With modules x images at or above what the card's HBM holds for the
internal frame generator (64 module frames in the HLS simulation, 1024 on
the card), frame_generator refuses the run and sends nothing, and the
collection hung in Measuring with no error. The host now checks this:
PrepareAction refuses such a run at the start (a 32-bit EIGER frame
counts twice, as in frame_generator.cpp), and SetInternalGeneratorFrame
refuses an image that would land there, each with an error that says
how many module frames fit. The capacity comes from hbm_size_bytes of
the card's status (the mock keeps its 64).
The check keeps the last module frame free as well: load_calibration.cpp
accepts module_id == 2 * hbm_size_bytes / module size (">" where ">="
was meant), one past the memory. The firmware is frozen, so the host
never sends that index instead.
The HLS simulation keeps frame_generator's return code: a refused
configuration cancels the data collection rather than waiting for frames,
and HLSSimulatedDevice::FPGA_EndAction throws with the code.
rc184 has no MockAcquisitionDevice: its part is left out, and the test checks HLSSimulatedDevice only, as HLSSimulatedDevice_InternalGeneratorMemory in FPGAIntegrationTest.
(ported from rc184-2 25542a9af)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HVeDQsU5a8AJUdJPgWevAK
leonarski_fandClaude Opus 5.5
f3be10f279