402864 - Number failures in testsuite when running on linux

Status: VERIFIED WORKSFORME

(Tamarin Graveyard :: Virtual Machine, defect, P4)

Description

[Brent Baker]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.9) Gecko/20071025 Firefox/2.0.0.9
Build Identifier: 

The following testcases are currently failing on linux:

FAILURES:
  ecma3/GlobalObject/e15_1_2_2_2.abc
  ecma3/GlobalObject/e15_1_2_6.abc
  ecma3/GlobalObject/e15_1_2_7.abc
  ecma3/TypeConversion/e9_3_1_3_rt.abc

Reproducible: Always

Steps to Reproduce:
1. cd tests
2. python runtests.py ecma3/GlobalObject/e15_1_2_2_2.as ecma3/GlobalObject/e15_1_2_6.as ecma3/GlobalObject/e15_1_2_7.as ecma3/TypeConversion/e9_3_1_3_rt.as
Actual Results:  
FAILURES:
  ecma3/GlobalObject/e15_1_2_2_2.abc : parseInt("123456789012345678") = 123456789012345680 FAILED! expected: 123456789012345700
  ecma3/GlobalObject/e15_1_2_6.abc : isNaN() = false FAILED! expected: true
  ecma3/GlobalObject/e15_1_2_7.abc : isFinite() = true FAILED! expected: false
  ecma3/TypeConversion/e9_3_1_3_rt.abc : parseInt("123456789012345678") = 123456789012345680 FAILED! expected: 123456789012345700

Comment 1

[Brent Baker]

Attachment: Mark testcases as expectedfail on linux until fixed

Comment 2

[arno renevier]

Attachment: output of python runtests.py ecma3/GlobalObject/e15_1_2_2_2.as ecma3/GlobalObject/e15_1_2_6.as ecma3/GlobalObject/e15_1_2_7.as ecma3/TypeConversion/e9_3_1_3_rt.as

those tests seem to work now.

Comment 3

[Leon Sha]

Attachment: patch

e15_1_2_6.abc and e15_1_2_7.abc are passed now. So remove these two case from expected failure list.

Comment 4

[Leon Sha]

Comment on attachment 333510 [details] [diff] [review]
patch

This patch already checked in by [e-mail].

Comment 5

[Brent Baker]

Update expected failures to only fail in release and release_debugger builds. Debug and debug_debugger produce the proper results.

ecma3/GlobalObject/e15_1_2_2_2:parseInt\("123456789012345678"\)
ecma3/TypeConversion/e9_3_1_3_rt:parseInt\("123456789012345678"\)

Pushed change 648:89e096550269

Comment 6

[Brent Baker]

Changeset 1241:9f9a9a0287a6 (Modify MathUtils bug #466272) seems to have fixed this issue. I have tested with gcc 4.1 and 4.2 and they are both passing. Removing the expected failures from the testconfig.txt

Pushed updated testconfig.txt 1245:a2dd51583214

Comment 7

[Tommy Reilly]

Happening again, smells like heisenbugs

Comment 8

[Dan Schaffer]

the release builds on linux are now failing for tests: 
ecma3/GlobalObject/e15_1_2_2_2:parseInt\("123456789012345678"\)
ecma3/TypeConversion/e9_3_1_3_rt:parseInt\("123456789012345678"\)

set tests to expectedfail for linux in acceptance testconfig.txt

Comment 9

[Brent Baker]

heisenbugs strikes again by disappearing. Looks like the change for bug# 546801 has removed this failure from linux release builds. Removing the expectedfail from the testconfig.

patch pushed to remove expectedfail:

tr-argo: 3757:fb2a6464082e
tr: 3923:fb2a6464082e

Marking as WORKSFORME

Comment 10

[Trevor Baker]

bulk verifying resolved !fixed issues