| --- Log opened Sat Sep 17 00:00:15 2016 | ||
| shorne | SMDhome: hello about running gdb against a program running in gdb | 04:36 |
|---|---|---|
| shorne | I am trying now, ideally it would be | 04:36 |
| shorne | sim -f ../openrisc/openrisc_os/or1ksim.cfg --srv=3333 | 04:36 |
| shorne | (starts the gdb server in or1ksim) | 04:37 |
| shorne | or1k-elf-gdb ./work/linux/vmlinux --eval-command='target remote localhost:3333' | 04:37 |
| shorne | (gdb connect to or1ksim debug port, you can then load the binary through there) | 04:37 |
| shorne | but when I run now I am getting an error ... Remote replied unexpectedly to 'vMustReplyEmpty': E01 | 04:38 |
| shorne | I think my gdb is too new | 04:40 |
| SMDhome | shorne: thanks! I should try it. Which sim are you using? | 04:41 |
| SMDhome | verilator or icarus | 04:42 |
| shorne | oh, its just or1ksim, right you were asking for one of the verilog simulators | 04:42 |
| shorne | they probably also support a gdbserver more? | 04:42 |
| shorne | *mode* | 04:42 |
| shorne | With icarus there is a blog here: https://balau82.wordpress.com/2010/08/07/debugging-openrisc-software-inside-rtl-simulation/ | 04:45 |
| SMDhome | thanks! | 04:46 |
| shorne | Similar details here: http://www.embecosm.com/appnotes/ean7/embecosm-gdb-verilator-ean7-issue-1.html#sec_jtag | 04:47 |
| shorne | The second one seems to be using verilator | 04:48 |
| shorne | Sorry if you already tried those, I havent tried either. But thats where I would start | 04:49 |
| wallento | olofk: About to PR my changes to fusesoc | 08:17 |
| wallento | found an acceptable way to cope with verilator being synth, I just use "usage = sim verilator" | 08:18 |
| wallento | :) | 08:18 |
| wallento | by the way: travis shows errors | 08:23 |
| wallento | something with attrs in fusesoc/satsolver and virtualenv or | 08:24 |
| olofk | SMDhome, shorne : About the embecosm app notes, we changed how to do gdb a few years ago (with the introduction of orpsocv3 which became FuseSoC) | 13:00 |
| olofk | Previously we implemented a gdbserver as a VPI library so that you could connect gdb to the sim | 13:01 |
| olofk | But now we only open up a socket that speaks _franck_'s jtag_vpi protocol. On the other side we connect openocd | 13:02 |
| olofk | The benefit is that we don't have to keep our internal gdbserver in synce with upstream changes. Already a few years ago, there were a lot of unimplemented features | 13:03 |
| olofk | So now we let openocd implement the gdbserver instead | 13:03 |
| olofk | mor1kx-generic implements jtag_vpi, so if you're running that just add --enable_jtag_vpi and connect openocd to port 5555 using the jtag_vpi protocol | 13:04 |
| olofk | Now you can connect gdb to the port that openocd exposes. Can't remember right away which one it is though | 13:05 |
| olofk | But if you're only interested in loading elf files, reading and writing memory and starting/stopping the CPU you don't need gdb. You can use the openocd telnet interface directly | 13:11 |
| olofk | I use it all the time now :) | 13:12 |
| olofk | Also discovered that you can create tcl scripts for it. Great for testing out new drivers | 13:12 |
| olofk | wallento: Thanks for the PR. I'm a bit tired now so I'll look at it later. Just ran 10km through mud up to my waist | 13:14 |
| olofk | But great that you found a way to cope with the verilator files. | 13:15 |
| olofk | And I know about the travis errors. I filed a bug against okonomiyaki, which is the lib that fails | 13:15 |
| olofk | But unfortunately it's really a pip bug (Oh god how I hate these stupid python package managers) | 13:16 |
| olofk | I think we could work around it by installing a known good version of attrs before FuseSoC, but I haven't looked into that yet | 13:16 |
| olofk | wallento: Fixed travis builds now | 16:16 |
| olofk | wallento: I think the cores root reordering is wrong | 16:24 |
| olofk | Cores found later will override those found earlier | 16:25 |
| olofk | And highest priority should be given to those specified on the command-line with --cores-root | 16:25 |
| olofk | So they need to be last | 16:26 |
| olofk | I pushed the other ones related to paths. Need to read through the DPI and Vivado patches when I get a bit more time | 16:29 |
| olofk | wallento: Pheew.. I just finished a review of the Vivado/XSIM patch | 17:11 |
| olofk | There are some bigger and smaller changes. I really like to see this merged soon, so hopefully we can fix this quickly together | 17:12 |
| olofk | Oh. I forgot a big thank you for all the work, in the review | 17:12 |
| olofk | My reviews tend to only focus on the negative parts, but all in all it's a great contribution | 17:13 |
| --- Log closed Sun Sep 18 00:00:17 2016 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!