Granted Patent B2

US 12,265,608 · App. 17/928,995 · Granted Apr 1, 2025

Inventors: Naveed Ahmed (Singapore, SG); Prasanna Hegde (Singapore, SG)

Assignee: THALES DIS FRANCE SAS

G06F21/54G06F21/554G06F2221/033

Abstract

Claims (41)

1. A method for protecting a program in a computer system, the method comprising:

when a subroutine of said program is called, pushing a return address on to a stack to start forming a stack frame;

when pushing said return address, generating a checksum for said stack frame;

each time a predetermined opcode is detected for said subroutine, updating said checksum according to an operand associated with said predetermined opcode;

if the predetermined opcode is a pop opcode, in addition to said updating, determining whether the operand associated with said pop opcode is said return address;

if it is determined that said operand is said return address, verifying said checksum before executing said predetermined opcode in order to detect an attack;

when generating said checksum, initializing said checksum to have a predetermined value and adding said return address's value to said predetermined value;

when updating said checksum;

if said predetermined opcode is a push opcode, adding said operand to the current value of said checksum;

if said predetermined opcode is a load opcode or a pop opcode, subtracting said operand from the current value of said checksum;

when verifying said checksum, checking whether the current value of said checksum is equal to said predetermined value; and

if said checksum is equal to said predetermined value, determining that there is no attack; otherwise, determining that there is an attack.

2. The method according to claim 1 , wherein said predetermined opcode is any one chosen from a group comprising a push opcode, a load opcode, and a pop opcode.

3. The method according to claim 1 , wherein the method further comprises:

introducing a pointer pointing to said checksum;

each time a new subroutine is called, operating said pointer to point to a new checksum generated for a new stack frame corresponding to said new subroutine;

when it is to update or verify the checksum, updating or verifying the checksum said pointer points to; and

each time after verifying the checksum, operating said pointer to point to the checksum of the stack frame of the caller of the subroutine.

4. The method according to claim 1 , wherein said checksum is generated and stored in a Random Access Memory RAM.

5. The method according to claim 1 , wherein said pointer is introduced in a Central Processing Unit CPU's internal register.

6. The method according to claim 1 , wherein the method further comprises:

if the checksum verification result is negative, stopping execution of said program.

7. A method for protecting a program in a computer system, the method comprising:

when a subroutine of said program is called, pushing a return address on to a stack to start forming a stack frame;

when pushing said return address, generating a checksum for said stack frame;

each time a predetermined opcode is detected for said subroutine, updating said checksum according to an operand associated with said predetermined opcode;

if the predetermined opcode is a pop opcode, in addition to said updating, determining whether the operand associated with said pop opcode is said return address;

if it is determined that said operand is said return address, verifying said checksum before executing said predetermined opcode in order to detect an attack;

creating new opcodes which extend said predetermined opcodes, the new opcodes comprising:

a new push opcode that extends a push opcode to generate the checksum and/or update the checksum;

a new load opcode that extends a load opcode to update the checksum;

a new pop opcode that extends a pop opcode to update the checksum and/or verify the checksum; and

applying the new opcodes to the entire program or sensitive part of said program.

8. The method according to claim 7 , wherein the new push opcode further extend the push opcode to operate a pointer to point to the currently activate checksum, and wherein the new pop opcode further extends the pop opcode to operate a pointer to point to the currently activate checksum.

9. A method for protecting a program in a computer system, the method comprising:

when a subroutine of said program is called, pushing a return address on to a stack to start forming a stack frame;

when pushing said return address, generating a checksum for said stack frame;

each time a predetermined opcode is detected for said subroutine, updating said checksum according to an operand associated with said predetermined opcode;

if the predetermined opcode is a pop opcode, in addition to said updating, determining whether the operand associated with said pop opcode is said return address;

if it is determined that said operand is said return address, verifying said checksum before executing said predetermined opcode in order to detect an attack; and

when detecting an opcode instructing to access data in a loop, loading the data from the stack only once, storing the data in a register, and accessing the data from the register in a loop.

Assignments (4)

Priority Claims (1)

Continuity (1)

References Cited (9)