[PATCH v5 0/5] Print AGESA version at bootup

[PATCH v5 0/5] Print AGESA version at bootup

[Mario Limonciello (AMD)]

The SMBIOS additional entries on AMD Zen4+ systems running an AGESA
based BIOS contain information about the AGESA version which can be
useful for matching the software stack when debugging an issue.

Add support for parsing this from SMBIOS tables and output it into
the logs at bootup.

v5:
 * Fix LKP and Yzen reported failure by returning ""
 * Drop prefix patch in amd.c
Mario Limonciello (AMD) (4):
  firmware: dmi: Correct an indexing error in dmi.h
  firmware: dmi: Adjust dmi_decode() to use enums
  firmware: dmi: Add missing DMI entry types
  firmware: dmi: Add pr_fmt() for dmi_scan.c

Yazen Ghannam (1):
  x86/CPU/AMD: Print AGESA string from DMI additional information entry

 arch/x86/kernel/cpu/amd.c   | 53 +++++++++++++++++++++++++++++++++++++
 drivers/firmware/dmi_scan.c | 34 +++++++++++++-----------
 include/linux/dmi.h         | 23 ++++++++++++++++
 3 files changed, 95 insertions(+), 15 deletions(-)

-- 
2.43.0

[PATCH v5 1/5] firmware: dmi: Correct an indexing error in dmi.h

[Mario Limonciello (AMD)]

The entries later in `enum dmi_entry_type` don't match the SMBIOS
specification.  The entry for type 33: `64-Bit Memory Error Information`
is not present and thus the index for all later entries is incorrect.

Add the missing type 33 entry.

Fixes: 93c890dbe5287 ("firmware: Add DMI entry types to the headers")
Link: https://www.dmtf.org/sites/default/files/standards/documents/DSP0134_3.4.0a.pdf
Reviewed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
---
v3:
 * Rename to DMI_ENTRY_64_MEM_ERROR (Jean)
---
 include/linux/dmi.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/include/linux/dmi.h b/include/linux/dmi.h
index 927f8a8b7a1dd..32b2529a73301 100644
--- a/include/linux/dmi.h
+++ b/include/linux/dmi.h
@@ -60,6 +60,7 @@ enum dmi_entry_type {
 	DMI_ENTRY_OOB_REMOTE_ACCESS,
 	DMI_ENTRY_BIS_ENTRY,
 	DMI_ENTRY_SYSTEM_BOOT,
+	DMI_ENTRY_64_MEM_ERROR,
 	DMI_ENTRY_MGMT_DEV,
 	DMI_ENTRY_MGMT_DEV_COMPONENT,
 	DMI_ENTRY_MGMT_DEV_THRES,
-- 
2.43.0

[PATCH v5 2/5] firmware: dmi: Adjust dmi_decode() to use enums

[Mario Limonciello (AMD)]

dmi_decode() has hardcoded values with comments for each DMI entry
type. The same information is already in dmi.h though, so drop the
comments and use the definitions instead.

Reviewed-by: Jean Delvare <jdelvare@suse.de>
Reviewed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
---
v3:
 * pick up tag (Jean)
---
 drivers/firmware/dmi_scan.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/firmware/dmi_scan.c b/drivers/firmware/dmi_scan.c
index 70d39adf50dca..80aded4c778bc 100644
--- a/drivers/firmware/dmi_scan.c
+++ b/drivers/firmware/dmi_scan.c
@@ -484,14 +484,14 @@ static void __init dmi_memdev_walk(void)
 static void __init dmi_decode(const struct dmi_header *dm, void *dummy)
 {
 	switch (dm->type) {
-	case 0:		/* BIOS Information */
+	case DMI_ENTRY_BIOS:
 		dmi_save_ident(dm, DMI_BIOS_VENDOR, 4);
 		dmi_save_ident(dm, DMI_BIOS_VERSION, 5);
 		dmi_save_ident(dm, DMI_BIOS_DATE, 8);
 		dmi_save_release(dm, DMI_BIOS_RELEASE, 21);
 		dmi_save_release(dm, DMI_EC_FIRMWARE_RELEASE, 23);
 		break;
-	case 1:		/* System Information */
+	case DMI_ENTRY_SYSTEM:
 		dmi_save_ident(dm, DMI_SYS_VENDOR, 4);
 		dmi_save_ident(dm, DMI_PRODUCT_NAME, 5);
 		dmi_save_ident(dm, DMI_PRODUCT_VERSION, 6);
@@ -500,33 +500,33 @@ static void __init dmi_decode(const struct dmi_header *dm, void *dummy)
 		dmi_save_ident(dm, DMI_PRODUCT_SKU, 25);
 		dmi_save_ident(dm, DMI_PRODUCT_FAMILY, 26);
 		break;
-	case 2:		/* Base Board Information */
+	case DMI_ENTRY_BASEBOARD:
 		dmi_save_ident(dm, DMI_BOARD_VENDOR, 4);
 		dmi_save_ident(dm, DMI_BOARD_NAME, 5);
 		dmi_save_ident(dm, DMI_BOARD_VERSION, 6);
 		dmi_save_ident(dm, DMI_BOARD_SERIAL, 7);
 		dmi_save_ident(dm, DMI_BOARD_ASSET_TAG, 8);
 		break;
-	case 3:		/* Chassis Information */
+	case DMI_ENTRY_CHASSIS:
 		dmi_save_ident(dm, DMI_CHASSIS_VENDOR, 4);
 		dmi_save_type(dm, DMI_CHASSIS_TYPE, 5);
 		dmi_save_ident(dm, DMI_CHASSIS_VERSION, 6);
 		dmi_save_ident(dm, DMI_CHASSIS_SERIAL, 7);
 		dmi_save_ident(dm, DMI_CHASSIS_ASSET_TAG, 8);
 		break;
-	case 9:		/* System Slots */
+	case DMI_ENTRY_SYSTEM_SLOT:
 		dmi_save_system_slot(dm);
 		break;
-	case 10:	/* Onboard Devices Information */
+	case DMI_ENTRY_ONBOARD_DEVICE:
 		dmi_save_devices(dm);
 		break;
-	case 11:	/* OEM Strings */
+	case DMI_ENTRY_OEMSTRINGS:
 		dmi_save_oem_strings_devices(dm);
 		break;
-	case 38:	/* IPMI Device Information */
+	case DMI_ENTRY_IPMI_DEV:
 		dmi_save_ipmi_device(dm);
 		break;
-	case 41:	/* Onboard Devices Extended Information */
+	case DMI_ENTRY_ONBOARD_DEV_EXT:
 		dmi_save_extended_devices(dm);
 	}
 }
-- 
2.43.0

[PATCH v5 3/5] firmware: dmi: Add missing DMI entry types

[Mario Limonciello (AMD)]

Type 43 through type 46 entry types are missing from the
definitions in the dmi_entry_type enum.

Link: https://www.dmtf.org/sites/default/files/standards/documents/DSP0134_3.7.1.pdf
Suggested-by: Yazen Ghannam <yazen.ghannam@amd.com>
Reviewed-by: Jean Delvare <jdelvare@suse.de>
Reviewed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
---
v4:
 * Add tag (Yazen)
 * Correct text (Yazen)
v3:
 * Add tag (Jean)
---
 include/linux/dmi.h | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/include/linux/dmi.h b/include/linux/dmi.h
index 32b2529a73301..2eedf44e68012 100644
--- a/include/linux/dmi.h
+++ b/include/linux/dmi.h
@@ -70,6 +70,10 @@ enum dmi_entry_type {
 	DMI_ENTRY_ADDITIONAL,
 	DMI_ENTRY_ONBOARD_DEV_EXT,
 	DMI_ENTRY_MGMT_CONTROLLER_HOST,
+	DMI_ENTRY_TPM_DEVICE,
+	DMI_ENTRY_PROCESSOR_ADDITIONAL,
+	DMI_ENTRY_FIRMWARE_INVENTORY,
+	DMI_ENTRY_STRING_PROPERTY,
 	DMI_ENTRY_INACTIVE = 126,
 	DMI_ENTRY_END_OF_TABLE = 127,
 };
-- 
2.43.0

[PATCH v5 4/5] firmware: dmi: Add pr_fmt() for dmi_scan.c

[Mario Limonciello (AMD)]

Several prints inconsistently use DMI: or dmi: or nothing.  To make
it clear which prints come from dmi_scan.c, add a pr_fmt macro.

Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
---
v4:
 * Adjust "DMI not present or invalid" message too (Yazen)
---
 drivers/firmware/dmi_scan.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/drivers/firmware/dmi_scan.c b/drivers/firmware/dmi_scan.c
index 80aded4c778bc..ed6235ac576b6 100644
--- a/drivers/firmware/dmi_scan.c
+++ b/drivers/firmware/dmi_scan.c
@@ -1,4 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0-only
+
+#define pr_fmt(fmt) "DMI: " fmt
+
 #include <linux/types.h>
 #include <linux/string.h>
 #include <linux/init.h>
@@ -634,7 +637,7 @@ static int __init dmi_present(const u8 *buf)
 					dmi_ver >> 16, (dmi_ver >> 8) & 0xFF);
 			}
 			dmi_format_ids(dmi_ids_string, sizeof(dmi_ids_string));
-			pr_info("DMI: %s\n", dmi_ids_string);
+			pr_info("%s\n", dmi_ids_string);
 			return 0;
 		}
 	}
@@ -663,7 +666,7 @@ static int __init dmi_smbios3_present(const u8 *buf)
 				dmi_ver >> 16, (dmi_ver >> 8) & 0xFF,
 				dmi_ver & 0xFF);
 			dmi_format_ids(dmi_ids_string, sizeof(dmi_ids_string));
-			pr_info("DMI: %s\n", dmi_ids_string);
+			pr_info("%s\n", dmi_ids_string);
 			return 0;
 		}
 	}
@@ -758,7 +761,7 @@ static void __init dmi_scan_machine(void)
 		dmi_early_unmap(p, 0x10000);
 	}
  error:
-	pr_info("DMI not present or invalid.\n");
+	pr_info("not present or invalid.\n");
 }
 
 static __ro_after_init BIN_ATTR_SIMPLE_ADMIN_RO(smbios_entry_point);
@@ -810,7 +813,7 @@ static int __init dmi_init(void)
 	kobject_del(tables_kobj);
 	kobject_put(tables_kobj);
  err:
-	pr_err("dmi: Firmware registration failed.\n");
+	pr_err("Firmware registration failed.\n");
 
 	return ret;
 }
@@ -831,7 +834,7 @@ void __init dmi_setup(void)
 		return;
 
 	dmi_memdev_walk();
-	pr_info("DMI: Memory slots populated: %d/%d\n",
+	pr_info("Memory slots populated: %d/%d\n",
 		dmi_memdev_populated_nr, dmi_memdev_nr);
 	dump_stack_set_arch_desc("%s", dmi_ids_string);
 }
-- 
2.43.0

[PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Mario Limonciello (AMD)]

From: Yazen Ghannam <yazen.ghannam@amd.com>

Type 40 entries (Additional Information) are summarized in section 7.41
as part of the SMBIOS specification.  Generally, these entries aren't
interesting to save.

However on some AMD Zen systems, the AGESA version is stored here. This
is useful to save to the kernel message logs for debugging. It can be
used to cross-reference issues.

Implement an iterator for the Additional Information entries. Use this
to find and print the AGESA string. Do so in AMD code, since the use
case is AMD-specific.

Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Co-developed-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
Signed-off-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
---
v5:
 * Return "" in !CONFIG_DMI case (LKP robot, Yazen)
v4:
 * New patch (based upon older versions though)
---
 arch/x86/kernel/cpu/amd.c   | 53 +++++++++++++++++++++++++++++++++++++
 drivers/firmware/dmi_scan.c |  3 ++-
 include/linux/dmi.h         | 18 +++++++++++++
 3 files changed, 73 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index bc94ff1e250ad..aa04a27aeb107 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0-only
 #include <linux/export.h>
 #include <linux/bitops.h>
+#include <linux/dmi.h>
 #include <linux/elf.h>
 #include <linux/mm.h>
 #include <linux/kvm_types.h>
@@ -1404,3 +1405,55 @@ static __init int print_s5_reset_status_mmio(void)
 	return 0;
 }
 late_initcall(print_s5_reset_status_mmio);
+
+static void __init amd_dmi_scan_additional(const struct dmi_header *d, void *p)
+{
+	struct dmi_a_info *info = (struct dmi_a_info *)d;
+	void *next, *end;
+
+	/*
+	 * DMI Additional Info table has a 'count' field. But it's not very
+	 * helpful since the entries are variable length. So don't use it.
+	 */
+	if (info->header.type != DMI_ENTRY_ADDITIONAL ||
+	    info->header.length < DMI_A_INFO_MIN_SIZE)
+		return;
+
+	/*
+	 * Get the first entry.
+	 * The minimum table size guarantees at least one entry is present.
+	 */
+	next = (void *)(info + 1);
+	end  = (void *)info + info->header.length;
+
+	do {
+		struct dmi_a_info_entry *entry;
+		const char *string_ptr;
+
+		entry = (struct dmi_a_info_entry *)next;
+
+		/*
+		 * Not much can be done to validate data. At least the entry
+		 * length shouldn't be 0.
+		 */
+		if (!entry->length)
+			return;
+
+		string_ptr = dmi_string_nosave(&info->header, entry->str_num);
+
+		/* Only one AGESA string is expected. */
+		if (!strncmp(string_ptr, "AGESA", 5)) {
+			pr_info("%s\n", string_ptr);
+			break;
+		}
+
+		next += entry->length;
+	} while (end - next >= DMI_A_INFO_ENT_MIN_SIZE);
+}
+
+static __init int print_dmi_agesa(void)
+{
+	dmi_walk(amd_dmi_scan_additional, NULL);
+	return 0;
+}
+late_initcall(print_dmi_agesa);
diff --git a/drivers/firmware/dmi_scan.c b/drivers/firmware/dmi_scan.c
index ed6235ac576b6..a3f7dabd49554 100644
--- a/drivers/firmware/dmi_scan.c
+++ b/drivers/firmware/dmi_scan.c
@@ -47,7 +47,7 @@ static struct dmi_memdev_info {
 static int dmi_memdev_nr;
 static int dmi_memdev_populated_nr __initdata;
 
-static const char * __init dmi_string_nosave(const struct dmi_header *dm, u8 s)
+const char *dmi_string_nosave(const struct dmi_header *dm, u8 s)
 {
 	const u8 *bp = ((u8 *) dm) + dm->length;
 	const u8 *nsp;
@@ -66,6 +66,7 @@ static const char * __init dmi_string_nosave(const struct dmi_header *dm, u8 s)
 
 	return dmi_empty_string;
 }
+EXPORT_SYMBOL_GPL(dmi_string_nosave);
 
 static const char * __init dmi_string(const struct dmi_header *dm, u8 s)
 {
diff --git a/include/linux/dmi.h b/include/linux/dmi.h
index 2eedf44e68012..c8700e6a694d1 100644
--- a/include/linux/dmi.h
+++ b/include/linux/dmi.h
@@ -91,6 +91,21 @@ struct dmi_device {
 	void *device_data;	/* Type specific data */
 };
 
+#define DMI_A_INFO_ENT_MIN_SIZE 0x6
+struct dmi_a_info_entry {
+	u8 length;
+	u16 handle;
+	u8 offset;
+	u8 str_num;
+	u8 value[];
+} __packed;
+
+#define DMI_A_INFO_MIN_SIZE	0xB
+struct dmi_a_info {
+	struct dmi_header header;
+	u8 count;
+} __packed;
+
 #ifdef CONFIG_DMI
 
 struct dmi_dev_onboard {
@@ -120,6 +135,7 @@ extern void dmi_memdev_name(u16 handle, const char **bank, const char **device);
 extern u64 dmi_memdev_size(u16 handle);
 extern u8 dmi_memdev_type(u16 handle);
 extern u16 dmi_memdev_handle(int slot);
+const char *dmi_string_nosave(const struct dmi_header *dm, u8 s);
 
 #else
 
@@ -153,6 +169,8 @@ static inline u8 dmi_memdev_type(u16 handle) { return 0x0; }
 static inline u16 dmi_memdev_handle(int slot) { return 0xffff; }
 static inline const struct dmi_system_id *
 	dmi_first_match(const struct dmi_system_id *list) { return NULL; }
+static inline const char *
+	dmi_string_nosave(const struct dmi_header *dm, u8 s) { return ""; }
 
 #endif
 
-- 
2.43.0

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Borislav Petkov]

On Fri, Jan 23, 2026 at 03:30:56PM -0600, Mario Limonciello (AMD) wrote:
> From: Yazen Ghannam <yazen.ghannam@amd.com>
> 
> Type 40 entries (Additional Information) are summarized in section 7.41
> as part of the SMBIOS specification.  Generally, these entries aren't
> interesting to save.
> 
> However on some AMD Zen systems, the AGESA version is stored here. This
> is useful to save to the kernel message logs for debugging. It can be
> used to cross-reference issues.
> 
> Implement an iterator for the Additional Information entries. Use this
> to find and print the AGESA string. Do so in AMD code, since the use
> case is AMD-specific.

Please do not explain the diff.

> Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
> Co-developed-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
> Signed-off-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
> ---
> v5:
>  * Return "" in !CONFIG_DMI case (LKP robot, Yazen)
> v4:
>  * New patch (based upon older versions though)
> ---
>  arch/x86/kernel/cpu/amd.c   | 53 +++++++++++++++++++++++++++++++++++++
>  drivers/firmware/dmi_scan.c |  3 ++-
>  include/linux/dmi.h         | 18 +++++++++++++
>  3 files changed, 73 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
> index bc94ff1e250ad..aa04a27aeb107 100644
> --- a/arch/x86/kernel/cpu/amd.c
> +++ b/arch/x86/kernel/cpu/amd.c
> @@ -1,6 +1,7 @@
>  // SPDX-License-Identifier: GPL-2.0-only
>  #include <linux/export.h>
>  #include <linux/bitops.h>
> +#include <linux/dmi.h>
>  #include <linux/elf.h>
>  #include <linux/mm.h>
>  #include <linux/kvm_types.h>
> @@ -1404,3 +1405,55 @@ static __init int print_s5_reset_status_mmio(void)
>  	return 0;
>  }
>  late_initcall(print_s5_reset_status_mmio);
> +
> +static void __init amd_dmi_scan_additional(const struct dmi_header *d, void *p)

It's a static - no need for the "amd_" prefix.

> +{
> +	struct dmi_a_info *info = (struct dmi_a_info *)d;
> +	void *next, *end;

This needs a 

	if (!IS_ENABLED(CONFIG_DMI))
		return;

> +
> +	/*
> +	 * DMI Additional Info table has a 'count' field. But it's not very
> +	 * helpful since the entries are variable length. So don't use it.
> +	 */
> +	if (info->header.type != DMI_ENTRY_ADDITIONAL ||
> +	    info->header.length < DMI_A_INFO_MIN_SIZE)
> +		return;
> +
> +	/*
> +	 * Get the first entry.
> +	 * The minimum table size guarantees at least one entry is present.
> +	 */
> +	next = (void *)(info + 1);
> +	end  = (void *)info + info->header.length;

This end needs to be corrected to some sane upper limit because we don't trust
firmware one bit.

> +
> +	do {
> +		struct dmi_a_info_entry *entry;
> +		const char *string_ptr;
> +
> +		entry = (struct dmi_a_info_entry *)next;
> +
> +		/*
> +		 * Not much can be done to validate data. At least the entry
> +		 * length shouldn't be 0.
> +		 */
> +		if (!entry->length)
> +			return;
> +
> +		string_ptr = dmi_string_nosave(&info->header, entry->str_num);
> +
> +		/* Only one AGESA string is expected. */

Superfluous comment.

> +		if (!strncmp(string_ptr, "AGESA", 5)) {
> +			pr_info("%s\n", string_ptr);

So on this Zen5 laptop here, this dumps:

[    0.840480] AGESA!V9 StrixPI-FP8 1.0.0.1d

I guess we can make that:

[    0.840480] AGESA: V9 StrixPI-FP8 1.0.0.1d

no?

Thx.

-- 
Regards/Gruss,
    Boris.

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

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Mario Limonciello (AMD) (kernel.org)]

On 1/26/2026 7:07 AM, Borislav Petkov wrote:
> On Fri, Jan 23, 2026 at 03:30:56PM -0600, Mario Limonciello (AMD) wrote:
>> From: Yazen Ghannam <yazen.ghannam@amd.com>
>>
>> Type 40 entries (Additional Information) are summarized in section 7.41
>> as part of the SMBIOS specification.  Generally, these entries aren't
>> interesting to save.
>>
>> However on some AMD Zen systems, the AGESA version is stored here. This
>> is useful to save to the kernel message logs for debugging. It can be
>> used to cross-reference issues.
>>
>> Implement an iterator for the Additional Information entries. Use this
>> to find and print the AGESA string. Do so in AMD code, since the use
>> case is AMD-specific.
> 
> Please do not explain the diff.
> 
👍
>> Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
>> Co-developed-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
>> Signed-off-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
>> ---
>> v5:
>>   * Return "" in !CONFIG_DMI case (LKP robot, Yazen)
>> v4:
>>   * New patch (based upon older versions though)
>> ---
>>   arch/x86/kernel/cpu/amd.c   | 53 +++++++++++++++++++++++++++++++++++++
>>   drivers/firmware/dmi_scan.c |  3 ++-
>>   include/linux/dmi.h         | 18 +++++++++++++
>>   3 files changed, 73 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
>> index bc94ff1e250ad..aa04a27aeb107 100644
>> --- a/arch/x86/kernel/cpu/amd.c
>> +++ b/arch/x86/kernel/cpu/amd.c
>> @@ -1,6 +1,7 @@
>>   // SPDX-License-Identifier: GPL-2.0-only
>>   #include <linux/export.h>
>>   #include <linux/bitops.h>
>> +#include <linux/dmi.h>
>>   #include <linux/elf.h>
>>   #include <linux/mm.h>
>>   #include <linux/kvm_types.h>
>> @@ -1404,3 +1405,55 @@ static __init int print_s5_reset_status_mmio(void)
>>   	return 0;
>>   }
>>   late_initcall(print_s5_reset_status_mmio);
>> +
>> +static void __init amd_dmi_scan_additional(const struct dmi_header *d, void *p)
> 
> It's a static - no need for the "amd_" prefix.
> 
>> +{
>> +	struct dmi_a_info *info = (struct dmi_a_info *)d;
>> +	void *next, *end;
> 
> This needs a
> 
> 	if (!IS_ENABLED(CONFIG_DMI))
> 		return;
> 
👍
>> +
>> +	/*
>> +	 * DMI Additional Info table has a 'count' field. But it's not very
>> +	 * helpful since the entries are variable length. So don't use it.
>> +	 */
>> +	if (info->header.type != DMI_ENTRY_ADDITIONAL ||
>> +	    info->header.length < DMI_A_INFO_MIN_SIZE)
>> +		return;
>> +
>> +	/*
>> +	 * Get the first entry.
>> +	 * The minimum table size guarantees at least one entry is present.
>> +	 */
>> +	next = (void *)(info + 1);
>> +	end  = (void *)info + info->header.length;
> 
> This end needs to be corrected to some sane upper limit because we don't trust
> firmware one bit.
> 
👍
>> +
>> +	do {
>> +		struct dmi_a_info_entry *entry;
>> +		const char *string_ptr;
>> +
>> +		entry = (struct dmi_a_info_entry *)next;
>> +
>> +		/*
>> +		 * Not much can be done to validate data. At least the entry
>> +		 * length shouldn't be 0.
>> +		 */
>> +		if (!entry->length)
>> +			return;
>> +
>> +		string_ptr = dmi_string_nosave(&info->header, entry->str_num);
>> +
>> +		/* Only one AGESA string is expected. */
> 
> Superfluous comment.

👍
> 
>> +		if (!strncmp(string_ptr, "AGESA", 5)) {
>> +			pr_info("%s\n", string_ptr);
> 
> So on this Zen5 laptop here, this dumps:
> 
> [    0.840480] AGESA!V9 StrixPI-FP8 1.0.0.1d
> 
> I guess we can make that:
> 
> [    0.840480] AGESA: V9 StrixPI-FP8 1.0.0.1d
> 
> no?

I think if slicing and dicing the string I would instead have us "match" 
"AGESA!V9" instead and then spit it out like this:

AGESA: StrixPI-FP8 1.0.0.1d

Then if there's a v10 sentinel some day or something like that we would 
add a new match then.

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Borislav Petkov]

On Mon, Jan 26, 2026 at 09:15:58AM -0600, Mario Limonciello (AMD) (kernel.org) wrote:
> I think if slicing and dicing the string I would instead have us "match"
> "AGESA!V9" instead and then spit it out like this:
> 
> AGESA: StrixPI-FP8 1.0.0.1d
> 
> Then if there's a v10 sentinel some day or something like that we would add
> a new match then.

Or, we can simply keep it simple and do:

[    0.840480] AGESA: AGESA!V9 StrixPI-FP8 1.0.0.1d

A little redundancy won't hurt - having the version is much more important.

Thx.

-- 
Regards/Gruss,
    Boris.

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

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Mario Limonciello (AMD) (kernel.org)]

On 1/26/2026 9:40 AM, Borislav Petkov wrote:
> On Mon, Jan 26, 2026 at 09:15:58AM -0600, Mario Limonciello (AMD) (kernel.org) wrote:
>> I think if slicing and dicing the string I would instead have us "match"
>> "AGESA!V9" instead and then spit it out like this:
>>
>> AGESA: StrixPI-FP8 1.0.0.1d
>>
>> Then if there's a v10 sentinel some day or something like that we would add
>> a new match then.
> 
> Or, we can simply keep it simple and do:
> 
> [    0.840480] AGESA: AGESA!V9 StrixPI-FP8 1.0.0.1d
> 
> A little redundancy won't hurt - having the version is much more important.
> 
> Thx.
> 

Well now you see why I was proposing the x86/amd: prefix in the first 
place.  It could be no slicing and dicing and with that prefix:

x86/amd: AGESA!V9 StrixPI-FP8 1.0.0.1d

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Borislav Petkov]

On Mon, Jan 26, 2026 at 10:41:21AM -0600, Mario Limonciello (AMD) (kernel.org) wrote:
> Well now you see why I was proposing the x86/amd: prefix in the first place.
> It could be no slicing and dicing and with that prefix:
> 
> x86/amd: AGESA!V9 StrixPI-FP8 1.0.0.1d

So the prefix tells so that it is somewhere from x86/.../amd.

The "AGESA:" prefix tells you it is AGESA.

I don't see how the "x86/amd" prefix works here...

-- 
Regards/Gruss,
    Boris.

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

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Yazen Ghannam]

On Mon, Jan 26, 2026 at 05:54:12PM +0100, Borislav Petkov wrote:
> On Mon, Jan 26, 2026 at 10:41:21AM -0600, Mario Limonciello (AMD) (kernel.org) wrote:
> > Well now you see why I was proposing the x86/amd: prefix in the first place.
> > It could be no slicing and dicing and with that prefix:
> > 
> > x86/amd: AGESA!V9 StrixPI-FP8 1.0.0.1d
> 
> So the prefix tells so that it is somewhere from x86/.../amd.
> 
> The "AGESA:" prefix tells you it is AGESA.
> 
> I don't see how the "x86/amd" prefix works here...
> 

Mario, what's your intended debug flow?

Do you want to grab any and all messages related to AMD and then go
through them? So here you'd like "x86/amd" or something similar as a
first pass filter.

Or do you look for specific pieces of data, as needed? So here you'd
just look for "AGESA" if you want that or "reset reason", etc.

Thanks,
Yazen

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Mario Limonciello]

On 1/28/26 8:39 AM, Yazen Ghannam wrote:
> On Mon, Jan 26, 2026 at 05:54:12PM +0100, Borislav Petkov wrote:
>> On Mon, Jan 26, 2026 at 10:41:21AM -0600, Mario Limonciello (AMD) (kernel.org) wrote:
>>> Well now you see why I was proposing the x86/amd: prefix in the first place.
>>> It could be no slicing and dicing and with that prefix:
>>>
>>> x86/amd: AGESA!V9 StrixPI-FP8 1.0.0.1d
>>
>> So the prefix tells so that it is somewhere from x86/.../amd.
>>
>> The "AGESA:" prefix tells you it is AGESA.
>>
>> I don't see how the "x86/amd" prefix works here...
>>
> 
> Mario, what's your intended debug flow?
> 
> Do you want to grab any and all messages related to AMD and then go
> through them? So here you'd like "x86/amd" or something similar as a
> first pass filter.
> 
> Or do you look for specific pieces of data, as needed? So here you'd
> just look for "AGESA" if you want that or "reset reason", etc.
> 
> Thanks,
> Yazen

Like all things - it's a case by case basis based upon the ticket and 
what's wrong.

I don't see any problem with just "grep AGESA" though.

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Borislav Petkov]

On Fri, Jan 23, 2026 at 03:30:56PM -0600, Mario Limonciello (AMD) wrote:
> From: Yazen Ghannam <yazen.ghannam@amd.com>
> 
> Type 40 entries (Additional Information) are summarized in section 7.41
> as part of the SMBIOS specification.  Generally, these entries aren't
> interesting to save.
> 
> However on some AMD Zen systems, the AGESA version is stored here. This
> is useful to save to the kernel message logs for debugging. It can be
> used to cross-reference issues.
> 
> Implement an iterator for the Additional Information entries. Use this
> to find and print the AGESA string. Do so in AMD code, since the use
> case is AMD-specific.

Look what I said the last time:

https://lore.kernel.org/r/20260126130755.GIaXdnK65MTIiXqg-l@fat_crate.local

I'll have to teach AI to give feedback for commit messages explaining the
diff.

> Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
> Co-developed-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
> Signed-off-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
> ---
> v5:
>  * Return "" in !CONFIG_DMI case (LKP robot, Yazen)
> v4:
>  * New patch (based upon older versions though)
> ---
>  arch/x86/kernel/cpu/amd.c   | 53 +++++++++++++++++++++++++++++++++++++
>  drivers/firmware/dmi_scan.c |  3 ++-
>  include/linux/dmi.h         | 18 +++++++++++++
>  3 files changed, 73 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
> index bc94ff1e250ad..aa04a27aeb107 100644
> --- a/arch/x86/kernel/cpu/amd.c
> +++ b/arch/x86/kernel/cpu/amd.c
> @@ -1,6 +1,7 @@
>  // SPDX-License-Identifier: GPL-2.0-only
>  #include <linux/export.h>
>  #include <linux/bitops.h>
> +#include <linux/dmi.h>
>  #include <linux/elf.h>
>  #include <linux/mm.h>
>  #include <linux/kvm_types.h>
> @@ -1404,3 +1405,55 @@ static __init int print_s5_reset_status_mmio(void)
>  	return 0;
>  }
>  late_initcall(print_s5_reset_status_mmio);
> +
> +static void __init amd_dmi_scan_additional(const struct dmi_header *d, void *p)
> +{
> +	struct dmi_a_info *info = (struct dmi_a_info *)d;
> +	void *next, *end;
> +
> +	/*
> +	 * DMI Additional Info table has a 'count' field. But it's not very
> +	 * helpful since the entries are variable length. So don't use it.
> +	 */
> +	if (info->header.type != DMI_ENTRY_ADDITIONAL ||
> +	    info->header.length < DMI_A_INFO_MIN_SIZE)
> +		return;
> +
> +	/*
> +	 * Get the first entry.
> +	 * The minimum table size guarantees at least one entry is present.
> +	 */
> +	next = (void *)(info + 1);
> +	end  = (void *)info + info->header.length;

From last time:

"This end needs to be corrected to some sane upper limit because we don't trust
firmware one bit."

Lemme stop review here until you address all review comments from the last
time.

-- 
Regards/Gruss,
    Boris.

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

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Mario Limonciello]

On 3/15/26 8:14 AM, Borislav Petkov wrote:
> On Fri, Jan 23, 2026 at 03:30:56PM -0600, Mario Limonciello (AMD) wrote:
>> From: Yazen Ghannam <yazen.ghannam@amd.com>
>>
>> Type 40 entries (Additional Information) are summarized in section 7.41
>> as part of the SMBIOS specification.  Generally, these entries aren't
>> interesting to save.
>>
>> However on some AMD Zen systems, the AGESA version is stored here. This
>> is useful to save to the kernel message logs for debugging. It can be
>> used to cross-reference issues.
>>
>> Implement an iterator for the Additional Information entries. Use this
>> to find and print the AGESA string. Do so in AMD code, since the use
>> case is AMD-specific.
> 
> Look what I said the last time:

FYI - you're replying to the v5 not the v6 [1].

Link: 
https://lore.kernel.org/all/20260307141024.819807-1-superm1@kernel.org/ [1]

> 
> https://lore.kernel.org/r/20260126130755.GIaXdnK65MTIiXqg-l@fat_crate.local
> 
> I'll have to teach AI to give feedback for commit messages explaining the
> diff.

Sorry about that one, I missed it even in v6.

> 
>> Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
>> Co-developed-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
>> Signed-off-by: "Mario Limonciello (AMD)" <superm1@kernel.org>
>> ---
>> v5:
>>   * Return "" in !CONFIG_DMI case (LKP robot, Yazen)
>> v4:
>>   * New patch (based upon older versions though)
>> ---
>>   arch/x86/kernel/cpu/amd.c   | 53 +++++++++++++++++++++++++++++++++++++
>>   drivers/firmware/dmi_scan.c |  3 ++-
>>   include/linux/dmi.h         | 18 +++++++++++++
>>   3 files changed, 73 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
>> index bc94ff1e250ad..aa04a27aeb107 100644
>> --- a/arch/x86/kernel/cpu/amd.c
>> +++ b/arch/x86/kernel/cpu/amd.c
>> @@ -1,6 +1,7 @@
>>   // SPDX-License-Identifier: GPL-2.0-only
>>   #include <linux/export.h>
>>   #include <linux/bitops.h>
>> +#include <linux/dmi.h>
>>   #include <linux/elf.h>
>>   #include <linux/mm.h>
>>   #include <linux/kvm_types.h>
>> @@ -1404,3 +1405,55 @@ static __init int print_s5_reset_status_mmio(void)
>>   	return 0;
>>   }
>>   late_initcall(print_s5_reset_status_mmio);
>> +
>> +static void __init amd_dmi_scan_additional(const struct dmi_header *d, void *p)
>> +{
>> +	struct dmi_a_info *info = (struct dmi_a_info *)d;
>> +	void *next, *end;
>> +
>> +	/*
>> +	 * DMI Additional Info table has a 'count' field. But it's not very
>> +	 * helpful since the entries are variable length. So don't use it.
>> +	 */
>> +	if (info->header.type != DMI_ENTRY_ADDITIONAL ||
>> +	    info->header.length < DMI_A_INFO_MIN_SIZE)
>> +		return;
>> +
>> +	/*
>> +	 * Get the first entry.
>> +	 * The minimum table size guarantees at least one entry is present.
>> +	 */
>> +	next = (void *)(info + 1);
>> +	end  = (void *)info + info->header.length;
> 
>  From last time:
> 
> "This end needs to be corrected to some sane upper limit because we don't trust
> firmware one bit."
> 

I did add a check for info->count < 1, but that's not what you were 
actually looking for.  Do you have a suggestion for an upper limit to 
bound things?

I was thinking something like this:

#define DMI_LIMIT 10
int limit = info -> count > DMI_LIMIT ? DMI_LIMIT : info->count;
end  = (void *)info + limit * info->header.length;

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Borislav Petkov]

On Sun, Mar 15, 2026 at 10:14:48AM -0500, Mario Limonciello wrote:
> FYI - you're replying to the v5 not the v6 [1].

Whoops, sorry about that.

It looks I've still used the right set tho:

v6_20260307_superm1_print_agesa_version_at_bootup.mbx

:)

> I did add a check for info->count < 1, but that's not what you were actually
> looking for.  Do you have a suggestion for an upper limit to bound things?
> 
> I was thinking something like this:
> 
> #define DMI_LIMIT 10
> int limit = info -> count > DMI_LIMIT ? DMI_LIMIT : info->count;
> end  = (void *)info + limit * info->header.length;

So that Zen5 laptop here says:

[    0.853836] dmi_scan_additional: next: 0xffffc9000051950c, end: 0xffffc90000519515, hdr length: 14
[    0.854364] entry 0: length: 9
[    0.854576] AGESA: StrixPI-FP8 1.0.0.1d

So after that first loop pass we have

	next = 0xffffc9000051950c + 9 = 0xffffc90000519515

And we should terminate. So I guess that's not a lot of entries.

In terms of limiting, I'd do something like:

/* A generous max size for the additional DMI entries */
#define MAX_DMI_ADDITIONAL_ENTRIES_LEN		256

	if (info->header.length > MAX_DMI_ADDITIONAL_ENTRIES_LEN)
		info->header.length = MAX_DMI_ADDITIONAL_ENTRIES_LEN;

And now you have a sane upper limit and we don't rely on any numbers the BIOS
is telling us.

Thx.

-- 
Regards/Gruss,
    Boris.

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

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Mario Limonciello]

On 3/15/26 3:56 PM, Borislav Petkov wrote:
> On Sun, Mar 15, 2026 at 10:14:48AM -0500, Mario Limonciello wrote:
>> FYI - you're replying to the v5 not the v6 [1].
> 
> Whoops, sorry about that.
> 
> It looks I've still used the right set tho:
> 
> v6_20260307_superm1_print_agesa_version_at_bootup.mbx
> 
> :)
> 
>> I did add a check for info->count < 1, but that's not what you were actually
>> looking for.  Do you have a suggestion for an upper limit to bound things?
>>
>> I was thinking something like this:
>>
>> #define DMI_LIMIT 10
>> int limit = info -> count > DMI_LIMIT ? DMI_LIMIT : info->count;
>> end  = (void *)info + limit * info->header.length;
> 
> So that Zen5 laptop here says:
> 
> [    0.853836] dmi_scan_additional: next: 0xffffc9000051950c, end: 0xffffc90000519515, hdr length: 14
> [    0.854364] entry 0: length: 9
> [    0.854576] AGESA: StrixPI-FP8 1.0.0.1d
> 
> So after that first loop pass we have
> 
> 	next = 0xffffc9000051950c + 9 = 0xffffc90000519515
> 
> And we should terminate. So I guess that's not a lot of entries.
> 
> In terms of limiting, I'd do something like:
> 
> /* A generous max size for the additional DMI entries */
> #define MAX_DMI_ADDITIONAL_ENTRIES_LEN		256
> 
> 	if (info->header.length > MAX_DMI_ADDITIONAL_ENTRIES_LEN)
> 		info->header.length = MAX_DMI_ADDITIONAL_ENTRIES_LEN;
> 
> And now you have a sane upper limit and we don't rely on any numbers the BIOS
> is telling us.
> 
> Thx.
> 

But that can't actually happen.  info->header.length is u8.

So we're never going to end up with that many entries such that it would 
exhaust the check.

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Borislav Petkov]

On Mon, Mar 16, 2026 at 09:33:04PM -0500, Mario Limonciello wrote:
> But that can't actually happen.  info->header.length is u8.
> 
> So we're never going to end up with that many entries such that it would
> exhaust the check.

Ah ok, I didn't see that. That's good then. I guess we can start queueing the
next version you send with all the feedback incorporated.

Thx.

-- 
Regards/Gruss,
    Boris.

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

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Mario Limonciello]

On 3/17/26 13:24, Borislav Petkov wrote:
> On Mon, Mar 16, 2026 at 09:33:04PM -0500, Mario Limonciello wrote:
>> But that can't actually happen.  info->header.length is u8.
>>
>> So we're never going to end up with that many entries such that it would
>> exhaust the check.
> 
> Ah ok, I didn't see that. That's good then. I guess we can start queueing the
> next version you send with all the feedback incorporated.
> 
> Thx.
> 

The only remaining feedback is commit message IIRC.  Do you want to just 
massage that yourself while committing or do you want me to spin again 
and do it?

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Borislav Petkov]

On Tue, Mar 17, 2026 at 01:28:02PM -0500, Mario Limonciello wrote:
> The only remaining feedback is commit message IIRC.  Do you want to just
> massage that yourself while committing or do you want me to spin again and
> do it?

Sure, I can do that.

@Jean, ack to the first 4 patches to go through tip?

Thx.

-- 
Regards/Gruss,
    Boris.

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

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Jean Delvare]

Hi Borislav,

On Tue, 17 Mar 2026 20:18:42 +0100, Borislav Petkov wrote:
> On Tue, Mar 17, 2026 at 01:28:02PM -0500, Mario Limonciello wrote:
> > The only remaining feedback is commit message IIRC.  Do you want to just
> > massage that yourself while committing or do you want me to spin again and
> > do it?  
> 
> Sure, I can do that.
> 
> @Jean, ack to the first 4 patches to go through tip?

I have reviewed the first 4 patches and they look good to me. You can
add:

Reviewed-by: Jean Delvare <jdelvare@suse.de>

to the patches which do not have it yet, and I am OK with them going
through tip.

Thanks Mario, Borislav and Yazen for all the work on this series!

-- 
Jean Delvare
SUSE L3 Support

Re: [PATCH v5 5/5] x86/CPU/AMD: Print AGESA string from DMI additional information entry

[Borislav Petkov]

On Tue, Mar 31, 2026 at 07:59:03PM +0200, Jean Delvare wrote:
> Hi Borislav,
> 
> On Tue, 17 Mar 2026 20:18:42 +0100, Borislav Petkov wrote:
> > On Tue, Mar 17, 2026 at 01:28:02PM -0500, Mario Limonciello wrote:
> > > The only remaining feedback is commit message IIRC.  Do you want to just
> > > massage that yourself while committing or do you want me to spin again and
> > > do it?  
> > 
> > Sure, I can do that.
> > 
> > @Jean, ack to the first 4 patches to go through tip?
> 
> I have reviewed the first 4 patches and they look good to me. You can
> add:
> 
> Reviewed-by: Jean Delvare <jdelvare@suse.de>
> 
> to the patches which do not have it yet, and I am OK with them going
> through tip.
> 
> Thanks Mario, Borislav and Yazen for all the work on this series!

Cool, thanks for the review!

I'll queue them soon.

-- 
Regards/Gruss,
    Boris.

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