add the se wipe and reorder the ErasePublicInfo to fix the power lose issue
What the developer wrote
Authored by aaron
add the se wipe and reorder the ErasePublicInfo to fix the power lose issue
What changed, and why it matters
This commit changes the factory-reset/wipe routine in a Keystone hardware wallet's startup self-check. It replaces one wipe step with a call to wipe the secure element (SE_WipeAll) and moves the 'ErasePublicInfo' step to occur after account destruction and flash erasure, right before the device reboots. The stated goal is to fix a problem where losing power during a wipe could leave sensitive data partially intact. The change is defensive, but because it is a partial patch and the commit message is terse, we cannot fully verify that the new ordering eliminates all power-loss recovery risks.
Treat as a security-hardening fix for a potential data-remanence/power-loss failure mode. Review the full wipe routine for atomic progress tracking, verify that SE_WipeAll and ErasePublicInfo cannot be interrupted in a way that leaves recoverable key material, and confirm whether this change corresponds to a disclosed vulnerability or bug report. If the device has a tamper/factory-reset trigger, ensure the new ordering does not introduce a window where an attacker can abort reset before the secure element is wiped.
Security signals we found
Secure-element wipe added to factory-reset path
Flash/public-info erasure reordered before final system reset
Commit message frames change as fixing power-loss-related data-remanence issue
No visible persistence of wipe-progress state or rollback guard in diff
Evidence from the diff
In src/driver/power_on_self_check.c, the PowerOnSelfCheck() function’s wipe branch now includes #include ‘se_manager.h’, calls SE_WipeAll() where ErasePublicInfo() was previously called, and moves ErasePublicInfo() to the end of the sequence after DestroyAccount(0..2) and Gd25FlashBlockErase loops but before NVIC_SystemReset(). The reordering is intended to address a ‘power lose issue’ during the erase process. No additional atomicity, rollback, or progress-state mechanism is visible in the diff, so the fix relies on sequence ordering alone.
Changed components
src/driver/power_on_self_check.cPowerOnSelfCheck() wipe/reset pathSecure element manager (se_manager.h / SE_WipeAll)Public info erase routine (ErasePublicInfo)Account destruction and SPI flash erase loopsInspect captured patch +4 / −2
diff --git a/src/driver/power_on_self_check.c b/src/driver/power_on_self_check.c
index f08bb93..6428a09 100644
--- a/src/driver/power_on_self_check.c
+++ b/src/driver/power_on_self_check.c
@@ -11,6 +11,7 @@
#include "keystore.h"
#include "draw_on_lcd.h"
#include "account_manager.h"
+#include "se_manager.h"
#include "drv_mpu.h"
#define GD25_FLASH_ID (0xC84018)
@@ -55,7 +56,7 @@ void PowerOnSelfCheck(void)
uint32_t c = 0x666666;
DrawStringOnLcd(170, 456, "About 1 minute", (uint16_t)(((c & 0xF80000) >> 16) | ((c & 0xFC00) >> 13) | ((c & 0x1C00) << 3) | ((c & 0xF8) << 5)), &openSans_20);
FpWipeManageInfo();
- ErasePublicInfo();
+ SE_WipeAll();
DestroyAccount(0);
DestroyAccount(1);
DestroyAccount(2);
@@ -63,6 +64,7 @@ void PowerOnSelfCheck(void)
Gd25FlashBlockErase(addr);
printf("flash erase address: %#x\n", addr);
}
+ ErasePublicInfo();
NVIC_SystemReset();
}
-}
\ No newline at end of file
+}
Why this scored 57/100
Community notes
Notes can correct, qualify, or add evidence to the AI analysis. Every note shown here has been validated by a human moderator.
The AI analysis stands alone for now. Submit a note if you can add evidence or important context.