1116366 - [FFOS7715 v2.1][system]Download failed，when switch SIM card data to wifi

Status: RESOLVED WONTFIX

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

Description

[zhaoyang]

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:36.0) Gecko/20100101 Firefox/36.0
Build ID: 20141119030200

Steps to reproduce:

 1.close the wifi and make sure SIM card data are available 
 2.open broswer and download one apk file 
 3.open wifi and connect the wifi
 4.download file failed


Actual results:

download file failed


Expected results:

download file continue

Comment 1

[zhaoyang]

Attachment: log.txt

Comment 2

[zhaoyang]

Hi paolo:
   
    I know you are expected in download, please help me to investigate. 
    thanks a lot. 

    ps: if you have any problem, please need info for me, i will reply quickly.

Comment 3

[:Paolo Amadini]

Hi! I believe this issue may be specific to Firefox OS networking or front-end, about which unfortunately I don't know much. I can't see anything obvious from the logs, thanks for attaching them anyways.

Francisco maybe may know more or can find the right people?

Comment 4

[zhaoyang]

(In reply to :Paolo Amadini from comment #3)
> Hi! I believe this issue may be specific to Firefox OS networking or
> front-end, about which unfortunately I don't know much. I can't see anything
> obvious from the logs, thanks for attaching them anyways.
> 
> Francisco maybe may know more or can find the right people?

Hi Paolo! 
Thanks! I am sorry, and I will try to find the right people.

Comment 5

[zhaoyang]

Hi francisco!

   As of 5 comments, you are expected！Can you help me to analyze the problem？

   Thanks a lot！

   ps: if you have any problem, please need info for me, i will reply quickly.

Comment 6

[zhaoyang]

when WiFi network to download a file, switch to the SIM card data download:

  if download video files (download Youku Video),download failed, error:2152398868; 
  if download the file format for.Apk or other documents, can continue to download;

Comment 7

[Francisco Jordano [:arcturus] [:francisco]]

Hi,

not my area of expertise, since I think this is necko related. When switching to wifi we are using a different interface and I have the feeling, by what I'm reading here that we shut down the telephony interface.

Not sure if we can do something, perhaps Fabrice could clarify a bit, or redirect to someone that could help.

Comment 8

[zhaoyang]

(In reply to Francisco Jordano [:arcturus] [:francisco] from comment #7)
> Hi,
> 
> not my area of expertise, since I think this is necko related. When
> switching to wifi we are using a different interface and I have the feeling,
> by what I'm reading here that we shut down the telephony interface.
> 
> Not sure if we can do something, perhaps Fabrice could clarify a bit, or
> redirect to someone that could help.

Hi,
Tank you very much!

Comment 9

[[:fabrice] Fabrice Desré]

We would need to debug more, but when we switch from Wifi to Data (or vice versa) it's similar to shutting down and restarting network. The download code is smart enough to use http range request when they are supported by the server side. That may be why we can resume the document downloads but not the streaming ones.

Comment 10

[zhaoyang]

(In reply to Fabrice Desré [:fabrice] from comment #9)
> We would need to debug more, but when we switch from Wifi to Data (or vice
> versa) it's similar to shutting down and restarting network. The download
> code is smart enough to use http range request when they are supported by
> the server side. That may be why we can resume the document downloads but
> not the streaming ones.

Hi Fabrice,
Could you tell your progress on this issure？

Thanks！

Comment 11

[[:fabrice] Fabrice Desré]

There's nothing I can do without knowing the urls to try to reproduce. Can you add them to this bug?

Comment 12

[zhaoyang]

Hi Fabrice,
sorry!

url:www.apk8.com
download one apk file,and switch wifi to data

Thanks！

Comment 13

[Vance Chen [:vchen][skype:vance.lucida][[e-mail]]]

Hi Fabrice, the url information is on Comment#12, could you help on this one?

Thanks

Comment 14

[Vance Chen [:vchen][skype:vance.lucida][[e-mail]]]

Hi Shawn -

Any one at device team side can also help to take a look at this one?

Thanks

Vance

Comment 15

[Vance Chen [:vchen][skype:vance.lucida][[e-mail]]]

Hi Steven.

Partner just want to clarify that if this is a per design behavior or it is indeed a bug. Could you help to find someone to clarify?

Thanks

Comment 16

[shawn ku [:sku]]

This issue should be resolved at Bug 1008091.
Please nominate 2.1s? if fix is really needed.

Similar bug - Bug 1106220.

Comment 17

[Steven Yang [:styang]]

(In reply to shawn ku [:sku] from comment #16)
> This issue should be resolved at Bug 1008091.
> Please nominate 2.1s? if fix is really needed.
> 
> Similar bug - Bug 1106220.

Yes, we need it for 2.1S. Vincent, could you help?

Comment 18

[zhaoyang]

Hi Vincent,

  we need it for 2.1，please help
  Thanks a lot

Comment 19

[Vincent Liu[:vliu]]

(In reply to Steven Yang [:styang] from comment #17)
> (In reply to shawn ku [:sku] from comment #16)
> > This issue should be resolved at Bug 1008091.
> > Please nominate 2.1s? if fix is really needed.
> > 
> > Similar bug - Bug 1106220.
> 
> Yes, we need it for 2.1S. Vincent, could you help?

I assume that we need the patch of Bug 1008091 for this bug. If so, I'd merged the patch of Bug 1008091.

Comment 20

[Steven Yang [:styang]]

Vincent, please update if the merge has been completed. thanks

Comment 21

[zhaoyang]

Attachment: download-log

Hi daniel,
   
   this is the bug I referred in the email,and I add the log when switch data to wifi,after apply your patch.please help give advices
   
   Thanks a lot

Comment 22

[Daniel Stenberg [:bagder]]

(In reply to zhaoyang from comment #21)
> Created attachment 8555142 [details]

>    this is the bug I referred in the email,and I add the log when switch
> data to wifi,after apply your patch.please help give advices

Thanks. Any chance you can enable more logging from some of the affected network modules? Like this: NSPR_LOG_MODULES=timestamp,nsHttp:5,nsSocketTransport:5,nsStreamPump:5,nsHostResolver:5,nsNotifyAddr:5

Further: I'm not sure how many things work in b2g land but there's a method in dom/system/gonk/NetworkManager.js that according to its comments "watches for network interfaces changing state and then adjusts routes etc. accordingly". Do we need to make something extra to make sure this happens?

Comment 23

[zhaoyang]

Attachment: download

Hi Daniel，

  there is new download log，after add NSPR_LOG_MODULES=nsStreamPump:5;NSPR_LOG_MODULES=nsNotifyAddr:5; NSPR_LOG_MODULES=nsSocketTransport:5
and open debug networkmanager.js 

would you get more from it？if you need others,please tell me,I will provide rapidly

Comment 24

[Daniel Stenberg [:bagder]]

(In reply to zhaoyang from comment #23)

>   there is new download log，after add
> NSPR_LOG_MODULES=nsStreamPump:5;NSPR_LOG_MODULES=nsNotifyAddr:5;
> NSPR_LOG_MODULES=nsSocketTransport:5
> and open debug networkmanager.js 
> 
> would you get more from it？if you need others,please tell me,I will provide
> rapidly

Thanks for your hard work on getting this. I think something was wrong in there since I couldn't see any messages at all from those requested log modules in the log file you attached.

Also note that I mentioned a few more modules than you said you enabled, like for example I was particularly interested in nsHostResolver as well (also, having the full request set will make the log better comparable to other Firefox runs in similar situations that may not be b2g).

Is that really how you are supposed to do it? You assign the variable several times with new values?

Finally, it'd be great if you just made a note at which time you switch the wifi on/off to give some good data points at where in the log stuff should occur.

About the problem, just to be perfectly clear: you first download without wifi and then you switch on wifi and you can't get to the file anymore, right? What is the exact error message/problem that you see when it fails (as detailed as possible please) ?

Comment 25

[zhaoyang]

Hi Daniel，

  sorry,my bad
  I add newdownloadlog.txt and PRdownloadlog.txt. newdownloadlog.txt is common log. PRdownloadlog.txt is about network modules.
  About the problem:first download without wifi and then switch on wifi and can't get to the file anymore, download failed. error:NS_ERROR_NET_TIMEOUT

  I can find from newdownloadlog.txt:

wifi connected:
01-28 22:06:01.285 I/Gecko   ( 5080): -*- DataConnectionHandler[0]: Disconnect data call when Wifi is connected.

downloaderror:
01-28 22:06:17.575 I/Gecko   ( 5080): -*- DownloadsAPI.jsm : onDownloadChanged ({totalBytes:17191793, currentBytes:522524, url:"http://d2.apk8.com:8020/game_g/xiaoledepenqibeibao.apk", path:"/storage/sdcard0/downloads/xiaoledepenqibeibao.apk", contentType:"application/vnd.android.package-archive", startTime:1422453942267, error:(new DownloadError("[Exception... \"The connection has timed out\"  nsresult: \"0x804b000e (NS_ERROR_NET_TIMEOUT)\"  location: \"JS frame :: resource://gre/modules/DownloadCore.jsm :: this.DownloadError :: line 1238\"  data: no]", "")), id:"download-1", state:"stopped"})
01-28 22:06:17.585 I/Gecko   ( 5080): -*- DownloadsIPC.jsm : message: Downloads:Changed
01-28 22:06:17.585 I/Gecko   ( 5080): -*- DownloadsIPC.jsm : notifyChanges notifying changes for download-1
01-28 22:06:17.595 I/Gecko   ( 5080): -*- DownloadsAPI.js : DOMDownloadImpl observe downloads-state-change-download-1
01-28 22:06:17.595 I/Gecko   ( 5080): -*- DownloadsAPI.js : update ({totalBytes:17191793, currentBytes:522524, url:"http://d2.apk8.com:8020/game_g/xiaoledepenqibeibao.apk", path:"/storage/sdcard0/downloads/xiaoledepenqibeibao.apk", contentType:"application/vnd.android.package-archive", startTime:(new Date(1422453942267)), id:"download-1", state:"stopped", error:{name:"DownloadError", result:2152398862, message:"[Exception... \"The connection has timed out\"  nsresult: \"0x804b000e (NS_ERROR_NET_TIMEOUT)\"  location: \"JS frame :: resource://gre/modules/DownloadCore.jsm :: this.DownloadError :: line 1238\"  data: no]", becauseSourceFailed:true, becauseTargetFailed:false, stack:"this.DownloadError@resource://gre/modules/DownloadCore.jsm:1266:16\nDLS_onTransferFinished@resource://gre/modules/DownloadCore.jsm:2012:33\nDLT_OSC_onDownload@jar:file:///system/b2g/omni.ja!/components/DownloadLegacy.js:176:9\nHandler.prototype.process@resource://gre/modules/Promise.jsm -> resource://gre/modules/Promise-backend.js:865:23\nthis.PromiseWalker.walkerLoop@reso
01-28 22:06:17.605 I/Gecko   ( 5080): -*- DownloadsAPI.js : Dispatching statechange event. state=stopped

Comment 26

[zhaoyang]

Attachment: newdownloadlog.txt

Comment 27

[zhaoyang]

Attachment: PRdownloadlog.txt

Comment 28

[zhaoyang]

in PRdownloadlog.txt:

9493896[b6a49880]: nsHttpConnection::Close [this=ab956c40 reason=80004004]
9493896[b6a49880]: nsHttpConnection::EndIdleMonitoring [this=ab956c40]
9493896[b6a49880]: Leaving Idle Monitoring Mode [this=ab956c40]
9493896[b6a49880]: nsSocketInputStream::AsyncWait [this=ae9c3dc8]
9493896[b6a49880]: nsSocketInputStream::Read [this=ae9c3dc8 count=4000]
9493896[b6a49880]:   calling PR_Read [count=4000]
9493896[b6a49880]:   PR_Read returned [n=-1]
9493896[b6a49880]: JIMB: ReleaseFD_Locked: mFDref = 2
9493896[b6a49880]: ErrorAccordingToNSPR [in=-5990 out=804b000e]
9493896[b6a49880]: nsSocketTransport::OnMsgInputClosed [this=ae9c3cd0 reason=804b000e]

I find 'reason=804b000e',it just the download failed error:NS_ERROR_NET_TIMEOUT

Comment 29

[Daniel Stenberg [:bagder]]

Yes, I see the 804b000e there. But what _isn't_ there is a logged event saying that the network changed (from nsNotifyAddr) which I would've expected - the question is really why it didn't and it could very well be a clue to this problem. I intend to get this version onto a local device of my own and see what I can dig up. May take a little while.

Comment 30

[zhaoyang]

Hi Daniel,

   Could you update your analysis progress？
   Thanks a lot

Comment 31

[Daniel Stenberg [:bagder]]

I haven't progressed much since, as I've been held up in other things. Can you double check that you had "nsNotifyAddr" as module specified when you asked for a log? Since there was no message from that module something is odd...

Comment 32

[zhaoyang]

Hi Ghislain Aus,
  
   what do you think of this bug,or could you give some guidences ?
 
   Thanks a lot

Comment 33

[Steven Yang [:styang]]

Seems been fixed on partner side. clear the blocking flag.

Comment 34

[Vincent Liu[:vliu]]

Clear the ni? since Comment 33

Comment 35

[Vance Chen [:vchen][skype:vance.lucida][[e-mail]]]

Close this one per Comment#33