tree 64fb195c6aebc36deeec0c8994490123bfbcf88a
parent 7d477c4b60f9e38c7f2f87ea37adbeab59d9e270
author Anthony Rocha <anthony.rocha@amd.com> 1790294652 -0700
committer Anthony Rocha <anthony.rocha@amd.com> 1790294652 -0700

pldm: use run_until to remove timeout-as-success ambiguity in host tests

base_host, inventory_host, unexpected_eid_host, and unexpected_eid_fw_host
all queue one command, run the FD, and check the response, then relied on
run_terminus timing out to signal "done" -- indistinguishable from a real
hang. Add a shared run_fd_step helper (common.rs) that uses run_until to
bound each of these to exactly one step, so every error they can now hit
is real.

firmware_update_host's own multi-step flow (an unbounded number of
RequestFirmwareData/TransferComplete/VerifyComplete/ApplyComplete round
trips per call) doesn't fit a fixed step count and keeps using
run_terminus directly; it already gained a direct run_until test in the
prior commit.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>