Skip to contentSkip to conversation

efi_vars: Implement SPI Flash storage for EFI

PATCH Superseded Michal Simek 4 participants 15 messages Jan 14 – Jan 23

v4 of 8 Superseded Latest is v8 →

lore
Share & tools
lore
#2 lore raw

PATCH v4 2/3 efi_var: Unify read/write access helper function

In reply to Michal Simek

efi_var_to/from_file() suggest method where variables are placed. But there is no reason for it and generic name can be used to wire also different locations for variables.

---

Changes in v4:
- New patch based on review comments from v3

 include/efi_variable.h        | 13 ++++---------
 lib/efi_loader/efi_var_file.c |  8 ++++----
 lib/efi_loader/efi_variable.c |  4 ++--
 3 files changed, 10 insertions(+), 15 deletions(-)
include/efi_variable.h +4 -9
index ee68fa4a885f..ae542c597129 100644
@@ -137,13 +137,11 @@ struct efi_var_file {
 };
 
 /**
- * efi_var_to_file() - save non-volatile variables as file
- *
- * File ubootefi.var is created on the EFI system partion.
+ * efi_var_write() - save non-volatile variables
  *
  * Return:	status code
  */
-efi_status_t efi_var_to_file(void);
+efi_status_t efi_var_write(void);
 
 /**
  * efi_var_collect() - collect variables in buffer
@@ -178,17 +176,14 @@ efi_status_t __maybe_unused efi_var_collect(struct efi_var_file **bufp, loff_t *
 efi_status_t efi_var_restore(struct efi_var_file *buf, bool safe);
 
 /**
- * efi_var_from_file() - read variables from file
- *
- * File ubootefi.var is read from the EFI system partitions and the variables
- * stored in the file are created.
+ * efi_var_read() - read variables
  *
  * In case the file does not exist yet or a variable cannot be set EFI_SUCCESS
  * is returned.
  *
  * Return:	status code
  */
-efi_status_t efi_var_from_file(void);
+efi_status_t efi_var_read(void);
 
 /**
  * efi_var_mem_init() - set-up variable list
lib/efi_loader/efi_var_file.c +4 -4
index d32edaac277d..01ac7cc20e1f 100644
@@ -43,13 +43,13 @@ static efi_status_t __maybe_unused efi_set_blk_dev_to_system_partition(void)
 }
 
 /**
- * efi_var_to_file() - save non-volatile variables as file
+ * efi_var_write() - save non-volatile variables as file
  *
  * File ubootefi.var is created on the EFI system partion.
  *
  * Return:	status code
  */
-efi_status_t efi_var_to_file(void)
+efi_status_t efi_var_write(void)
 {
 	efi_status_t ret;
 	struct efi_var_file *buf;
@@ -85,7 +85,7 @@ out:
 }
 
 /**
- * efi_var_from_file() - read variables from file
+ * efi_var_read() - read variables from file
  *
  * File ubootefi.var is read from the EFI system partitions and the variables
  * stored in the file are created.
@@ -98,7 +98,7 @@ out:
  *
  * Return:	status code
  */
-efi_status_t efi_var_from_file(void)
+efi_status_t efi_var_read(void)
 {
 	struct efi_var_file *buf;
 	loff_t len;
lib/efi_loader/efi_variable.c +2 -2
index 9394ecf6994c..be670a8e7c25 100644
@@ -402,7 +402,7 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
 	 */
 	if (attributes & EFI_VARIABLE_NON_VOLATILE) {
 #if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
-		efi_var_to_file();
+		efi_var_write();
 #else
 		return EFI_NOT_READY;
 #endif
@@ -600,7 +600,7 @@ efi_status_t efi_init_variables(void)
 		return ret;
 
 #if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
-	ret = efi_var_from_file();
+	ret = efi_var_read();
 	if (ret != EFI_SUCCESS)
 		return ret;
 #endif
Signature · 1 line
2.43.0
#0 lore raw

v4 cover efi_vars: Implement SPI Flash storage for EFI

This is updated series based on v3 version sent here
https://lore.kernel.org/all/[email protected]/

that's why I am continuing on v4 instead of starting from scratch.

Tested on kv260 with saving variables to location where User MTD partition is.
CONFIG_EFI_VARIABLE_SF_STORE=y
CONFIG_EFI_RT_VOLATILE_STORE=y CONFIG_EFI_VARIABLE_SF_OFFSET=0x22a0000

Thanks,
Michal

Changes in v4:
- New patch based on review comments from v3
- use unify methods for reading/writing variable

Changes in v3:
- Fixed compiler warnings.

Changes in v2:
- Refactored efi_var_file to move common parts out as requested
- Changed ifdefs to use CONFIG_IS_DEFINED
- Fixed typos

Michal Simek (1):
efi_var: Unify read/write access helper function

Shantur Rathore (2):
efi_var_file: refactor to move buffer functions
efi_vars: Implement SPI Flash store

 include/efi_variable.h          | 18 +++----
 lib/efi_loader/Kconfig          | 26 +++++++++-
 lib/efi_loader/Makefile         |  3 +-
 lib/efi_loader/efi_var_common.c | 44 ++++++++++++++++
 lib/efi_loader/efi_var_file.c   | 65 ++---------------------
 lib/efi_loader/efi_var_sf.c     | 91 +++++++++++++++++++++++++++++++++
 lib/efi_loader/efi_variable.c   | 16 ++++--
 7 files changed, 187 insertions(+), 76 deletions(-)
 create mode 100644 lib/efi_loader/efi_var_sf.c
Signature · 4 lines
2.43.0

base-commit: 4aa15db6ff8b2bbdcdb082e547967bf4a96fee52
branch: master-next-test
#1 lore raw

PATCH v4 1/3 efi_var_file: refactor to move buffer functions

From: Shantur Rathore <[email protected]>

Currently efi_var_file.c has functions to store/read EFI variables to/from memory buffer. These functions can be used with other EFI variable stores so move them out to efi_var_common.c

---

(no changes since v1)

 include/efi_variable.h          |  5 +++
 lib/efi_loader/Makefile         |  2 +-
 lib/efi_loader/efi_var_common.c | 44 +++++++++++++++++++++++++
 lib/efi_loader/efi_var_file.c   | 57 ---------------------------------
 lib/efi_loader/efi_variable.c   | 10 +++++-
 5 files changed, 59 insertions(+), 59 deletions(-)
include/efi_variable.h +5 -0
index 4065cf45ecaf..ee68fa4a885f 100644
@@ -161,6 +161,11 @@ efi_status_t efi_var_to_file(void);
 efi_status_t __maybe_unused efi_var_collect(struct efi_var_file **bufp, loff_t *lenp,
 					    u32 check_attr_mask);
 
+/* GUID used by Shim to store the MOK database */
+#define SHIM_LOCK_GUID \
+	EFI_GUID(0x605dab50, 0xe046, 0x4300, \
+		 0xab, 0xb6, 0x3d, 0xd8, 0x10, 0xdd, 0x8b, 0x23)
+
 /**
  * efi_var_restore() - restore EFI variables from buffer
  *
lib/efi_loader/Makefile +1 -1
index bfa607c88274..07ae048f2e0c 100644
@@ -53,7 +53,7 @@ ifeq ($(CONFIG_EFI_MM_COMM_TEE),y)
 obj-y += efi_variable_tee.o
 else
 obj-y += efi_variable.o
-obj-y += efi_var_file.o
+obj-$(CONFIG_EFI_VARIABLE_FILE_STORE) += efi_var_file.o
 obj-$(CONFIG_EFI_VARIABLES_PRESEED) += efi_var_seed.o
 endif
 obj-y += efi_watchdog.o
lib/efi_loader/efi_var_common.c +44 -0

53 lines not on this page · raw message

lib/efi_loader/efi_var_file.c +0 -57

89 lines not on this page · raw message

lib/efi_loader/efi_variable.c +9 -1
index f3533f4def3a..9394ecf6994c 100644
@@ -400,8 +400,13 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
 	 * Write non-volatile EFI variables to file
 	 * TODO: check if a value change has occured to avoid superfluous writes
 	 */
-	if (attributes & EFI_VARIABLE_NON_VOLATILE)
+	if (attributes & EFI_VARIABLE_NON_VOLATILE) {
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
 		efi_var_to_file();
+#else
+		return EFI_NOT_READY;
+#endif
+	}
 
 	return EFI_SUCCESS;
 }
@@ -594,9 +599,12 @@ efi_status_t efi_init_variables(void)
 	if (ret != EFI_SUCCESS)
 		return ret;
 
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
 	ret = efi_var_from_file();
 	if (ret != EFI_SUCCESS)
 		return ret;
+#endif
+
 	if (IS_ENABLED(CONFIG_EFI_VARIABLES_PRESEED)) {
 		ret = efi_var_restore((struct efi_var_file *)
 				      __efi_var_file_begin, true);
Signature · 1 line
2.43.0
#3 lore raw

PATCH v4 3/3 efi_vars: Implement SPI Flash store

In reply to Michal Simek

From: Shantur Rathore <[email protected]>

Currently U-boot uses ESP as storage for EFI variables. Devices with SPI Flash are used for storing environment with this commit we allow EFI variables to be stored on SPI Flash.

---

Changes in v4:
- use unify methods for reading/writing variable

Changes in v3:
- Fixed compiler warnings.

Changes in v2:
- Refactored efi_var_file to move common parts out as requested
- Changed ifdefs to use CONFIG_IS_DEFINED
- Fixed typos

 lib/efi_loader/Kconfig        | 26 +++++++++-
 lib/efi_loader/Makefile       |  1 +
 lib/efi_loader/efi_var_sf.c   | 91 +++++++++++++++++++++++++++++++++++
 lib/efi_loader/efi_variable.c |  6 +--
 4 files changed, 120 insertions(+), 4 deletions(-)
 create mode 100644 lib/efi_loader/efi_var_sf.c
lib/efi_loader/Kconfig +25 -1
index 13e44be1d067..e753a3aef36d 100644
@@ -123,6 +123,16 @@ config EFI_VARIABLE_FILE_STORE
 	  Select this option if you want non-volatile UEFI variables to be
 	  stored as file /ubootefi.var on the EFI system partition.
 
+config EFI_VARIABLE_SF_STORE
+	bool "Store non-volatile UEFI variables in SPI Flash"
+	depends on SPI_FLASH
+	help
+	  Select this option if you want non-volatile UEFI variables to be
+	  stored in SPI Flash.
+	  Define CONFIG_EFI_VARIABLE_SF_OFFSET as offset in SPI Flash to use as
+	  the storage for variables. CONFIG_EFI_VAR_BUF_SIZE defines the space
+	  needed.
+
 config EFI_MM_COMM_TEE
 	bool "UEFI variables storage service via the trusted world"
 	depends on OPTEE
@@ -152,7 +162,7 @@ endchoice
 
 config EFI_RT_VOLATILE_STORE
 	bool "Allow variable runtime services in volatile storage (e.g RAM)"
-	depends on EFI_VARIABLE_FILE_STORE
+	depends on EFI_VARIABLE_FILE_STORE || EFI_VARIABLE_SF_STORE
 	help
 	  When EFI variables are stored on file we don't allow SetVariableRT,
 	  since the OS doesn't know how to write that file. At the same time
@@ -193,6 +203,20 @@ config FFA_SHARED_MM_BUF_ADDR
 	  the MM SP in secure world.
 	  It is assumed that the MM SP knows the address of the shared MM communication buffer.
 
+config EFI_VARIABLE_SF_OFFSET
+	hex "EFI variables in SPI flash offset"
+	depends on EFI_VARIABLE_SF_STORE
+	help
+	  Offset from the start of the SPI Flash where EFI variables will be stored.
+	  This should be aligned to the sector size of SPI Flash.
+
+config EFI_VARIABLE_SF_DEVICE_INDEX
+	int "Device Index for target SPI Flash"
+	default 0
+	help
+	  The index of SPI Flash device used for storing EFI variables. This would be
+	  needed if there are more than 1 SPI Flash devices available to use.
+
 config EFI_VARIABLES_PRESEED
 	bool "Initial values for UEFI variables"
 	depends on !COMPILE_TEST
lib/efi_loader/Makefile +1 -0
index 07ae048f2e0c..0eb9826c7ed8 100644
@@ -54,6 +54,7 @@ obj-y += efi_variable_tee.o
 else
 obj-y += efi_variable.o
 obj-$(CONFIG_EFI_VARIABLE_FILE_STORE) += efi_var_file.o
+obj-$(CONFIG_EFI_VARIABLE_SF_STORE) += efi_var_sf.o
 obj-$(CONFIG_EFI_VARIABLES_PRESEED) += efi_var_seed.o
 endif
 obj-y += efi_watchdog.o
lib/efi_loader/efi_var_sf.c new file +91 -0

91 lines not on this page · raw message

lib/efi_loader/efi_variable.c +3 -3
index be670a8e7c25..feab212b245b 100644
@@ -397,11 +397,11 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
 		ret = EFI_SUCCESS;
 
 	/*
-	 * Write non-volatile EFI variables to file
+	 * Write non-volatile EFI variables to file or SPI Flash
 	 * TODO: check if a value change has occured to avoid superfluous writes
 	 */
 	if (attributes & EFI_VARIABLE_NON_VOLATILE) {
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)
 		efi_var_write();
 #else
 		return EFI_NOT_READY;
@@ -599,7 +599,7 @@ efi_status_t efi_init_variables(void)
 	if (ret != EFI_SUCCESS)
 		return ret;
 
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)
 	ret = efi_var_read();
 	if (ret != EFI_SUCCESS)
 		return ret;
Signature · 1 line
2.43.0
#13 lore raw
In reply to Michal Simek

Hi Michal I have some user-facing questions about this concept.

Quoted 34 lines · Michal Simek

This is updated series based on v3 version sent here
https://lore.kernel.org/all/[email protected]/

that's why I am continuing on v4 instead of starting from scratch.

Tested on kv260 with saving variables to location where User MTD partition is.
CONFIG_EFI_VARIABLE_SF_STORE=y
CONFIG_EFI_RT_VOLATILE_STORE=y CONFIG_EFI_VARIABLE_SF_OFFSET=0x22a0000

Thanks,
Michal

Changes in v4:
- New patch based on review comments from v3
- use unify methods for reading/writing variable

Changes in v3:
- Fixed compiler warnings.

Changes in v2:
- Refactored efi_var_file to move common parts out as requested
- Changed ifdefs to use CONFIG_IS_DEFINED
- Fixed typos

Michal Simek (1):
efi_var: Unify read/write access helper function

Shantur Rathore (2):
efi_var_file: refactor to move buffer functions
efi_vars: Implement SPI Flash store

Quoted 10 lines
 include/efi_variable.h          | 18 +++----
 lib/efi_loader/Kconfig          | 26 +++++++++-
 lib/efi_loader/Makefile         |  3 +-
 lib/efi_loader/efi_var_common.c | 44 ++++++++++++++++
 lib/efi_loader/efi_var_file.c   | 65 ++---------------------
 lib/efi_loader/efi_var_sf.c     | 91 +++++++++++++++++++++++++++++++++
 lib/efi_loader/efi_variable.c   | 16 ++++--
 7 files changed, 187 insertions(+), 76 deletions(-)
 create mode 100644 lib/efi_loader/efi_var_sf.c

So far as I'm aware support for EFI variable storage is always integral to a filesystem as the EFI System Partition, even in the case of the specially-handled MMC boot partition facility. With presence of many possible ESP is there then only one active ESP, or do all ESP become part of a pool of available locations? I'm not understanding this clearly what is the purpose of selecting an ESP if the variable storage is not integral to the ESP.

Does having EFI variable storage in SPI Flash always take priority over other ESP presence that might have additional variable storage as a file?

Can (or should?) there be a runtime configurable setting to decide two preferences; does one depend on the other or are these independent?:

1. Preferred ESP
2. Preferred EFI variable storage strategy

Consider not wanting to have a perceived ABI breakage as the SPI Flash layout is previously codified by user expectations of what the vendor initially shipped with the product and having been codified in upstream Linux kernel devicetree definition. For the example where this would be useful for boards where there is a well-known SPI Flash layout, I am thinking of StarFive VisionFive 2 where there is {U-Boot SPL, U-Boot env storage, U-Boot Main at 1MiB offset}; does your implementation of EFI variable storage in SPI Flash have a suggested co-existence with the "gap" in the area that is currently U-Boot env storage?

The use situations I am thinking of:

A. Installing a new OS or booting purpose-built image, where it is intended that having (for example) a USB mass storage class or SD Card of the all-in-one image depends on U-Boot in SPI Flash deciding both that the EFI System Partition on the removable media is a priority over other similar media (other different USB boot devices that may be present, existing eMMC storage module possibly with ESP in addition to the inserted removable SD Card).

B. Preventing the ambiguous selection of any additional removable storage containing an EFI System Partition from breaking the desired boot strategy. In many situations a user wants assurance that the board will boot exactly the same every time regardless of what additional removable media containing EFI System Partition might be connected.

C. Corrupt or unusable selected EFI System Partition prevents any recovery and there is no runtime mechanism to attempt an alternative

Is there a possible mechanism for adding a runtime check (GPIO state, key press, network event) to make the decision between some immutable boot preference, or a user-configured boot preference?

Overall I want to get an understanding of how it might be recommended to make use of this for StarFive VisionFive 2 board target(s). Many hours of troubleshooting headache have been the result of EFI System Partition confusion on board that have a wide variety of storage subsystems (MMC, USB, NVMe, ...) which result in some really strange outcomes with installers of GNU/Linux OS distros; particularly as the devicetree dtb search path currently points to storage subsystems that are typically removable media, or the failure mode where an OS installer is using the selected EFI System Partition that "leads on a path to nowhere" and is not the intended installation target.

-E

#14 lore raw
Quoted 49 lines · E Shattow

Hi Michal I have some user-facing questions about this concept.

Quoted 34 lines · Michal Simek

This is updated series based on v3 version sent here
https://lore.kernel.org/all/[email protected]/

that's why I am continuing on v4 instead of starting from scratch.

Tested on kv260 with saving variables to location where User MTD partition is.
CONFIG_EFI_VARIABLE_SF_STORE=y
CONFIG_EFI_RT_VOLATILE_STORE=y CONFIG_EFI_VARIABLE_SF_OFFSET=0x22a0000

Thanks,
Michal

Changes in v4:
- New patch based on review comments from v3
- use unify methods for reading/writing variable

Changes in v3:
- Fixed compiler warnings.

Changes in v2:
- Refactored efi_var_file to move common parts out as requested
- Changed ifdefs to use CONFIG_IS_DEFINED
- Fixed typos

Michal Simek (1):
efi_var: Unify read/write access helper function

Shantur Rathore (2):
efi_var_file: refactor to move buffer functions
efi_vars: Implement SPI Flash store

Quoted 10 lines
 include/efi_variable.h          | 18 +++----
 lib/efi_loader/Kconfig          | 26 +++++++++-
 lib/efi_loader/Makefile         |  3 +-
 lib/efi_loader/efi_var_common.c | 44 ++++++++++++++++
 lib/efi_loader/efi_var_file.c   | 65 ++---------------------
 lib/efi_loader/efi_var_sf.c     | 91 +++++++++++++++++++++++++++++++++
 lib/efi_loader/efi_variable.c   | 16 ++++--
 7 files changed, 187 insertions(+), 76 deletions(-)
 create mode 100644 lib/efi_loader/efi_var_sf.c
Quoted 7 lines

So far as I'm aware support for EFI variable storage is always integral to a filesystem as the EFI System Partition, even in the case of the specially-handled MMC boot partition facility. With presence of many possible ESP is there then only one active ESP, or do all ESP become part of a pool of available locations? I'm not understanding this clearly what is the purpose of selecting an ESP if the variable storage is not integral to the ESP.

EFI variables have nothing to do with ESP filesystems. Most implementations of EFI variables don't use ESP partitions to write variables, U-Boot did that basically as a work around. If you look at the average UEFI implementation on a Intel device the EFI vars are written to the flash storage where they can be signed and validated to ensure they're not changed without the firmware knowing.

Quoted 2 lines

Does having EFI variable storage in SPI Flash always take priority over other ESP presence that might have additional variable storage as a file?

That would presumably depend on the U-Boot build options.

Quoted 2 lines

Can (or should?) there be a runtime configurable setting to decide two preferences; does one depend on the other or are these independent?:

No. It would be a device decision.

Quoted 12 lines

1. Preferred ESP
2. Preferred EFI variable storage strategy

Consider not wanting to have a perceived ABI breakage as the SPI Flash layout is previously codified by user expectations of what the vendor initially shipped with the product and having been codified in upstream Linux kernel devicetree definition. For the example where this would be useful for boards where there is a well-known SPI Flash layout, I am thinking of StarFive VisionFive 2 where there is {U-Boot SPL, U-Boot env storage, U-Boot Main at 1MiB offset}; does your implementation of EFI variable storage in SPI Flash have a suggested co-existence with the "gap" in the area that is currently U-Boot env storage?

The board maintainer would decide that, or the vendor, or if the user wants to building their own.

Quoted 9 lines

The use situations I am thinking of:

A. Installing a new OS or booting purpose-built image, where it is intended that having (for example) a USB mass storage class or SD Card of the all-in-one image depends on U-Boot in SPI Flash deciding both that the EFI System Partition on the removable media is a priority over other similar media (other different USB boot devices that may be present, existing eMMC storage module possibly with ESP in addition to the inserted removable SD Card).

Well by storing the EFI vars on the SPI flash the user can erase one of the storage options without necessarily impacting the EFI vars without knowing.

The storage of EFI vars on the flash of the firmware is what you get on x86 devices.

Quoted 8 lines

B. Preventing the ambiguous selection of any additional removable storage containing an EFI System Partition from breaking the desired boot strategy. In many situations a user wants assurance that the board will boot exactly the same every time regardless of what additional removable media containing EFI System Partition might be connected.

C. Corrupt or unusable selected EFI System Partition prevents any recovery and there is no runtime mechanism to attempt an alternative

No it doesn't there's EFI fall back options or you could erase the vars from the u-boot cmd line etc.

Quoted 16 lines

Is there a possible mechanism for adding a runtime check (GPIO state, key press, network event) to make the decision between some immutable boot preference, or a user-configured boot preference?

Overall I want to get an understanding of how it might be recommended to make use of this for StarFive VisionFive 2 board target(s). Many hours of troubleshooting headache have been the result of EFI System Partition confusion on board that have a wide variety of storage subsystems (MMC, USB, NVMe, ...) which result in some really strange outcomes with installers of GNU/Linux OS distros; particularly as the devicetree dtb search path currently points to storage subsystems that are typically removable media, or the failure mode where an OS installer is using the selected EFI System Partition that "leads on a path to nowhere" and is not the intended installation target.

-E

#4 lore raw

Re: [PATCH v4 3/3] efi_vars: Implement SPI Flash store

In reply to Michal Simek

Hi Michal

Thanks for taking the time.

[...]

Quoted 19 lines
lib/efi_loader/efi_variable.c +2 -2
index be670a8e7c25..feab212b245b 100644
@@ -397,11 +397,11 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
                ret = EFI_SUCCESS;
 
        /*
-        * Write non-volatile EFI variables to file
+        * Write non-volatile EFI variables to file or SPI Flash
         * TODO: check if a value change has occured to avoid superfluous writes
         */
        if (attributes & EFI_VARIABLE_NON_VOLATILE) {
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)
                efi_var_write();
 #else
                return EFI_NOT_READY;

lib/efi_loader/efi_variable.c efi_status_t efi_init_variables(void) · line 602

Quoted 5 lines
lib/efi_loader/efi_variable.c +1 -1
@@ -599,7 +599,7 @@ efi_status_t efi_init_variables(void)
        if (ret != EFI_SUCCESS)
                return ret;
 
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)

So I think this is might to come back and bite us in the future. The name of the file is a bit misleading, but efi_variable.c is supposed to handle the variables with a file backed storage and efi_variable_tee.c is handling the variables when the storage is isolated in the secure world.
The problem is that each case defines the boottime and runtime services. The functions that are common across cases live in efi_var_common.c. That files defines the boottime and runtime calls, by calling the *_int() variants.

The file backed and SPI flash storage seem to have a lot in common. e.g they both use the memory backend to expose the vairables the the OS, query the available storage etc. But I am not they will end up being 100% identical. If they do this approach is ok. But ifthey don't it's better to expand the efi_var_sf.c you added and add code for efi_variables_boot_exit_notify(), and any *_int() functions that are different.

Heinrich any opinions? I'll need to think through the SPI case inclduing runtime support before taking this in

Thanks
/Ilias

Quoted 6 lines
lib/efi_loader/efi_variable.c +0 -1
        ret = efi_var_read();
        if (ret != EFI_SUCCESS)
                return ret;
--

2.43.0

#5 lore raw

Re: [PATCH v4 3/3] efi_vars: Implement SPI Flash store

Quoted 50 lines · Ilias Apalodimas

Hi Michal

Thanks for taking the time.

[...]

Quoted 19 lines
lib/efi_loader/efi_variable.c +2 -2
index be670a8e7c25..feab212b245b 100644
@@ -397,11 +397,11 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
                ret = EFI_SUCCESS;
 
        /*
-        * Write non-volatile EFI variables to file
+        * Write non-volatile EFI variables to file or SPI Flash
         * TODO: check if a value change has occured to avoid superfluous writes
         */
        if (attributes & EFI_VARIABLE_NON_VOLATILE) {
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)
                efi_var_write();
 #else
                return EFI_NOT_READY;

lib/efi_loader/efi_variable.c efi_status_t efi_init_variables(void) · line 602

Quoted 5 lines
lib/efi_loader/efi_variable.c +1 -1
@@ -599,7 +599,7 @@ efi_status_t efi_init_variables(void)
        if (ret != EFI_SUCCESS)
                return ret;
 
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)

So I think this is might to come back and bite us in the future. The name of the file is a bit misleading, but efi_variable.c is supposed to handle the variables with a file backed storage and efi_variable_tee.c is handling the variables when the storage is isolated in the secure world.
The problem is that each case defines the boottime and runtime services. The functions that are common across cases live in efi_var_common.c. That files defines the boottime and runtime calls, by calling the *_int() variants.

The file backed and SPI flash storage seem to have a lot in common. e.g they both use the memory backend to expose the vairables the the OS, query the available storage etc. But I am not they will end up being 100% identical. If they do this approach is ok. But ifthey don't it's better to expand the efi_var_sf.c you added and add code for efi_variables_boot_exit_notify(), and any *_int() functions that are different.

Quoted 2 lines

Heinrich any opinions? I'll need to think through the SPI case inclduing runtime support before taking this in

Do we also need to consider the case where the SPI flash remains owned by the firmware/secure world, and the case where it's handed over to the OS?

#6 lore raw

Re: [PATCH v4 3/3] efi_vars: Implement SPI Flash store

Quoted 53 lines · Peter Robinson
Quoted 50 lines · Ilias Apalodimas

Hi Michal

Thanks for taking the time.

[...]

Quoted 19 lines
lib/efi_loader/efi_variable.c +2 -2
index be670a8e7c25..feab212b245b 100644
@@ -397,11 +397,11 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
                ret = EFI_SUCCESS;
 
        /*
-        * Write non-volatile EFI variables to file
+        * Write non-volatile EFI variables to file or SPI Flash
         * TODO: check if a value change has occured to avoid superfluous writes
         */
        if (attributes & EFI_VARIABLE_NON_VOLATILE) {
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)
                efi_var_write();
 #else
                return EFI_NOT_READY;

lib/efi_loader/efi_variable.c efi_status_t efi_init_variables(void) · line 602

Quoted 5 lines
lib/efi_loader/efi_variable.c +1 -1
@@ -599,7 +599,7 @@ efi_status_t efi_init_variables(void)
        if (ret != EFI_SUCCESS)
                return ret;
 
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)

So I think this is might to come back and bite us in the future. The name of the file is a bit misleading, but efi_variable.c is supposed to handle the variables with a file backed storage and efi_variable_tee.c is handling the variables when the storage is isolated in the secure world.
The problem is that each case defines the boottime and runtime services. The functions that are common across cases live in efi_var_common.c. That files defines the boottime and runtime calls, by calling the *_int() variants.

The file backed and SPI flash storage seem to have a lot in common. e.g they both use the memory backend to expose the vairables the the OS, query the available storage etc. But I am not they will end up being 100% identical. If they do this approach is ok. But ifthey don't it's better to expand the efi_var_sf.c you added and add code for efi_variables_boot_exit_notify(), and any *_int() functions that are different.

Quoted 6 lines
Quoted 2 lines

Heinrich any opinions? I'll need to think through the SPI case inclduing runtime support before taking this in

Do we also need to consider the case where the SPI flash remains owned by the firmware/secure world, and the case where it's handed over to the OS?

No, we dont have to care about this. If the SPI moves to the secure world, then efi_variable_tee.c will take care of it. This file is basically a glue layer between U-Boot and StMM running in op-tee. So the majority of the works to support this is in OP-TEE & StMM -- add a spi bus driver, spi flash driver etc. I dont expect to change the u-boot part substantially

Cheers
/Ilias

#7 lore raw

Re: [PATCH v4 3/3] efi_vars: Implement SPI Flash store

In reply to Michal Simek

Hi Michal,

[...]

lib/efi_loader/Kconfig endchoice · line 165

Quoted 5 lines

config EFI_RT_VOLATILE_STORE
bool "Allow variable runtime services in volatile storage (e.g RAM)"

lib/efi_loader/Kconfig +1 -1
-       depends on EFI_VARIABLE_FILE_STORE
+       depends on EFI_VARIABLE_FILE_STORE || EFI_VARIABLE_SF_STORE

Will this work on nands as well? They got a much shrter lifetime that spi flashes.

Quoted 3 lines

help
When EFI variables are stored on file we don't allow SetVariableRT, since the OS doesn't know how to write that file. At the same time

I am not sure we need to allow this for now. At least not until we've talked to the efitool maintainers and make sure they will accept another 'special' case.

The problem with allowing this is that if people enable it, boot a linux and do a setvariable, it will return a success. But none of the variables will be preserved after a reboot unless someone manually updates the serial flash contents. In theory, we can preserve the driver model and the spi drivers in EFI runtime services and allow 'proper' setvariable at runtime.
However, I think this is not very useful. Having an unprotected to store authenticated EFI variables, is dangerous. Someone can erase the SPI flash and efectively disable secure boot. Due to that, I prefer the current file based approach for EFI variables -- which doesn't store/restore authenticated EFI variables (and which this patch implements). The obvious downside is that enable setvariable at runtime is tricky once again....

[...]

Quoted 5 lines
lib/efi_loader/efi_variable.c +0 -0
index be670a8e7c25..feab212b245b 100644

lib/efi_loader/efi_variable.c efi_status_t efi_set_variable_int(const u16 *variable_name, · line 404

Quoted 10 lines
lib/efi_loader/efi_variable.c +2 -2
@@ -397,11 +397,11 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
                ret = EFI_SUCCESS;
 
        /*
-        * Write non-volatile EFI variables to file
+        * Write non-volatile EFI variables to file or SPI Flash
         * TODO: check if a value change has occured to avoid superfluous writes
         */
        if (attributes & EFI_VARIABLE_NON_VOLATILE) {
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)

Do we need the ifdefery here? efi_variable.o is basically compiled whenever we have the variables managed by the non-secure world and this function will exist either with a file back storage or a SPI flash

lib/efi_loader/efi_variable.c efi_status_t efi_init_variables(void) · line 602

Quoted 9 lines
lib/efi_loader/efi_variable.c +1 -1
                efi_var_write();
 #else
                return EFI_NOT_READY;
@@ -599,7 +599,7 @@ efi_status_t efi_init_variables(void)
        if (ret != EFI_SUCCESS)
                return ret;
 
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)

ditto

Quoted 6 lines
lib/efi_loader/efi_variable.c +0 -1
        ret = efi_var_read();
        if (ret != EFI_SUCCESS)
                return ret;
--

2.43.0

Cheers
/Ilias

#8 lore raw

Re: [PATCH v4 3/3] efi_vars: Implement SPI Flash store

Hi,

Quoted 5 lines · Ilias Apalodimas

Hi Michal,

[...]

Quoted 1 line
Quoted 7 lines

lib/efi_loader/Kconfig endchoice · line 165

Quoted 4 lines

config EFI_RT_VOLATILE_STORE
bool "Allow variable runtime services in volatile storage (e.g RAM)"

lib/efi_loader/Kconfig +1 -1
-       depends on EFI_VARIABLE_FILE_STORE
+       depends on EFI_VARIABLE_FILE_STORE || EFI_VARIABLE_SF_STORE

Will this work on nands as well? They got a much shrter lifetime that spi flashes.

I don't think this is valid argument here. Users have options to decide where to put them. The same argument can be used for u-boot variables saved in NAND. Obviously I can update Kconfig option with saying that you should be careful about it but it is user choice.

Quoted 8 lines
Quoted 4 lines

help
When EFI variables are stored on file we don't allow SetVariableRT, since the OS doesn't know how to write that file. At the same time

I am not sure we need to allow this for now. At least not until we've talked to the efitool maintainers and make sure they will accept another 'special' case.

Not a problem for me to disable it for now.

Quoted 7 lines

The problem with allowing this is that if people enable it, boot a linux and do a setvariable, it will return a success. But none of the variables will be preserved after a reboot unless someone manually updates the serial flash contents. In theory, we can preserve the driver model and the spi drivers in EFI runtime services and allow 'proper' setvariable at runtime.

Wasn't this the same case for file based one? I mean u-boot had it but efi

Yours efivar patch was merged Jun 23, 2025
But this symbol was introduce in U-Boot April 18, 2024.

Quoted 7 lines

However, I think this is not very useful. Having an unprotected to store authenticated EFI variables, is dangerous. Someone can erase the SPI flash and efectively disable secure boot. Due to that, I prefer the current file based approach for EFI variables -- which doesn't store/restore authenticated EFI variables (and which this patch implements). The obvious downside is that enable setvariable at runtime is tricky once again....

Isn't the same happening with file store too?

From implementation perspective it should be the same as file based. Obviously efivar part is missing.

Quoted 10 lines

[...]

Quoted 7 lines
lib/efi_loader/efi_variable.c +0 -0
index be670a8e7c25..feab212b245b 100644
@@ -397,11 +397,11 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
                 ret = EFI_SUCCESS;
 
Quoted 13 lines

lib/efi_loader/efi_variable.c efi_status_t efi_set_variable_int(const u16 *variable_name, · line 404

Quoted 8 lines
lib/efi_loader/efi_variable.c +2 -2
         /*
-        * Write non-volatile EFI variables to file
+        * Write non-volatile EFI variables to file or SPI Flash
          * TODO: check if a value change has occured to avoid superfluous writes
          */
         if (attributes & EFI_VARIABLE_NON_VOLATILE) {
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)

Do we need the ifdefery here? efi_variable.o is basically compiled whenever we have the variables managed by the non-secure world and this function will exist either with a file back storage or a SPI flash

It can be reverted and return EFI_NOT_READY for EFI_VARIABLE_NO_STORE.

Thanks,
Michal

#9 lore raw

Re: [PATCH v4 3/3] efi_vars: Implement SPI Flash store

Quoted 14 lines · Michal Simek

Hi,

Quoted 10 lines · Ilias Apalodimas

Hi Michal,

[...]

Quoted 5 lines

config EFI_RT_VOLATILE_STORE
bool "Allow variable runtime services in volatile storage (e.g RAM)"

lib/efi_loader/Kconfig +1 -1
-       depends on EFI_VARIABLE_FILE_STORE
+       depends on EFI_VARIABLE_FILE_STORE || EFI_VARIABLE_SF_STORE
Quoted 7 lines
Quoted 2 lines

Will this work on nands as well? They got a much shrter lifetime that spi flashes.

I don't think this is valid argument here. Users have options to decide where to put them. The same argument can be used for u-boot variables saved in NAND. Obviously I can update Kconfig option with saying that you should be careful about it but it is user choice.

Yes I think that's a good idea.

Quoted 20 lines
Quoted 9 lines
Quoted 5 lines

help
When EFI variables are stored on file we don't allow SetVariableRT, since the OS doesn't know how to write that file. At the same time

I am not sure we need to allow this for now. At least not until we've talked to the efitool maintainers and make sure they will accept another 'special' case.

Not a problem for me to disable it for now.

Quoted 7 lines

The problem with allowing this is that if people enable it, boot a linux and do a setvariable, it will return a success. But none of the variables will be preserved after a reboot unless someone manually updates the serial flash contents. In theory, we can preserve the driver model and the spi drivers in EFI runtime services and allow 'proper' setvariable at runtime.

Quoted 4 lines

Wasn't this the same case for file based one? I mean u-boot had it but efi

Yours efivar patch was merged Jun 23, 2025
But this symbol was introduce in U-Boot April 18, 2024.

Yes but the patches were sent a long time ago in efivar. If you look at the original PR date it was a lot eariler [0]. It just took a long time to discuss and merge.
Apart from that the case for file stored variables is by itself pretty unique and convincing the maintainers to accept the patch had some basis. I don't know if they want the same workaround for spi flashes.

Quoted 10 lines
Quoted 9 lines

However, I think this is not very useful. Having an unprotected to store authenticated EFI variables, is dangerous. Someone can erase the SPI flash and efectively disable secure boot. Due to that, I prefer the current file based approach for EFI variables -- which doesn't store/restore authenticated EFI variables (and which this patch implements). The obvious downside is that enable setvariable at runtime is tricky once again....

Quoted 4 lines

Isn't the same happening with file store too?

From implementation perspective it should be the same as file based. Obviously
efivar part is missing.

Yes. However, that commit included instructions on how to set runtime variables with dd, so people could use it until the efivar changes were merged.

Quoted 25 lines
Quoted 11 lines

[...]

Quoted 7 lines
lib/efi_loader/efi_variable.c +0 -0
index be670a8e7c25..feab212b245b 100644
@@ -397,11 +397,11 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
                 ret = EFI_SUCCESS;
 
Quoted 13 lines

lib/efi_loader/efi_variable.c efi_status_t efi_set_variable_int(const u16 *variable_name, · line 404

Quoted 8 lines
lib/efi_loader/efi_variable.c +2 -2
         /*
-        * Write non-volatile EFI variables to file
+        * Write non-volatile EFI variables to file or SPI Flash
          * TODO: check if a value change has occured to avoid superfluous writes
          */
         if (attributes & EFI_VARIABLE_NON_VOLATILE) {
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)

Do we need the ifdefery here? efi_variable.o is basically compiled whenever we have the variables managed by the non-secure world and this function will exist either with a file back storage or a SPI flash

Quoted 1 line

It can be reverted and return EFI_NOT_READY for EFI_VARIABLE_NO_STORE.

Yes I *thinkk* that will look cleaner.

Quoted 1 line

[0] https://github.com/rhboot/efivar/pull/267

Cheers
/Ilias

Quoted 2 lines

Thanks,
Michal

#10 lore raw

Re: [PATCH v4 3/3] efi_vars: Implement SPI Flash store

Quoted 50 lines · Ilias Apalodimas
Quoted 14 lines · Michal Simek

Hi,

Quoted 10 lines · Ilias Apalodimas

Hi Michal,

[...]

Quoted 5 lines

config EFI_RT_VOLATILE_STORE
bool "Allow variable runtime services in volatile storage (e.g RAM)"

lib/efi_loader/Kconfig +1 -1
-       depends on EFI_VARIABLE_FILE_STORE
+       depends on EFI_VARIABLE_FILE_STORE || EFI_VARIABLE_SF_STORE
Quoted 7 lines
Quoted 2 lines

Will this work on nands as well? They got a much shrter lifetime that spi flashes.

I don't think this is valid argument here. Users have options to decide where to put them. The same argument can be used for u-boot variables saved in NAND. Obviously I can update Kconfig option with saying that you should be careful about it but it is user choice.

Yes I think that's a good idea.

Quoted 20 lines
Quoted 9 lines
Quoted 5 lines

help
When EFI variables are stored on file we don't allow SetVariableRT, since the OS doesn't know how to write that file. At the same time

I am not sure we need to allow this for now. At least not until we've talked to the efitool maintainers and make sure they will accept another 'special' case.

Not a problem for me to disable it for now.

Quoted 7 lines

The problem with allowing this is that if people enable it, boot a linux and do a setvariable, it will return a success. But none of the variables will be preserved after a reboot unless someone manually updates the serial flash contents. In theory, we can preserve the driver model and the spi drivers in EFI runtime services and allow 'proper' setvariable at runtime.

Quoted 4 lines

Wasn't this the same case for file based one? I mean u-boot had it but efi

Yours efivar patch was merged Jun 23, 2025
But this symbol was introduce in U-Boot April 18, 2024.

Quoted 6 lines

Yes but the patches were sent a long time ago in efivar. If you look at the original PR date it was a lot eariler [0]. It just took a long time to discuss and merge.
Apart from that the case for file stored variables is by itself pretty unique and convincing the maintainers to accept the patch had some basis. I don't know if they want the same workaround for spi flashes.

Correct June, 2024
https://github.com/rhboot/efivar/pull/267

Quoted 10 lines
Quoted 10 lines
Quoted 9 lines

However, I think this is not very useful. Having an unprotected to store authenticated EFI variables, is dangerous. Someone can erase the SPI flash and efectively disable secure boot. Due to that, I prefer the current file based approach for EFI variables -- which doesn't store/restore authenticated EFI variables (and which this patch implements). The obvious downside is that enable setvariable at runtime is tricky once again....

Quoted 8 lines
Quoted 4 lines

Isn't the same happening with file store too?

From implementation perspective it should be the same as file based. Obviously
efivar part is missing.

Yes. However, that commit included instructions on how to set runtime variables with dd, so people could use it until the efivar changes were merged.

I don't think it is going to be problem too.

What about?
dd if=/sys/firmware/efi/efivars/VarToFile-b2ac5fc9-92b7-4acd-aeac-11e818c3130c \
of=/tmp/vars skip=4 bs=1
flashcp /tmp/vars /dev/mtdX

Quoted 26 lines
Quoted 26 lines
Quoted 25 lines

[...]

Quoted 5 lines
lib/efi_loader/efi_variable.c +0 -0
index be670a8e7c25..feab212b245b 100644

lib/efi_loader/efi_variable.c efi_status_t efi_set_variable_int(const u16 *variable_name, · line 404

Quoted 10 lines
lib/efi_loader/efi_variable.c +2 -2
@@ -397,11 +397,11 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
                  ret = EFI_SUCCESS;
 
          /*
-        * Write non-volatile EFI variables to file
+        * Write non-volatile EFI variables to file or SPI Flash
           * TODO: check if a value change has occured to avoid superfluous writes
           */
          if (attributes & EFI_VARIABLE_NON_VOLATILE) {
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)

Do we need the ifdefery here? efi_variable.o is basically compiled whenever we have the variables managed by the non-secure world and this function will exist either with a file back storage or a SPI flash

Quoted 3 lines
Quoted 1 line

It can be reverted and return EFI_NOT_READY for EFI_VARIABLE_NO_STORE.

Yes I *thinkk* that will look cleaner.

ok.

Thanks,
Michal

#11 lore raw

Re: [PATCH v4 3/3] efi_vars: Implement SPI Flash store

Quoted 84 lines · Michal Simek
Quoted 50 lines · Ilias Apalodimas
Quoted 14 lines · Michal Simek

Hi,

Quoted 10 lines · Ilias Apalodimas

Hi Michal,

[...]

Quoted 5 lines

config EFI_RT_VOLATILE_STORE
bool "Allow variable runtime services in volatile storage (e.g RAM)"

lib/efi_loader/Kconfig +1 -1
-       depends on EFI_VARIABLE_FILE_STORE
+       depends on EFI_VARIABLE_FILE_STORE || EFI_VARIABLE_SF_STORE
Quoted 7 lines
Quoted 2 lines

Will this work on nands as well? They got a much shrter lifetime that spi flashes.

I don't think this is valid argument here. Users have options to decide where to put them. The same argument can be used for u-boot variables saved in NAND. Obviously I can update Kconfig option with saying that you should be careful about it but it is user choice.

Yes I think that's a good idea.

Quoted 20 lines
Quoted 9 lines
Quoted 5 lines

help
When EFI variables are stored on file we don't allow SetVariableRT, since the OS doesn't know how to write that file. At the same time

I am not sure we need to allow this for now. At least not until we've talked to the efitool maintainers and make sure they will accept another 'special' case.

Not a problem for me to disable it for now.

Quoted 7 lines

The problem with allowing this is that if people enable it, boot a linux and do a setvariable, it will return a success. But none of the variables will be preserved after a reboot unless someone manually updates the serial flash contents. In theory, we can preserve the driver model and the spi drivers in EFI runtime services and allow 'proper' setvariable at runtime.

Quoted 4 lines

Wasn't this the same case for file based one? I mean u-boot had it but efi

Yours efivar patch was merged Jun 23, 2025
But this symbol was introduce in U-Boot April 18, 2024.

Quoted 6 lines

Yes but the patches were sent a long time ago in efivar. If you look at the original PR date it was a lot eariler [0]. It just took a long time to discuss and merge.
Apart from that the case for file stored variables is by itself pretty unique and convincing the maintainers to accept the patch had some basis. I don't know if they want the same workaround for spi flashes.

Correct June, 2024
https://github.com/rhboot/efivar/pull/267

Quoted 10 lines
Quoted 10 lines
Quoted 9 lines

However, I think this is not very useful. Having an unprotected to store authenticated EFI variables, is dangerous. Someone can erase the SPI flash and efectively disable secure boot. Due to that, I prefer the current file based approach for EFI variables -- which doesn't store/restore authenticated EFI variables (and which this patch implements). The obvious downside is that enable setvariable at runtime is tricky once again....

Quoted 8 lines
Quoted 4 lines

Isn't the same happening with file store too?

From implementation perspective it should be the same as file based. Obviously
efivar part is missing.

Yes. However, that commit included instructions on how to set runtime variables with dd, so people could use it until the efivar changes were merged.

Quoted 6 lines

I don't think it is going to be problem too.

What about?

dd if=/sys/firmware/efi/efivars/VarToFile-b2ac5fc9-92b7-4acd-aeac-11e818c3130c \
    of=/tmp/vars skip=4 bs=1

flashcp /tmp/vars /dev/mtdX

If you manage to test this then yes, please include it on the commit message and in the future we can add it to the documentation

Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 26 lines

[...]

Quoted 5 lines
lib/efi_loader/efi_variable.c +0 -0
index be670a8e7c25..feab212b245b 100644

lib/efi_loader/efi_variable.c efi_status_t efi_set_variable_int(const u16 *variable_name, · line 404

Quoted 10 lines
lib/efi_loader/efi_variable.c +2 -2
@@ -397,11 +397,11 @@ efi_status_t efi_set_variable_int(const u16 *variable_name,
                  ret = EFI_SUCCESS;
 
          /*
-        * Write non-volatile EFI variables to file
+        * Write non-volatile EFI variables to file or SPI Flash
           * TODO: check if a value change has occured to avoid superfluous writes
           */
          if (attributes & EFI_VARIABLE_NON_VOLATILE) {
-#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE)
+#if CONFIG_IS_ENABLED(EFI_VARIABLE_FILE_STORE) || CONFIG_IS_ENABLED(EFI_VARIABLE_SF_STORE)

Do we need the ifdefery here? efi_variable.o is basically compiled whenever we have the variables managed by the non-secure world and this function will exist either with a file back storage or a SPI flash

Quoted 5 lines
Quoted 3 lines
Quoted 1 line

It can be reverted and return EFI_NOT_READY for EFI_VARIABLE_NO_STORE.

Yes I *thinkk* that will look cleaner.

ok.

Thanks
/Ilias

Quoted 3 lines

Thanks,
Michal

#12 lore raw

Re: [PATCH v4 3/3] efi_vars: Implement SPI Flash store

Quoted 87 lines · Ilias Apalodimas
Quoted 86 lines · Michal Simek
Quoted 50 lines · Ilias Apalodimas
Quoted 14 lines · Michal Simek

Hi,

Quoted 10 lines · Ilias Apalodimas

Hi Michal,

[...]

Quoted 5 lines

config EFI_RT_VOLATILE_STORE
bool "Allow variable runtime services in volatile storage (e.g RAM)"

lib/efi_loader/Kconfig +1 -1
-       depends on EFI_VARIABLE_FILE_STORE
+       depends on EFI_VARIABLE_FILE_STORE || EFI_VARIABLE_SF_STORE
Quoted 7 lines
Quoted 2 lines

Will this work on nands as well? They got a much shrter lifetime that spi flashes.

I don't think this is valid argument here. Users have options to decide where to put them. The same argument can be used for u-boot variables saved in NAND. Obviously I can update Kconfig option with saying that you should be careful about it but it is user choice.

Yes I think that's a good idea.

Quoted 20 lines
Quoted 9 lines
Quoted 5 lines

help
When EFI variables are stored on file we don't allow SetVariableRT, since the OS doesn't know how to write that file. At the same time

I am not sure we need to allow this for now. At least not until we've talked to the efitool maintainers and make sure they will accept another 'special' case.

Not a problem for me to disable it for now.

Quoted 7 lines

The problem with allowing this is that if people enable it, boot a linux and do a setvariable, it will return a success. But none of the variables will be preserved after a reboot unless someone manually updates the serial flash contents. In theory, we can preserve the driver model and the spi drivers in EFI runtime services and allow 'proper' setvariable at runtime.

Quoted 4 lines

Wasn't this the same case for file based one? I mean u-boot had it but efi

Yours efivar patch was merged Jun 23, 2025
But this symbol was introduce in U-Boot April 18, 2024.

Quoted 6 lines

Yes but the patches were sent a long time ago in efivar. If you look at the original PR date it was a lot eariler [0]. It just took a long time to discuss and merge.
Apart from that the case for file stored variables is by itself pretty unique and convincing the maintainers to accept the patch had some basis. I don't know if they want the same workaround for spi flashes.

Correct June, 2024
https://github.com/rhboot/efivar/pull/267

Quoted 10 lines
Quoted 10 lines
Quoted 9 lines

However, I think this is not very useful. Having an unprotected to store authenticated EFI variables, is dangerous. Someone can erase the SPI flash and efectively disable secure boot. Due to that, I prefer the current file based approach for EFI variables -- which doesn't store/restore authenticated EFI variables (and which this patch implements). The obvious downside is that enable setvariable at runtime is tricky once again....

Quoted 8 lines
Quoted 4 lines

Isn't the same happening with file store too?

From implementation perspective it should be the same as file based. Obviously
efivar part is missing.

Yes. However, that commit included instructions on how to set runtime variables with dd, so people could use it until the efivar changes were merged.

I don't think it is going to be problem too.

Quoted 7 lines
Quoted 4 lines

What about?

dd if=/sys/firmware/efi/efivars/VarToFile-b2ac5fc9-92b7-4acd-aeac-11e818c3130c \
     of=/tmp/vars skip=4 bs=1

flashcp /tmp/vars /dev/mtdX

If you manage to test this then yes, please include it on the commit message and in the future we can add it to the documentation

yes, I have also tested it on SOM.

Let me send v5 with discussed changes and also with logs.

Thanks,
Michal