[Libre-soc-dev] daily kan-ban update 22dec2021

lkcl luke.leighton at gmail.com
Wed Dec 22 20:04:59 GMT 2021

early this morning i had a first microwatt mmu.bin test pass, after updating multiple obscure bugs including MTMSRD reading the wrong X-Form L field because X-Form has 3 completely different named L fields.

(what that has to do with the MMU is of course that the 0x300 DSI trap handler sets the MSR).

the second mmu.bin test is proving trickier: the Page Directory Entry is being read correctly but the Page Table Entry is not. i am staring at several thousand trace state changes and have no clue.

therefore i need to revive the verilator DMI single-stepping but TestIssuer's FSMs cannot cope with single-step, so i have to fix it.

microwatt however i have confirmed is ok to single-step, and i was able to get a full debug execution trace.

once libresoc singlestep is fixed i can do a diff -u and identify exactly and very quickly which instruction is going wrong.


More information about the Libre-soc-dev mailing list