This repository has been archived by the owner on Feb 28, 2024. It is now read-only.
[BUG-3095] Viewer freezes in infinate loop and never recovers when editing certain rigged meshes at altitude - Octree insertion failed, starting over from root! #11859
Labels
Steps To Reproduce
Grab this demo mesh item https://marketplace.secondlife.com/p/ZED-Mens-Multi-Colour-Demo-of-Rigged-MESH-Pad-Locked-Shoulder-Harness-in-Hex-Leather-Latex-DEMO/4691763
Wear one of these mesh harnesses.
Teleport somewhere at altitude eg) http://maps.secondlife.com/secondlife/Australia/169/80/3729
Right click -> Edit on any rezzed object then left click on the rigged mesh harness to edit it.
EDIT TO ADD: Another repro item for this crash: https://marketplace.secondlife.com/p/rMPosture-Collar-No04-Rigged-Mesh/4541375
Observed Behaviour
World view freezes and then viewer goes not responsing with a black screen.
Viewer will never recover.
Viewer logs are spammed with this continuously:
|2013-07-05T12:37:59Z WARNING: LLOctreeNode::insert: Octree insertion failed, starting over from root!
This does not seem to reproduce at ground level. It only reproduces at altitude.
It does not appear to matter whether other objects are nearby or not at ground level or altitide.
So far I cannot reproduce this crash on empty or busy regions at ground level and I can 100% reproduce this crash at empty or busy locations at altitude.
It can appear that editing on non-mesh items attached near the repro objects also reproduce the same crash but if you enable Octree under Develop -> Render Metadata, clicking anywhere within the Octree of the repro object will crash you.
Logs attached.
Viewer does produce a dmp file & send in a crash report.
The stack varies though.
Java<br> <br>Thread 0 (crashed) <br> 0 msvcr100.dll!_free_osfhnd + 0x7a <br> eip = 0x63c9a747 esp = 0x00043000 ebp = 0x00043018 ebx = 0x63d33740 <br> esi = 0x00000003 edi = 0x000000c0 eax = 0x00000000 ecx = 0x00000000 <br> edx = 0x00ce5400 efl = 0x00210202 <br> Found by: given as instruction pointer in context<br>
Java<br>Thread 0 (crashed) <br> 0 kernel32.dll + 0x3fc9e <br> eip = 0x7505fc9e esp = 0x000a2fd8 ebp = 0x000a3000 ebx = 0x63d33740 <br> esi = 0x00000003 edi = 0xffffffff eax = 0x00000000 ecx = 0x000a3024 <br> edx = 0x00ce5400 efl = 0x00210293 <br> Found by: given as instruction pointer in context<br>
Java<br> <br>Operating system: Windows NT <br> 6.1.7601 Service Pack 1 <br>CPU: x86 <br> GenuineIntel family 6 model 58 stepping 9 <br> 4 CPUs <br> <br>Crash reason: EXCEPTION_STACK_OVERFLOW <br>Crash address: 0x6e604f47 <br> <br>Thread 0 (crashed) <br> 0 msvcr100.dll!zzz_AsmCodeRange_Begin [chkstk.asm : 99 + 0x0] <br> eip = 0x6e604f47 esp = 0x00204a88 ebp = 0x00204a90 ebx = 0x00000003 <br> esi = 0x000000c0 edi = 0x6e6a3740 eax = 0x00202000 ecx = 0x00202fa8 <br> edx = 0x01176150 efl = 0x00210206 <br> Found by: given as instruction pointer in context <br>
#### Expected Behaviour
Not to freeze and crash.
#### Other Information
Reproduces on:
|
|-|
Second Life 3.6.1 (278007) Jun 27 2013 12:41:07 (Second Life Release)
Second Life 3.6.2 (278133) Jul 1 2013 20:56:48 (Second Life Beta Viewer)
When the crash reporter tries to send in its report about this crash, it just hangs & never seems to finish. I can login normally when the crash reporter is still sending a report but it never progresses beyond "Sending to server, try 1".
When reproducing other crashes, the crash report sends normally. Only this crash seems unable to send properly.
Attachments
Original Jira Fields
The text was updated successfully, but these errors were encountered: