macromedia players flash: Bug Report

Summary : Flash runs for a short time but almost inevitably segfaults.

Browsers affected : Firefox 1.5.x, 2.0.0.1, Mozilla and Seamonkey. Opera is
unsupported, but its behaviour may be instructive and is described below.

System Hardware : CPU = 2 x PIII smp. 768 MB ram. Graphics Card = nVidia
GeForce FX 5500.

Kernel: vanilla 2.6.19.2 (previously 2.6.15)
X-Windows : Xorg 6.9.0
nVidia driver 1.0.9631 (current)

Libraries : I will supply details of the precise library versions to
libflashplayer.so is linked if that will assist.

Distro: Nominally Slackware 9.1 (has been very extensively updated). glibc =
2.3.2. (Users of the more recent Slackware 10.2 are also reporting crashes on
alt.os.linux.slackware).

Details: Both the beta and the release almost inevitably terminate in a
segfault under all Mozilla.org browsers. Using youtube.com as a test site,
videos load and run normally, including sound, for a few seconds (usually
between 5 and 30 seconds), but almost inevitably segfault. Once or twice in at
least 30 attempts display of a video has completed. Until the segfault arises
all flash controls work normally. Very simple flash advertising content will
sometimes play and loop without causing the segfault. I have created a test
user with his own copy of the browsers and no extensions / themes other than
libnullplugin.so and libflashplayer.so. The segfaults occur just as in the
case of my own user.

For what it may be worth, when Opera is run from a terminal one sees :

(process:4982): GLib-GObject-CRITICAL **: gtype.c:2242: initialization
assertion failed, use g_type_init() prior to this function

(process:4982): Gtk-CRITICAL **: gtk_clipboard_get_for_display: assertion
`GDK_IS_DISPLAY (display)' failed
Adobe FlashPlayer: gtk_clipboard_get(GDK_SELECTION_PRIMARY); failed. Trying to
call gtk_init(0,0);

The above errors do not cause any apparent problems. After the usual few
seconds of play on youtube.com, however, the flash screen goes grey with the
following error output to the terminal:

opera: Plug-in 4982 is not responding. It will be closed.
opera: Define environment variable OPERA_KEEP_BLOCKED_PLUGIN to keep blocked
plug-ins.

It would appear that Opera continues to load the flash content even after
this, but nothing is displayed.

I experience the same thing. I have found Flash 9 Seg Faults consistently and
predictably whenever a window (or tab) exits/closes that has Flash 9 running on
it. There is one curious exception. After running certain Flash animations
once (I am still trying to document a specific example) then Flash 9 no longer
Seg Faults.

Slackware 10.2
Kernel 2.6.16
Firefox 2.0

Flash 9 (for Linux) in Firefox Seg Faults every time the window running flash
is closed, or example, when following a link which replaces the page running
Flash. This behavior is consistent and reproducible. It has exhibited this
behavior since version 8.

The following is the tail end results of running strace on Firefox 2.0.0.1
with Flash 9. www.adobe.com was accessed and the menu bar clicked so that the
window would close and be replaced by the linked page. Note that it was an
munmap call that threw the Segmentation Fault. As noted in
http://www.opengroup.org/onlinepubs/007908799/xsh/munmap.html, after releasing
the memory mapping "Further references to these pages result in the generation
of a SIGSEGV signal to the process."

...

read(3, "\1\30\7\354\20\1\0\0\0\0\0\0O\377\16\10\36\0\0\0`\367\263"..., 32) =
32
readv(3, [{"\377\377\377\0\377\377\377\0\377\377\377\0\377\377\377"..., 1088},
{"", 0}], 2 ) = 1088
write(3, "H\2\26\1\5#\340\0026\7\340\2\21\0\20\0\26\4\7\0\0\30\6"..., 2492) =
2492
ioctl(3, FIONREAD, [0]) = 0
poll([{fd=3, events=POLLIN}, {fd=10, events=POLLIN}, {fd=14,
events=POLLIN|POLLPRI}, {fd=1 6, events=POLLIN|POLLPRI}, {fd=17,
events=POLLIN|POLLPRI}, {fd=18, events=POLLIN|POLLPRI}, {fd=5, events=POLLIN,
revents=POLLIN}], 7, -1) = 1
futex(0x9573860, FUTEX_WAKE, 1) = 1
futex(0x957385c, FUTEX_WAKE, 1) = 1
futex(0x9573860, FUTEX_WAKE, 1) = 1
futex(0x957385c, FUTEX_WAKE, 1) = 1
futex(0x9573860, FUTEX_WAKE, 1) = 1
futex(0x957385c, FUTEX_WAKE, 1) = 1
munmap(0xb0876000, 16777216) = 0
munmap(0xb1876000, 7854972) = 0
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
rt_sigaction(SIGSEGV, NULL, {0xb622be14, [], 0}, 8) = 0
times({tms_utime=1582, tms_stime=34, tms_cutime=0, tms_cstime=0}) = 1718733705
rt_sigaction(SIGSEGV, {0x8082db6, ~[KILL STOP RTMIN RT_1], 0}, NULL, 8) = 0
tgkill(8127, 8127, SIGSEGV) = 0
sigreturn() = ? (mask now [])
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
unlink("/home/pcrichtn/.mozilla/firefox/default.fwz/lock") = 0
rt_sigaction(SIGSEGV, {SIG_DFL}, NULL, 8) = 0
rt_sigprocmask(SIG_UNBLOCK, [SEGV], NULL, 8) = 0
tgkill(8127, 8127, SIGSEGV) = 0
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
+++ killed by SIGSEGV +++

Replying to my own reply ... is that like talking to myself?

Here's a twist to the window closing segmentation fault - It appears it only
happens if it is the last window Flash 9 window to close. In other words, it
seems to run as long as there is another window/tab still open.

This maybe related to the other bug reported with Flash showing only on the top.
It was mentioned this might be a graphics card problem and not a flash problem.

Not sure yet, hoping someone can figure it out.