833764 - [B2G] target is null after tapping on cards view

Status: VERIFIED FIXED

(Remote Protocol :: Marionette, defect)

Description

[Zac C (:zac)]

Attachment: TC for target is null

In this test case, loading 3 apps, then loading cards view then switching to one of them causes marionette to lose its way.

Any commands to marionette will fail with "TypeError: target is null"

Marionette should stay on the homescreen frame.

Comment 1

[Florin Strugariu [:Bebe]]

I added a break point at line 
        self.marionette.tap(second_app_card) 

and fallowed the log cat:
http://pastebin.mozilla.org/2080604


This is what we get after the tap:

I/Gecko   (  110): 1358949065265        Marionette      DEBUG Got request: execute, data: {"to":"conn27.marionette1","specialPowers":false,"session":"6-b2g","newSandbox":true,"args":[],"type":"executeScript","value":"return typeof window.wrappedJSObject.MouseEventShim === 'undefined';"}, id: {59354758-e962-42e0-9bac-38167d59eb49}
I/Gecko   (  110): 1358949065355        Marionette      INFO  sendToClient: {"from":"conn27.marionette1","value":true}, {59354758-e962-42e0-9bac-38167d59eb49}, {59354758-e962-42e0-9bac-38167d59eb49}
I/Gecko   (  110): 1358949065363        Marionette      DEBUG Got request: execute, data: {"to":"conn27.marionette1","specialPowers":false,"session":"6-b2g","newSandbox":true,"args":[{"ELEMENT":"{3786edc5-fc84-465a-b23a-a10be3f0956b}"},true],"type":"executeScript","value":"SyntheticGestures.tap(arguments[0], null, null, null, null, arguments[1]);"}, id: {f7b9ad06-1765-4f08-a896-13e97a9c3f65}
I/Gecko   (  110): 1358949065438        Marionette      INFO  sendToClient: {"from":"conn27.marionette1","value":null}, {f7b9ad06-1765-4f08-a896-13e97a9c3f65}, {f7b9ad06-1765-4f08-a896-13e97a9c3f65}
E/GeckoConsole(  110): [JavaScript Error: "TypeError: target is null" {file: "chrome://marionette/content/marionette-listener.js" line: 902}]
D/memalloc(  110): /dev/pmem: Allocated buffer base:0x4ad5a000 size:614400 offset:1843200 fd:157
I/Gecko   (  110): 1358949065627        Marionette      INFO  sendToClient: {"from":"conn27.marionette1","error":{"message":"TypeError: target is null","status":17,"stacktrace":null}}, undefined, null
I/Gecko   (  110): 1358949065628        Marionette      WARN  got a response with no command_id
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4ae86000 size:614400 offset:1228800 fd:102
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4b028000 size:614400 offset:2940928 fd:160
D/memalloc(  110): /dev/pmem: Allocated buffer base:0x4ad5a000 size:614400 offset:1228800 fd:102
D/memalloc(  110): /dev/pmem: Allocated buffer base:0x4ad5a000 size:614400 offset:2940928 fd:160
D/memalloc( 1879): /dev/pmem: Mapped buffer base:0x44400000 size:3555328 offset:2940928 fd:27
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4ae86000 size:614400 offset:1228800 fd:102
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4af1c000 size:614400 offset:1843200 fd:157
D/memalloc(  397): /dev/pmem: Unmapping buffer base:0x44eeb000 size:2707456 offset:2699264
D/memalloc(  397): /dev/pmem: Unmapping buffer base:0x45208000 size:2805760 offset:2707456
D/memalloc(  397): /dev/pmem: Unmapping buffer base:0x454ea000 size:2818048 offset:2805760
D/memalloc(  397): /dev/pmem: Unmapping buffer base:0x44c58000 size:2699264 offset:2576384
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4afb2000 size:8192 offset:2457600 fd:127
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4afb4000 size:98304 offset:2465792 fd:130
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4afcc000 size:12288 offset:2564096 fd:134
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4b00a000 size:122880 offset:2818048 fd:154
D/memalloc(  397): /dev/pmem: Unmapping buffer base:0x44200000 size:2465792 offset:2457600
D/memalloc(  397): /dev/pmem: Unmapping buffer base:0x44503000 size:2564096 offset:2465792
D/memalloc(  397): /dev/pmem: Unmapping buffer base:0x44900000 size:2576384 offset:2564096
D/memalloc(  397): /dev/pmem: Unmapping buffer base:0x457c0000 size:2940928 offset:2818048
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4afed000 size:8192 offset:2699264 fd:142
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4afef000 size:98304 offset:2707456 fd:146
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4b007000 size:12288 offset:2805760 fd:150
D/memalloc(  110): /dev/pmem: Freeing buffer base:0x4afcf000 size:122880 offset:2576384 fd:138
I/Gecko   (  110): 1358949068345        Marionette      INFO  sendToClient: {"from":"conn27.marionette1","value":"{6bb1c1c6-a570-4c9a-8b23-4859ab9a991e}"}, {d799776e-5714-467a-b856-e85cf473483f}, {d799776e-5714-467a-b856-e85cf473483f}
I/Gecko   (  110): 1358949070377        Marionette      INFO  sendToClient: {"from":"conn27.marionette1","ok":true}, {047cdd6e-4327-4f55-918f-7d5735d0a658}, {047cdd6e-4327-4f55-918f-7d5735d0a658}
E/GeckoConsole(  110): [JavaScript Error: "[Exception... "Component returned failure code: 0x80470002 (NS_BASE_STREAM_CLOSED) [nsIOutputStream.write]"  nsresult: "0x80470002 (NS_BASE_STREAM_CLOSED)"  location: "JS frame :: chrome://global/content/devtools/dbg-transport.js :: DT_onOutputStreamReady :: line 94"  data: no]" {file: "chrome://global/content/devtools/dbg-transport.js" line: 94}]

Comment 2

[Florin Strugariu [:Bebe]]

also any command run after that tap gets the target is null error.

The command run after that gets as a response a mix or the output of the previous and/or the current command.

So if we run a page source after the tap it gets  the target is null error.
if we run a self.marionette.get_active_element() after the page source the returned object will "contain" a mix between the pages_source output and the the get_active_element() output

Here is a screencast http://www.youtube.com/watch?v=DPUYeZkIeAk&feature=youtu.be

Comment 3

[David Burns :automatedtester]

hurray replicated, now to debug

Comment 4

[David Burns :automatedtester]

It appears we are not returning the error straight away. Logcat is erroring on the tap saying its not there and then complaining that the ID is missing

I/Gecko   (  105): 1359127255815	Marionette	DEBUG	Got request: execute, data: {"to":"conn0.marionette1","specialPowers":false,"session":"6-b2g","newSandbox":true,"args":[{"ELEMENT":"{fd866d85-10bf-46b5-835d-9a2cabf24a0e}"},true],"type":"executeScript","value":"SyntheticGestures.tap(arguments[0], null, null, null, null, arguments[1]);"}, id: {704cea17-c28e-403d-b3c9-031c17c7afb9}
I/Gecko   (  105): 1359127255910	Marionette	INFO	sendToClient: {"from":"conn0.marionette1","value":null}, {704cea17-c28e-403d-b3c9-031c17c7afb9}, {704cea17-c28e-403d-b3c9-031c17c7afb9}
E/GeckoConsole(  105): [JavaScript Error: "TypeError: target is null" {file: "chrome://marionette/content/marionette-listener.js" line: 902}]
D/memalloc(  105): /dev/pmem: Allocated buffer base:0x4a05a000 size:614400 offset:2371584 fd:106
I/Gecko   (  105): 1359127256096	Marionette	INFO	sendToClient: {"from":"conn0.marionette1","error":{"message":"TypeError: target is null","status":17,"stacktrace":null}}, undefined, null
I/Gecko   (  105): 1359127256096	Marionette	WARN	got a response with no command_id

Comment 5

[David Burns :automatedtester]

it is appearing that the element is being passed through to the synthetic gestures library but then it thinks it is null. I am wondering if the issue is down to context not being what we think it us when tap() is finally called.

Comment 6

[David Burns :automatedtester]

i think that this is similar to bug 833761 but need more proof

Comment 7

[Andreea Matei [:AndreeaMatei]]

I have the same issue when running gaia test_wallpaper.py, which needs to tap on an element that is not in view. It is down in the list of settings. If I manually scroll fast enough, the test manages to tap on the element and finally pass.

Comment 9

[David Burns :automatedtester]

test case in bug 852985 is much prettier :)

def test_issue_550(self):
  self.lockscreen.lock()
  self.marionette.tap(self.marionette.find_element('id', 'lockscreen-area-camera'))

Comment 10

[Jonathan Griffin (:jgriffin)]

(In reply to David Burns :automatedtester from comment #9)
> test case in bug 852985 is much prettier :)
> 
> def test_issue_550(self):
>   self.lockscreen.lock()
>   self.marionette.tap(self.marionette.find_element('id',
> 'lockscreen-area-camera'))

Both this and the attached test pass for me using mozilla-inbound; I'm guessing this report was made against mozilla-b2g18, though.  I'll try reproducing it there.  It's likely some patches that need to be uplifted.

Comment 11

[Jonathan Griffin (:jgriffin)]

Yep, reproduced it against mozilla-b2g18.  So it's just a matter of figuring out what patches are missing and need to be uplifted.

Comment 12

[Zac C (:zac)]

Thanks for looking into this jgriffin!

Comment 13

[Jonathan Griffin (:jgriffin)]

Uplifting the patches from bugs 814768, 834266 and 845625 seemed to do the trick.  Let me know if you run into this problem again.

Comment 14

[Zac C (:zac)]

Validated this with both test cases against 0.5.22, v1-train Gaia build 2013032507xx and it's nice and stable. great!