Skip to contentSkip to conversation

platform/x86: think-lmi: add debug_cmd

PATCH In review Mark Pearson 2 participants 3 messages Aug 17, 2021 – Aug 18, 2021

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 platform/x86: think-lmi: add debug_cmd

Many Lenovo BIOS's support the ability to send a debug command which is useful for debugging and testing unreleased or early features.

Adding support for this feature as a module parameter.

---
Changes in v2:
- Updated to use debug_support parameter to enable feature
- remove changes to pending_reboot as that was integrated separately

 .../testing/sysfs-class-firmware-attributes   | 11 +++
 drivers/platform/x86/think-lmi.c              | 80 +++++++++++++++++++
 drivers/platform/x86/think-lmi.h              |  1 +
 3 files changed, 92 insertions(+)
Documentation/ABI/testing/sysfs-class-firmware-attributes +11 -0
index 0b43997b76e3..90fdf935aa5e 100644
@@ -272,3 +272,14 @@ Description:
 
 		Note that any changes to this attribute requires a reboot
 		for changes to take effect.
+
+What:		/sys/class/firmware-attributes/*/attributes/debug_cmd
+Date:		July 2021
+KernelVersion:	5.14
+Contact:	Mark Pearson <[email protected]>
+Description:
+		This write only attribute can be used to send debug commands to the BIOS.
+		This should only be used when recommended by the BIOS vendor. Vendors may
+		use it to enable extra debug attributes or BIOS features for testing purposes.
+
+		Note that any changes to this attribute requires a reboot for changes to take effect.
drivers/platform/x86/think-lmi.c +80 -0

116 lines not on this page · raw message

drivers/platform/x86/think-lmi.h +1 -0
index eb598846628a..f8e26823075f 100644
@@ -61,6 +61,7 @@ struct think_lmi {
 	bool can_set_bios_password;
 	bool can_get_password_settings;
 	bool pending_changes;
+	bool can_debug_cmd;
 
 	struct tlmi_attr_setting *setting[TLMI_SETTINGS_COUNT];
 	struct device *class_dev;
Signature · 1 line
2.31.1
#1 lore raw

Hi,

Quoted 40 lines · Mark Pearson

Many Lenovo BIOS's support the ability to send a debug command which is useful for debugging and testing unreleased or early features.

Adding support for this feature as a module parameter.

---
Changes in v2:
- Updated to use debug_support parameter to enable feature
- remove changes to pending_reboot as that was integrated separately

 .../testing/sysfs-class-firmware-attributes   | 11 +++
 drivers/platform/x86/think-lmi.c              | 80 +++++++++++++++++++
 drivers/platform/x86/think-lmi.h              |  1 +
 3 files changed, 92 insertions(+)
Documentation/ABI/testing/sysfs-class-firmware-attributes +11 -0
index 0b43997b76e3..90fdf935aa5e 100644
@@ -272,3 +272,14 @@ Description:
 
 		Note that any changes to this attribute requires a reboot
 		for changes to take effect.
+
+What:		/sys/class/firmware-attributes/*/attributes/debug_cmd
+Date:		July 2021
+KernelVersion:	5.14
+Contact:	Mark Pearson <[email protected]>
+Description:
+		This write only attribute can be used to send debug commands to the BIOS.
+		This should only be used when recommended by the BIOS vendor. Vendors may
+		use it to enable extra debug attributes or BIOS features for testing purposes.
+
+		Note that any changes to this attribute requires a reboot for changes to take effect.
drivers/platform/x86/think-lmi.c +0 -0
index 6cfed4427fb0..4d246bfe4907 100644

drivers/platform/x86/think-lmi.c line 24

Quoted 5 lines
drivers/platform/x86/think-lmi.c +2 -0
@@ -20,6 +20,10 @@
 #include "firmware_attributes_class.h"
 #include "think-lmi.h"
 
+static bool debug_support;
+module_param(debug_support, bool, 0644);

0644 implies the setting can be changed at runtime, but it is checked only at driver bind + unbind time, so this should be 0444.

Since the rest of the patch looks good, I've fixed this up while merging the patch.

###

Thank you for your patch, I've applied this patch to my review-hans branch:
https://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86.git/log/?h=review-hans

Once I've run some tests on this branch the patches there will be added to the platform-drivers-x86/for-next branch and eventually will be included in the pdx86 pull-request to Linus for the next merge-window.

Regards,

Hans

Quoted 129 lines
drivers/platform/x86/think-lmi.c +78 -0
+MODULE_PARM_DESC(debug_support, "Enable debug command support");
+
 /*
  * Name:
  *  Lenovo_BiosSetting
@@ -116,6 +120,14 @@
  */
 #define LENOVO_GET_BIOS_SELECTIONS_GUID	"7364651A-132F-4FE7-ADAA-40C6C7EE2E3B"
 
+/*
+ * Name:
+ *  Lenovo_DebugCmdGUID
+ * Description
+ *  Debug entry GUID method for entering debug commands to the BIOS
+ */
+#define LENOVO_DEBUG_CMD_GUID "7FF47003-3B6C-4E5E-A227-E979824A85D1"
+
 #define TLMI_POP_PWD (1 << 0)
 #define TLMI_PAP_PWD (1 << 1)
 #define to_tlmi_pwd_setting(kobj)  container_of(kobj, struct tlmi_pwd_setting, kobj)
@@ -660,6 +672,64 @@ static ssize_t pending_reboot_show(struct kobject *kobj, struct kobj_attribute *
 
 static struct kobj_attribute pending_reboot = __ATTR_RO(pending_reboot);
 
+/* ---- Debug interface--------------------------------------------------------- */
+static ssize_t debug_cmd_store(struct kobject *kobj, struct kobj_attribute *attr,
+				const char *buf, size_t count)
+{
+	char *set_str = NULL, *new_setting = NULL;
+	char *auth_str = NULL;
+	char *p;
+	int ret;
+
+	if (!tlmi_priv.can_debug_cmd)
+		return -EOPNOTSUPP;
+
+	new_setting = kstrdup(buf, GFP_KERNEL);
+	if (!new_setting)
+		return -ENOMEM;
+
+	/* Strip out CR if one is present */
+	p = strchrnul(new_setting, '\n');
+	*p = '\0';
+
+	if (tlmi_priv.pwd_admin->valid && tlmi_priv.pwd_admin->password[0]) {
+		auth_str = kasprintf(GFP_KERNEL, "%s,%s,%s;",
+				tlmi_priv.pwd_admin->password,
+				encoding_options[tlmi_priv.pwd_admin->encoding],
+				tlmi_priv.pwd_admin->kbdlang);
+		if (!auth_str) {
+			ret = -ENOMEM;
+			goto out;
+		}
+	}
+
+	if (auth_str)
+		set_str = kasprintf(GFP_KERNEL, "%s,%s", new_setting, auth_str);
+	else
+		set_str = kasprintf(GFP_KERNEL, "%s;", new_setting);
+	if (!set_str) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = tlmi_simple_call(LENOVO_DEBUG_CMD_GUID, set_str);
+	if (ret)
+		goto out;
+
+	if (!ret && !tlmi_priv.pending_changes) {
+		tlmi_priv.pending_changes = true;
+		/* let userland know it may need to check reboot pending again */
+		kobject_uevent(&tlmi_priv.class_dev->kobj, KOBJ_CHANGE);
+	}
+out:
+	kfree(auth_str);
+	kfree(set_str);
+	kfree(new_setting);
+	return ret ?: count;
+}
+
+static struct kobj_attribute debug_cmd = __ATTR_WO(debug_cmd);
+
 /* ---- Initialisation --------------------------------------------------------- */
 static void tlmi_release_attr(void)
 {
@@ -673,6 +743,8 @@ static void tlmi_release_attr(void)
 		}
 	}
 	sysfs_remove_file(&tlmi_priv.attribute_kset->kobj, &pending_reboot.attr);
+	if (tlmi_priv.can_debug_cmd && debug_support)
+		sysfs_remove_file(&tlmi_priv.attribute_kset->kobj, &debug_cmd.attr);
 	kset_unregister(tlmi_priv.attribute_kset);
 
 	/* Authentication structures */
@@ -737,6 +809,11 @@ static int tlmi_sysfs_init(void)
 	if (ret)
 		goto fail_create_attr;
 
+	if (tlmi_priv.can_debug_cmd && debug_support) {
+		ret = sysfs_create_file(&tlmi_priv.attribute_kset->kobj, &debug_cmd.attr);
+		if (ret)
+			goto fail_create_attr;
+	}
 	/* Create authentication entries */
 	tlmi_priv.authentication_kset = kset_create_and_add("authentication", NULL,
 								&tlmi_priv.class_dev->kobj);
@@ -793,6 +870,9 @@ static int tlmi_analyze(void)
 	if (wmi_has_guid(LENOVO_BIOS_PASSWORD_SETTINGS_GUID))
 		tlmi_priv.can_get_password_settings = true;
 
+	if (wmi_has_guid(LENOVO_DEBUG_CMD_GUID))
+		tlmi_priv.can_debug_cmd = true;
+
 	/*
 	 * Try to find the number of valid settings of this machine
 	 * and use it to create sysfs attributes.
drivers/platform/x86/think-lmi.h +1 -0
index eb598846628a..f8e26823075f 100644
@@ -61,6 +61,7 @@ struct think_lmi {
 	bool can_set_bios_password;
 	bool can_get_password_settings;
 	bool pending_changes;
+	bool can_debug_cmd;
 
 	struct tlmi_attr_setting *setting[TLMI_SETTINGS_COUNT];
 	struct device *class_dev;
#2 lore raw

Hi Hans

Quoted 46 lines · Hans de Goede

Hi,

Quoted 43 lines · Mark Pearson

Many Lenovo BIOS's support the ability to send a debug command which is useful for debugging and testing unreleased or early features.

Adding support for this feature as a module parameter.

---
Changes in v2:
- Updated to use debug_support parameter to enable feature
- remove changes to pending_reboot as that was integrated separately

  .../testing/sysfs-class-firmware-attributes   | 11 +++
  drivers/platform/x86/think-lmi.c              | 80 +++++++++++++++++++
  drivers/platform/x86/think-lmi.h              |  1 +
  3 files changed, 92 insertions(+)
Documentation/ABI/testing/sysfs-class-firmware-attributes +11 -0
index 0b43997b76e3..90fdf935aa5e 100644
@@ -272,3 +272,14 @@ Description:
 
  		Note that any changes to this attribute requires a reboot
  		for changes to take effect.
+
+What:		/sys/class/firmware-attributes/*/attributes/debug_cmd
+Date:		July 2021
+KernelVersion:	5.14
+Contact:	Mark Pearson <[email protected]>
+Description:
+		This write only attribute can be used to send debug commands to the BIOS.
+		This should only be used when recommended by the BIOS vendor. Vendors may
+		use it to enable extra debug attributes or BIOS features for testing purposes.
+
+		Note that any changes to this attribute requires a reboot for changes to take effect.
drivers/platform/x86/think-lmi.c +0 -0
index 6cfed4427fb0..4d246bfe4907 100644
@@ -20,6 +20,10 @@
  #include "firmware_attributes_class.h"
  #include "think-lmi.h"
 
Quoted 5 lines

drivers/platform/x86/think-lmi.c line 24

Quoted 2 lines
drivers/platform/x86/think-lmi.c +2 -0
+static bool debug_support;
+module_param(debug_support, bool, 0644);

0644 implies the setting can be changed at runtime, but it is checked only at driver bind + unbind time, so this should be 0444.

Good catch - thanks!

Quoted 3 lines

Since the rest of the patch looks good, I've fixed this up while merging the patch.

Awesome - thank you!

Mark