platform/x86: think-lmi: Keep pending save state on failure
Subsystems (4)
PATCH 1/1 platform/x86: think-lmi: Keep pending save state on failure
In save_settings_store(), the opcode and legacy password paths do not check if saving BIOS settings failed, but instead unconditionally clear save_required and set reboot_required, losing the pending save state.
Check the save result in those paths too.
--- drivers/platform/x86/lenovo/think-lmi.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/platform/x86/lenovo/think-lmi.c +4 -0
@@ -1347,6 +1347,8 @@ static ssize_t save_settings_store(struct kobject *kobj, struct kobj_attribute * goto out; } ret =tlmi_save_bios_settings ("" ); +if (ret) + goto out; }else {/* old non-opcode based authentication method (deprecated) */ if (tlmi_priv.pwd_admin->pwd_enabled && tlmi_priv.pwd_admin->password[0 ]) { auth_str = kasprintf(GFP_KERNEL, "%s,%s,%s;",
@@ -1363,6 +1365,8 @@ static ssize_t save_settings_store(struct kobject *kobj, struct kobj_attribute * ret =tlmi_save_bios_settings (auth_str); else ret =tlmi_save_bios_settings ("" ); +if (ret) + goto out; } tlmi_priv.save_required = false; tlmi_priv.reboot_required = true;
Mark Pearson
maintainerHi Thorsten,
Quoted 39 lines · Thorsten Blum
In
save_settings_store(), the opcode and legacy password paths do not check if saving BIOS settings failed, but instead unconditionally clearsave_requiredand setreboot_required, losing the pending save state.Check the save result in those paths too.
Fixes 318d97849fc2 ("platform/x86: think-lmi: Add bulk save feature")Signed-off-by Thorsten Blum <[email protected]>--- drivers/platform/x86/lenovo/think-lmi.c | 4 ++++ 1 file changed, 4 insertions(+)drivers/platform/x86/lenovo/think-lmi.c +2
-0@@ -1347,6 +1347,8 @@ static ssize_t save_settings_store(struct kobject *kobj, struct kobj_attribute * goto out; } ret =tlmi_save_bios_settings ("" ); +if (ret) + goto out; }else {/* old non-opcode based authentication method (deprecated) */ if (
tlmi_priv.pwd_admin->pwd_enabled&&
tlmi_priv.pwd_admin->password[0]) {drivers/platform/x86/lenovo/think-lmi.c +2
-0auth_str = kasprintf(GFP_KERNEL, "%s,%s,%s;",@@ -1363,6 +1365,8 @@ static ssize_t save_settings_store(struct kobject *kobj, struct kobj_attribute * ret =tlmi_save_bios_settings (auth_str); else ret =tlmi_save_bios_settings ("" ); +if (ret) + goto out; } tlmi_priv.save_required = false;
Quoted 1 line
tlmi_priv.reboot_required = true;
Curious - have you actually hit this condition?
My main thought is that if the save fails - chances are that something is pretty messed up and doing another save isn't going to fix it. Most likely you're going to have to reboot.
I'll see if I can get an answer from the FW team on expectations - but getting straight answers for error paths without a repro mechanism can be tricky.
Mark
Quoted 37 lines · Mark Pearson
Hi Thorsten,
Quoted 34 lines · Thorsten Blum
In
save_settings_store(), the opcode and legacy password paths do not check if saving BIOS settings failed, but instead unconditionally clearsave_requiredand setreboot_required, losing the pending save state.Check the save result in those paths too.
Fixes 318d97849fc2 ("platform/x86: think-lmi: Add bulk save feature")Signed-off-by Thorsten Blum <[email protected]>--- drivers/platform/x86/lenovo/think-lmi.c | 4 ++++ 1 file changed, 4 insertions(+)drivers/platform/x86/lenovo/think-lmi.c +2
-0@@ -1347,6 +1347,8 @@ static ssize_t save_settings_store(struct kobject *kobj, struct kobj_attribute * goto out; } ret =tlmi_save_bios_settings ("" ); +if (ret) + goto out; }else {/* old non-opcode based authentication method (deprecated) */ if (
tlmi_priv.pwd_admin->pwd_enabled&&
tlmi_priv.pwd_admin->password[0]) {drivers/platform/x86/lenovo/think-lmi.c +0
-0auth_str = kasprintf(GFP_KERNEL, "%s,%s,%s;",@@ -1363,6 +1365,8 @@ static ssize_t save_settings_store(struct kobject *kobj, struct kobj_attribute * ret =tlmi_save_bios_settings (auth_str); else
Quoted 8 lines
Quoted 6 lines
drivers/platform/x86/lenovo/think-lmi.c +2
-0ret =tlmi_save_bios_settings ("" ); +if (ret) + goto out; } tlmi_priv.save_required = false;tlmi_priv.reboot_required = true;
Curious - have you actually hit this condition?
No, I was just reading the code and wondered why those return values aren't checked for errors.