Created attachment 547 [details] upper 16 channels with random content (which is not in the .sr input file) File gpib/serial/cbm1571/cbm_1571_read_status.sr comes with inconsistent metadata, although there is only logic data for 8 channels and unit size is 1, 16 channels are listed. Which results in random data display which changes with zoom and pan (and may depend on the platform, compile options, etc).
IRC discussion suggests the issue may result from not masking out absent data in the optimized code path for unaligned access.
Strictly speaking the cbm_1571_read_status.sr file's metadata need not be inconsistent, but it seems to be unexpected, and is a good candidate to trigger the reported issue. :) Is both the set of 16 channels, and the unitsize 1 and thus data availability for the first 8 channels only, available to the pulseview application code? So that traces' data can get masked out when their input data is absent?