GNR101 at all?

> +	 * header (GNR101).
> +	 *
> +	 * Prevent loading 0x1000405 or later unless the CPU has already been
> +	 * updated to 0x1000405 or later.
> +	 */
> +	if (vfm == INTEL_GRANITERAPIDS_X &&
> +	    x86_stepping(sig->sig) == 1 &&
> +	    sig->pf & 0x95 &&
> +	    sig->rev < 0x1000405 &&
> +	    rev >= 0x1000405) {

You don't really need to test rev here - it is enough that sig->rev is
< 0x1000405 - that already makes you susceptible and then you can check rev
inside the { }.

> +		if (rev == 0x1000405)
> +			pr_err_once("Erratum GNR98: 0x1000405 is not loadable.\n");
> +		else
> +			pr_err_once("Erratum GNR98: 0x1000405 is required before 0x%x.\n", rev);
> +		pr_err_once("Please update the system BIOS or firmware.\n");

This is useless most of the time because client won't usually get BIOS
updates. You can tell people they should update their microcode packages
instead. That's where we can really help.

> +		return true;
> +	}
> +
> +	return false;
> +}
> +
>  /* Scan blob for microcode matching the boot CPUs family, model, stepping */
>  static __init struct microcode_intel *scan_microcode(void *data, size_t size,
>  						     struct ucode_cpu_info *uci,
> @@ -330,6 +362,9 @@ static __init struct microcode_intel *scan_microcode(void *data, size_t size,
>  		if (!intel_find_matching_signature(data, &uci->cpu_sig))
>  			continue;
>  
> +		if (is_loading_denied(&uci->cpu_sig, mc_header->rev))
> +			continue;
> +
>  		/*
>  		 * For saving the early microcode, find the matching revision which
>  		 * was loaded on the BSP.
> @@ -878,6 +913,9 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
>  		if (!intel_find_matching_signature(mc, &uci->cpu_sig))
>  			continue;
>  
> +		if (is_loading_denied(&uci->cpu_sig, mc_header.rev))
> +			continue;
> +
>  		is_safe = ucode_validate_minrev(&mc_header);
>  		if (force_minrev && !is_safe)
>  			continue;
> @@ -905,7 +943,7 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
>  	return UCODE_ERROR;
>  }
>  
> -static bool is_blacklisted(unsigned int cpu)
> +static bool is_late_loading_denied(unsigned int cpu)
>  {
>  	struct cpuinfo_x86 *c = &cpu_data(cpu);
>  
> @@ -936,7 +974,7 @@ static enum ucode_state request_microcode_fw(int cpu, struct device *device)
>  	struct kvec kvec;
>  	char name[30];
>  
> -	if (is_blacklisted(cpu))
> +	if (is_late_loading_denied(cpu))

Aaaah, you wanna be politically correct and can't use "blacklisted" anymore.

Well, you're not introducing new usage so you don't have to touch old usage.
And "is denied" does not express the situation properly. Try a better one.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [PATCH 1/8] x86/microcode/intel: Reject problematic loading on GNR systems

[Chang S. Bae]

On 9/2/2026 6:51 PM, Borislav Petkov wrote:
> On Tue, Sep 01, 2026 at 11:16:26PM +0000, Chang S. Bae wrote:
>> +static bool is_loading_denied(struct cpu_signature *sig, u32 rev)
> 
> This naming is not better, sorry.
> 
> Is loading denied means the loading in general is denied because <raisin> or
> are you trying to check whether this particular revision should not be loaded?
> 
> I think it is latter.
> 
> So you wanna say
> 
> 	revision_blacklisted()
> 
> or so.

Maybe revision_banned()?

> 
>> +{
>> +	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
>> +
>> +	/*
>> +	 * Revision 0x1000405 contains prerequisite changes for subsequent
>> +	 * microcode updates on Granite Rapids systems. Updates directly from
>> +	 * an older revision to this or a newer one can result in #MC (GNR98).
>> +	 *
>> +	 * This dependency can be indicated from the minimum revision field.
>> +	 * However, revision 0x1000423 has an incorrect minimum revision in its
> 
> So you lost me here: 0x1000423 is not tested anywhere - just mentioned here.
> 
> So what I understand is: usually, dependencies like that can be expressed with
> minrev but *in addition* to the current issue, patch 0x1000423 has minrev
> wrong so that dependency cannot be upheld there either.
> 
> But then why even mention it if you're not going to test it?
> 
> Why do we care about GNR101 at all?
> 

My original intention was to explain why rejection is need for the late 
loading path. There is a minrev check, but broken with that case.

But to preserve the legacy behavior, the minrev check isn't enforced by 
default. So what's the point of mentioning? Yes, if we need to ban 
particular revisions, we need to do it everywhere.

So mentioning minrev is just distracting readers as long as rejection is 
based on revision number in the first place. Let me remove minrev 
wording from this patch.

>> +	 * header (GNR101).
>> +	 *
>> +	 * Prevent loading 0x1000405 or later unless the CPU has already been
>> +	 * updated to 0x1000405 or later.
>> +	 */
>> +	if (vfm == INTEL_GRANITERAPIDS_X &&
>> +	    x86_stepping(sig->sig) == 1 &&
>> +	    sig->pf & 0x95 &&
>> +	    sig->rev < 0x1000405 &&
>> +	    rev >= 0x1000405) {
> 
> You don't really need to test rev here - it is enough that sig->rev is
> < 0x1000405 - that already makes you susceptible and then you can check rev
> inside the { }.

Yeah, I'd like to simplify like that, though. Intel repository looks to 
have three published revisions before 0x1000405. So there can be a valid 
update between those revisions.

So still need to allow:
   old_rev < 0x1000405 -> new_rev < 0x1000405

while rejecting:
   old_rev < 0x1000405 -> new_rev >= 0x1000405

> 
>> +		if (rev == 0x1000405)
>> +			pr_err_once("Erratum GNR98: 0x1000405 is not loadable.\n");
>> +		else
>> +			pr_err_once("Erratum GNR98: 0x1000405 is required before 0x%x.\n", rev);
>> +		pr_err_once("Please update the system BIOS or firmware.\n");
> 
> This is useless most of the time because client won't usually get BIOS
> updates. You can tell people they should update their microcode packages
> instead. That's where we can really help.
> 

I see.

I think anyone seeing this message has probably already tried updating 
microcode. With `GNR98` in the message, they would still need to decode 
it by looking up the erratum and its workaround if want to follow up on it:

   Workaround: None identified. Software should avoid updating MCU
   versions earlier than 0x1000405 to 0x1000405 or later until the system
   is Firmware Interface Table (FIT) loaded with UEFI FW/BIOS to
   0x1000405 or later.

Then there isn't much point of the additional message, I suppose. I'll 
remove it.

>>   
>> -	if (is_blacklisted(cpu))
>> +	if (is_late_loading_denied(cpu))
> 
> Aaaah, you wanna be politically correct and can't use "blacklisted" anymore.
> 
> Well, you're not introducing new usage so you don't have to touch old usage.
> And "is denied" does not express the situation properly. Try a better one.
Okay. I'll leave the existing "is_blacklisted(cpu)" as it-is since it 
already exists and probably descriptive enough. Also assuming it will 
remain distinctive from the new one, e.g. revision_banned(...).

Thanks,
Chang

[PATCH v2] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Chang S. Bae]

Microcode updates can usually jump revisions. However, there is an
erratum on Granite Rapids systems. If they "jump over" to revision
0x1000405 or later, they result in #MC.

Prevent loading 0x1000405 or later unless the running revision is already
at least 0x1000405. Apply this blocking to both early- and late-loading
paths.

Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Cc: <stable@vger.kernel.org>
---
V1 -> V2:
* Cut the code comments and print messages (Boris)
* Rename the new function and keep the old function as it-is (Boris)
* Rewrote the changelog (Dave)
* Add `revision` in the error messages (Sohil)
---
 arch/x86/kernel/cpu/microcode/intel.c | 32 +++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
index 1142183c950c..61ad280497e9 100644
--- a/arch/x86/kernel/cpu/microcode/intel.c
+++ b/arch/x86/kernel/cpu/microcode/intel.c
@@ -309,6 +309,32 @@ static void save_microcode_patch(struct microcode_intel *patch)
 		pr_err("Unable to allocate microcode memory size: %u\n", size);
 }
 
+static bool revision_banned(struct cpu_signature *sig, u32 rev)
+{
+	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
+
+	/*
+	 * Revision 0x1000405 contains prerequisite changes for subsequent
+	 * microcode updates on Granite Rapids systems. Updates directly from
+	 * an older revision to this or a newer one can result in #MC. This is
+	 * documented item GNR98, #835486 (Intel Xeon 6900/6700/6500-Series
+	 * Processors with P-Cores).
+	 */
+	if (vfm == INTEL_GRANITERAPIDS_X &&
+	    x86_stepping(sig->sig) == 1 &&
+	    sig->pf & 0x95 &&
+	    sig->rev < 0x1000405 &&
+	    rev >= 0x1000405) {
+		if (rev == 0x1000405)
+			pr_err_once("Erratum GNR98: revision 0x1000405 is not loadable.\n");
+		else
+			pr_err_once("Erratum GNR98: revision 0x1000405 is required before 0x%x.\n", rev);
+		return true;
+	}
+
+	return false;
+}
+
 /* Scan blob for microcode matching the boot CPUs family, model, stepping */
 static __init struct microcode_intel *scan_microcode(void *data, size_t size,
 						     struct ucode_cpu_info *uci,
@@ -330,6 +356,9 @@ static __init struct microcode_intel *scan_microcode(void *data, size_t size,
 		if (!intel_find_matching_signature(data, &uci->cpu_sig))
 			continue;
 
+		if (revision_banned(&uci->cpu_sig, mc_header->rev))
+			continue;
+
 		/*
 		 * For saving the early microcode, find the matching revision which
 		 * was loaded on the BSP.
@@ -878,6 +907,9 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
 		if (!intel_find_matching_signature(mc, &uci->cpu_sig))
 			continue;
 
+		if (revision_banned(&uci->cpu_sig, mc_header.rev))
+			continue;
+
 		is_safe = ucode_validate_minrev(&mc_header);
 		if (force_minrev && !is_safe)
 			continue;
-- 
2.53.0

Re: [PATCH v2] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Dave Hansen]

On 9/8/26 15:32, Chang S. Bae wrote:
> Microcode updates can usually jump revisions. However, there is an
> erratum on Granite Rapids systems. If they "jump over" to revision
> 0x1000405 or later, they result in #MC.
> 
> Prevent loading 0x1000405 or later unless the running revision is already
> at least 0x1000405. Apply this blocking to both early- and late-loading
> paths.

This is a kinda wonky-looking patch but I don't know a nicer way to do
it, so:

Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>

> +static bool revision_banned(struct cpu_signature *sig, u32 rev)
> +{
> +	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
> +
> +	/*
> +	 * Revision 0x1000405 contains prerequisite changes for subsequent
> +	 * microcode updates on Granite Rapids systems. Updates directly from
> +	 * an older revision to this or a newer one can result in #MC. This is
> +	 * documented item GNR98, #835486 (Intel Xeon 6900/6700/6500-Series
> +	 * Processors with P-Cores).
> +	 */

I honestly still prefer that this comment be more short and sweet. Does
this miss anything?

	/*
	 * Erratum GNR98 can cause #MC's if "jumping over"
	 * revision 0x1000405. Avoid the jumps.
	 */

Also, we don't need to put whole document titles or numbers in here.
They're fine, but if you can Google "erratum GNR98" and get the doc for
the first hit, what does it matter?

Leave the precision and long chit chat for the changelog, I say.

But we can edit this in place when it gets applied.

> +	if (vfm == INTEL_GRANITERAPIDS_X &&
> +	    x86_stepping(sig->sig) == 1 &&
> +	    sig->pf & 0x95 &&
> +	    sig->rev < 0x1000405 &&
> +	    rev >= 0x1000405) {
> +		if (rev == 0x1000405)
> +			pr_err_once("Erratum GNR98: revision 0x1000405 is not loadable.\n");
> +		else
> +			pr_err_once("Erratum GNR98: revision 0x1000405 is required before 0x%x.\n", rev);
> +		return true;
> +	}
> +
> +	return false;
> +}
> +
>  /* Scan blob for microcode matching the boot CPUs family, model, stepping */
>  static __init struct microcode_intel *scan_microcode(void *data, size_t size,
>  						     struct ucode_cpu_info *uci,
> @@ -330,6 +356,9 @@ static __init struct microcode_intel *scan_microcode(void *data, size_t size,
>  		if (!intel_find_matching_signature(data, &uci->cpu_sig))
>  			continue;
>  
> +		if (revision_banned(&uci->cpu_sig, mc_header->rev))
> +			continue;
> +
>  		/*
>  		 * For saving the early microcode, find the matching revision which
>  		 * was loaded on the BSP.
> @@ -878,6 +907,9 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
>  		if (!intel_find_matching_signature(mc, &uci->cpu_sig))
>  			continue;
>  
> +		if (revision_banned(&uci->cpu_sig, mc_header.rev))
> +			continue;
> +
>  		is_safe = ucode_validate_minrev(&mc_header);
>  		if (force_minrev && !is_safe)
>  			continue;

Re: [PATCH v2] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Borislav Petkov]

On Tue, Sep 08, 2026 at 10:32:09PM +0000, Chang S. Bae wrote:
> Microcode updates can usually jump revisions. However, there is an
> erratum on Granite Rapids systems. If they "jump over" to revision
> 0x1000405 or later, they result in #MC.
>
> Prevent loading 0x1000405 or later unless the running revision is already
> at least 0x1000405. Apply this blocking to both early- and late-loading
> paths.

People have got to stop explaining the patch in the commit message. That
should be obvious from the diff. If you have to explain it then there is
something very non-obvious here which I don't see it...

> Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
> Cc: <stable@vger.kernel.org>
> ---
> V1 -> V2:
> * Cut the code comments and print messages (Boris)
> * Rename the new function and keep the old function as it-is (Boris)
> * Rewrote the changelog (Dave)
> * Add `revision` in the error messages (Sohil)
> ---
>  arch/x86/kernel/cpu/microcode/intel.c | 32 +++++++++++++++++++++++++++
>  1 file changed, 32 insertions(+)
> 
> diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
> index 1142183c950c..61ad280497e9 100644
> --- a/arch/x86/kernel/cpu/microcode/intel.c
> +++ b/arch/x86/kernel/cpu/microcode/intel.c
> @@ -309,6 +309,32 @@ static void save_microcode_patch(struct microcode_intel *patch)
>  		pr_err("Unable to allocate microcode memory size: %u\n", size);
>  }
>  
> +static bool revision_banned(struct cpu_signature *sig, u32 rev)

I like Andy's naming:

https://lore.kernel.org/xen-devel/20260908171525.3196765-1-andrew.cooper3@citrix.com/T/#u

...is_safe is much better than banned.

> +{
> +	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
> +
> +	/*
> +	 * Revision 0x1000405 contains prerequisite changes for subsequent
> +	 * microcode updates on Granite Rapids systems. Updates directly from
> +	 * an older revision to this or a newer one can result in #MC. This is
> +	 * documented item GNR98, #835486 (Intel Xeon 6900/6700/6500-Series
> +	 * Processors with P-Cores).
> +	 */

What dhansen said - keep this short'n'sweet.

> +	if (vfm == INTEL_GRANITERAPIDS_X &&
> +	    x86_stepping(sig->sig) == 1 &&
> +	    sig->pf & 0x95 &&
> +	    sig->rev < 0x1000405 &&
> +	    rev >= 0x1000405) {
> +		if (rev == 0x1000405)

I also like Andy's testing of the patch revs:

+         ((cpu_sig->rev < 0x01000380 && mc->rev >= 0x01000405) ||
+          (cpu_sig->rev < 0x01000405 && mc->rev >  0x01000405)) )

> +			pr_err_once("Erratum GNR98: revision 0x1000405 is not loadable.\n");
> +		else
> +			pr_err_once("Erratum GNR98: revision 0x1000405 is required before 0x%x.\n", rev);

And you don't need those semi-identical strings here.

> +		return true;
> +	}
> +
> +	return false;
> +}

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [PATCH v2] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Chang S. Bae]

On 9/8/2026 5:13 PM, Borislav Petkov wrote:
> On Tue, Sep 08, 2026 at 10:32:09PM +0000, Chang S. Bae wrote:
>> Microcode updates can usually jump revisions. However, there is an
>> erratum on Granite Rapids systems. If they "jump over" to revision
>> 0x1000405 or later, they result in #MC.
>>
>> Prevent loading 0x1000405 or later unless the running revision is already
>> at least 0x1000405. Apply this blocking to both early- and late-loading
>> paths.
> 
> People have got to stop explaining the patch in the commit message. That
> should be obvious from the diff. If you have to explain it then there is
> something very non-obvious here which I don't see it...

... If they "jump over" to revision 0x1000405 or later, they result in 
#MC. Avoid it.

>>   
>> +static bool revision_banned(struct cpu_signature *sig, u32 rev)
> 
> I like Andy's naming:
> 
> https://lore.kernel.org/xen-devel/20260908171525.3196765-1-andrew.cooper3@citrix.com/T/#u
> 
> ...is_safe is much better than banned.

s/revision_banned/revision_is_safe/ in V3.

> 
>> +{
>> +	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
>> +
>> +	/*
>> +	 * Revision 0x1000405 contains prerequisite changes for subsequent
>> +	 * microcode updates on Granite Rapids systems. Updates directly from
>> +	 * an older revision to this or a newer one can result in #MC. This is
>> +	 * documented item GNR98, #835486 (Intel Xeon 6900/6700/6500-Series
>> +	 * Processors with P-Cores).
>> +	 */
> 
> What dhansen said - keep this short'n'sweet.

Will replace the above with Dave's.

> 
>> +	if (vfm == INTEL_GRANITERAPIDS_X &&
>> +	    x86_stepping(sig->sig) == 1 &&
>> +	    sig->pf & 0x95 &&
>> +	    sig->rev < 0x1000405 &&
>> +	    rev >= 0x1000405) {
>> +		if (rev == 0x1000405)
> 
> I also like Andy's testing of the patch revs:
> 
> +         ((cpu_sig->rev < 0x01000380 && mc->rev >= 0x01000405) ||
> +          (cpu_sig->rev < 0x01000405 && mc->rev >  0x01000405)) )

Let me double-check this with ucode folks, first. Then will take.

> 
>> +			pr_err_once("Erratum GNR98: revision 0x1000405 is not loadable.\n");
>> +		else
>> +			pr_err_once("Erratum GNR98: revision 0x1000405 is required before 0x%x.\n", rev);
 > > And you don't need those semi-identical strings here.

pr_err_once("Erratum GNR98: skipping revision 0x%x.\n", rev);

Yeah, overall I can hear "keep it simple & short" messages.

Thanks,
Chang

Re: [PATCH v2] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Andrew Cooper]

On 08/09/2026 11:32 pm, Chang S. Bae wrote:
> diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
> index 1142183c950c..61ad280497e9 100644
> --- a/arch/x86/kernel/cpu/microcode/intel.c
> +++ b/arch/x86/kernel/cpu/microcode/intel.c
> @@ -309,6 +309,32 @@ static void save_microcode_patch(struct microcode_intel *patch)
>  		pr_err("Unable to allocate microcode memory size: %u\n", size);
>  }
>  
> +static bool revision_banned(struct cpu_signature *sig, u32 rev)
> +{
> +	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
> +
> +	/*
> +	 * Revision 0x1000405 contains prerequisite changes for subsequent
> +	 * microcode updates on Granite Rapids systems. Updates directly from
> +	 * an older revision to this or a newer one can result in #MC. This is
> +	 * documented item GNR98, #835486 (Intel Xeon 6900/6700/6500-Series
> +	 * Processors with P-Cores).
> +	 */
> +	if (vfm == INTEL_GRANITERAPIDS_X &&
> +	    x86_stepping(sig->sig) == 1 &&
> +	    sig->pf & 0x95 &&
> +	    sig->rev < 0x1000405 &&
> +	    rev >= 0x1000405) {
> +		if (rev == 0x1000405)
> +			pr_err_once("Erratum GNR98: revision 0x1000405 is not loadable.\n");
> +		else
> +			pr_err_once("Erratum GNR98: revision 0x1000405 is required before 0x%x.\n", rev);
> +		return true;
> +	}

This logic is going to need editing when the muti-blob finally appears.

This is what I'm doing for Xen:
https://lore.kernel.org/xen-devel/20260908171525.3196765-1-andrew.cooper3@citrix.com/T/#u

Notably, 0x1000405 is safe to load if 0x1000380 is in FIT, and this is a
necessary hoop to jump through in due course.

~Andrew

[PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Chang S. Bae]

Microcode updates can usually jump revisions. However, there is an
erratum on Granite Rapids systems. If they "jump over" revision
0x1000405, they result in #MC. Avoid it.

Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: <stable@vger.kernel.org>
---
V2 -> V3:
* Shorten the changelog and rename the function (Boris)
* Reduce the code comment (Dave)
* Allow 0x1000405 loading. Thanks to Andrew, this fix got attention.
* Collect Dave review tag. Thanks, Dave!

Note:
* GNR98 currently describes loading 0x1000405 itself is unsafe, but it
  will be updated to say okay with that. I will watch out the GNR98
  changes.
* Jumping from < 0x1000380 to 0x1000405 was identified as an issue, but
  0x1000380 is the first revision as GNR products. So loading 0x1000405
  in production systems should be okay.
---
 arch/x86/kernel/cpu/microcode/intel.c | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
index 1142183c950c..30a22388d4b1 100644
--- a/arch/x86/kernel/cpu/microcode/intel.c
+++ b/arch/x86/kernel/cpu/microcode/intel.c
@@ -309,6 +309,26 @@ static void save_microcode_patch(struct microcode_intel *patch)
 		pr_err("Unable to allocate microcode memory size: %u\n", size);
 }
 
+static bool revision_is_safe(struct cpu_signature *sig, u32 rev)
+{
+	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
+
+	/*
+	 * Erratum GNR98 can cause #MC's if "jumping over" revision 0x1000405.
+	 * Avoid the jumps.
+	 */
+	if (vfm == INTEL_GRANITERAPIDS_X &&
+	    x86_stepping(sig->sig) == 1 &&
+	    sig->pf & 0x95 &&
+	    sig->rev < 0x1000405 &&
+	    rev > 0x1000405) {
+		pr_err_once("Erratum GNR98: skipping revision 0x%x.\n", rev);
+		return false;
+	}
+
+	return true;
+}
+
 /* Scan blob for microcode matching the boot CPUs family, model, stepping */
 static __init struct microcode_intel *scan_microcode(void *data, size_t size,
 						     struct ucode_cpu_info *uci,
@@ -330,6 +350,9 @@ static __init struct microcode_intel *scan_microcode(void *data, size_t size,
 		if (!intel_find_matching_signature(data, &uci->cpu_sig))
 			continue;
 
+		if (!revision_is_safe(&uci->cpu_sig, mc_header->rev))
+			continue;
+
 		/*
 		 * For saving the early microcode, find the matching revision which
 		 * was loaded on the BSP.
@@ -878,6 +901,9 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
 		if (!intel_find_matching_signature(mc, &uci->cpu_sig))
 			continue;
 
+		if (!revision_is_safe(&uci->cpu_sig, mc_header.rev))
+			continue;
+
 		is_safe = ucode_validate_minrev(&mc_header);
 		if (force_minrev && !is_safe)
 			continue;
-- 
2.53.0

Re: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[David Laight]

On Wed, 16 Sep 2026 22:59:39 +0000
"Chang S. Bae" <chang.seok.bae@intel.com> wrote:

> Microcode updates can usually jump revisions. However, there is an
> erratum on Granite Rapids systems. If they "jump over" revision
> 0x1000405, they result in #MC. Avoid it.

A probably silly question.
Is it valid to downgrade microcode?

David

> 
> Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
> Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>
> Cc: <stable@vger.kernel.org>
> ---
> V2 -> V3:
> * Shorten the changelog and rename the function (Boris)
> * Reduce the code comment (Dave)
> * Allow 0x1000405 loading. Thanks to Andrew, this fix got attention.
> * Collect Dave review tag. Thanks, Dave!
> 
> Note:
> * GNR98 currently describes loading 0x1000405 itself is unsafe, but it
>   will be updated to say okay with that. I will watch out the GNR98
>   changes.
> * Jumping from < 0x1000380 to 0x1000405 was identified as an issue, but
>   0x1000380 is the first revision as GNR products. So loading 0x1000405
>   in production systems should be okay.
> ---
>  arch/x86/kernel/cpu/microcode/intel.c | 26 ++++++++++++++++++++++++++
>  1 file changed, 26 insertions(+)
> 
> diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
> index 1142183c950c..30a22388d4b1 100644
> --- a/arch/x86/kernel/cpu/microcode/intel.c
> +++ b/arch/x86/kernel/cpu/microcode/intel.c
> @@ -309,6 +309,26 @@ static void save_microcode_patch(struct microcode_intel *patch)
>  		pr_err("Unable to allocate microcode memory size: %u\n", size);
>  }
>  
> +static bool revision_is_safe(struct cpu_signature *sig, u32 rev)
> +{
> +	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
> +
> +	/*
> +	 * Erratum GNR98 can cause #MC's if "jumping over" revision 0x1000405.
> +	 * Avoid the jumps.
> +	 */
> +	if (vfm == INTEL_GRANITERAPIDS_X &&
> +	    x86_stepping(sig->sig) == 1 &&
> +	    sig->pf & 0x95 &&
> +	    sig->rev < 0x1000405 &&
> +	    rev > 0x1000405) {
> +		pr_err_once("Erratum GNR98: skipping revision 0x%x.\n", rev);
> +		return false;
> +	}
> +
> +	return true;
> +}
> +
>  /* Scan blob for microcode matching the boot CPUs family, model, stepping */
>  static __init struct microcode_intel *scan_microcode(void *data, size_t size,
>  						     struct ucode_cpu_info *uci,
> @@ -330,6 +350,9 @@ static __init struct microcode_intel *scan_microcode(void *data, size_t size,
>  		if (!intel_find_matching_signature(data, &uci->cpu_sig))
>  			continue;
>  
> +		if (!revision_is_safe(&uci->cpu_sig, mc_header->rev))
> +			continue;
> +
>  		/*
>  		 * For saving the early microcode, find the matching revision which
>  		 * was loaded on the BSP.
> @@ -878,6 +901,9 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
>  		if (!intel_find_matching_signature(mc, &uci->cpu_sig))
>  			continue;
>  
> +		if (!revision_is_safe(&uci->cpu_sig, mc_header.rev))
> +			continue;
> +
>  		is_safe = ucode_validate_minrev(&mc_header);
>  		if (force_minrev && !is_safe)
>  			continue;

Re: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Andrew Cooper]

On 17/09/2026 11:14 am, David Laight wrote:
> On Wed, 16 Sep 2026 22:59:39 +0000
> "Chang S. Bae" <chang.seok.bae@intel.com> wrote:
>
>> Microcode updates can usually jump revisions. However, there is an
>> erratum on Granite Rapids systems. If they "jump over" revision
>> 0x1000405, they result in #MC. Avoid it.
> A probably silly question.
> Is it valid to downgrade microcode?

Yes, and it does happen in practice.

Intel microcode contains a field called the Security Version Number. 
The SVN bumps as infrequently as possible, but does bump when necessary.

There is a hard block on the SVN going backwards, but as long as you're
not violating this constraint, you can upgrade or downgrade microcode as
desired.

~Andrew

RE: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Van De Ven, Arjan]

> > Microcode updates can usually jump revisions. However, there is an
> > erratum on Granite Rapids systems. If they "jump over" revision
> > 0x1000405, they result in #MC. Avoid it.
> 
> A probably silly question.
> Is it valid to downgrade microcode?


Within SVN in theory it can be done

In practice, unless you have very special circumstances (and check with Intel if the exact downgrade you have in mind has been tested for downgrading) I would very strongly recommend against doing so. It may appear to work, but note that 

A -> B -> A 

Is NOT identical to just

A  (staying at A)

as part of the microcode load, some microcode runs as "setup" which may change settings in the CPU (versus runtime behavior changes) -- and that setup step is not undone as you go back to A.... which means you run A with a (partial or whole) setup of B. 

Re: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Sohil Mehta]

>  /* Scan blob for microcode matching the boot CPUs family, model, stepping */
>  static __init struct microcode_intel *scan_microcode(void *data, size_t size,
>  						     struct ucode_cpu_info *uci,
> @@ -330,6 +350,9 @@ static __init struct microcode_intel *scan_microcode(void *data, size_t size,
>  		if (!intel_find_matching_signature(data, &uci->cpu_sig))
>  			continue;
>  
> +		if (!revision_is_safe(&uci->cpu_sig, mc_header->rev))
> +			continue;
> +
>  		/*
>  		 * For saving the early microcode, find the matching revision which
>  		 * was loaded on the BSP.
> @@ -878,6 +901,9 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
>  		if (!intel_find_matching_signature(mc, &uci->cpu_sig))
>  			continue;
>  
> +		if (!revision_is_safe(&uci->cpu_sig, mc_header.rev))
> +			continue;
> +
>  		is_safe = ucode_validate_minrev(&mc_header);
>  		if (force_minrev && !is_safe)
>  			continue;

TL;dr: Should there be a revision_is_safe() check during
__apply_microcode()?

The patch only adds the revision check during blob selection. Have we
evaluated the corner cases hinted by sashiko when this check might be
bypassed? It talks about suspend/resume and the CPU hotplug cases.

v1:
https://sashiko.dev/#/patchset/20260901231634.714144-1-chang.seok.bae%40intel.com?part=1

v3:
https://sashiko.dev/#/patchset/20260916225939.1144524-1-chang.seok.bae%40intel.com?part=1

The suspend/resume path probably doesn't matter for GNR servers and most
of the CPU hotplug flows also seem to be covered. But, what about ucode
update on cores that are not enabled at boot? If there are brought
online later, would the ucode update skip the above check?

For example, maxcpus=N prevents certain CPUs from showing up in
cpus_booted_once_mask. So the checks in setup_cpus() during late-loading
would not catch them. IIUC, the BIOS version can be < 0x1000405, early
load can bump the booted cpus to 0x1000405, and then late-loading can
load newer versions and only cache the latest ucode revision.

All this while, the non-booted cores would be stuck at a revision less
than 0x1000405. So, when they are brought online later, could they
directly jump to a revision greater than 0x1000405?

Would it be safer to add the revision_is_safe() check in
__apply_microcode() so that all of such cases are covered?

Maybe sashiko just made me paranoid. I only started looking at this
because it complained! :(

Re: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Chang S. Bae]

On 9/17/2026 10:52 AM, Sohil Mehta wrote:
> 
> TL;dr: Should there be a revision_is_safe() check during
> __apply_microcode()?
> 
> The patch only adds the revision check during blob selection. Have we
> evaluated the corner cases hinted by sashiko when this check might be
> bypassed? It talks about suspend/resume and the CPU hotplug cases.
> 
> v1:
> https://sashiko.dev/#/patchset/20260901231634.714144-1-chang.seok.bae%40intel.com?part=1
> 
> v3:
> https://sashiko.dev/#/patchset/20260916225939.1144524-1-chang.seok.bae%40intel.com?part=1
> 
> The suspend/resume path probably doesn't matter for GNR servers and most
> of the CPU hotplug flows also seem to be covered. But, what about ucode
> update on cores that are not enabled at boot? If there are brought
> online later, would the ucode update skip the above check?
> 
> For example, maxcpus=N prevents certain CPUs from showing up in
> cpus_booted_once_mask. So the checks in setup_cpus() during late-loading
> would not catch them. IIUC, the BIOS version can be < 0x1000405, early
> load can bump the booted cpus to 0x1000405, and then late-loading can
> load newer versions and only cache the latest ucode revision.

Currently, what setup_cpus() does as its comment says is first ensure 
all CPUs that are present and has been booted up have their primary 
threads online. It just allows its secondary thread offline with nosmt. 
I don't think the current logic is broken there.

Then, those sibling threads assuming late-loading while soft-offlined 
will see an updated revision on its bringup because the loading scope is 
per-core by default, meaning the update performed by the primary thread 
also applies to its sibling.

Also, the NMI stop-machine rendezvous includes those soft-offlined CPUs. 
They are brought into the rendezvous and wait there while the update is 
being performed.

Now, I think it could be misleading if we put the revision check toward 
the end right before the application.

For example, suppose a multi-blob image that is bundled with revision 
0x1000405 and later ones and currently running < 0x1000405. With the 
check during blob selection, the parser can identify 0x1000405 as the 
loadable revision and reject later revisions.

If we move the check to just before application, the parser may instead 
select a revision newer than 0x1000405. The application would then be 
rejected, and the same blob would be selected again on the next attempt, 
resulting in the loading process repeatedly aborting without ever making 
progress.

So the revision check needs to remain part of blob selection, where it 
can affect which revision is considered loadable, rather than being only 
an application-time check.

Thanks,
Chang

RE: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Van De Ven, Arjan]

 
> If we move the check to just before application, the parser may instead
> select a revision newer than 0x1000405. The application would then be
> rejected, and the same blob would be selected again on the next attempt,
> resulting in the loading process repeatedly aborting without ever making
> progress.
> 
> So the revision check needs to remain part of blob selection, where it
> can affect which revision is considered loadable, rather than being only
> an application-time check.


Agreed that we need the early check.
But we could, on the actual application, at least check if the ucode already got upgraded to EXACTLY the target version, and skip the actual application if it has been upgraded. That is a simple check with basically no policy ("identity") but with huge savings

Re: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Chang S. Bae]

On 9/17/2026 11:43 AM, Van De Ven, Arjan wrote:
> 
> But we could, on the actual application, at least check if the ucode 
 > already got upgraded to EXACTLY the target version, and skip the
 > actual application if it has been upgraded. That is a simple check
 > with basically no policy ("identity") but with huge savings

Yes, I think that' what we do now:

__apply_microcode(..., struct microcode_intel *mc, ...)
{
	u32 rev;

	if (!mc)
		return UCODE_NFOUND;

	*cur_rev = intel_get_microcode_revision();
                    ^ read MSR0x8b (aka MSR_IA32_UCODE_REV)
	if (*cur_rev >= mc->hdr.rev) {
		uci->cpu_sig.rev = *cur_rev;
		return UCODE_OK;
		^ we skip the loading if updated
	}

	/* write microcode via MSR 0x79 */
	native_wrmsrq(MSR_IA32_UCODE_WRITE, (unsigned long)mc->bits);
	...
}

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/x86/kernel/cpu/microcode/intel.c#n646

Thanks,
Chang

[tip: x86/urgent] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[tip-bot2 for Chang S. Bae]

The following commit has been merged into the x86/urgent branch of tip:

Commit-ID:     e7d3e2f46dd5a69046e6d95a0f189155a5516b93
Gitweb:        https://git.kernel.org/tip/e7d3e2f46dd5a69046e6d95a0f189155a5516b93
Author:        Chang S. Bae <chang.seok.bae@intel.com>
AuthorDate:    Wed, 16 Sep 2026 22:59:39 
Committer:     Borislav Petkov (AMD) <bp@alien8.de>
CommitterDate: Thu, 17 Sep 2026 16:54:20 -07:00

x86/microcode/intel: Reject problematic loading on Granite Rapids systems

Microcode updates can usually jump revisions. However, there is an erratum on
Granite Rapids systems. If they "jump over" revision 0x1000405, they result in
an #MC. Avoid it.

Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: stable@vger.kernel.org
Link: https://patch.msgid.link/20260916225939.1144524-1-chang.seok.bae@intel.com
---
 arch/x86/kernel/cpu/microcode/intel.c | 26 ++++++++++++++++++++++++++-
 1 file changed, 26 insertions(+)

diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
index 1142183..9f09d38 100644
--- a/arch/x86/kernel/cpu/microcode/intel.c
+++ b/arch/x86/kernel/cpu/microcode/intel.c
@@ -309,6 +309,26 @@ static void save_microcode_patch(struct microcode_intel *patch)
 		pr_err("Unable to allocate microcode memory size: %u\n", size);
 }
 
+static bool revision_is_safe(struct cpu_signature *sig, u32 rev)
+{
+	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
+
+	/*
+	 * Erratum GNR98 can cause #MCs if "jumping over" revision 0x1000405.
+	 * Avoid the jumps.
+	 */
+	if (vfm == INTEL_GRANITERAPIDS_X &&
+	    x86_stepping(sig->sig) == 1 &&
+	    sig->pf & 0x95 &&
+	    sig->rev < 0x1000405 &&
+	    rev > 0x1000405) {
+		pr_err_once("Erratum GNR98: skipping revision 0x%x.\n", rev);
+		return false;
+	}
+
+	return true;
+}
+
 /* Scan blob for microcode matching the boot CPUs family, model, stepping */
 static __init struct microcode_intel *scan_microcode(void *data, size_t size,
 						     struct ucode_cpu_info *uci,
@@ -330,6 +350,9 @@ static __init struct microcode_intel *scan_microcode(void *data, size_t size,
 		if (!intel_find_matching_signature(data, &uci->cpu_sig))
 			continue;
 
+		if (!revision_is_safe(&uci->cpu_sig, mc_header->rev))
+			continue;
+
 		/*
 		 * For saving the early microcode, find the matching revision which
 		 * was loaded on the BSP.
@@ -878,6 +901,9 @@ static enum ucode_state parse_microcode_blobs(int cpu, struct iov_iter *iter)
 		if (!intel_find_matching_signature(mc, &uci->cpu_sig))
 			continue;
 
+		if (!revision_is_safe(&uci->cpu_sig, mc_header.rev))
+			continue;
+
 		is_safe = ucode_validate_minrev(&mc_header);
 		if (force_minrev && !is_safe)
 			continue;

Re: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Ingo Molnar]

* Chang S. Bae <chang.seok.bae@intel.com> wrote:

> Microcode updates can usually jump revisions. However, there is an
> erratum on Granite Rapids systems. If they "jump over" revision
> 0x1000405, they result in #MC. Avoid it.
> 
> Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
> Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>
> Cc: <stable@vger.kernel.org>
> ---
> V2 -> V3:
> * Shorten the changelog and rename the function (Boris)
> * Reduce the code comment (Dave)
> * Allow 0x1000405 loading. Thanks to Andrew, this fix got attention.
> * Collect Dave review tag. Thanks, Dave!
> 
> Note:
> * GNR98 currently describes loading 0x1000405 itself is unsafe, but it
>   will be updated to say okay with that. I will watch out the GNR98
>   changes.
> * Jumping from < 0x1000380 to 0x1000405 was identified as an issue, but
>   0x1000380 is the first revision as GNR products. So loading 0x1000405
>   in production systems should be okay.
> ---
>  arch/x86/kernel/cpu/microcode/intel.c | 26 ++++++++++++++++++++++++++
>  1 file changed, 26 insertions(+)
> 
> diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
> index 1142183c950c..30a22388d4b1 100644
> --- a/arch/x86/kernel/cpu/microcode/intel.c
> +++ b/arch/x86/kernel/cpu/microcode/intel.c
> @@ -309,6 +309,26 @@ static void save_microcode_patch(struct microcode_intel *patch)
>  		pr_err("Unable to allocate microcode memory size: %u\n", size);
>  }
>  
> +static bool revision_is_safe(struct cpu_signature *sig, u32 rev)
> +{
> +	u32 vfm = IFM(x86_family(sig->sig), x86_model(sig->sig));
> +
> +	/*
> +	 * Erratum GNR98 can cause #MC's if "jumping over" revision 0x1000405.
> +	 * Avoid the jumps.
> +	 */
> +	if (vfm == INTEL_GRANITERAPIDS_X &&
> +	    x86_stepping(sig->sig) == 1 &&
> +	    sig->pf & 0x95 &&
> +	    sig->rev < 0x1000405 &&
> +	    rev > 0x1000405) {
> +		pr_err_once("Erratum GNR98: skipping revision 0x%x.\n", rev);

So this was explained in a really confusing way, I had to read the
changelog and comments trice and then the code to figure out what's
going on:

 - There's a microcode bug that makes it unsafe to apply current
   post-0x1000405 revisions on Granite Rapid CPUs if the current
   microcode version is below 0x1000405. The interim 0x1000405
   version *must* be applied first for it to be safe to upgrade
   GNR CPUs. Will this be a problem perpetually? Will it be unsafe
   to have an older GNR CPU and simply apply fresh microcode to it,
   without first loading the interim 0x1000405 version? Will GNR
   microcode upgrades on pre-0x1000405 CPUs will always be a
   two-step process?

 - The new code doesn't declare it, but this patch creates a hidden,
   permanent microcode version upgrade barrier if user-space
   firmware/microcode tools do not provide the 0x1000405
   microcode version reliably and implement the two-step upgrade
   workaround.

 - The message the kernel prints is rather passive-aggressive as well:

		Erratum GNR98: skipping revision 0x%x

   It does not explain *why* the fresh microcode upgrade is skipped,
   and if firmware tooling does not apply 0x1000405 then the kernel
   stays in this state indefinitely.

   It should at minimum say something like:

      Erratum GNR98: new revision %x is unsafe to apply until 0x1000405 is applied first, skipping it. Please upgrade firmware tools.

   Because, presumably, this scenario should not be possible with
   new user-space tooling, right?

   The comment and changelog should be updated accordingly as well.

So this patch is still problematic IMO.

Thanks,

	Ingo

Re: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Borislav Petkov]

On Fri, Sep 18, 2026 at 08:54:01AM +0200, Ingo Molnar wrote:
>    It should at minimum say something like:
> 
>       Erratum GNR98: new revision %x is unsafe to apply until 0x1000405 is applied first, skipping it. Please upgrade firmware tools.

Even if you put prose in a warning message, it still doesn't explain the issue
fully. For stuff like that you need some properly written longer explanation
in the documentation.

Which I'm hoping we'll get eventually once the dust has settled down.

That's what the rest of this thread is trying to do but I haven't reviewed
yet.

> So this patch is still problematic IMO.

This patch is a minimal stable fix to stop it from #MCing. The real stuff
comes later. It all has been discussed on this very thread.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [PATCH v3] x86/microcode/intel: Reject problematic loading on Granite Rapids systems

[Chang S. Bae]

On 9/16/2026 3:59 PM, Chang S. Bae wrote:
> 
> * GNR98 currently describes loading 0x1000405 itself is unsafe, but it
>    will be updated to say okay with that. I will watch out the GNR98
>    changes.

The erratum documentation [*] appears to have been updated since last 
week. GNR98 now describes correctly:

   ... Due to this erratum, an Early OS Microcode Update from a microcode
   version earlier than 0x1000405 to a release _after_ 0x1000405 may
   result in a Machine Check Error (MCE).

[*] 
https://www.intel.com/content/www/us/en/content-details/835486/intel-xeon-6900-6700-6500-series-processors-with-p-cores-specification-update.html

Thanks,
Chang

[PATCH 2/8] x86/microcode: Solidify base_rev= option parsing

[Chang S. Bae]

After base_rev= has been parsed, the remainder of the token is still
compared against the other option names. Along with this, the substring
check strstr() may also produce a spurious match.

Move on to the next token once the option has been parsed, and use
str_has_prefix() instead.

Fixes: 43181a47263d ("x86/microcode: Add microcode loader debugging functionality")
Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Cc: <stable@vger.kernel.org>
---
 arch/x86/kernel/cpu/microcode/core.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/microcode/core.c b/arch/x86/kernel/cpu/microcode/core.c
index 0dd0c7241c57..303b0d0b4573 100644
--- a/arch/x86/kernel/cpu/microcode/core.c
+++ b/arch/x86/kernel/cpu/microcode/core.c
@@ -146,10 +146,11 @@ static void __init early_parse_cmdline(void)
 	if (cmdline_find_option(boot_command_line, "microcode", cmd_buf, sizeof(cmd_buf)) > 0) {
 		while ((s = strsep(&p, ","))) {
 			if (IS_ENABLED(CONFIG_MICROCODE_DBG)) {
-				if (strstr(s, "base_rev=")) {
+				if (str_has_prefix(s, "base_rev=")) {
 					/* advance to the option arg */
 					strsep(&s, "=");
 					if (kstrtouint(s, 16, &base_rev)) { ; }
+					continue;
 				}
 			}
 
-- 
2.53.0

[PATCH 3/8] x86/microcode: Accept a boolean for force_minrev parameter

[Chang S. Bae]

The microcode= command-line option allows controlling all microcode
loading options. However, force_minrev can only enable minimum revision
enforcement, not disable it.

The previous microcode.force_minrev= option accepted both boolean values,
and the current behavior differs from the documentation in
kernel-parameters.txt.

Add `force_minrev=` as a sub-option to explicitly enable or disable the
minimum revision enforcement. Update the documentation accordingly.

While doing so, wrap strsep() in a descriptively named helper.

Fixes: 632ff6170647 ("x86/microcode: Add microcode= cmdline parsing")
Reported-by: Sohil Mehta <sohil.mehta@intel.com>
Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Cc: <stable@vger.kernel.org>
---
 Documentation/admin-guide/kernel-parameters.txt |  7 ++++---
 arch/x86/kernel/cpu/microcode/core.c            | 15 ++++++++++++---
 2 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 68647ff4bdd2..cde092017cd8 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -4071,9 +4071,10 @@ Kernel parameters
 
 			dis_ucode_ldr: disable the microcode loader
 
-			force_minrev:
-			Enable or disable the microcode minimal revision
-			enforcement for the runtime microcode loader.
+			force_minrev[=<bool>]:
+			Enable or disable microcode minimal revision enforcement
+			for the runtime microcode loader according to <bool>. If
+			<bool> is not given, enable the enforcement.
 
 	mini2440=	[ARM,HW,KNL]
 			Format:[0..2][b][c][t]
diff --git a/arch/x86/kernel/cpu/microcode/core.c b/arch/x86/kernel/cpu/microcode/core.c
index 303b0d0b4573..af7196d94540 100644
--- a/arch/x86/kernel/cpu/microcode/core.c
+++ b/arch/x86/kernel/cpu/microcode/core.c
@@ -138,6 +138,11 @@ bool __init microcode_loader_disabled(void)
 	return dis_ucode_ldr;
 }
 
+static inline void advance_option_argument(char **s)
+{
+	strsep(s, "=");
+}
+
 static void __init early_parse_cmdline(void)
 {
 	char cmd_buf[64] = {};
@@ -147,15 +152,19 @@ static void __init early_parse_cmdline(void)
 		while ((s = strsep(&p, ","))) {
 			if (IS_ENABLED(CONFIG_MICROCODE_DBG)) {
 				if (str_has_prefix(s, "base_rev=")) {
-					/* advance to the option arg */
-					strsep(&s, "=");
+					advance_option_argument(&s);
 					if (kstrtouint(s, 16, &base_rev)) { ; }
 					continue;
 				}
 			}
 
-			if (!strcmp("force_minrev", s))
+			if (!strcmp("force_minrev", s)) {
 				force_minrev = true;
+			} else if (str_has_prefix(s, "force_minrev=")) {
+				advance_option_argument(&s);
+				if (kstrtobool(s, &force_minrev)) { ; }
+				continue;
+			}
 
 			if (!strcmp(s, "dis_ucode_ldr"))
 				dis_ucode_ldr = true;
-- 
2.53.0