576518 - e10s: InitDBFile and NS_GetSpecialDirectory fail in nsNavHistory

Status: RESOLVED FIXED

(Core :: IPC, defect)

Description

[Egor Starkov]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.0.19) Gecko/2010031218 Firefox/3.5
Build Identifier: 

While loading http://bash.org.ru get these warnings:
WARNING: NS_ENSURE_SUCCESS(rv, rv) failed with result 0x80004005: file mozilla/toolkit/components/places/src/nsNavHistory.cpp, line 540

the code is:

  nsresult rv = NS_GetSpecialDirectory(NS_APP_USER_PROFILE_50_DIR,
                                       getter_AddRefs(profDir));
  NS_ENSURE_SUCCESS(rv, rv);

in
  nsNavHistory::InitDBFile(PRBool aForceInit)

Second warning is:
WARNING: NS_ENSURE_SUCCESS(rv, rv) failed with result 0x80004005: file /home/egor/harmattan_hg1/microb-engine/build-tree/mozilla/toolkit/components/places/src/nsNavHistory.cpp, line 438

the code is:
  rv = InitDBFile(PR_FALSE);
  NS_ENSURE_SUCCESS(rv, rv);
in:
  nsNavHistory::Init()

I believe that there should be no assertions in Init code always.

Reproducible: Always

Comment 1

[Boris Zbarsky [:bzbarsky]]

Seems like nav history should be failing out early on attempts to init it in the content process.

Comment 2

[Shawn Wilsher :sdwilsh]

(In reply to comment #1)
> Seems like nav history should be failing out early on attempts to init it in
> the content process.
I thought we shouldn't have been initializing history in the content process.

Comment 4

[Oleg Romashin (:romaxa)]

Attachment: Stop history initialization on child process

Also we are calling nsIWebBrowserSetup::SETUP_ALLOW_DNS_PREFETCH:, and that initializing global history because of missing "break"..

Comment 5

[Oleg Romashin (:romaxa)]

and missing break coming from https://bug492196.bugzilla.mozilla.org/attachment.cgi?id=377824 ... is it expected or just bug?

Comment 6

[Shawn Wilsher :sdwilsh]

Is this the same as bug 549887?

Comment 7

[Oleg Romashin (:romaxa)]

don't think so, it is about stopping history init in child process

Comment 8

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 460515 [details] [diff] [review]
Stop history initialization on child process

OK, I buy that, I think.

Comment 9

[Oleg Romashin (:romaxa)]

Is it blocking IPC fennec?

Comment 10

[Stuart Parmenter]

is this ready to land?

Comment 11

[Oleg Romashin (:romaxa)]

http://hg.mozilla.org/projects/electrolysis/rev/a72857bca7e6

Comment 12

[Simon Montagu :smontagu]

This broke --disable-ipc builds: it needs an #ifdef MOZ_IPC