491550 - Deadlock detector breaking mac fennec build

Status: RESOLVED FIXED

(Firefox for Android Graveyard :: General, defect)

Description

[Nino D'Aversa (:ndaversa)]

see http://pastebin.mozilla.org/646896 for build output

Comment 1

[Al Billings [:abillings - ex-MoCo]]

This is weird. I just finished building an hour before this on my OS X 10.5.6 machine.

Comment 2

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

(In reply to comment #1)
> This is weird. I just finished building an hour before this on my OS X 10.5.6
> machine.

You mean it worked an hour ago, but doesn't any more?  Or is it still working for you?

Comment 3

[Al Billings [:abillings - ex-MoCo]]

It worked an hour before and it worked just now when I tried (and succeeded) in building again. No error messages.

I'm not building debug though. Perhaps that is the issue?

Comment 4

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

(In reply to comment #3)
> It worked an hour before and it worked just now when I tried (and succeeded) in
> building again. No error messages.
> 
> I'm not building debug though. Perhaps that is the issue?

Yes, this code is not built for DEBUG.

Comment 5

[Jeremy Lea]

I get the similar errors on FreeBSD and Linux (Ubuntu).  Building DEBUG.  Does this need a clobber build?

Comment 6

[Jeremy Lea]

Sorry, that should have said, building firefox (not fennec)

Comment 7

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

(In reply to comment #5)
> I get the similar errors on FreeBSD and Linux (Ubuntu).  Building DEBUG.  Does
> this need a clobber build?

Can you try a clobber build and see if this goes away?  I don't have much idea right now what's wrong.

Comment 8

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Most likely an --enable-libxul issue. I think that changes whether or not the tests are built with the MOZILLA_INTERNAL_API flag.

Comment 9

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Attachment: Fix

Nino or Jeremy, can you please post your .mozconfigs so I can test this patch?

If I can't test today, this may remain broken until next Tuesday because I'm going on vacation.

Comment 10

[Stuart Parmenter]

It will need to be backed out

Comment 11

[Ben Turner (not reading bugmail, use the needinfo flag!)]

chris, to test, just add --enable-libxul to your debug firefox build.

Comment 12

[Nino D'Aversa (:ndaversa)]

I just did a clobber build and everything worked fine.

This is the mozconfig i'm using...
http://pastebin.mozilla.org/646974

Comment 13

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

(In reply to comment #12)
> I just did a clobber build and everything worked fine.
> 

A clobber build, but without applying the patch above?  If you didn't apply the patch, and --enable-libxul was the issue, it still shouldn't have worked ...

Comment 14

[Nino D'Aversa (:ndaversa)]

i just updated my tree before i built, I didn't apply the patch here, nor did I have --enable-libxul, but it builds and runs for me, I was just using it.

Let me know what I can do to help debug.

Comment 15

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

(In reply to comment #14)
> i just updated my tree before i built, I didn't apply the patch here, nor did I
> have --enable-libxul, but it builds and runs for me, I was just using it.
> 

Does the option "mobile --with-libxul-sdk=..." imply "--enable-libxul"?

Comment 16

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Never mind.  I repro'd this with --enable-libxul, seeing if my fix is correct ...

Comment 17

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

OK, it works.  I'm going to see if I can check this in without bsmedberg's r+ ...

Comment 18

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Advised not to check in without review.  The workaround until then is to apply the patch here.

Comment 19

[Benjamin Smedberg]

Attachment: Account for API differences between internal and external string.Find()

I'm going to commit this alternate fix: there are slight variations in the signature of .Find() between the internal and external string API... we should build the test in both configurations, and just account for the differences.

Comment 20

[Benjamin Smedberg]

http://hg.mozilla.org/mozilla-central/rev/5b6ce32aeab3