Welcome to the main channel on the development of MoarVM, a virtual machine for NQP and Rakudo (moarvm.org). This channel is being logged for historical purposes.
Set by lizmat on 24 May 2021.
00:09 reportable6 left 00:10 reportable6 joined 00:50 nativecallable6 joined, committable6 joined, benchable6 joined 00:52 coverable6 joined, bloatable6 joined 01:48 evalable6 joined 01:52 notable6 joined 02:50 linkable6 joined 02:51 statisfiable6 joined 02:52 bisectable6 joined 06:08 reportable6 left, reportable6 joined 07:08 sourceable6 left, coverable6 left, reportable6 left, statisfiable6 left, bloatable6 left, linkable6 left, notable6 left, evalable6 left, bisectable6 left, unicodable6 left, nativecallable6 left, benchable6 left, tellable6 left, quotable6 left, shareable6 left, squashable6 left, releasable6 left, committable6 left, greppable6 left, greppable6 joined 07:09 sourceable6 joined 07:10 statisfiable6 joined, linkable6 joined, coverable6 joined, bisectable6 joined 07:11 reportable6 joined, shareable6 joined, releasable6 joined 07:59 frost joined 08:09 squashable6 joined 08:11 tellable6 joined 09:09 bloatable6 joined, nativecallable6 joined 09:10 evalable6 joined 09:11 benchable6 joined, committable6 joined, notable6 joined 10:11 quotable6 joined 10:55 RakuIRCLogger left, RakuIRCLogger joined 11:01 RakuIRCLogger left 11:02 RakuIRCLogger joined 12:08 reportable6 left 12:11 reportable6 joined 13:02 frost left 14:12 unicodable6 joined
Geth MoarVM/fix_gc_issue_in_materialization: b5d04d754f | (Stefan Seifert)++ | src/spesh/deopt.c
Fix possible access to fromspace after deopt materialization

We stored the newly allocated objects we create during materialization in an array, for future reference so we could store the same objects into multiple target registers if required. Allocating these objects however could trigger a GC run which then missed the pointers in the materialized array. This would then lead to such outdated pointers getting written to frame registers. ... (5 more lines)
14:22
MoarVM/fix_gc_issue_in_materialization: 2883ba7547 | (Stefan Seifert)++ | src/spesh/deopt.c
Fix possible access to fromspace after deopt materialization

We stored the newly allocated objects we create during materialization in an array, for future reference so we could store the same objects into multiple target registers if required. Allocating these objects however could trigger a GC run which then missed the pointers in the materialized array. This would then lead to such outdated pointers getting written to frame registers. ... (7 more lines)
14:23
MoarVM: niner++ created pull request #1630:
Fix possible access to fromspace after deopt materialization
nine I didn't like my first attempt (using temp roots) of a fix and it turned out, it was utterly broken. Second attempt (storing materialized objects in a VMArray) worked but was a bit heavier on performance. This one I like best. Only downside is that it will break on frames with 65535 registers. 14:24
timo as long as we don't store dates in the number-of-registers slot, we should be safe 15:47
lizmat that sounds like an Excel lent idea :-) 16:05
timo let's Exchange more brilliant ideas like this 16:10
Geth MoarVM/new_unsigned_ops: 17 commits pushed by (Stefan Seifert)++
review: github.com/MoarVM/MoarVM/compare/a...d2c4d5ea35
16:40
MoarVM: niner++ created pull request #1631:
New unsigned ops
16:44
18:08 reportable6 left 20:05 Kaipi joined 20:08 [Coke]_ joined 20:12 RakuIRCLogger_ joined, Kaipi left 20:13 unicodable6 left 20:14 gfldex_ joined
Nicholas splitters! 20:15
20:15 benchable6 joined, releasable6 joined, quotable6 joined, shareable6 joined, Kaipi left, TempIRCLogger left, harrow left, RakuIRCLogger__ left, gfldex left, Voldenet left, nativecallable6 joined, gfldex_ is now known as gflsex 20:16 gflsex is now known as gfldex 20:17 [Coke] joined, RakuIRCLogger_ left
lizmat the logger reconnected by itself! whee! 20:18
but missed Nicholas' remark at 21:15 20:19
20:19 TempIRCLogger joined, squashable6 joined, unicodable6 joined, linkable6 joined, statisfiable6 joined
Nicholas I assume that this was because it looks like it pinged out within 60 seconds of my remark, and the ping timeout was reported as 240 seconds, so I'm inferring it was lagged by >240 seconds, so hadn't recieved it 20:20
lizmat yeah, I guess
20:21 AlexDaniel left
lizmat it wasn't able to reconnect for ~ 10 minutes from 20:10-20:20 UTC 20:21
20:23 psydroid left 20:25 Geth joined 20:28 [Coke]_ joined, [Coke] left 20:29 harrow joined 20:35 bartolin joined 21:07 notable6 joined 21:08 committable6 joined, bisectable6 joined, coverable6 joined 21:23 [Coke]_ is now known as [Coke] 21:29 AlexDaniel joined 21:48 AlexDaniel left 21:50 AlexDaniel joined 21:52 RakuIRCLogger left 22:01 psydroid joined 22:06 greppable6 joined, bloatable6 joined 22:09 reportable6 joined 22:15 AlexDaniel left 22:57 MasterDuke joined 23:06 evalable6 joined 23:07 tellable6 joined