Closed smunaut closed 2 years ago
Thanks @smunaut! The code changes make sense and I've been using it the last fews days in different configurations over JTAGBone and Etherbone without troubles so we can merge.
@smunaut: I just did a minor change with https://github.com/enjoy-digital/litescope/commit/13813457d74152e8c21f899508636ca374897500 to simplify the code and fix the case where pad_bits == 0
.
Instead of using a very wide CSR, we force it to max 32 bits and read each captured words as several sub-words.
Also, instead of checking for 'valid' flag every time, we just read the 'level' of the memory buffer. Given the way LiteScope works, we know capture is done and this is how many words there is to read.
All in all this means that reading the data off the buffer is just reading the same address over and over meaning we can use very long bursts which helps _alot to speed things up.
Signed-off-by: Sylvain Munaut tnt@246tNt.com