Benchmark example times out on UBLOX EVK ODIN W2, on GCC_ARM

[k-stachowiak]

The example times out while performing benchmark for dh-3072. Surrounding log:

[CONN][RXD]   DHE-3072                 :    8571 ms/handshake
[HTST][INF] test suite run finished after 600.36 sec...
[HTST][ERR] Expected output [\s+DH-3072\s*:\s*\d+ ms/handshake] not received in log.

[ciarmcom]

ARM Internal Ref: IOTSSL-2611

[k-stachowiak]

Note that it is similar to #210

[simonbutcher]

GCC_ARM is now passing, but IAR is failing at the same point, as can be seen in the log for build #816.

[RonEld]

@k-stachowiak I don't have a board to test, but could you check if setting MBEDTLS_MPI_MAX_SIZE to 512 fixes this issue? I have noticed that the default size of 1024 bytes causes stack overflow in many cases.

[k-stachowiak]

@sbutcher-arm This issue seems to be fixed by #213, but that has not been approved yet. If it is an acceptable approach to remove the tests, then the issue may be considered resolved (even though this is a workaround, rather than a fix).

@RonEld I don't have the board either, but I raised a test PR (#219) to see the result in the CI run.

[RonEld]

@k-stachowiak Thank you for running the test PR.

If reducing the MPI size to 512 will work, I will make a PR changing the default MPI size

[k-stachowiak]

@RonEld The CI results are already in. There doesn't seem to be any timeout issue (besides ones stemming from a missing(?) network interface). However, there is still one failure because of lack of memory.

[RonEld]

@k-stachowiak Yes, I noticed, thanks The lack of memory probably needs further investigation, but at least it fixes the timeout

[k-stachowiak]

@sbutcher-arm @Patater Based on #231 I conclude, that the problem doesn't occur anymore. I suggest closing this issue.