Closed Bug 957769 Opened 12 years ago Closed 12 years ago

FTU Network Scanning causes infinite Scanning For Networks screen

Categories

(Firefox OS Graveyard :: Gaia::First Time Experience, defect)

x86_64
Linux
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED
1.3 C2/1.4 S2(17jan)

People

(Reporter: qdot, Assigned: mikehenrty)

References

Details

(Keywords: regression, Whiteboard: [systemsfe])

Attachments

(1 file)

Repro: - Turn off wifi capabilities on computer - Create a clean gaia profile with ftu enabled - Start B2G Desktop - Choose Language Expected: Scanning For Networks screen quickly pops up then goes away. Actual: Scanning For Networks screen stays up forever. Stopping and restarting B2G process sometimes seems to unwedge it. This error message seems to pop up whenever it hits the infinite spinlock state, not sure if relevant: [22268] WARNING: Failed to terminate!: file /share/mozbuild/gecko-dev/dom/workers/WorkerPrivate.cpp, line 2621 [22268] WARNING: A control runnable was posted to a worker that is already shutting down!: file /share/mozbuild/gecko-dev/dom/workers/WorkerPrivate.cpp, line 2236 [22268] WARNING: A control runnable was posted to a worker that is already shutting down!: file /share/mozbuild/gecko-dev/dom/workers/WorkerPrivate.cpp, line 2236
gwagner says error message was already there so may not be relevant. Changing title to not infer cause.
Summary: Race Condition in FTU Network Scanning causes infinite Scanning For Networks screen → FTU Network Scanning causes infinite Scanning For Networks screen
Keywords: regression
Whiteboard: [systemsfe]
Target Milestone: --- → 1.3 C2/1.4 S2(17jan)
Blocks: 926347
Is it possible for this bug to reproduce on device?
(In reply to Jason Smith [:jsmith] from comment #3) > Is it possible for this bug to reproduce on device? Good point. It only reproduces with debug builds. We could insert some timeout for testing to make sure that this can never happen on the device.
The problem stems from two things, first: we are showing and then hiding the "Scanning" overlay twice, and second: hiding the overlay can _sometimes_ be asynchronous (depending on animation state). We show/hide this overlay twice when we skip the "Enable Cell Data" step (happens when no SIM is present). This in turn causes the navigation logic to set the hash twice (once before skipping once after), which then causes two wifi scans, and two overlay show/hides. AFAICT, this double show/hide has always happened, but my patch changed the timing between the two show/hides which caused this bug. If this is true, the fix is easy: don't set the hash twice. I'll upload a patch for this and ask Kyle to see if that fixes the issue for him too.
Attached file Test Solution
Kyle, could you test and see if this patch fixes it for you?
Attachment #8358129 - Flags: feedback?(kyle)
Comment on attachment 8358129 [details] Test Solution Looks like this is still broken, clearing the FB from Kyle until I come up with something better.
Attachment #8358129 - Flags: feedback?(kyle)
Comment on attachment 8358129 [details] Test Solution Lets try this again, does this work for you Kyle or Gregor?
Attachment #8358129 - Flags: feedback?(kyle)
Attachment #8358129 - Flags: feedback?(anygregor)
Comment on attachment 8358129 [details] Test Solution Fernando, could you take a look at this patch for fixing infinite wifi scanning? Cause of the bug is listed in comment 5.
Attachment #8358129 - Attachment mime type: text/plain → text/x-github-pull-request
Attachment #8358129 - Flags: review?(fernando.campo)
Comment on attachment 8358129 [details] Test Solution Seems to work well for me. Clicked through navigation in a few different ways and couldn't break it on b2g-desktop on a machine with no wifi.
Attachment #8358129 - Flags: feedback?(kyle) → feedback+
Comment on attachment 8358129 [details] Test Solution Updated the pull request to address a case where shimless B2G-desktop still caused this issue.
Attachment #8358129 - Flags: feedback?(anygregor) → feedback+
Comment on attachment 8358129 [details] Test Solution Patch looks nice, and after testing it I'm unable to repro. R+ with couple of minimal nits. Thanks for fixing it.
Attachment #8358129 - Flags: review?(fernando.campo) → review+
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: