847963 - sometimes cannot transfer files to a mac

Status: RESOLVED WORKSFORME

(Firefox OS Graveyard :: Gaia::Bluetooth, defect)

Description

[Dietrich Ayala (:dietrich)]

STR:

1. pair unagi to mac
2. ensure bluetooth file sharing is enabled on mac
3. share a file from gallery to mac

Transfer fails without a detailed error message.

However, I tried again the next day and it worked!

Perhaps first time after pairing?

Comment 1

[[e-mail], [e-mail]]

Dietrich,
What the build version do you occur the issue?

Comment 2

[Dietrich Ayala (:dietrich)]

Nightly channel, from March 4, when I reported the bug.

Comment 3

[Dietrich Ayala (:dietrich)]

Walter, were you able to reproduce it?

I'm not sure this is enough to request blocking any release on, given the lack of repeatable steps.

I did see it again today though.

Comment 4

[Walter Chen[:ypwalter][:wachen]]

I didn't have a device for me to try it on yet.
However, I saw this issue on and off, and I was able to reproduce it only for once long time ago.

Comment 5

[Michael Vines [:m1] [:evilmachines]]

(clearing tef? until we have more info here)

Comment 6

[John Hammink]

Tested on 
Gecko  http://hg.mozilla.org/releases/mozilla-b2g18/rev/99680a32f297
Gaia   de3e5b9205e6cb1a6bd0858a98d159272ad96d11
BuildID 20130312070202
Version 18.0
And file transfer works correctly and consistently (both ways) with 10 trials.

Also tested on
Gecko  http://hg.mozilla.org/releases/mozilla-b2g18_v1_0_1/rev/cdf058e47197
Gaia   1a6e1cd7715a5192b32db4b7127c5ae5b8162a7a
BuildID 20130312070203
Version 18.0
And file transfer works correctly and consistently (both ways) with 10 trials.

Comment 7

[Dietrich Ayala (:dietrich)]

I reproduce this almost every single time I try to transfer from Unagi on nightly channel to Macbook Air 11" running OS X 10.8.2.

Comment 8

[Eric Chou [:ericchou] [:echou]]

I'll take a look.

Comment 9

[Gina Yeh [:gyeh] [:ginayeh]]

I tried nightly build and had no luck to reproduce it.

Here's the build information extracted from sources.xml:

Build id: 20130502130649

Device: unagi

<!-- Mercurial-Information: <project name="releases/mozilla-b2g18_v1_0_1" path="gecko" remote="hgmozillaorg" revision="cbfd20cafcfb"/> --><project name="gecko.git" path="gecko" remote="mozillaorg" revision="25be785425d50165fef5cd06b171bfd7d79be70d"/>

<project name="gaia.git" path="gaia" remote="mozillaorg" revision="716c192821f6528dfaa198c1e1526c464ed42ba9"/>

Comment 10

[Gina Yeh [:gyeh] [:ginayeh]]

Dietrich, can you help to try the latest build?

Comment 11

[Dietrich Ayala (:dietrich)]

Tried the 5/12 build on Unagi (nightly channel, which is v1-train IIRC).

Unagi -> Mac
* First try failed. Hung the whole UI. Eventually came back and flashed a bunch of screens in quick succession, so couldn't see what happened.
* Second try trailed with no hang, and showed the error screen.
* Third try, hung for about 20 seconds, came back with error screen.

Mac -> Unagi
* First try sending a PDF file 0.5mb, crashed the phone, caused B2G restart.

Comment 12

[Eric Chou [:ericchou] [:echou]]

Hi Dietrich,

There is another bug (Bug 862654) that Daniel Holbert seems to have the same kind of 'weird devices'. I was thinking that maybe Kyle could help us with debugging this since we still can't reproduce on our devices.

Hi Kyle,

Unfortunately we had no luck reproducing it on our devices, but Bluetooth file transfer seems not work at all on Dietrich's device. Could you help us with this? Thanks in advance, and please feel free to let me know if you have any questions.

Eric

(In reply to Dietrich Ayala (:dietrich) from comment #11)
> Tried the 5/12 build on Unagi (nightly channel, which is v1-train IIRC).
> 
> Unagi -> Mac
> * First try failed. Hung the whole UI. Eventually came back and flashed a
> bunch of screens in quick succession, so couldn't see what happened.
> * Second try trailed with no hang, and showed the error screen.
> * Third try, hung for about 20 seconds, came back with error screen.
> 
> Mac -> Unagi
> * First try sending a PDF file 0.5mb, crashed the phone, caused B2G restart.

Comment 13

[Kyle Machulis [:qdot] [:kmachulis] (INACTIVE)]

Ok, sorry about the delay on this, my mbp drive died last friday so I had a rather nasty reinstall process.

My experiences with 1.1 production build 2013-05-20:

- Can pair with mac reliably
- Phone always shows as disconnected from mac. Cannot send file from phone to mac, nor can I find a way to get the mac to connect to the phone. Always get not connected error screen on phone, nothing on mac
- When sending from mac to phone: connection happens, tried to send png, crashed the phone.

I guess I'm seeing pretty much what Dietrich is. I'll try to get some stacks to figure this out.

Comment 14

[Kyle Machulis [:qdot] [:kmachulis] (INACTIVE)]

Ok, my crash on mac -> unagi transfer is intermittent too, but I did at least catch a stack.

#0  0x4186043a in mozalloc_abort (msg=<value optimized out>) at /share/mozbuild/B2G/gecko/memory/mozalloc/mozalloc_abort.cpp:30
#1  0x413e981c in Abort (aSeverity=<value optimized out>, aStr=0x418e944c "corrupted actor state", aExpr=<value optimized out>, aFile=<value optimized out>, aLine=35) at /share/mozbuild/B2G/gecko/xpcom/base/nsDebugImpl.cpp:423
#2  NS_DebugBreak_P (aSeverity=<value optimized out>, aStr=0x418e944c "corrupted actor state", aExpr=<value optimized out>, aFile=<value optimized out>, aLine=35) at /share/mozbuild/B2G/gecko/xpcom/base/nsDebugImpl.cpp:380
#3  0x4137a3c0 in mozilla::dom::bluetooth::PBluetoothRequest::Transition (from=<value optimized out>, trigger=..., next=<value optimized out>) at /share/mozbuild/B2G/gecko/objdir-gonk-debug/ipc/ipdl/PBluetoothRequest.cpp:35
#4  0x412e8490 in mozilla::dom::bluetooth::PBluetoothRequestParent::Send__delete__ (actor=0x48a52c40, response=...) at /share/mozbuild/B2G/gecko/objdir-gonk-debug/ipc/ipdl/PBluetoothRequestParent.cpp:87
#5  0x410491b0 in mozilla::dom::bluetooth::BluetoothRequestParent::ReplyRunnable::Run (this=0x48af3020) at /share/mozbuild/B2G/gecko/dom/bluetooth/ipc/BluetoothParent.cpp:50
#6  0x413e5566 in nsThread::ProcessNextEvent (this=0x40409940, mayWait=<value optimized out>, result=0xbecc97df) at /share/mozbuild/B2G/gecko/xpcom/threads/nsThread.cpp:620
#7  0x413c593e in NS_ProcessNextEvent_P (thread=0x23, mayWait=false) at /share/mozbuild/B2G/gecko/objdir-gonk-debug/xpcom/build/nsThreadUtils.cpp:237
#8  0x412d8404 in mozilla::ipc::MessagePump::Run (this=0x40402460, aDelegate=0x404310c0) at /share/mozbuild/B2G/gecko/ipc/glue/MessagePump.cpp:82
#9  0x41407410 in MessageLoop::RunInternal (this=0x1000000) at /share/mozbuild/B2G/gecko/ipc/chromium/src/base/message_loop.cc:219
#10 0x414074ba in MessageLoop::RunHandler (this=0x404310c0) at /share/mozbuild/B2G/gecko/ipc/chromium/src/base/message_loop.cc:212
#11 MessageLoop::Run (this=0x404310c0) at /share/mozbuild/B2G/gecko/ipc/chromium/src/base/message_loop.cc:186
#12 0x4125d2dc in nsBaseAppShell::Run (this=0x404b08e0) at /share/mozbuild/B2G/gecko/widget/xpwidgets/nsBaseAppShell.cpp:163
#13 0x411c14e0 in nsAppStartup::Run (this=0x42e83bb0) at /share/mozbuild/B2G/gecko/toolkit/components/startup/nsAppStartup.cpp:290
#14 0x40bebdfa in XREMain::XRE_mainRun (this=0xbecc999c) at /share/mozbuild/B2G/gecko/toolkit/xre/nsAppRunner.cpp:3794
#15 0x40bee464 in XREMain::XRE_main (this=0xbecc999c, argc=<value optimized out>, argv=0xbeccbb84, aAppData=<value optimized out>) at /share/mozbuild/B2G/gecko/toolkit/xre/nsAppRunner.cpp:3860
#16 0x40bee5b0 in XRE_main (argc=1, argv=0xbeccbb84, aAppData=0x1f170, aFlags=<value optimized out>) at /share/mozbuild/B2G/gecko/toolkit/xre/nsAppRunner.cpp:3935
#17 0x0000999e in do_main (argc=1, argv=0xbeccbb84) at /share/mozbuild/B2G/gecko/b2g/app/nsBrowserApp.cpp:168
#18 main (argc=1, argv=0xbeccbb84) at /share/mozbuild/B2G/gecko/b2g/app/nsBrowserApp.cpp:261

Working on a repro now. I think this may only happen on a first transfer after pairing?

Comment 15

[Kyle Machulis [:qdot] [:kmachulis] (INACTIVE)]

Ok, I can't find any stable repro for this. I've tried reflashing the phone, unpairing and repairing, and combinations of all of that. It's sending files fine again. At least we have a stack now, but I can't give any good steps to make that happen anywhere else.

Comment 16

[Kyle Machulis [:qdot] [:kmachulis] (INACTIVE)]

Attachment: logcat during bluetooth file transfer crash

Caught a logcat while the crash was happening. Lines around crash (removing idleservice/memalloc/oncrpc calls):

W/Gonk    (  109): Could not open boot animation
I/Gecko:MemoryInfoDumper(  109): FifoWatcher::OpenFifo unlink failed; errno=2.  Continuing despite error.
E/        (  131): cby, conn_id = 0, client_id = 1, service_id = 224
I/Gonk    (  109): Setting nice for pid 109 to 0
I/Gonk    (  109): Changed nice for pid 109 from 0 to 0.
I/Gonk    (  109): Setting nice for pid 527 to 18
I/Gonk    (  109): Changed nice for pid 527 from 0 to 18.
I/GeckoDump(  109): XXX FIXME : Got a mozContentEvent: activity-choice
I/Gonk    (  109): Setting nice for pid 527 to 1
I/Gonk    (  109): Changed nice for pid 527 from 18 to 1.
E/GeckoConsole(  109): Content JS LOG at app://system.gaiamobile.org/js/window_manager.js:662 in anonymous: Window Manager: No screenshot in database. This is expected from a fresh installed app.
E/profiler(  527): Registering start signal
E/GeckoConsole(  527): [JavaScript Warning: "Unknown property '-moz-align-self'.  Declaration dropped." {file: "resource://gre-resources/ua.css" line: 44}]
I/Gecko   (  527): ###################################### forms.js loaded
I/Gecko   (  527): ############################### browserElementPanning.js loaded
I/Gecko   (  527): ######################## BrowserElementChildPreload.js loaded
I/Gonk    (  109): dbus_func_args_timeout_valist: D-Bus error in GetServiceAttributeValue: org.bluez.Error.Failed (GetServiceAttribute Failed)
E/GeckoConsole(  527): [JavaScript Warning: "XUL box for p element contained an inline #text child, forcing all its children to be wrapped in a block." {file: "app://bluetooth.gaiamobile.org/transfer.html" line: 0}]
I/Gonk    (  109): Setting nice for pid 557 to 18
I/Gonk    (  109): Changed nice for pid 557 from 0 to 18.
E/profiler(  557): Registering start signal
E/GeckoConsole(  557): [JavaScript Warning: "Unknown property '-moz-align-self'.  Declaration dropped." {file: "resource://gre-resources/ua.css" line: 44}]
I/Gecko   (  557): ###################################### forms.js loaded
I/Gecko   (  557): ############################### browserElementPanning.js loaded
I/Gecko   (  557): ######################## BrowserElementChildPreload.js loaded
I/Gecko   (  109): Warning: No fanotify support in this device's kernel.
F/libc    (  109): Fatal signal 11 (SIGSEGV) at 0x00000005 (code=1)
I/Gecko   (  430): [Child 430] ###!!! ABORT: ActorDestroy by IPC channel failure at ShadowLayersChild: file /share/mozbuild/B2G/gecko/gfx/layers/ipc/ShadowLayersChild.cpp, line 68
E/Gecko   (  430): mozalloc_abort: [Child 430] ###!!! ABORT: ActorDestroy by IPC channel failure at ShadowLayersChild: file /share/mozbuild/B2G/gecko/gfx/layers/ipc/ShadowLayersChild.cpp, line 68
F/libc    (  430): Fatal signal 11 (SIGSEGV) at 0x00000000 (code=1)
I/DEBUG   (  116): debuggerd committing suicide to free the zombie!
I/DEBUG   (  572): debuggerd: May 20 2013 17:52:33

Comment 17

[Kyle Machulis [:qdot] [:kmachulis] (INACTIVE)]

Talked to bent, actor is disappearing for some reason. Need to run valgrind against the phone while trying file transfer to see if there's destruction happening in a way we don't expect.

Comment 18

[Eric Chou [:ericchou] [:echou]]

Hey Kyle,

Thanks for debugging!

I saw that you used 0520 v1.1 build for testing, but v1.1 had two bugs because of DeviceStorage interface changing, bug 872238 and bug 873065. Although they were solved about a week, they were merged into v1.1 just 2 days ago since we couldn't get leo+. Let me try the latest build and I'll comment later.

Comment 19

[Eric Chou [:ericchou] [:echou]]

Close this bug as WFM since the last comment was written 7 months ago and no similar bugs have been reported since then.

Comment 20

[Preeti Raghunath(:Preeti)]

Please review bug 1011110.

Comment 21

[Preeti Raghunath(:Preeti)]

Eric

Can you please take a look?

Comment 22

[Shawn Huang [:shawnjohnjr][:shuang] (as a happy gecko contributor)]

Base on Comment 15 and 17, there is no easy way to hit the problem on Comment 14. I suggest keep bug 1011110. Since this bug (bug 847963) aims to address file transfer but not for Comment 14 crash bug.

Comment 23

[Preeti Raghunath(:Preeti)]

Closing per comment 22