136936 - TestInterfaceInfo is out of date - nsIServiceManager

Status: RESOLVED FIXED

(Core :: XPConnect, defect)

Description

[timeless]

QNX6.0a & FreeBSD4.4

I'm going to attach my logging info because I don't think I'll have time to 
figure this one out today and I don't want to flood the report with three 
stacks.

Comment 1

[timeless]

Attachment: log (assert, assert, crash)

Comment 2

[David Bradley]

This is an XPCOM issue. Basically the code is triggering recursion that begins
with XPTI_GetInterfaceInfoManager which gets called a second time by NS_InitXPCOM2.

I wonder if the test should not initialize XPCOM before making this call?

Comment 3

[Doug Turner (:dougt)]

the problem is that you are NOT init'ing xpcom before calling
XPTI_GetInterfaceInfoManager().

see
http://lxr.mozilla.org/seamonkey/source/xpcom/reflect/xptinfo/tests/TestInterfaceInfo.cpp#54

Comment 4

[David Bradley]

timeless, can you try adding the NS_InitXPCOM2 call and see if that takes care
of the problem?

Comment 5

[timeless]

that fixes assert #1. and leaves me with assert #2 and crash.

Comment 6

[David Bradley]

The test is broken, nsIServiceManager has changed but no one updated the test.
The test assumes 7 methods but there are only 6 now. Changing summary to reflect.

Comment 7

[Doug Turner (:dougt)]

david, can you post a fix?  If not, bounce it back to me and I will add it to my
plate.

Comment 8

[David Bradley]

Attachment: Fixes the test

This patch makes the test current. It's not ideal, but gets us back to a
working test.

Timeless can you verify this fixes the problem. It did on my system, but needs
a double check.

Comment 9

[timeless]

Comment on attachment 79269 [details] [diff] [review]
Fixes the test

This didn't work for me
I get an assert from
#7  0x08048da3 in main (argc=1, argv=0xbfbff808) at
/home/timeless/mozilla/xpcom/reflect/xptinfo/tests/TestInterfaceInfo.cpp:127
and i crash at
#1  0x08048db5 in main (argc=1, argv=0xbfbff808) at
/home/timeless/mozilla/xpcom/reflect/xptinfo/tests/TestInterfaceInfo.cpp:129
0x2816640c in nsID::ToString (this=0x1) at
/home/timeless/mozilla/xpcom/base/nsID.cpp:136

Comment 10

[David Bradley]

Attachment: Patch that uses nsIComponentManager

Ok, this doesn't quite test the same thing as the original test, but it does
test a bit more. So I think it's as good. If anyone has a better interface to
test than nsIComponentManager let me know and I'll use it instead.

Comment 11

[David Bradley]

Setting milestone and priority.

Comment 12

[David Bradley]

*** Bug 157699 has been marked as a duplicate of this bug. ***

Comment 13

[David Bradley]

Taking suggestions on better interfaces to use on this test. If you feel my last
patch is ok, then looking for an r/sr=

Comment 14

[David Bradley]

Moving out to 1.3. If this needs to be in before 1.3 please comment.

Comment 15

[David Bradley]

Moving to 1.4 Alpha

Comment 16

[Brant Gurganus]

By the definitions on <http://bugzilla.mozilla.org/bug_status.html#severity> and
<http://bugzilla.mozilla.org/enter_bug.cgi?format=guided>, crashing and dataloss
bugs are of critical or possibly higher severity.  Only changing open bugs to
minimize unnecessary spam.  Keywords to trigger this would be crash, topcrash,
topcrash+, zt4newcrash, dataloss.

Comment 17

[timeless]

*** Bug 11809 has been marked as a duplicate of this bug. ***

Comment 18

[David Bradley]

Moving out

Comment 19

[David Bradley]

Moving out, speak up if you believe this needs to be considered for 1.5b

Comment 20

[David Bradley]

Moving out

Comment 21

[Christian :Biesinger (don't email me, ping me on IRC)]

Comment on attachment 79611 [details] [diff] [review]
Patch that uses nsIComponentManager

still applies w/o even line mismatches, and fixes the crash.

Comment 22

[Doug Turner (:dougt)]

Comment on attachment 79611 [details] [diff] [review]
Patch that uses nsIComponentManager

no longer applies.  sorry for the delay.

Comment 23

[Doug Turner (:dougt)]

Attachment: patch v.2

timeless, this is your patch, updated to the tip.  It looks fine.

Comment 24

[Doug Turner (:dougt)]

Comment on attachment 367048 [details] [diff] [review]
patch v.2

sorry.  not this patch. :-)

Comment 25

[Doug Turner (:dougt)]

Attachment: patch v.2 [Checkin: Comment 27]

for realz this time.

Comment 26

[Doug Turner (:dougt)]

Comment on attachment 367049 [details] [diff] [review]
patch v.2
[Checkin: Comment 27]

this doesn't improve the test, it just makes it so that it doesn't crash against an old interface assumption.

Comment 27

[Serge Gautherie (:sgautherie)]

Comment on attachment 367049 [details] [diff] [review]
patch v.2
[Checkin: Comment 27]


http://hg.mozilla.org/mozilla-central/rev/fa9d6990f1a9

Comment 28

[Serge Gautherie (:sgautherie)]

You may want to file a follow-up bug to make this test be automatically executed by |make check|...

Comment 29

[Serge Gautherie (:sgautherie)]

This test could be rewritten as an xpcshell-test, couldn't it?