484729 - NPN_Write always returns zero no matter what

Status: RESOLVED FIXED

(Core Graveyard :: Plug-ins, defect)

Description

[Gurmeet Kochar]

User-Agent:       Opera/9.63 (Windows NT 5.1; U; en) Presto/2.1.1
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.0.7) Gecko/2009021910 Firefox/3.0.7

I am trying to stream out some HTML (text/html) data from my plugin to Firefox but seems like NPN_Write has a buggy implementation or something as it always gives 0 as the return value, the result of which is that a zero-byte file named "_blank" gets created in the TEMP folder and Firefox then tries to display that file.

Reproducible: Always

Steps to Reproduce:
1. NPN_NewStream(instance, "text/html", "_blank", &pStream);
2. NPN_Write(instance, pStream, streamDataLen, (void*)streamData);
3. NPN_DestroyStream(instance, pStream, NPRES_DONE);
Actual Results:  
NPN_Write returns 0.

Expected Results:  
* negative value in case NPN_Write fails with an error.
* positive value (> 0 I guess) in case NPN_Write succeeds.

OS: Windows XP Home Edition SP3
Gecko Plugin SDK Version: 1.9.0.7
moztools: http://ftp.mozilla.org/pub/mozilla.org/mozilla/libraries/win32/moztools-static.zip
Microsoft Visual Studio: 2008 Professional

Comment 1

[Gurmeet Kochar]

I've noticed that the stream is closed right away ( mOutputStream->Close() ) in the constructor of nsPluginStreamToFile class ( http://mxr.mozilla.org/firefox/source/modules/plugin/base/src/nsPluginInstancePeer.cpp#194 ) which, most probably, is the reason of this bug. While debugging, if the above mentioned statement is skipped, the NPN_Write code works fine. It still returns 0 for some reason but successfully sends the stream data to Firefox, i.e. it just reports the number of bytes copied wrong.

Comment 2

[Jonathan Griffin (:jgriffin)]

Confirming this bug.  nsPluginStreamToFile::Write() always returns NS_OK, instead of the actual number of bytes written, as the documentation states that it should.

Comment 3

[Jonathan Griffin (:jgriffin)]

Attachment: patch

This patch fixes NPN_Write so that it works as documented.  I verified this using the changes to the application/x-test plugin I've been working on as part of bug 517078.

Comment 4

[Josh Aas]

Comment on attachment 401506 [details] [diff] [review]
patch

yikes - patch looks good to me

Comment 5

[Josh Aas]

blocking1.9.2+ as this looks really bad, regression or not

Comment 6

[Johnny Stenback (:jst)]

Josh, I was talking to Jonathan about this on irc and it doesn't seem like this is a regression. I think I'd be happier shipping 3.6 w/o this fix, taking it on the trunk now, and letting it be part of 3.7. No plugins out there depend on this, and few, if any, have complained, so there's IMO not much reason to risk taking this as it's certainly possible that some plugins depend on our current broken behavior.

Comment 7

[Jonathan Griffin (:jgriffin)]

http://hg.mozilla.org/mozilla-central/rev/42b60706dc9c

Comment 8

[Johnny Stenback (:jst)]

This can be marked fixed now, right?

Comment 9

[Jonathan Griffin (:jgriffin)]

Fixed on trunk yes.

Comment 10

[Benjamin Smedberg]

http://hg.mozilla.org/projects/firefox-lorentz/rev/7695f9013530

Comment 11

[Mike Beltzner [:beltzner, not reading bugmail]]

Blanket approval for Lorentz merge to mozilla-1.9.2
a=beltzner for 1.9.2.4 - please make sure to mark status1.9.2:.4-fixed

Comment 12

[Benjamin Smedberg]

Merged into 1.9.2 at http://hg.mozilla.org/releases/mozilla-1.9.2/rev/84ba4d805430