Closed Bug 124006 Opened 23 years ago Closed 22 years ago

Hack for using static libflashplayer.so (Flash 5.0) and Java under gcc-3 system

Categories

(Core Graveyard :: Plug-ins, defect)

Product:
Core Graveyard
Component:
Plug-ins
Platform:
x86
Linux
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
VERIFIED DUPLICATE of bug 154206
Milestone:
mozilla1.2alpha

People

(Reporter: kernd, Assigned: srgchrpv)

References

(
URL
)

Details

Sorry, couldn't find a better place to do this...  Many people have had problems
using the current Macromedia Flash plugin with mozilla when running under a
gcc-3.0.x system.  In addition, I've heard a similar/same problem regarding the
java plugin.

This will affect anyone who is running under the new libstdc++v3, and does not
have multiple gcc's installed.

The problem manifests itself as a browser crash, caused by libflashplayer.so
failing to dynamically load.  The libflashplayer.so binary depends on
__builtin_vec_new and __builtin_vec_delete which has been removed/moved/mangled
(hopefully temporarily) in libstdc++v3.

To fix, it is necessary to make the needed functions available to the plugin.

Start a new file, libc++mem.c, containing the following:
///
void *__builtin_new(int size)
{
   return (void*)malloc(size)
}
void *__builtin_vec_new(int size)
{
   return __builtin_new(size)
}
void *__builtin_delete(void *ptr)
{
   free(ptr);
}
void *__builtin_vec_delete(void *ptr)
{
   __builtin_delete(ptr)
}
///EOF
Compile and link with the following:
gcc -shared -fPIC -o libc++mem.so libc++mem.c

Copy the resulting libc++mem.so file into /usr/local/lib (or other any other
location).

Before starting mozilla with libflashplayer.so installed set
LD_PRELOAD=/usr/local/lib/libc++mem.so

This is obviously a hack, but it works - once Macromedia releases a binary
linked against the new libstdc++, this will be unneeded.

Please forward this to the maintainer of your help tips - there are alot of
people still trying to get this to work who probably don't like the idea of
downgrading to gcc-2.

Thanks,

David L. Kern, Jr.
Lead Developer
Streamedia Productions, LLC
Thanks David, 
I'll try your suggestion.
Assignee: av → serge
confirming
Status: UNCONFIRMED → NEW
Ever confirmed: true
Blocks: 91470
yes, it works, at least I can say it about flash plugin.
No longer blocks: 91470
Target Milestone: --- → mozilla1.0
I have compiled mozilla with gcc-3.0.3 with -O3 and with this hack.

For me, Flash works, but Java plugin from.
ftp://ftp.netscape.com/pub/netscape6/english/6.2/unix/linux22/plugins/jre131i.xpi
fails:
undefined symbols:
__vt_17nsGetServiceByCID
if I create stub function, I get another missing ref:
__vt_16nsQueryInterface
assign_from_helper__13nsCOMPtr_baseRC15nsCOMPtr_helperRC4nsID
_._13nsCOMPtr_base
etc.

undefined symbol: __vt_17nsGetServiceByCID bug 116444
Moving Netscape owned 0.9.9 and 1.0 bugs that don't have an nsbeta1, nsbeta1+,
topembed, topembed+, Mozilla0.9.9+ or Mozilla1.0+ keyword.  Please send any
questions or feedback about this to adt@netscape.com.  You can search for
"Moving bugs not scheduled for a project" to quickly delete this bugmail.
Target Milestone: mozilla1.0 → mozilla1.2
*** Bug 124499 has been marked as a duplicate of this bug. ***
*** Bug 115721 has been marked as a duplicate of this bug. ***
*** Bug 130782 has been marked as a duplicate of this bug. ***
I must be doing something wrong- is that *all* that is needed of this hack file?

I get this error when I try exactly what it says to do:

---------
libc++mem.c: In function `__builtin_new':
libc++mem.c:4: parse error before '}' token
libc++mem.c: In function `__builtin_vec_new':
libc++mem.c:8: parse error before '}' token
libc++mem.c: In function `__builtin_vec_delete':
libc++mem.c:16: parse error before '}' token
---------

Is the file missing some includes or something?
; are missing from line 4, 8 and 16

correct code is :

void *__builtin_new(int size)
{
  return (void*)malloc(size);
}
void *__builtin_vec_new(int size)
{
  return __builtin_new(size),
}
void *__builtin_delete(void *ptr)
{
   free(ptr);
}
void *__builtin_vec_delete(void *ptr)
{
  __builtin_delete(ptr);
}

*** Bug 144977 has been marked as a duplicate of this bug. ***
A bug has been raised with Sun w.r.t. their JRE (Bug Id: 4689085). The problem
is more generic though as it also occurs with Blackdown. I have yet to try the
IBM JRE. I've seen a few places where errors like:

LoadPlugin: failed to initialize shared library
/usr/java/j2sdk1.4.0/jre/plugin/i386/ns610/libjavaplugin_oji140.so
[/usr/java/j2sdk1.4.0/jre/plugin/i386/ns610/libjavaplugin_oji140.so: undefined
symbol: GetGlobalServiceManager__16nsServiceManagerPP17nsIServiceManager]

Are appearing in newgroups and mailing lists. Perhaps a pointer in the mozilla
release notes/faq to gcc incompatabilities when running a browser and plugins
compiled by different gccs would at least give people an idea of what is going on.

Now raised as Bug 4687814 with Sun (Make Java 1.4 plugin work with gcc3 compiled
mozilla). See http://developer.java.sun.com/developer/bugParade/bugs/4687814.html
Sun turned down the request rather bluntly :-( 
'we do not support Mozilla'. Way to go, Sun...
Sun of a gun ;)
I noticed this issue is discussed in bug 116444 as well. I think a dependency
should be added.
Depends on: 116444
*** Bug 145207 has been marked as a duplicate of this bug. ***
*** Bug 150132 has been marked as a duplicate of this bug. ***
This bug is also related to bug #124815 (Evangelism bug to ask Macromedia to
release a Flash plugin which works with recent C++ libraries).
  I can't see an obvious blocks / depends on relationship though.

Same for RealPlayer plugin (but the workaround does not help):

LoadPlugin: failed to initialize shared library /usr/lib/mozilla/plugins/rpnp.so
[/usr/lib/mozilla/plugins/rpnp.so: undefined symbol: __pure_virtual]
Serge: try adding this:

 void __pure_virtual(void) {
    extern void __cxa_pure_virtual(void);

    __cxa_pure_virtual();
  }
It seems to be ok, at least for startup - no error messages. about:plugins shows
the presence of the RealPlayer plugin.
Yeah, it works too.  See also bug 154206, which has a lot of overlap with this
one.  There's a patch there containing these workarounds and one more, almost
ready to land.
Since bug 154206 should land soon for Flash and Real plugins, and we've already
got bug 116444 for Java, I'm gonna mark this as a DUP of 154206.  If this is a
dumb idea, feel free to re-open, etc.

*** This bug has been marked as a duplicate of 154206 ***
Status: NEW → RESOLVED
Closed: 22 years ago
Resolution: --- → DUPLICATE
mass duplicate verifications . For filtering purposes, pls use keywd
"massdupverification"
Status: RESOLVED → VERIFIED
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.