Repository navigation
Conversation
wolfSSL-Fenrir-bot
left a comment
There was a problem hiding this comment.
Fenrir Automated Review — PR #980
Scan targets checked: wolfboot-src, wolfboot-bugs
Not reviewed (no scan target covers them): docs/PolarFire.md, docs/Targets.md, hal/mpfs250.h, test-app/Makefile, test-app/RISCV64-mpfs250.ld, test-app/vector_riscv.S
Coverage: 5 of 5 in-scope changed file(s) opened by the reviewer
Fenrir result: Approved ✅
No new issues found in the changed files.
Advisory only — this automated result does not count as a GitHub approval.
Review tier: Max
…PT_BENCHMARK with per-size offload timing
aa85c8e to
b71229c
Compare
wolfSSL-Fenrir-bot
left a comment
There was a problem hiding this comment.
Fenrir Automated Review — PR #980
Scan targets checked: none
Unchanged since last review (not re-run): wolfboot-src, wolfboot-bugs
Not reviewed (no scan target covers them): docs/PolarFire.md, docs/Targets.md, hal/mpfs250.h, test-app/Makefile, test-app/RISCV64-mpfs250.ld, test-app/vector_riscv.S
Fenrir result: Approved ✅
No new issues found in the changed files.
Advisory only — this automated result does not count as a GitHub approval.
Review tier: Lite
Description
With
MPFS_ATHENA=1the PolarFire SoC test application brings the engine up itself, so the stock wolfCrypt test and benchmark run against it bare metal, in the M-mode environment wolfBoot uses.test-app/app_mpfs250.c,test-app/Makefile: registration and self-test, the benchmark once per call size inMPFS_BENCH_SIZES, and the engine call count at the end, since a passing test proves nothing about offload on its own.include/user_settings.h:WOLFSSL_USE_ALIGNforARCH_RISCV64, as the harts trap on a misaligned access with nothing to emulate it; a static pool large enough for the benchmark's buffers; and an__ASSEMBLER__guard on the two test-app function declarations, which the assembler otherwise reads when wolfSSL's RISC-V assembly includes this header.test-app/startup_riscv.c,test-app/vector_riscv.S,hal/mpfs250.c,hal/mpfs250.h: the M-mode trap handler prints the cause and halts instead of resuming at the fault; a machine-timer interrupt refreshes the five MSS watchdogs, which expire well inside a test run, with the HAL's MTIME enable and CLINT timer registers made available to the app.test-app/wolfcrypt_support.c,test-app/RISCV64-mpfs250.ld,test-app/Makefile: the timebase follows the HAL's live clock (the DDR configuration runs the E51 at the PLL rate), and the stack the benchmark's SP math frames need is substituted into the linker script only for aWOLFCRYPT_TESTorWOLFCRYPT_BENCHMARKbuild, since the template is shared with the L2 scratchpad configurations that have no room to reserve it.docs/PolarFire.md,docs/Targets.md: how to build and read it, with the measured offload.Testing
MPFS250TS Video Kit, DDR configuration:
wolfcrypt_test()complete and the benchmark at every size, each with and without the engine. The M-mode QSPI configuration runs the benchmark app withWOLFBOOT_L2SCRATCH_SIZE=512k.