1443084 - Crash in _alldiv

Status: RESOLVED WORKSFORME

(Core :: DOM: Events, defect, P2)

Description

[Calixte Denizet (:calixte)]

This bug was filed from the Socorro interface and is
report bp-4eff918b-8868-4a21-b50c-3272b0180305.
=============================================================

Top 8 frames of crashing thread:

0 xul.dll _alldiv f:/dd/vctools/crt/vcstartup/src/helper/i386/lldiv.asm:121
1 xul.dll mozilla::nsRFPService::RandomMidpoint toolkit/components/resistfingerprinting/nsRFPService.cpp:340
2 xul.dll mozilla::nsRFPService::ReduceTimePrecisionImpl toolkit/components/resistfingerprinting/nsRFPService.cpp:489
3 xul.dll mozilla::nsRFPService::ReduceTimePrecisionAsUSecsWrapper toolkit/components/resistfingerprinting/nsRFPService.cpp:521
4 xul.dll NowAsMillis js/src/jsdate.cpp:1312
5 xul.dll DateNoArguments js/src/jsdate.cpp:3123
6 xul.dll js::DateConstructor js/src/jsdate.cpp:3263
7 xul.dll EnterJit js/src/jit/Jit.cpp:99

=============================================================

There is 1 crash in nightly 60 with buildid 20180304220118. In analyzing the backtrace, the regression may have been introduced by patch [1] to fix bug 1425462.

[1] https://hg.mozilla.org/mozilla-central/rev?node=3052a1cf3b1c

Comment 1

[Tom Ritter [:tjr]]

Thanks! Yea that's definitely me. I'm not sure how it's happening.

line 340 :

> long long extraClampedTime = (aClampedTimeUSec / reducedResolution) * reducedResolution;

Socorro says 'EXCEPTION_INT_DIVIDE_BY_ZERO'


reducedResolution comes from:

> long long reducedResolution = aResolutionUSec * kClampTimesPerDigest;

Since kClampTimesPerDigest is a constant, aResolutionUSec must be zero or impossibly large to overflow a long long.


RandomMidpoint is only called from ReduceTimePrecisionImpl. At the top of that function we have the following which would avoid the call to RandomMidpoint.

> if (!IsTimerPrecisionReductionEnabled(aType) || aResolutionUSec <= 0)
>    return aTime;

We do the following cast from double -> long long later:

> long long resolutionAsInt = aResolutionUSec;

Maybe a decimal to avoid the top check, and get truncated later? But you can't set the preference to a decimal, prefs doesn't allow it. (Unless you manually hack at the JS file??)



I'm not sure how to proceed on this, so I guess I'll wait to see if there are any more crashes...?

Comment 2

[Marcia Knous [:marcia]]

The one crash on 60 has this url: https://donate.libreoffice.org/home/dl/win-x86/6.0.2/ja/LibreOffice_6.0.2_Win_x86.msi

Comment 3

[BugBot [:suhaib / :marco]]

Since the crash volume is low (less than 5 per week), the severity is downgraded to S3. Feel free to change it back if you think the bug is still critical.

For more information, please visit auto_nag documentation.

Comment 4

[BugBot [:suhaib / :marco]]

Closing because no crashes reported for 12 weeks.