Normalizing precession problem

[Tal500]

If I omit this line in LinearMath/btQuaternion.h:

/**@brief Inversely scale this quaternion
 * @param s The scale factor */
btQuaternion& operator/=(const btScalar& s) 
{
    btAssert(s != btScalar(0.0));// <<<===Omitting this line cause the bug
    return *this *= btScalar(1.0) / s;
}

And compiling Bullet3 on debug64 mode(gmake, gcc 4.9.2, linux64), the "Constraints" demo in ExampleBrowser crush, here is gdb:

$ gdb ./App_ExampleBrowser_gmake_x64_debug 
GNU gdb (Debian 7.7.1+dfsg-5) 7.7.1
Copyright (C) 2014 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from ./App_ExampleBrowser_gmake_x64_debug...done.
(gdb) run
Starting program: /home/tal/src/bullet3/bin/App_ExampleBrowser_gmake_x64_debug 
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
[New Thread 0x7fffef83b700 (LWP 13067)]
Creating context
Created GL 3.0 context
Direct GLX rendering context obtained
Making context current
GL_VENDOR=X.Org
GL_RENDERER=Gallium 0.4 on AMD JUNIPER
GL_VERSION=3.3 (Core Profile) Mesa 10.3.2
GL_SHADING_LANGUAGE_VERSION=3.30
pthread_getconcurrency()=0
b3Printf: Selected demo: Constraints
ven = X.Org
App_ExampleBrowser_gmake_x64_debug: ../../src/LinearMath/btVector3.h:300: btVector3& btVector3::normalize(): Assertion `!fuzzyZero()' failed.

Program received signal SIGABRT, Aborted.
0x00007ffff65bc107 in __GI_raise (sig=sig@entry=6)
    at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
56  ../nptl/sysdeps/unix/sysv/linux/raise.c: No such file or directory.
(gdb) bt
#0  0x00007ffff65bc107 in __GI_raise (sig=sig@entry=6)
    at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
#1  0x00007ffff65bd4e8 in __GI_abort () at abort.c:89
#2  0x00007ffff65b5226 in __assert_fail_base (
    fmt=0x7ffff66ebce8 "%s%s%s:%u: %s%sAssertion `%s' failed.\n%n", 
    assertion=assertion@entry=0x723480 "!fuzzyZero()", 
    file=file@entry=0x723368 "../../src/LinearMath/btVector3.h", 
    line=line@entry=300, 
    function=function@entry=0x7cf000 <_ZZN9btVector39normalizeEvE19__PRETTY_FUNCTION__> "btVector3& btVector3::normalize()") at assert.c:92
#3  0x00007ffff65b52d2 in __GI___assert_fail (
    assertion=0x723480 "!fuzzyZero()", 
    file=0x723368 "../../src/LinearMath/btVector3.h", line=300, 
    function=0x7cf000 <_ZZN9btVector39normalizeEvE19__PRETTY_FUNCTION__> "btVector3& btVector3::normalize()") at assert.c:101
#4  0x000000000040b1c0 in btVector3::normalize (this=0x7fffffffda10)
    at ../../src/LinearMath/btVector3.h:300
#5  0x000000000057e946 in normalize (this=0x7fffffffda10)
    at ../../src/BulletDynamics/ConstraintSolver/btConeTwistConstraint.cpp:891
#6  btConeTwistConstraint::computeConeLimitInfo (this=0x1260a30, qCone=..., 
    swingAngle=@0x7fffffffd9e8: 0.000690533954, vSwingAxis=..., 
    swingLimit=@0x7fffffffd9ec: 0)
    at ../../src/BulletDynamics/ConstraintSolver/btConeTwistConstraint.cpp:838
---Type <return> to continue, or q <return> to quit---
#7  0x0000000000581210 in btConeTwistConstraint::calcAngleInfo2 (
    this=this@entry=0x1260a30, transA=..., transB=..., invInertiaWorldA=..., 
    invInertiaWorldB=...)
    at ../../src/BulletDynamics/ConstraintSolver/btConeTwistConstraint.cpp:675
#8  0x0000000000581d3c in btConeTwistConstraint::getInfo1 (this=0x1260a30, 
    info=0x12ed510)
    at ../../src/BulletDynamics/ConstraintSolver/btConeTwistConstraint.cpp:90
#9  0x000000000058e07f in btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup (this=0x1259470, bodies=<optimized out>, 
    numBodies=<optimized out>, manifoldPtr=0x12ed730, numManifolds=21, 
    constraints=0x12ed570, numConstraints=15, infoGlobal=..., 
    debugDrawer=0x1254d70)
    at ../../src/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.cpp:1332
#10 0x00000000005886b5 in btSequentialImpulseConstraintSolver::solveGroup (
    this=0x1259470, bodies=0x12ed610, numBodies=18, manifoldPtr=0x12ed730, 
    numManifolds=21, constraints=0x12ed570, numConstraints=15, infoGlobal=..., 
    debugDrawer=0x1254d70)
    at ../../src/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.cpp:1917
#11 0x0000000000577c18 in processConstraints (this=0x1259860)
    at ../../src/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.cpp:192
#12 btDiscreteDynamicsWorld::solveConstraints (this=0x12595e0, solverInfo=...)
---Type <return> to continue, or q <return> to quit---
    at ../../src/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.cpp:736
#13 0x00000000005768d2 in btDiscreteDynamicsWorld::internalSingleStepSimulation
    (this=0x12595e0, timeStep=0.0166666675)
    at ../../src/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.cpp:506
#14 0x000000000056f9c2 in btDiscreteDynamicsWorld::stepSimulation (
    this=0x12595e0, timeStep=0.0377589986, maxSubSteps=<optimized out>, 
    fixedTimeStep=0.0166666675)
    at ../../src/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.cpp:456
#15 0x0000000000417ac5 in OpenGLExampleBrowser::update (this=<optimized out>, 
    deltaTime=0.0377589986)
    at ../../examples/ExampleBrowser/OpenGLExampleBrowser.cpp:871
#16 0x000000000040ea80 in main (argc=<optimized out>, argv=0x7fffffffe348)
    at ../../examples/ExampleBrowser/main.cpp:34
(gdb) 

After analyzing the problem, it is caused when normalizing quaternion, which call the "/=" operator that I changed above. The "/=" op has a precession error in this situation, which yields to epsilon quaternion, which is shown in the back trace.

I've seen this issue while on my Bullet3 C API fork (while merging original branch).

It is weird that btAssert() is saving the precession.
This might even be a compiler bug, but I'm not sure.
It might be the cause for #180

Please try my change test and tell me if you have crush in debug64 mode.