DiscussionsIssue archiveOCCT:Modeling Algorithms

Archived issue #0027194

Possible division by zero in IntPatch_WLineTool

Open CASCADEOCCT:Modeling Algorithmsclosed17 public notes

Search issues

Description

This issue is observed in scope of the #0026329 bug.

aStepCoeff = Min(aStepOnS1, aStepOnS2) / Max(aStepOnS1, aStepOnS2);

Such division may lead to the problems when aStepOnS1 and aStepOnS2 are equal to 0.0.

It is necessary to investigate how this situation may happen and elaborate solution. Problem can be reproduced on the CR26329 branch only.

Steps to reproduce

test bugs moddata_2 bug469

Public activity

17 archived notes

Participants are labeled by their role within this record.

01Commenter 3
Branch [archived branch] has been created by Author.

[revision removed]


Detailed log of new commits:

Author: Author
Date: Sat Feb 20 11:54:44 2016 +0300

    0027194: Possible division by zero in IntPatch_WLineTool
    
    Correct handling for division by zero is added. This prevents exception when FPE is enabled
02Commenter 2
Dear msv,
Please check current state of the branch CR27194.
03Commenter 3
Reviewed.
04Commenter 3
Branch [archived branch] has been updated forcibly by Participant.

[revision removed]
05Commenter 5
Dear Commenter 1,
Branch CR27194 was rebased on current master of occt git-repository.
[revision removed]
06Commenter 6
Dear Commenter 1,
Branch CR27194 from occt git-repository (and master from products git-repository) was compiled on Linux, MacOS and Windows platforms and tested on Release mode.
[revision removed]

Number of compiler warnings:

occt component :
Linux: 0 (0 on master)
Windows: 0 (0 on master)
MacOS : 0 (0 on master)

products component :
Linux: 41 (41 on master)
Windows: 0 (0 on master)

Regressions/Differences/Improvements:
No regressions/differences

Testing cases:
Not needed

Testing on Linux:
occt component :
Total MEMORY difference: 90157515 / 90534753 [-0.42%]
Total CPU difference: 19392.780000000017 / 19576.420000000107 [-0.94%]
products component :
Total MEMORY difference: 25923801 / 25872807 [+0.20%]
Total CPU difference: 5361.569999999991 / 5316.139999999985 [+0.85%]

Testing on Windows:
occt component :
Total MEMORY difference: 57375213 / 57106337 [+0.47%]
Total CPU difference: 18744.31575509885 / 17710.10712559875 [+5.84%]
products component :
Total MEMORY difference: 17493005 / 17395600 [+0.56%]
Total CPU difference: 5411.159886699964 / 5095.257861699954 [+6.20%]

There are no differences in images found by testdiff.
07Commenter 7
Dear Commenter 1,
Branch CR27194 is TESTED.
08Commenter 8
Remarks on proposed fix:

1. Please avoid comparing real numbers for equality using operator == or != -- this will not work in general case. For instance, AFAIK, -0 != +0

2. In order to protect against division by zero, it would be sufficient to check the value of Max(aStepOnS1, aStepOnS2), so why checking both aStepOnS1 and aStepOnS2 individually?

3. What happens if aStepOnS1=1e-308 and aStepOnS2 = 10? Should not the behavior be the same as when aStepOnS1=0. and aStepOnS2=1.? (currently it will be different).
09Commenter 9
1. Please avoid comparing real numbers for equality using operator == or != -- this will not work in general case. For instance, AFAIK, -0 != +0

if (+0.0 == -0.0)
 cout << "EQUAL";

Returns true in MSVC2010. Here only +0.0 is used due to values nature.

2. In order to protect against division by zero, it would be sufficient to check the value of Max(aStepOnS1, aStepOnS2), so why checking both aStepOnS1 and aStepOnS2 individually?

Max(aStepOnS1, aStepOnS2) is not computed directly.

3. What happens if aStepOnS1=1e-308 and aStepOnS2 = 10? Should not the behavior be the same as when aStepOnS1=0. and aStepOnS2=1.? (currently it will be different).

Denormal numbers will be used.
10Commenter 10
1. I agree. It is better to compare with gp::Resolution(), that is RealSmall().

2. Checking of Max only is not enough, we must check also Min.

3. If we check both values this will not happen.
11Commenter 3
Branch [archived branch] has been created by Author.

[revision removed]


Detailed log of new commits:

Author: Author
Date: Sat Feb 20 11:54:44 2016 +0300

    0027194: Possible division by zero in IntPatch_WLineTool
    
    Correct handling for division by zero is added. This prevents exception when FPE is enabled
12Commenter 12
Dear abv,
Please check current state of the CR27194_2 branch.
13Commenter 13
No remarks, please test
14Commenter 14
Dear Commenter 1,
Branch CR27194_2 from occt git-repository (and master from products git-repository) was compiled on Linux, MacOS and Windows platforms and tested on Release mode.
[revision removed]

Number of compiler warnings:

occt component :
Linux: 0 (0 on master)
Windows: 0 (0 on master)
MacOS : 0 (0 on master)

products component :
Linux: 72 (72 on master)
Windows: 4 (4 on master)
MacOS : 1144

Regressions/Differences/Improvements:
No regressions/differences

Testing cases:
Not needed

Testing on Linux:
occt component :
Total MEMORY difference: 88283719 / 87385561 [+1.03%]
Total CPU difference: 19284.279999999915 / 19257.83 [+0.14%]
products component :
Total MEMORY difference: 27790493 / 27718261 [+0.26%]
Total CPU difference: 5231.579999999983 / 5200.899999999987 [+0.59%]

Testing on Windows:
occt component :
Total MEMORY difference: 55779341 / 55791306 [-0.02%]
Total CPU difference: 18062.716185898837 / 17814.222192998932 [+1.39%]
products component :
Total MEMORY difference: 18925839 / 18889957 [+0.19%]
Total CPU difference: 5175.707577399963 / 4958.179782999945 [+4.39%]

There are no differences in images found by testdiff.
15Commenter 15
Dear Commenter 1,
Branch CR27194_2 is TESTED.
16Commenter 3
Branch [archived branch] has been deleted by Participant.

[revision removed]
17Commenter 3
Branch [archived branch] has been deleted by Participant.

[revision removed]

Related records