16519 - NPP_DestroyStream() invoked too often for 4.0 plugin

Status: VERIFIED FIXED

(Core Graveyard :: Plug-ins, defect, P3)

Description

[Peter Grandmaison]

Using Shockwave, I loaded a movie whose SRC was 24380 bytes long (a local file),
and it aborted the stream before downloading the entire file. Logging the plugin
methods called, I find :

NPP_NewStream ( returns NPERR_NO_ERROR )
NPP_WriteReady ( returns 32678 )
NPP_Write ( offset 0, len 8192, returns 8192 )
NPP_DestroyStream ( movie won't play cause it ain't all there ! )

If I skip handling NPP_DestroyStream() I get

NPP_NewStream ( returns NPERR_NO_ERROR )
NPP_WriteReady ( returns 32678 )
NPP_Write ( offset 0, len 8192, returns 8192 )
NPP_DestroyStream ( do nothing )
NPP_WriteReady ( returns 32678 )
NPP_Write ( offset 8192, len 8192, returns 8192 )
NPP_DestroyStream ( do nothing )
NPP_WriteReady ( returns 32678 )
NPP_Write ( offset 16384, len 8192, returns 8192 )
NPP_DestroyStream ( do nothing )
NPP_WriteReady ( returns 32678 )
NPP_Write ( offset 16384, len 7996, returns 7996 )
NPP_DestroyStream ( does the real stream end and movie plays )

Furthermore, if I alter my NPP_WriteReady() I get

NPP_WriteReady ( returns 8000 )
NPP_Write ( offset 0, len 8000, returns 8000 )
NPP_WriteReady ( returns 8000 )
NPP_Write ( offset 8000, len 192, returns 192 )
NPP_DestroyStream ( altered to do nothing )
...

And so, I think NPP_DestroyStream() is being called when your buffer (8192) is
consumed, in addition to when the stream en

Comment 1

[Eli Goldberg]

Shrirang is now QA owner for Plug-ins; QA assigning all of my Plug-ins bugs over
to him.

Comment 2

[Masatoshi Kimura [:emk]]

This seems to be a side effect of a workaround for the bug 13122.

Comment 3

[Masatoshi Kimura [:emk]]

Attachment: patch: temporary fix for both bug 13122 and this bug

Comment 4

[av (gone)]

Right, that's what I was looking at too. Let me do more testing before I check
this in.

Comment 5

[Masatoshi Kimura [:emk]]

NPP_DestroyStream() is no longer invoked too often,
but I will leave this open since it should be reviewd
when AsFileOnly Plugin support arrives.

Comment 6

[av (gone)]

Latering

Comment 7

[Masatoshi Kimura [:emk]]

Reopening since disk cache has arrived, and marking dup of 23667.

Comment 8

[Masatoshi Kimura [:emk]]

*** This bug has been marked as a duplicate of 23667 ***

Comment 9

[Masatoshi Kimura [:emk]]

Reopening and adding dependency to bug 23667. See my comment in 23667.

Comment 10

[David Krause]

Mass moving old milestone bugs to M16

Comment 11

[Mike]

M16 has been out for a while now, these bugs target milestones need to be 
updated.

Comment 12

[sean echevarria]

Peter, can you check this bug against a recent build to see if it is still valid?

Comment 13

[ekrock's old account (dead)]

Not a Netscape 6 RTM blocker. FUTURE. This bug has been marked Future because
the Netscape engineer it is assigned to is overburdened.

Comment 14

[Peter Grandmaison]

From the original comment "... aborted the stream before downloading the 
entire ...". This was fixed a long time ago (summer) ... clearly streams are 
downloading fully or none of the content would play.

Comment 15

[Peter Lubczynski]

This bug has been FIXED (probably a long time ago). NPP_DestroyStream should be 
called correctly now or 4.x plugins wouldn't be working.

Comment 16

[shrirang khanzode]

marking verif based on earlier comments.

Comment 17

[Pulsebot]

Pushed by ffxbld-merge:
https://hg.mozilla.org/releases/mozilla-beta/rev/a6375d1f6883
[fenix] For https://github.com/mozilla-mobile/fenix/issues/16519: Add space above Delete History button (https://github.com/mozilla-mobile/fenix/pull/17492)