Fixups: Implement sign extension + Fixes

[stffrdhrn]

Working on fixing up some of the test cases in or1k-tests. I found l.ext* instructions were never implemented. I implemented it using case because I thought its easy to read like that. But if others have suggestions let me know.

[stffrdhrn]

This looks to fix #53

[stffrdhrn]

I added a commit the this PR to fix #48 too.

[olofk]

The core file and the shorter fixes look fine to me. I'm afraid I don't know the code well enough to give a proper review of the ext stuff, but it looks like it does the job. Only thing I'm wondering is if this will put more logic in the alu_result_o path even when FEATURE_EXT is disabled? Please take a look at that, but I'm happy to see these patches merged anway

[stffrdhrn]

Thanks for the review. It's a good point in the extra logic on alu_result_o. I tried to model it like the other optional features which I assumed would be optimized out when disabled. I will check what I can see in the netlist, I might find I need to fix other optional functions as well.

[bandvig]

I think you shiuld add round brackets for decode_op_ext_o (in mor1kx_decode) expression in the way:
assign decode_op_ext_o = opc_insn == OR1K_OPCODE_ALU && (opc_alu == OR1K_ALU_OPC_EXTBH ||
opc_alu == `OR1K_ALU_OPC_EXTW);

[stffrdhrn]

Thank you, I made the change to clean up decode_op_ext_o logic as suggested by Andrey and this should also help to remove the logic from alu_result_o when extension is disabled. But I am away from my desktop and don't know how to check the synthesis results from the command line.

Also, one strange thing, I am trying to test the mor1kx build with de0_nano-multicore. But Its failing to fit. It seems something has come into mor1kx or another core which is causing us to go over the resource limit now. I am looking into it. Once its fixed I will merge this.

Also, I added one more commit to pull the BRANCH PREDICTION parameter up to the mor1kx top.

[stffrdhrn]

This also fixes #66 now, after reverting the true_pram change from @imphil. Philipp, let me know if you know if there is a better solution.

[stffrdhrn]

I updated the fix for #66 as discussed with Philipp, I am hoping switching to 2 clocks will work for xilinx, it works for altera.

[stffrdhrn]

I am merging this. or1k-test results are looking OK.

Running or1k-alignillegalinsn                               PASS
Running or1k-backtoback_jmp                                 PASS
Running or1k-basic                                          PASS
Running or1k-cmov                                           PASS
Running or1k-cy                                             FAIL  - failing as mor1kx doesn't support multiply carry
Running or1k-dsx                                            PASS
Running or1k-dsxinsn                                        PASS
Running or1k-ext                                            PASS
Running or1k-ffl1                                           PASS
Running or1k-icache                                         PASS
Running or1k-illegalinsn                                    PASS
Running or1k-illegalinsndelayslot                           PASS
Running or1k-insnfetchalign                                 PASS
Running or1k-insnfetcherror                                 PASS
Running or1k-intloop                                        PASS
Running or1k-intmulticycle                                  PASS
Running or1k-intsyscall                                     PASS
Running or1k-inttickloop                                    PASS
Running or1k-jmp                                            PASS
Running or1k-jr                                             PASS
Running or1k-lsu                                            PASS
Running or1k-lsualign                                       PASS
Running or1k-lsualigndelayslot                              PASS
Running or1k-lsuerror                                       PASS
Running or1k-lsuerrordelayslot                              PASS
Running or1k-lwjr                                           PASS
Running or1k-msync                                          PASS
Running or1k-mul-basic                                      PASS
Running or1k-ov                                             PASS
Running or1k-regjmp                                         PASS
Running or1k-rfe                                            PASS
Running or1k-sf                                             PASS
Running or1k-sfbf                                           PASS
Running or1k-shiftopts                                      PASS
Running or1k-shortbranch                                    PASS
Running or1k-shortjump                                      PASS
Running or1k-systemcall                                     PASS
Running or1k-tickloop                                       PASS
Running or1k-tickrfforward                                  PASS
Running or1k-ticksyscall                                    PASS
Running or1k-timer                                          PASS
Running or1k-trap                                           PASS
Running or1k-trapdelayslot                                  PASS
Results                                                     Total:  43  FAIL:   1

Creating issue for multiply carry limitation.

[juliusbaxter]

This all looks good to me. Nice work folks!