Bug 83072 - VIEWING: segfault on particular slide in a .pptx.
Summary: VIEWING: segfault on particular slide in a .pptx.
Status: RESOLVED WORKSFORME
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: Impress (show other bugs)
Version:
(earliest affected)
4.4.0.0.alpha0+ Master
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Not Assigned
URL:
Whiteboard:
Keywords: bibisected, haveBacktrace, regression
Depends on:
Blocks:
 
Reported: 2014-08-25 19:22 UTC by Terrence Enger
Modified: 2015-12-15 11:03 UTC (History)
1 user (show)

See Also:
Crash report or crash signature:


Attachments
gdb on core file, daily dbgutil commit 0202d213 (4.70 KB, text/plain)
2014-08-25 19:26 UTC, Terrence Enger
Details
gdb on core file, master f74a633 (11.24 KB, text/plain)
2014-08-25 19:30 UTC, Terrence Enger
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Terrence Enger 2014-08-25 19:22:54 UTC
These observations are on a version of LibreOffice from the daily
dbgutil bibisect, commit 0202d213... 2014-07-16.


STR

(1) Download file ICCP14_upload.pptx attached to bug 81078.

(2) Remove the lock file .~lock.ICCP14_upload.pptx# if it is left from
    previous test.

(3) Run LibreOffice naming ICCP14_upload.pptx on the command line.
    Program presents window "ICCP14_upload.pptx..."; the middle pane
    is open to tab Normal, and it shows Slide 1 "A Switchable Light
    Field Camera".

(4) In a moderately quick tempo, type <Page Down> 15 time.  The middle
    pane shows Slide 16 "Angular Response" with placeholders for two
    images, each labelled "Picture <unreadable smudge>".

(5) Wait ten or twenty seconds.  Program terminates with a
    segmentation fault.  Backtrace show PC = 0x0, called from
    GraphicObject::ImplAutoSwapOutHdl.


From git bisect:

    commit 0202d213f4018a42c883fa125456b39400ae1161
    Author: Miklos Vajna <vmiklos@collabora.co.uk>
    Date:   Wed Jul 16 09:12:14 2014 +0200

        2014-07-16

    :100644 100644 fbf5aa0fad629816226efe2226168f6fed291dd1 2b56e43be25f0f8f98c8c58fd64d16b9b08903d6 M	build-info.txt
    :040000 040000 ca9906d2895eecb7f8f5053f4e0c73e7523d9637 c8f238255df448b3159d4c732393814faedb399a M	opt

and git bisect log:

    # bad: [59a8d89d4255a1cbcab4fcfeff9270dd56c62aed] 2014-08-12
    # good: [b3130c846de5cf1b4be48b48dfc780bb369549fa] 2014-05-21
    git bisect start '59a8d89' 'oldest'
    # good: [51d0b5cdc80ce7b0a3e0694e91547d3e7942f19f] 2014-07-01
    git bisect good 51d0b5cdc80ce7b0a3e0694e91547d3e7942f19f
    # bad: [1d3584c0317743305d4720d30e66335f12f39d9d] 2014-07-21
    git bisect bad 1d3584c0317743305d4720d30e66335f12f39d9d
    # good: [eb37477c7906b4c4b7eb11433c7566f669a040d1] 2014-07-11
    git bisect good eb37477c7906b4c4b7eb11433c7566f669a040d1
    # bad: [0202d213f4018a42c883fa125456b39400ae1161] 2014-07-16
    git bisect bad 0202d213f4018a42c883fa125456b39400ae1161
    # good: [a5750a3c0c5d3b975a787f844d7ba60db53a765e] 2014-07-13
    git bisect good a5750a3c0c5d3b975a787f844d7ba60db53a765e
    # good: [d435a3233c58183c13ccf8efee43e48039a235cf] 2014-07-15
    git bisect good d435a3233c58183c13ccf8efee43e48039a235cf
    # first bad commit: [0202d213f4018a42c883fa125456b39400ae1161] 2014-07-16

IIUC, this places the start of the problem between source commits
3a2010c and 6bd087b.
Comment 1 Terrence Enger 2014-08-25 19:26:18 UTC
Created attachment 105251 [details]
gdb on core file, daily dbgutil commit 0202d213
Comment 2 Terrence Enger 2014-08-25 19:30:55 UTC
Created attachment 105252 [details]
gdb on core file, master f74a633

My debug build of master also fails on approximately the same STR.
Note especially *approximately*: I first (by accident) saw the
backtrace attached here; subsequent attempts to reproduce the failure
produced segfaults at different places, sometimes with a backtrace
like this attachment and sometimes with a backtrace like the one from
the daily dbgutil version.  This was before I noticed that the timing
of the <Page Down> operations is significant.

Short version: this = this=0x9999999999999999 in Timer::Start,
vcl/source/app/timer.cxx:240.

Anyway, if this is deemed to be a different bug, I shall be challenged
to write STR to reproduce it consistently.
Comment 3 Yousuf Philips (jay) (retired) 2014-09-29 07:47:27 UTC
Hi Terrence,

I tested 4.4 and it didnt crash for me. Can you test if it still happens for you.

Version: 4.4.0.0.alpha0+
Build ID: df73f4115cfe4d07e4159adf087571687eb173ec
TinderBox: Linux-rpm_deb-x86@45-TDF, Branch:master, Time: 2014-09-25_23:36:54
Comment 4 Terrence Enger 2014-09-29 13:52:24 UTC
Thank you, Jay, for the reminder.  The crash is gone again.

Working in reverse with the daily dbgutil bibisect repository, I see
that crash disappears in the version from 2014-09-01
Comment 5 Robinson Tryon (qubit) 2015-12-15 11:03:10 UTC
Migrating Whiteboard tags to Keywords: (bibisected)
[NinjaEdit]