|
github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm Set by AlexDaniel on 12 June 2018. |
|||
| ugexe | jnthn: ci.appveyor.com/project/rakudo/rak...ta9j#L4681 | 01:21 | |
| looks like moarvm changes in master might have broken rakudo make test? | |||
| on windows | |||
|
03:14
elronnd joined,
elronnd is now known as Elronnd
06:43
domidumont joined
07:42
brrt joined
|
|||
| brrt | \o | 07:42 | |
| nwc10 | o/ | 07:43 | |
| brrt | I'm going to get rid of `reg_type` in the linear scan allocator | 07:53 | |
| oh hang on, reg_type means something differnet than I thought it di | 07:55 | ||
| I mean, reg_spec | 08:05 | ||
| Geth | MoarVM/jit-expr-float: 17 commits pushed by (Bart Wiegmans)++ review: github.com/MoarVM/MoarVM/compare/b...cfcf0cdf21 |
||
| brrt | fastcrash! | 08:32 | |
|
08:35
brrt left
08:37
brrt joined
|
|||
| brrt | hmm, debugging on windows isn't very easy | 08:56 | |
| timotimo | soon windows will ship a linux kernel, then debugging will be easier :P | 08:58 | |
|
09:01
brrt left
10:31
lizmat joined
11:05
lizmat left
|
|||
| jnthn | timotimo: Yeah, but will it have the same bugs? ;) | 12:39 | |
|
12:45
dogbert17 joined
13:29
harrow left
13:35
harrow joined
|
|||
| Geth | MoarVM/configurable-subsystems: 6 commits pushed by (Timo Paulssen)++ | 13:40 | |
| timotimo | (mostly a rebase) | ||
|
14:16
pamplemousse joined
14:47
pamplemousse left
15:02
pamplemousse joined
|
|||
| pamplemousse | Hi! I'm currently trying to reproduce this issue: github.com/MoarVM/MoarVM/issues/1068 and when I run the gdb command referenced in the issue, I get a message saying, "Reading symbols from /home/madeleine/sandbox/perl6/install/bin/moar...(no debugging symbols found)...done." I configured MoarVM with the --debug flag when I was setting everything up, do you guys have any ideas what I'm missing in trying to reproduce | 15:53 | |
| this? | |||
| timotimo | pamplemousse: bin/moar usually has barely any debug symbols, but libmoar.so will have a load of 'em | 16:02 | |
|
16:03
ggoebel joined
|
|||
| timotimo | the most important is that you get names, filenames, and line numbers when you "bt" | 16:05 | |
| annoyingly, you can't put in breakpoints before the program has started | |||
| well, you can. you just have to click through a prompt every time | 16:09 | ||
| and you don't get tab completion | |||
| pamplemousse | Thank you! Using libmoar.so worked. | 16:13 | |
| MasterDuke | pamplemousse: looks like i didn't need to in this case, but you often need to disable the jit (e.g., MVM_JIT_DISABLE=1 gdb ...) to get good backtraces | 16:17 | |
| pamplemousse | Getting a bit further, but it doesn't get to .gdbinit. I'm not super comfortable with gdb so it might be that. I needed chmod $HOME/perl6/install/share/perl6/runtime/perl6.moarvm and $HOME/perl/lib/libmoar.so but now get the symbols read. | 16:24 | |
| timotimo | that's a little odd | 16:26 | |
| i'm not sure the ".gdbinit: running app" message is necessary | |||
| what do you mean it "doesn't get to" that? | |||
| MasterDuke | how did you build moarvm? | ||
| pamplemousse | The .gdbinit: running app message doesn't happen and it doesn't start up any of the threads and I can't backtrace. I'm executing the equivalent gdb command for: github.com/MoarVM/MoarVM/issues/1068 | 16:29 | |
| timotimo | do you perhaps have to "run" before anything happens? | ||
| oh! | 16:30 | ||
| you know, if you have a very up to date rakudo, you don't have to have the long command | |||
| and in fact | |||
| it may not work | |||
| i believe the new runner will forward --full-cleanup for you from the perl6 command | |||
| so maybe this works: gdb --args perl6 --full-cleanup -e 'blah blah code here blah' | 16:31 | ||
| pamplemousse | MasterDuke: From the head of the current source for MoarVm with ./Configure.pl --prefix="appropriate prefix" --debug; make install | ||
| timotimo | i often not only --debug but also --optimize=1 or even --optimize=0 so that debugging gets easier | ||
| MasterDuke | or --debug=3 | 16:32 | |
| timotimo | i'm not actually completely certain if that makes a difference (but i do pass it always) | ||
| pamplemousse | I'll try that, thanks | 16:33 | |
| timotimo | BBL | 16:36 | |
| pamplemousse | MasterDuke: Are you currently able to use the gdb command from #1068 to reproduce the same output? This is my output: pastebin.com/4SRLf2Ae | 16:49 | |
| MasterDuke | pamplemousse: did you type 'run' in gdb? | 16:52 | |
| pamplemousse | I did and I got an unhandled exception: index start offset out of range at gen/moar/main.nqp:1 | 16:55 | |
| MasterDuke | i changed the command to `gdb --args /home/dan/Source/perl6/install/bin/perl6 --full-cleanup -e 'for ^2 { my $cmd = run « echo 1 », :out; $cmd.out.close }'` and it works | 16:57 | |
| but i now get `Unhandled exception: Cannot find method 'name' on object of type Perl6::Metamodel::ClassHOW` instead of the segv | 16:58 | ||
| but valgrind still complains about `invalid read` and `Process terminating with default action of signal 6 (SIGABRT): dumping core` | 16:59 | ||
| woah! if i configure moar with --valgrind i get tons of errors | 17:01 | ||
| pamplemousse | Just tried running the command again and I'm now getting the same unhandled exception command as you | 17:02 | |
| MasterDuke | afk for a bit, but it looks like you're making progress now | 17:03 | |
| pamplemousse | Thanks for your help! | 17:04 | |
| Grabbing lunch, BBL | 17:08 | ||
|
17:13
pamplemousse left
17:32
ggoebel left
17:43
zakharyas joined
17:54
domidumont left
18:08
Altai-man_ joined
18:11
sena_kun left
18:55
zakharyas left
18:58
zakharyas joined
21:27
mgoebel joined,
mgoebel left
21:38
zakharyas left
21:58
lizmat joined
22:55
lizmat left,
lizmat joined
23:05
lizmat left
23:45
Altai-man_ left
|
|||