github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
samcv going through and tagging all issues if they're uni related, and closing some pretty old ones that have been fixed 08:24
Geth_ ¦ MoarVM: samcv self-assigned build on x32 has "pointer cast size mismatch" github.com/MoarVM/MoarVM/issues/444 09:07
dogbert11 oh, a SEGV 15:45
#0 0x00007f68f633b0be in MVM_multi_cache_find_spesh (tc=0x139aef0, cache_obj=0x47aa390, arg_info=0x7f68f4fa6450, type_tuple=0x7f68ec118990) at src/6model/reprs/MVMMultiCache.c:501 15:46
501 known_type_st = type_tuple[tt_offset].type->st;
when running the included example in R#1951 15:49
synopsebot_ R#1951 [open]: github.com/rakudo/rakudo/issues/1951 Improve empty hash creation performance with {}
dogbert11 gist for those interested: gist.github.com/dogbert17/c627e75b...33c1a1eced 15:53
dogbert17 reported as github.com/MoarVM/MoarVM/issues/996 20:15
AlexDaniel Please someone take a look at github.com/rakudo/rakudo/issues/2477 21:54
there are modules failing because of it
lucasb m: sub f1 { hash a=>1 }; sub f2 { hash b=>2 }; sub f3 { hash c=>3 }; f1 for ^10000; f2 for ^10000; f3 for ^10000 21:57
camelia (signal SEGV)
lucasb ^^ dogbert17 I tried golf it a little 21:58
on my 32-bit 1000 iterations is enough to segfault
*machine 21:59
dogbert17 lucasb: cool, if it fails everytime perhaps the bisectbot might be able to furnish additional info 22:00
.seen AlexDaniel 22:01
yoleaux I saw AlexDaniel 21:54Z in #moarvm: <AlexDaniel> there are modules failing because of it
dogbert17 :)
AlexDaniel . 22:03
bisect: sub f1 { hash a=>1 }; sub f2 { hash b=>2 }; sub f3 { hash c=>3 }; f1 for ^10000; f2 for ^10000; f3 for ^10000
bisectable6 AlexDaniel, On both starting points (old=2015.12 new=534d8f6) the exit code is 0 and the output is identical as well
AlexDaniel, Output on both points: «»
AlexDaniel argh 22:04
bisect: sub f1 { hash a=>1 }; sub f2 { hash b=>2 }; sub f3 { hash c=>3 }; f1 for ^100000; f2 for ^100000; f3 for ^100000
lucasb hehe tumuch 22:05
bisectable6 AlexDaniel, On both starting points (old=2015.12 new=534d8f6) the exit code is 0 and the output is identical as well
AlexDaniel, Output on both points: «»
AlexDaniel bisect: for ^10 { exit 42 if run(<perl6 -e>, ｢sub f1 { hash a=>1 }; sub f2 { hash b=>2 }; sub f3 { hash c=>3 }; f1 for ^10000; f2 for ^10000; f3 for ^10000｣).signal != 0 } 22:06
dogbert17 ah, nice trick
unless it times out ofc
bisectable6 AlexDaniel, Bisecting by exit code (old=2015.12 new=534d8f6). Old exit code: 0
dogbert17 I'm reasonably certain that the problem is newer than 2918.09 22:07
AlexDaniel oh wow
dogbert17 *2018
bisectable6 AlexDaniel, bisect log: gist.github.com/cfe99f2eebc7aeaa59...bdd540041b
AlexDaniel, (2018-09-25) github.com/rakudo/rakudo/commit/ac...5b952f1093
AlexDaniel if they still have segfaults in 2918 that's sad
dogbert17 indeed :)
AlexDaniel okay that's probably it
let's confirm
c: ace87cb5d9^,ace87cb5d9 for ^10 { say run(<perl6 -e>, ｢sub f1 { hash a=>1 }; sub f2 { hash b=>2 }; sub f3 { hash c=>3 }; f1 for ^10000; f2 for ^10000; f3 for ^10000｣).signal } 22:08
committable6 AlexDaniel, ¦ace87cb5d9^: «0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤» ¦ace87cb: «11␤0␤11␤0␤11␤0␤11␤11␤0␤11␤»
AlexDaniel c: ace87cb5d9^,ace87cb5d9 for ^10 { say run(<perl6 -e>, ｢sub f1 { hash a=>1 }; sub f2 { hash b=>2 }; sub f3 { hash c=>3 }; f1 for ^10000; f2 for ^10000; f3 for ^10000｣).signal }
committable6 AlexDaniel, ¦ace87cb5d9^: «0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤» ¦ace87cb: «11␤11␤11␤0␤0␤11␤11␤0␤0␤0␤»
AlexDaniel yeah that's definitely it
lucasb: can you file a ticket please?
dogbert17 or update the existing one
lucasb I think dogbert17 already did
dogbert17 lucasb++
AlexDaniel which one is it?
dogbert17 996 22:09
AlexDaniel R#996
synopsebot_ R#996 [closed]: github.com/rakudo/rakudo/pull/996 [JVM] Don't use --nqp-lib=blib for evalserver
AlexDaniel M#996
synopsebot_ M#996 [open]: github.com/MoarVM/MoarVM/issues/996 SEGV when running example code from R#1951
lucasb dogbert17++ AlexDaniel++ thanks 22:10
AlexDaniel cool 22:11
dogbert17 indeed
the other thing, i.e. R#2477 is probably something for brrt 22:12
synopsebot_ R#2477 [open]: github.com/rakudo/rakudo/issues/2477 [⚠ blocker ⚠] MoarVM panic: Register types do not match between value and node
jnthn wonders if accidentally snorting chili will help his cold... 22:15
I'll probably have chance to investigate the SEGV tomorrow 22:16
lizmat jnthn: snorting lukewarm water with a little salt, helps me every now and then
tadzik jnthn: "accidentally"? ;) 22:18
jnthn I got distracted when cooking, and then couldn't remember if the thing I'd picked out of the spice rack was the chili of paprika, so decided to sniff it, except I've got a cold and sniffed harder, and then I knew...because of the burning sensation :P 22:21
*or 22:22
tadzik ouch :D
jnthn 8501e2e Fix off-by-one in multi cache spesh lookup 22:23
Is in the bump the bissect found
And guess what's on top of the stack when it SEGVs?
MVM_multi_cache_find_spesh (tc=0x68eae0, cache_obj=0x7ffff3347368, arg_info=0x7ffff6237450, type_tuple=0x7fffec207590)
ASAN or valgrind can probably tell us what kind of naughty memory access it is. 22:24

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!