Dead code in mbedtls_mpi_exp_mod()

[pkolbus]

Description

• Type: Bug/Question
• Priority: Minor

Bug

OS
FreeRTOS

mbed TLS build:
Version: 2.14.1
OS version: 9.0
Configuration: MBEDTLS_MPI_WINDOW_SIZE = 6 (default)
Compiler and options (if you used a pre-built binary, please indicate how you obtained it): arm-none-eabi-gcc 6 2017-q2-update
Additional environment information: Static analyzer (Coverity)

Actual behavior
In mbedtls_mpi_exp_mod(), the comparison of wsize is always false. Coverity reports a potential issue. (line 1673, as of commit b9eb786).

Admittedly, the compiler should be able to optimize this out...

Question

Would it be appropriate to add a preprocessor guard #if (MBEDTLS_MPI_WINDOW_SIZE < 6) here? Or should the Coverity finding be treated as an intentional practice?

If a preprocessor guard is appropriate, I'm happy to create a PR.

[RonEld]

@pkolbus Thank you for raising this issue!
I agree that when MBEDTLS_MPI_WINDOW_SIZE is 6, this comparison is always false, and coverity might warn on this, therefore marking this as a bug.
You are welcome to raise a PR, assuming you have accepted our CLA.
Thanks again for your interest in Mbed TLS!

[ciarmcom]

ARM Internal Ref: IOTSSL-2681