[J-core] [Possible bug] It seems only decode_table_reverse.vhd works out of the decode implementations

Robert Ou rqou at robertou.com
Mon Jul 18 05:44:39 EDT 2016


Hi,

It seems to me that only the decode_table_reverse.vhd implementation
actually works correctly. If you take my MyHDL demonstration sent to
the list earlier and change the ghdl command line to use one of the
other decode implementations, you will get:

with decode_table_rom.vhd:
The CPU continuously fetches instructions sequentially. It never seems
to actually execute them. Some poking around seems to show that the
line signal never changes.

with decode_table_simple.vhd:
Nothing ever happens externally. I didn't investigate what happens
inside the core.

Of course, these issues could very well be with my particular
demonstration setup. Can someone confirm that these other decode
implementations are/are not working with the normal J-core build
system? I am using the 20160418 source dump, md5
f6876049560136ab0a6b21914a939b95.

Robert


More information about the J-core mailing list