116839 - zero disk cache capacity brokes flush plugin

Status: VERIFIED DUPLICATE of bug 116562

(Core Graveyard :: Plug-ins, defect)

Description

[km]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; SunOS sun4u; en-US; rv:0.9.7) Gecko/20011221
BuildID:    200112204

The Sparc/Solaris shockwave flash plugin (5.0r52) fails on
Milestone 0.97. According to about:plugins it loads correctly.
However when it executes it just shows the white background with
no sign of graphics or sound. On exposure event you can briefly
see the grey background which is quickly filled with empty white.

This does not occur on Milestone 0.96 Sparc/Solaris with the
same plugin.

It does not occur on Linux or Windows Milestone 0.97.

I manually copy libflashplayer.so to plugins directory before
starting mozilla

The URL cited is the macromedia test page.

Ignore the popup for Shockwave player (none for Solaris).
Its the box below it for Shockwave Flash that is of concern.
There is no error popup for flash, just no output.

Reproducible: Always
Steps to Reproduce:
1.As above
2.
3.

Comment 1

[av (gone)]

Over to serge for investigation.

Comment 2

[serge (gone)]

ccing [e-mail]
Sergi, you are done a lot of moziila debugging/coding on solaris,
would you please take a look on this. Thanks.

Comment 3

[Sergi]

Mozilla: Mozilla/5.0 (X11; U; SunOS sun4u; en-US; rv:0.9.7) Gecko/20020125
Workstation: SunOS shiva 5.8 Generic_108528-03 sun4u sparc SUNW,Ultra-1

Works for me. Page contains two plugin's areas, one for
application/x-director mime type and second for application/x-shockwave-flash
mime type. First shows 'default plugin' window (to download plugin for
x-director), second shows flash normally.

Propose to close bug as WFM.

Comment 4

[shrirang khanzode]

thank you,verified !

Comment 5

[km]

I'm still having the problem. Just to be clear I just downloaded

mozilla-sparc-sun-solaris2.8-0.9.7.tar.gz

untarred it, and stuck libflashplayer.so in the plugins directory.
My libflashplayer.so sums as 

     38312 2237 libflashplayer.so

about:plugins shows it as Shockwave Flash 5.0 r52

I then visit http://www.macromedia.com/shockwave/welcome/
and get the same problem results I reported originally

This is on

     SunOS flare 5.8 Generic_108528-11 sun4u sparc SUNW,Ultra-4

I have gotten the same results on several other Sun's and with the
dailys.

Any suggestions about why I have the problem and you don't?

Comment 6

[Sergi]

Do you have ShockwaveFlash.class file in same directory as libflashplayer.so?
It's necessary to run flashes. If copying of this file fix your problem,
I'll close bug.

Comment 7

[serge (gone)]

[e-mail], could you try mv ~/.mozilla/appreg ~/.mozilla/appreg.OLD?

Comment 8

[km]

Removing the appreg solved the problem.

Was there something about shockwave that it was caching
from 0.96 that broke 0.97?

Comment 9

[km]

I was wrong when I said the appreg solved my problem. It happened that when I
deleted the appreg I got a different profile with a different prefs.js.
It was one of my prefs that caused the problem.

Specifically it was

    user_pref("browser.dom.window.dump.enabled", false);

In 0.96 this didn't affect shockwave, and in 0.97 it did.

Comment 10

[km]

Sorry, one (hopefully last) correction. It was not the 
browser.dom.window.dump.enabled preference that caused
the problem. It was

     user_pref("browser.cache.disk.capacity", 0); 

I normally use this to disable disk caching, but apparently that causes
trouble for shockwave in 0.97

Comment 11

[serge (gone)]

Thank you, I can see it now on linux.
Reopening and investigating...

Comment 12

[Sergi]

Looks like we never have OnDataAvailable event in case of zero disk cache
capacity :(
Anyway, bug can be reproduced, switch status to NEW.

Propose to change summary something like this:
"zero disk cache capacity brokes old plugins"

Comment 13

[Sergi]

Now I know what the reason of this bug.

Look at file mozilla/netwerk/cache/src/nsCacheService.cpp

Difference between 0.9.6 and 0.9.7 in method
nsCacheService::EnsureEntryHasDevice(nsCacheEntry * entry)

lines of 0.9.6:

---
      if (!device && mEnableMemoryDevice) {
         NS_ASSERTION(entry->IsAllowedInMemory(), "oops.. bad flags");
---

was changed to line of 0.9.7:

---
    if (!device && mEnableMemoryDevice && entry->IsAllowedInMemory()) {
---

But entry->IsAllowedInMemory() still returns false.

Return to 0.9.6 code, and you'll see flash working :)

Reassign to me, please, I'll provide fix soon.

Comment 14

[shrirang khanzode]

Thx for the finding....this is dup of 116562, i think

*** This bug has been marked as a duplicate of 116562 ***

Comment 15

[shrirang khanzode]

v