Bug 113100

Summary:	Crash in: libc-2.19.so
Product:	LibreOffice	Reporter:	jeffhoo <jeffhoo.bugzilla>
Component:	Impress	Assignee:	Not Assigned <libreoffice-bugs>
Status:	RESOLVED WORKSFORME
Severity:	normal	CC:	ilmari.lauhakangas, jeffhoo.bugzilla, raal, serval2412, xiscofauli
Priority:	medium	Keywords:	bibisectRequest, haveBacktrace, regression
Version:	5.2.0.0.alpha0+
Hardware:	All
OS:	All
Whiteboard:
Crash report or crash signature:	["libc-2.19.so"]	Regression By:
Attachments:	Impress ODP with some 3D rectangles and applied bitmaps gdb backtrace

Description jeffhoo 2017-10-13 16:50:48 UTC

Created attachment 136957 [details]
Impress ODP with some 3D rectangles and applied bitmaps

This bug was filed from the crash reporting server and is br-6557ec4f-4c83-463d-a814-c199e3c7165e.
=========================================

When editing a a 3D object which has a bitmap (PNG) applied for texture, some bitmaps cause the entire suite to crash when rotated more than a few degrees about any axis.

Attached is a sample file which has increasingly complex bitmaps applied to demo the problem.   The first with a default bitmap appears to work all day no prob.   The second has a smaller bitmap tiled, and it can spin a couple of time before crashing.   The third is a larger PNG and it can only spin a few degrees about any axis before crashing.
 
I have not found a solid combination of bitmap textures (sizes, formats, tiling/scaling) that induce this.   But it appears to be something with either the size or format of the CUSTOM applied bitmap texture?   Or memory size/settings of graphics system?

I have tried Enabling and disabling GL in the View menu.   Same result.

I have not tried Windows variant or other h/w.

I see that it may be poss to upgrade GL driver, but too risky for this particular machine for this particular prblem.

System:  Linux Mint 17.3

CPU~Dual core Intel Core i7-4600U (-HT-MCP-) speed/max~3088/3300 MHz Kernel~3.19.0-32-generic x86_64 Up~21:09 Mem~2212.8/7890.2MB HDD~320.1GB(49.6% used) Procs~218 Client~Shell inxi~2.2.28  
$ inxi -Gx
Graphics:  Card: Intel Haswell-ULT Integrated Graphics Controller
           bus-ID: 00:02.0
           Display Server: X.Org 1.15.1 driver: N/A
           Resolution: 1366x768@60.0hz
           GLX Renderer: Gallium 0.4 on llvmpipe (LLVM 3.4, 256 bits)
           GLX Version: 2.1 Mesa 10.1.3 Direct Rendering: Yes

Comment 1 jeffhoo 2017-10-13 17:25:17 UTC

Update:  Works ok on a different Linux Mint 17.3 install:

Libreoffice:
Version: 5.0.3.2
Build ID: 1:5.0.3~rc2-0ubuntu1~trusty2
Locale: en-US (en_US.UTF-8)

$ inxi
CPU~Dual core Intel Core i3-4012Y (-HT-MCP-) speed/max~1499/1500 MHz Kernel~3.19.0-32-generic i686 Up~29 days Mem~595.5/3939.6MB HDD~120.0GB(43.0% used) Procs~189 Client~Shell inxi~2.2.28  
$ inxi -Gx
Graphics:  Card: Intel Device 0a1e bus-ID: 00:02.0
           Display Server: X.Org 1.17.1 drivers: intel (unloaded: fbdev,vesa)
           Resolution: 1366x768@60.0hz
           GLX Renderer: Mesa DRI Intel Haswell x86/MMX/SSE2
           GLX Version: 3.0 Mesa 10.5.9 Direct Rendering: Yes

Comment 2 Xisco Faulí 2017-10-13 21:43:39 UTC

Confirmed in

Version: 6.0.0.0.alpha0+
Build ID: 3672cdd35985201ea87463cf032fedd02c052f4d
CPU threads: 4; OS: Linux 4.10; UI render: default; VCL: gtk3; 
Locale: ca-ES (ca_ES.UTF-8); Calc: group

and

Version: 5.2.0.0.alpha1+
Build ID: 5b168b3fa568e48e795234dc5fa454bf24c9805e
CPU Threads: 4; OS Version: Linux 4.8; UI Render: default; 
Locale: ca-ES (ca_ES.UTF-8)

but not in

Version: 5.0.0.0.alpha1+
Build ID: 0db96caf0fcce09b87621c11b584a6d81cc7df86
Locale: ca-ES (ca_ES.UTF-8)

Comment 3 Julien Nabet 2017-10-14 08:37:51 UTC

On pc Debian x86-64 with master sources updated today, I only succeeded to reproduce the crash once at random.
I tested again with gdb to retrieve a bt but couldn't reproduce it again.

It'd be helpful to have a bt with debug symbols on recent master.

Comment 4 Xisco Faulí 2017-10-15 21:13:35 UTC

Created attachment 137004 [details]
gdb backtrace

Comment 5 Buovjaga 2018-05-25 17:27:14 UTC

Weird: on Win I can repro with 5.3.0 alpha1, but not with master commit of bibisect repo 5.2 or 5.3!! So no way to bibisect ... I wonder what is different in the builds.

Comment 6 raal 2018-06-11 20:53:11 UTC

Cannot reproduce with Version: 6.2.0.0.alpha0+
Build ID: c4c56de1b0e62ec866b519b2b24c5e805f0a86d3
CPU threads: 4; OS: Linux 4.4; UI render: default; VCL: gtk3; 

Please could you retest with actual master? 

http://dev-builds.libreoffice.org/daily/master/

Comment 7 Buovjaga 2018-06-13 16:42:01 UTC

Huh, yeah, I wonder if I failed to test with a recent version. I can't even repro with 6.0.4

Version: 6.0.4.2 (x64)
Build ID: 9b0d9b32d5dcda91d2f1a96dc04c645c450872bf
CPU threads: 4; OS: Windows 10.0; UI render: default; 
Locale: fi-FI (fi_FI); Calc: group

Comment 8 Xisco Faulí 2018-06-13 16:48:56 UTC

I can't reproduce it in now in

Version: 5.2.0.0.alpha0+
Build ID: 3ca42d8d51174010d5e8a32b96e9b4c0b3730a53
Threads 4; Ver: 4.13; Render: default; 

I'm wondering if it was an external bug fixed now...