--- Log opened Sat Jan 12 00:00:24 2019 | ||
shorne | I am working on fixing up the or1k-tests cases | 16:10 |
---|---|---|
shorne | I first got them running with cappuccino pipeline, all are pretty much passing | 16:11 |
shorne | Now I am looking at espresso | 16:11 |
shorne | I figured I would document the status here | 16:11 |
shorne | Please feel free to comment / help | 16:11 |
shorne | https://github.com/openrisc/mor1kx/issues/71 | 16:12 |
tpb | Title: Espresso Tests Failing · Issue #71 · openrisc/mor1kx · GitHub (at github.com) | 16:12 |
shorne | First test failing is, https://github.com/openrisc/or1k-tests/blob/master/native/or1k/or1k-alignillegalinsn.S | 16:12 |
tpb | Title: or1k-tests/or1k-alignillegalinsn.S at master · openrisc/or1k-tests · GitHub (at github.com) | 16:12 |
shorne | tracing the VCD... I see 1. exection get to the first l.sh, which causes alignment | 16:13 |
shorne | exception. | 16:13 |
shorne | 2. alignment handler runs, and calls l.rfe | 16:14 |
shorne | 3. l.sh runs again with an address that shouldnt cause alignment exception, but everything stops there | 16:14 |
shorne | PC jumps to 0x0 | 16:14 |
shorne | Issue, seems now is setting NPC, when advancing pc_fetch_i is 0x0 | 16:23 |
shorne | I think I fixed it | 16:40 |
shorne | rtl/verilog/mor1kx_ctrl_espresso.v | 16:41 |
shorne | @@ -469,7 +469,7 @@ module mor1kx_ctrl_espresso | 16:41 |
shorne | if (rst) | 16:41 |
shorne | take_exception <= 0; | 16:41 |
shorne | else | 16:41 |
shorne | - take_exception <= (exception_pending | exception_r | doing_rfe_r) & | 16:41 |
shorne | + take_exception <= (exception_pending | exception_r) & | 16:41 |
shorne | (fetch_advance | | 16:41 |
shorne | // Cause exception to always be 'taken' if stepping | 16:41 |
shorne | Don't take the exception PC during RFE, we take the next PC | 16:42 |
shorne | Test gets further now dieing on first l.j | 16:42 |
--- Log closed Sun Jan 13 00:00:26 2019 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!