github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
01:59 frost-lab joined 03:46 evalable6 left 03:50 evalable6 joined 06:59 domidumont joined
nwc10 Good Friday, #moarvm 08:36
09:33 frost-lab left 09:57 linkable6 left 09:58 linkable6 joined
dogbert11 good afternoon 10:30
regarding the SEGV that nwc10 found, all I can see is that when running the loop which evaluates the guards, the following seems to happen: 10:49
current_node agn->op
0 MVM_SPESH_GUARD_OP_CALLSITE
2 MVM_SPESH_GUARD_OP_LOAD_ARG
3 MVM_SPESH_GUARD_OP_STABLE_TYPE
4 MVM_SPESH_GUARD_OP_LOAD_ARG
5 MVM_SPESH_GUARD_OP_STABLE_CONC # SEGV here at github.com/MoarVM/MoarVM/blob/mast...ard.c#L496
nwc10, nine: is the above something which makes sense? 10:53
nine dogbert11: I think the segfault happens because that LOAD_ARG step loads a NULL, right? 11:21
dogbert11 nine: I think so if my debug printf's are correct :) 11:35
I don't know how to get backwards from this point though 11:38
nine rr? 11:40
Where would it load that arg from? 11:41
11:43 linkable6 left 11:45 linkable6 joined
dogbert11 nine: sadly I can't use rr, last time I checked Virtualbox didn't support it :( 12:00
nine: your services is in high demand today :) 12:01
s/is/are/ 12:02
nine you have no idea... 12:07
MasterDuke good friday 12:12
12:14 brrt joined
brrt \o 12:14
dogbert11 hello brrt and MasterDuke 12:16
brrt ohai dogbert11 12:19
dogbert11 brrt, anything interesting going on? 12:27
nine: could the following tail end of the spesh log give any further clues? gist.github.com/dogbert17/1f3da75c...01f50a62e2 12:28
MasterDuke especially now that our CI is being worked on, it'd be nice to get some extra kinds of testing added (e.g., run some stuff with --full-cleanup under valgrind to detect if we add any new leaks)
nine gotta go 12:30
brrt dogbert11: not really 12:33
12:41 linkable6 left
lizmat feels like there's still a gremlin in the import process 12:43
t/spec/S11-modules/import-multi.rakudo.t just flapped for me in test 6 12:44
12:44 linkable6 joined
MasterDuke well github.com/MoarVM/MoarVM/pull/1454 is now green, because i reran one of the windows pipelines and t/02-rakudo/repl.t apparently is a flapper (maybe just on windows?) 12:49
12:50 linkable6 left
lizmat ok, so merge that and bump MoarVM / NQP ? 12:51
12:53 linkable6 joined
MasterDuke i wouldn't necessarily say it deserves/requires a bump, but sure 12:53
12:54 linkable6 left 12:55 linkable6 joined
MasterDuke i also wouldn't object if someone gave it a once over 12:58
i'm pretty sure it's correct in that it doesn't do anything wrong, but maybe there's more it could do
13:00 linkable6 left 13:02 linkable6 joined
lizmat ah... ok, not much happened since the last bump 13:05
well, for bisectability I would like to see smaller bumps :-) 13:06
13:06 linkable6 left 13:09 linkable6 joined
dogbert11 brrt: perhaps something will come up later 13:09
13:20 linkable6 left 13:21 linkable6 joined 13:22 brrt left 13:29 linkable6 left 13:31 linkable6 joined 13:35 linkable6 left 13:37 linkable6 joined 13:38 linkable6 left, brrt joined 13:40 linkable6 joined 13:44 linkable6 left, linkable6 joined 13:46 linkable6 left 13:47 linkable6 joined 14:08 linkable6 left 14:10 linkable6 joined
dogbert11 www.anandtech.com/show/16584/arm-a...chitecture 14:56
brrt fun 14:58
do they have a different ABI?
dogbert11 brrt: haven't actually read it yet, just noticed it 15:01
15:32 brrt left 17:11 sena_kun left, Util left 17:13 sena_kun joined 17:18 domidumont left 17:24 domidumont joined 17:28 domidumont left 18:04 sena_kun left
lizmat trying to upgrade 'dd' to do this: glot.io/snippets/fx9cjgi62o 18:27
Stage mast : Serialization Error: missing static code ref for closure '' (gen/moar/World.nqp:2597) 18:28
now, I can put that code into a module, and export the "ic" and it works fine
so I'm a bit perplexed it doesn't work (at the end of) the setting 18:29
t/spec/S11-modules/import-multi.rakudo.t just flapped for me in test 6 *again* 18:46
can't reproduce outside of spectest :( 18:48
18:58 zakharyas joined
MasterDuke don't think i've ever seen that one flap 19:34
lizmat: btw, re the "Serialization Error: missing static code ref for closure" i assume you've seencolabti.org/irclogger/irclogger_lo...-04-02#l83 ? 19:42
[Coke] jnthn: would you like someone to remove the "email you about moarvm" note in the readme? 19:51
Geth MoarVM: coke++ created pull request #1455:
Minor README updates
20:03
nine dogbert11: oh, it looks like that spesh segfault is already fixed by my various_fixes branch :) 20:04
I guess it's about time to merge that
Funny how a week after I did that another case pops up where the mis-spesh leads to NULLs in registers
dogbert11 nine: that sounds like very good news indeed 20:08
20:50 zakharyas left 21:11 dogbert17 joined 21:14 dogbert11 left
jnthn [Coke]: hah, that has probably been there from the start. I don't really mind in so far as pretty much nobody does...but yeah, given they're likely to get a more timely response on IRC, it's probably a better experience :) 21:21
23:14 evalable6 left, linkable6 left 23:15 linkable6 joined 23:16 evalable6 joined
[Coke] ok, will remove it. Cheers! 23:31
(PR updated) 23:34