Table 18. RAM_SM_2
SM CODE RAM_SM_2
Description Stack hardening for Application software
Ownership End user
Detailed implementation
The stack hardening method is used to enhance Application software robustness to SRAM
faults that affect the address decoder. The method is based on source code modification,
introducing information redundancy in the stack-passed information to the called functions.
Method contribution is relevant in case the combination between the final Application software
structure and the compiler settings requires a significant use of the stack for passing function
parameters.
Implementation is the same as method CPU_SM_4.
Error reporting Refer to CPU_SM_4
Fault detection time Refer to CPU_SM_4
Addressed fault model Refer to CPU_SM_4
Dependency on Device configuration Refer to CPU_SM_4
Initialization Refer to CPU_SM_4
Periodicity Refer to CPU_SM_4
Test for the diagnostic Refer to CPU_SM_4
Multiple-fault protection Refer to CPU_SM_4
Recommendations and known limitations Refer to CPU_SM_4
Table 19. RAM_SM_3
SM CODE RAM_SM_3
Description Information redundancy for safety-related variables in Application software
Ownership End user
Detailed implementation
To address transient faults affecting SRAM controller, it is required to implement information
redundancy on the safety-related system variables stored in the RAM.
The guidelines for the implementation of this method are the following:
• The system variables that are safety-related (in the sense that a wrong value due to
a failure in reading on the RAM affects the safety functions) are well-identified and
documented.
• The arithmetic computation or decision based on such variables are executed twice and
the two final results are compared.
• Safety-related variables are stored and updated in two redundant locations, and
comparison is checked before consuming data.
• Enumerated fields must use non-trivial values, checked for coherence with the same
frequency as for periodically executed diagnostics (see
(1)
in Section 3.6 Hardware and
software diagnostics).
• Data vectors stored in SRAM must be protected by a encoding checksum (such as
CRC).
Error reporting Depends on implementation
Fault detection time Depends on implementation
Addressed fault model Permanent/transient
Dependency on Device configuration None
Initialization Depends on implementation
Periodicity On demand
Test for the diagnostic Not applicable
Multiple-fault protection CPU_SM_0: Periodic core self-test software
UM2305
Hardware and software diagnostics
UM2305 - Rev 10
page 19/110
To Next Page
Loading...
Need help?
General
