833723 - test_bug455311.js fails on Linux locally

Status: RESOLVED FIXED

(Core :: Networking, defect, P5)

Description

[Georg Koppen]

User Agent: Mozilla/5.0 (Windows NT 6.1; rv:17.0) Gecko/20100101 Firefox/17.0

Steps to reproduce:

I tried to run the xpcshell testsuite both on a desktop (Ubuntu 12.04) and on a server (Ubuntu 10.04).


Actual results:

test_bug455311.js failed with the following output:

TEST-UNEXPECTED-FAIL | /home/groeg/JonDoBrowser/stable/testing/mozilla-esr17/obj-i686-pc-linux-gnu/_tests/xpcshell/netwerk/test/unit/test_bug455311.js | test failed (with xpcshell return code: 0), see following log:
>>>>>>>

TEST-INFO | (xpcshell/head.js) | test 1 pending

TEST-INFO | (xpcshell/head.js) | test 2 pending

TEST-UNEXPECTED-FAIL | /home/groeg/JonDoBrowser/stable/testing/mozilla-esr17/obj-i686-pc-linux-gnu/_tests/xpcshell/netwerk/test/unit/test_bug455311.js | [xpconnect wrapped nsIURI] == [xpconnect wrapped nsIURI] - See following stack:
JS frame :: /home/groeg/JonDoBrowser/stable/testing/mozilla-esr17/testing/xpcshell/head.js :: do_throw :: line 451
JS frame :: /home/groeg/JonDoBrowser/stable/testing/mozilla-esr17/testing/xpcshell/head.js :: _do_check_eq :: line 545
JS frame :: /home/groeg/JonDoBrowser/stable/testing/mozilla-esr17/testing/xpcshell/head.js :: do_check_eq :: line 566
JS frame :: /home/groeg/JonDoBrowser/stable/testing/mozilla-esr17/obj-i686-pc-linux-gnu/_tests/xpcshell/netwerk/test/unit/test_bug455311.js :: run_test :: line 122
JS frame :: /home/groeg/JonDoBrowser/stable/testing/mozilla-esr17/testing/xpcshell/head.js :: _execute_test :: line 315
JS frame :: -e :: <TOP_LEVEL> :: line 1

TEST-INFO | (xpcshell/head.js) | exiting test
<<<<<<<


Expected results:

The test shouldn't have failed.

Comment 1

[Georg Koppen]

linkURI.spec is: file:///home/groeg/JonDoBrowser/stable/testing/mozilla-esr17/obj-i686-pc-linux-gnu/_tests/xpcshell/netwerk/test/unit/test_link.desktop

while channel.URI.spec is: /home/groeg/JonDoBrowser/stable/testing/mozilla-esr17/netwerk/test/unit/test_link.desktop

Looking a bit deeper https://mxr.mozilla.org/mozilla-esr17/source/netwerk/protocol/file/nsFileChannel.cpp#243ff. causes the former to get resolved to the latter. So the question is, where does the test go wrong?

1) Should the former should not get resolved to the latter?
2) Does getLinkFile() (https://mxr.mozilla.org/mozilla-esr17/source/netwerk/test/unit/test_bug455311.js#116) give the wrong thing back and the error would thus be somewhere in  do_get_file()?
3) ...?

Comment 2

[Georg Koppen]

https://bugzilla.mozilla.org/show_bug.cgi?id=482085 could be relevant here. But still not sure what the expected behavior/where the bug is.

Comment 3

[Georg Koppen]

I've finally bisected my way down to bug 670514 which causes the test failure. Given the nature of that bug I guess the fix to this bug will adapt the unit test although that might not safe. I am willing to come up with a patch if someone points me in the right direction (see my previous comments). Boris?

Comment 4

[Boris Zbarsky [:bzbarsky]]

Fixing the test makes sense, but why is this not failing all over?  I thought we symlinked this stuff in general on Linux/Mac...

Comment 5

[Georg Koppen]

(In reply to Boris Zbarsky (:bz) from comment #4)
> Fixing the test makes sense, but why is this not failing all over?  I
> thought we symlinked this stuff in general on Linux/Mac...

How do you think that test should get fixed:

1) We should delete the offending lines in it and make the test just pass (that makes me actually a bit nervous as those lines/tests are probably there for some reason...).
2) We should try to avoid the symlink issue leaving the tests itself untouched.
3) ...

Maybe we should investigate why this test is not failing all over first which could lead to solution 2) mentioned above. Not sure whom to talk to in this case, though.

Comment 6

[Boris Zbarsky [:bzbarsky]]

> Maybe we should investigate why this test is not failing all over first

Yes.

So I just checked, and it fails locally for me too on Linux.  It passes on tinderbox, because tinderbox runs on a packaged build, not directly out of an objdir, so doesn't have a symlink.

So the right fix for the test is probably to fix this line:

  do_check_eq(chan.URI, linkURI);

to use a URI on the RHS that has resolved the file to its target in cases when it's a symlink...

Comment 7

[Firefox Bug Husbandry Bot]

Bulk change to priority: https://bugzilla.mozilla.org/show_bug.cgi?id=1399258

Comment 8

[(no longer active)]

I'm tired of ignoring this test failure locally all the time, so I'd like to fix it.

Comment 9

[(no longer active)]

Attachment: Bug 833723 - Make test_bug455311.js pass locally on Linux by resolving symlinks;

Comment 10

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/89552e3aa8be
Make test_bug455311.js pass locally on Linux by resolving symlinks; r=bzbarsky

Comment 11

[Noemi Erli[:noemi_erli]]

Backed out changeset 89552e3aa8be (Bug 833723) for xpcshell failures in test_bug455311.js CLOSED TREE

Push with failure: https://treeherder.mozilla.org/#/jobs?repo=autoland&selectedJob=268021940&resultStatus=testfailed%2Cbusted%2Cexception&revision=89552e3aa8beb2f5ed6e6a559ccdab501ac0f686

Failure log: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=268021940&repo=autoland&lineNumber=1532

Backout: https://hg.mozilla.org/integration/autoland/rev/7bbe1131aaf545c833547d2440d40d0e21f040a4

Comment 12

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/41422a29d1c5
Make test_bug455311.js pass locally on Linux by resolving symlinks; r=bzbarsky

Comment 13

[(no longer active)]

Attachment: Bug 833723 - Part 2: Fix typo

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/a0a4bb79d1f1
Part 2: Fix typo

Comment 15

[Cosmin Sabou [:CosminS]]

https://hg.mozilla.org/mozilla-central/rev/41422a29d1c5
https://hg.mozilla.org/mozilla-central/rev/a0a4bb79d1f1