From gb.public at free.fr Sun Nov 2 16:41:58 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Sun, 2 Nov 2008 17:41:58 +0100 Subject: [PATCH] Fix XEMBED support Message-ID: <9C02C5CA-A609-43FC-960C-4710D783B9E0@free.fr> Hi, Here is a patch I committed to trunk and that fixes Ketan's problem, among others. I will backport it to 1.0-branch if it survives more testing. Actually, this patch workarounds Gtk and Firefox bugs. What happened was that Firefox sometimes killed the plugin window indirectly, without calling NPP_Destroy() or NPP_SetWindow(). Why? nspluginwrapper creates a new GtkPlug and GtkSocket in another process. As such, the GtkSocket in the browser gets notified of the new window and creates a GDK_WINDOW_FOREIGN window and adds it its hierarchy. When the page is reloaded, and if DoStopPlugin() is called for a delayed stop, the whole frame is destroyed. g_object_unref()'ing the MozDrawingArea will nuke all the children. And because the plugin window (seen in the browser) is a GDK_WINDOW_FOREIGN, it's also destroyed and the underlying X window reparented to anything. On the viewer side, this is equivalent to killing it. That's why you also saw errors (failed assertions for GTK_IS_WIDGET() et al.) in the viewer during an NPP_Destroy(). Flash then gets confused because Firefox killed its window. GtkSocket::"plug_removed" comes too late since the windows are destroyed at the Gdk level without any possible notification. Here is the interesting trace (Firefox side): #0 IA__gdk_window_reparent (window=0x170bb10, new_parent=0x0, x=0, y=0) at gdkwindow-x11.c:1860 #1 0x00007f9024959ca5 in _gdk_windowing_window_destroy_foreign (window=0x170bb10) at gdkwindow-x11.c:1178 #2 0x00007f902492aa95 in _gdk_window_destroy_hierarchy (window=0x170bb10, recursing=1, foreign_destroy=0) at gdkwindow.c:390 #3 0x00007f902492a9b1 in _gdk_window_destroy_hierarchy (window=0x170b9f0, recursing=1, foreign_destroy=0) at gdkwindow.c:435 #4 0x00007f902492a9b1 in _gdk_window_destroy_hierarchy (window=0x170b840, recursing=0, foreign_destroy=0) at gdkwindow.c:435 #5 0x00007f902492aabd in IA__gdk_window_destroy (window=0x170bb10) at gdkwindow.c:487 #6 0x00007f9028fdc8aa in moz_drawingarea_finalize (object=0x1a57f90) at mozdrawingarea.c:192 #7 0x00007f9025fb21d8 in g_object_unref () from /usr/lib64/ libgobject-2.0.so.0 #8 0x00007f9028fed4d3 in nsWindow::Destroy (this=0x16355a0) at nsWindow.cpp:497 #9 0x00007f9028bd06be in ~nsView (this=0x1ab1960) at nsView.cpp:272 #10 0x00007f90288feab4 in nsFrame::Destroy (this=0x1ab28c0) at nsFrame.cpp:516 [...] * Gtk problem in _gdk_window_destroy_hierarchy(). Actually, as the comment mentions: gdkwindow.c: if (GDK_WINDOW_TYPE (window) == GDK_WINDOW_FOREIGN && ! foreign_destroy) { /* Logically, it probably makes more sense to send * a "destroy yourself" message to the foreign window * whether or not it's in our heirarchy; but for historical * reasons, we only send "destroy yourself" messages to * foreign windows in our heirarchy. */ if (private->parent) _gdk_windowing_window_destroy_foreign (window); [...] } I believe we wouldn't run into problems if the "logical" way was implemented. ;-) * Firefox problem in his NPP_SetWindow() implementation. ns4xPluginInstance.cpp: // XXX 4.x plugins don't want a SetWindow(NULL). if (!window || !mStarted) return NS_OK; By stepping through DoStopPlugin(), the plugin was actually given a chance to destroy its windows through NPP_SetWindow(instance, NULL) or NPP_SetWindow(instance, &(NPWindow){ .window = NULL }). However, since NPP_SetWindow() actually ignores thos cases... For Firefox, I believe it is safe to allow NULL windows in NPP_SetWindow() for plugins supporting at least the XEMBED protocol. They are recent enough and people should actually follow the NPAPI specs that "now" have this clearly defined. IIRC, WebKit does that for some (if not all) plugins too. * A workaround at the nspluginwrapper level. Since we have to live with those bugs for "historical" reasons, I chose an alternative. The idea was to break the GtkPlug/GtkSocket chain. On the browser side, we act as if the plugin doesn't support XEMBED. A GtkXtBin is then created and its XID exposed through NPP_SetWindow(). On the viewer side, a "fake" GtkWindow (GTK_WINDOW_POPUP) is created instead of the GtkPlug to serve as the toplevel. The window is then reparented as we would in the Xt path. Note that this somewhat re-introduces the Debian bug #435912 for (stolen key events, for the record). However, that was also Flash Player's fault. I checked *released* Flash Plugins (9.0.124 and 10.0.12.36) and the behaviour is correct. However it's not with an earlier Flash 10 beta or RC and even 9.0.48 (I don't remember whether that one was a beta or not). I checked other Gtk2 plugins and the following were fine too: Adobe Reader 7 and 8, mplayerplugin. I think I will explicitly reject Flash Player 9.0.48 plugin. People who still want to use Flash9 should move to the latest stable release of that branch (9.0.124, or 9.0.125?). Now, the patch in attachment. Regards, Gwenol?. -------------- next part -------------- A non-text attachment was scrubbed... Name: nspluginwrapper-xembed.patch Type: application/octet-stream Size: 3482 bytes Desc: not available URL: -------------- next part -------------- From mtrainer at central-data.net Mon Nov 3 03:31:34 2008 From: mtrainer at central-data.net (Murray Trainer) Date: Mon, 3 Nov 2008 11:31:34 +0800 (GMT+08:00) Subject: Citrix ICA plugin on Ubuntu Hardy In-Reply-To: <9915203.815211225430781775.JavaMail.root@mailstore01.gopc.net> Message-ID: <4738660.828611225683094610.JavaMail.root@mailstore01.gopc.net> Hi, I have just installed nspluginwrapper 1.1.2 from here: deb http://ppa.launchpad.net/psyke83/ubuntu hardy main and I still don't get the Citrix plugin in my list of plugins in Firefox 3.0.3 after re-installing the Citrix /usr/lib/ICAclient/npica.so plugin with nspluginwrapper. Does anyone have this working on Ubuntu Hardy 64-bit? Thanks Murray ----- Forwarded Message ----- From: "Murray Trainer" To: "NSPluginWrapper List" Sent: Friday, 31 October, 2008 2:26:21 PM GMT +08:00 Perth Subject: Citrix ICA plugin on Ubuntu Hardy Hi, I had the Citrix ICA plugin working fine on Ubuntu Gutsy (7.10) 64-bit using Firefox 2 and the default nspluginwrapper on Gutsy (0.9.x from memory). Now I have upgraded to Ubuntu Hardy (8.04) 64-bit and Firefox 3.0.3. The ICA plugin installs and lists fine with nspluginwrapper but doesn't appear in the about:plugins list of plugins in Firefox 3.0.3. I have tried nspluginwrapper 0.9.91 and 1.0.0-1 with the same problem. Is this a known issue with Firefox 3? Any suggestions on a fix or degugging the problem? Thanks Murray Trainer -------------- next part -------------- An HTML attachment was scrubbed... URL: From gb.public at free.fr Mon Nov 3 05:39:15 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Mon, 3 Nov 2008 06:39:15 +0100 Subject: Citrix ICA plugin on Ubuntu Hardy In-Reply-To: <4738660.828611225683094610.JavaMail.root@mailstore01.gopc.net> References: <4738660.828611225683094610.JavaMail.root@mailstore01.gopc.net> Message-ID: Hi, Le 3 nov. 08 ? 04:31, Murray Trainer a ?crit : > and I still don't get the Citrix plugin in my list of plugins in > Firefox 3.0.3 > after re-installing the Citrix /usr/lib/ICAclient/npica.so plugin > with nspluginwrapper. > Does anyone have this working on Ubuntu Hardy 64-bit? I have just installed the ICA Client 10.6 and it lists fine in Firefox 3.0.3. If `nspluginwrapper' lists the plugin, then there is no reason it wouldn't list in Firefox too. Well, the difference is probably Firefox is NP_Initialize()'ing it and you are missing a symbol? Please report the output of: $ nspluginwrapper -l $ firefox ; # then about:plugins i.e. run firefox from the command line and report if you see any error from there after you listed the plugins with about:plugins Regards, Gwenol?. From mtrainer at central-data.net Mon Nov 3 06:15:46 2008 From: mtrainer at central-data.net (mtrainer at central-data.net) Date: Mon, 3 Nov 2008 14:15:46 +0800 (GMT+08:00) Subject: Citrix ICA plugin on Ubuntu Hardy In-Reply-To: <12153069.830931225692702171.JavaMail.root@mailstore01.gopc.net> Message-ID: <32110685.830971225692945981.JavaMail.root@mailstore01.gopc.net> > and I still don't get the Citrix plugin in my list of plugins in > Firefox 3.0.3 > after re-installing the Citrix /usr/lib/ICAclient/npica.so plugin > with nspluginwrapper. > Does anyone have this working on Ubuntu Hardy 64-bit? I have just installed the ICA Client 10.6 and it lists fine in Firefox 3.0.3. If `nspluginwrapper' lists the plugin, then there is no reason it wouldn't list in Firefox too. Well, the difference is probably Firefox is NP_Initialize()'ing it and you are missing a symbol? Please report the output of: $ nspluginwrapper -l $ firefox ; # then about:plugins i.e. run firefox from the command line and report if you see any error from there after you listed the plugins with about:plugins Regards, Gwenol?. Hi Gwenole, Output below. Thanks Murray $ nspluginwrapper -l /usr/lib/mozilla/plugins/npwrapper.libflashplayer.so Original plugin: /usr/lib/mozilla/plugins/libflashplayer.so Wrapper version string: 1.1.2 /usr/lib/mozilla/plugins/npwrapper.npica.so Original plugin: /usr/lib/ICAClient/npica.so Wrapper version string: 1.1.2 /usr/lib/mozilla/plugins/nppdf.so Original plugin: /usr/lib32/Adobe/Reader8/Browser/intellinux/nppdf.so Wrapper version string: 1.1.2 /usr/lib64/mozilla/plugins/npwrapper.libflashplayer.so Original plugin: /usr/lib/mozilla/plugins/libflashplayer.so Wrapper version string: 1.1.2 /usr/lib64/mozilla/plugins/npwrapper.npica.so Original plugin: /usr/lib/ICAClient/npica.so Wrapper version string: 1.1.2 /usr/lib64/mozilla/plugins/nppdf.so Original plugin: /usr/lib32/Adobe/Reader8/Browser/intellinux/nppdf.so Wrapper version string: 1.1.2 /usr/lib/firefox/plugins/npwrapper.libflashplayer.so Original plugin: /usr/lib/mozilla/plugins/libflashplayer.so Wrapper version string: 1.1.2 /usr/lib/firefox/plugins/npwrapper.npica.so Original plugin: /usr/lib/ICAClient/npica.so Wrapper version string: 1.1.2 /usr/lib/firefox/plugins/nppdf.so Original plugin: /usr/lib32/Adobe/Reader8/Browser/intellinux/nppdf.so Wrapper version string: 1.1.2 /usr/lib64/firefox/plugins/npwrapper.libflashplayer.so Original plugin: /usr/lib/mozilla/plugins/libflashplayer.so Wrapper version string: 1.1.2 /usr/lib64/firefox/plugins/npwrapper.npica.so Original plugin: /usr/lib/ICAClient/npica.so Wrapper version string: 1.1.2 /usr/lib64/firefox/plugins/nppdf.so Original plugin: /usr/lib32/Adobe/Reader8/Browser/intellinux/nppdf.so Wrapper version string: 1.1.2 $ firefox ERROR: ld.so: object '/usr/$LIB/libartsdsp.so.0' from LD_PRELOAD cannot be preloaded: ignored. GCJ PLUGIN: thread 0x621e20: NP_GetMIMEDescription GCJ PLUGIN: thread 0x621e20: NP_GetMIMEDescription return GCJ PLUGIN: thread 0x621e20: NP_GetValue GCJ PLUGIN: thread 0x621e20: NP_GetValue: returning plugin name. GCJ PLUGIN: thread 0x621e20: NP_GetValue return GCJ PLUGIN: thread 0x621e20: NP_GetValue GCJ PLUGIN: thread 0x621e20: NP_GetValue: returning plugin description. GCJ PLUGIN: thread 0x621e20: NP_GetValue return _______________________________________________ Nspluginwrapper-devel-list mailing list Nspluginwrapper-devel-list at redhat.com https://www.redhat.com/mailman/listinfo/nspluginwrapper-devel-list From gb.public at free.fr Mon Nov 3 23:17:28 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Tue, 4 Nov 2008 00:17:28 +0100 Subject: [PATCH] Fix for invalid NPP_Write() calls Message-ID: <17F621F0-0A46-4CE1-ACF0-CE2BEA72AB2A@free.fr> Hi, The following patch fixes a problem that could happen with non-Mozilla browsers like Konqueror and Google Chrome. Interestingly, WebKit really mimics the Mozilla behaviour. A plugin that exhibits this problem is DiamondX. I have yet to find a "real world" plugin showing off that problem though. I noticed this while debugging XEMBED issues with Konqueror and since it doesn't hurt after all, I have committed it. Commit to 1.0-branch is also planned. -------------- next part -------------- A non-text attachment was scrubbed... Name: nspluginwrapper-fix-invalid-NPP_Write-calls.patch Type: application/octet-stream Size: 1600 bytes Desc: not available URL: -------------- next part -------------- From gb.public at free.fr Mon Nov 3 23:23:09 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Tue, 4 Nov 2008 00:23:09 +0100 Subject: [PATCH] Fix XEMBED support (take 2) In-Reply-To: <9C02C5CA-A609-43FC-960C-4710D783B9E0@free.fr> References: <9C02C5CA-A609-43FC-960C-4710D783B9E0@free.fr> Message-ID: Hi, > * A workaround at the nspluginwrapper level. > > Since we have to live with those bugs for "historical" reasons, I > chose an alternative. The idea was to break the GtkPlug/GtkSocket > chain. On the browser side, we act as if the plugin doesn't support > XEMBED. A GtkXtBin is then created and its XID exposed through > NPP_SetWindow(). On the viewer side, a "fake" GtkWindow > (GTK_WINDOW_POPUP) is created instead of the GtkPlug to serve as the > toplevel. The window is then reparented as we would in the Xt path. That strategy fails with Konqueror. The new rule is to do this trick only for Gtk2-based browsers, assuming they are exposing a GtkSocket, which is the case for Mozilla (obviously) and WebKit. Otherwise, the GtkPlug/GtkSocket trick is used as before. I have attached the cumulative patch. -------------- next part -------------- A non-text attachment was scrubbed... Name: nspluginwrapper-xembed.patch Type: application/octet-stream Size: 6361 bytes Desc: not available URL: -------------- next part -------------- From mtrainer at central-data.net Tue Nov 4 05:14:49 2008 From: mtrainer at central-data.net (Murray Trainer) Date: Tue, 4 Nov 2008 13:14:49 +0800 (GMT+08:00) Subject: Citrix ICA plugin on Ubuntu Hardy In-Reply-To: <32110685.830971225692945981.JavaMail.root@mailstore01.gopc.net> Message-ID: <22392514.843821225775689041.JavaMail.root@mailstore01.gopc.net> ----- Original Message ----- From: mtrainer at central-data.net To: "nspluginwrapper development discussion list" Sent: Monday, 3 November, 2008 3:15:46 PM GMT +08:00 Perth Subject: Re: Citrix ICA plugin on Ubuntu Hardy > and I still don't get the Citrix plugin in my list of plugins in > Firefox 3.0.3 > after re-installing the Citrix /usr/lib/ICAclient/npica.so plugin > with nspluginwrapper. > Does anyone have this working on Ubuntu Hardy 64-bit? I have just installed the ICA Client 10.6 and it lists fine in Firefox 3.0.3. If `nspluginwrapper' lists the plugin, then there is no reason it wouldn't list in Firefox too. Well, the difference is probably Firefox is NP_Initialize()'ing it and you are missing a symbol? Please report the output of: $ nspluginwrapper -l $ firefox ; # then about:plugins i.e. run firefox from the command line and report if you see any error from there after you listed the plugins with about:plugins Regards, Gwenol?. Hi Gwenole, I figured out the problem. It was some missing links in /usr/lib/firefox-3.0.3/plugins. I did the following and the citrix and nsdejavu plugins now appear and work. Thanks for you help. Regards Murray Trainer From gb.public at free.fr Tue Nov 4 23:47:03 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Wed, 5 Nov 2008 00:47:03 +0100 Subject: nspluginwrapper-1.1.2 invisible video In-Reply-To: <48FBBFAB.6070003@redhat.com> References: <48FBBFAB.6070003@redhat.com> Message-ID: <992B1333-47C1-4613-BCB4-FC24A4D76741@free.fr> Hi, Le 20 oct. 08 ? 01:15, Warren Togami a ?crit : > flash-plugin-10.0.12.36-release.i386 > nspluginwrapper-1.1.2-1.fc9.i386 > nspluginwrapper-1.1.2-1.fc9.x86_64 > firefox-3.0.2-1.fc9.x86_64 > > http://www.theonion.com/content/video/john_mccain_accidentally_left_on > If you reload a few times (or sometimes the first time), it somehow > gets into a state where you can hear the audio playing, but there is > a blank white area where the Flash video should be. Right-clicking > the area where the video should be shows an outline of the video, > but no menus or anything. Reloading the page fails to recover the > video view. Finally, I think this is independent from nspluginwrapper. I was just looking at how much CPU nspluginwrapper+Firefox is taking against Firefox alone[*] and I noticed the same thing happens with a standard 32-bit Firefox too. Both builds are of the same version- release (from openSUSE 11.0 updates) but different architectures (i586, x86_64). Your link is a good one for testing and exhibits this behaviour easily. [*] BTW, has anyone noticed nspluginwrapper taking 100% CPU and that is not caused by the plugin? I mean, I saw (many?) people complaining that nspluginwrapper is using the whole CPU but why don't they realize that now that the plugin runs out of the browser process, it thus also accounts for the actual plugin needs? e.g.: - 32-bit firefox alone: 25 to 30% on my system with the above- mentionned link - 64-bit firefox + nspluginwrapper: 5% for firefox and 20 to 26% for nspluginwrapper (rough `top d 1' measures with the implied inacuracy but relative comparison can hold). I don't do busy wait, so I don't see how nspluginwrapper could consume that much CPU itself. Regards, Gwenole. From wtogami at redhat.com Tue Nov 4 23:57:19 2008 From: wtogami at redhat.com (Warren Togami) Date: Tue, 04 Nov 2008 18:57:19 -0500 Subject: nspluginwrapper-1.1.2 invisible video In-Reply-To: <992B1333-47C1-4613-BCB4-FC24A4D76741@free.fr> References: <48FBBFAB.6070003@redhat.com> <992B1333-47C1-4613-BCB4-FC24A4D76741@free.fr> Message-ID: <4910E15F.4000700@redhat.com> Gwenole Beauchesne wrote: > Hi, > > Le 20 oct. 08 ? 01:15, Warren Togami a ?crit : > >> flash-plugin-10.0.12.36-release.i386 >> nspluginwrapper-1.1.2-1.fc9.i386 >> nspluginwrapper-1.1.2-1.fc9.x86_64 >> firefox-3.0.2-1.fc9.x86_64 >> >> http://www.theonion.com/content/video/john_mccain_accidentally_left_on >> If you reload a few times (or sometimes the first time), it somehow >> gets into a state where you can hear the audio playing, but there is a >> blank white area where the Flash video should be. Right-clicking the >> area where the video should be shows an outline of the video, but no >> menus or anything. Reloading the page fails to recover the video view. > > Finally, I think this is independent from nspluginwrapper. > > I was just looking at how much CPU nspluginwrapper+Firefox is taking > against Firefox alone[*] and I noticed the same thing happens with a > standard 32-bit Firefox too. Both builds are of the same version-release > (from openSUSE 11.0 updates) but different architectures (i586, x86_64). > Your link is a good one for testing and exhibits this behaviour easily. > Are you suggesting this is a bug in either Flash or Firefox? Warren From gb.public at free.fr Wed Nov 5 06:11:00 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Wed, 5 Nov 2008 07:11:00 +0100 Subject: nspluginwrapper-1.1.2 invisible video In-Reply-To: <4910E15F.4000700@redhat.com> References: <48FBBFAB.6070003@redhat.com> <992B1333-47C1-4613-BCB4-FC24A4D76741@free.fr> <4910E15F.4000700@redhat.com> Message-ID: <49FC3F98-F718-4082-B71F-88B0BD3BD1EC@free.fr> Le 5 nov. 08 ? 00:57, Warren Togami a ?crit : >> Finally, I think this is independent from nspluginwrapper. >> I was just looking at how much CPU nspluginwrapper+Firefox is >> taking against Firefox alone[*] and I noticed the same thing >> happens with a standard 32-bit Firefox too. Both builds are of the >> same version-release (from openSUSE 11.0 updates) but different >> architectures (i586, x86_64). Your link is a good one for testing >> and exhibits this behaviour easily. > > Are you suggesting this is a bug in either Flash or Firefox? Yes, though I don't know which one yet. However, I know Firefox is sending the stream and Flash consumes it. It could be either Flash is not writing back to the offscreen buffer (NPP_SetWindow() and NPP_HandleEvent()[GraphicsExpose] sequences happen though), or Firefox is not merging back that offscreen buffer (NPN_InvalidateRect() happens too). Some things to examine: - The contents of the pixmap XID - Or whether Flash is actually writing to the pixmap (e.g. intercepting XShmPutImage()?) Regards, Gwenole. From alexeskin at yahoo.com Thu Nov 6 01:54:09 2008 From: alexeskin at yahoo.com (Alex Eskin) Date: Wed, 5 Nov 2008 17:54:09 -0800 (PST) Subject: Failure on some streams (no crash) Message-ID: <560429.27986.qm@web51703.mail.re2.yahoo.com> (Sorry if this gets posted twice.) On some sites the flash video stream does not play, even though there is no crash. The same stream works with a native flash plugin. The only test case I have at the moment is www.cbs.com/video (which is AFAIK not accessible outside the US). What does not work: fedora 9 with all the latest updates on x86_64, i.e. nspluginwrapper-1.1.2-2.fc9.x86_64 nspluginwrapper-1.1.2-2.fc9.i386 firefox-3.0.2-1.fc9.x86_64 flash-plugin-9.0.124.0-release.i386 libflashsupport-000-0.5.svn20070904.x86_64 libflashsupport-000-0.5.svn20070904.i386 If I install flash-plugin-10 and remove libflashsupport, I get exactly the same symptoms. What does work: ubuntu 8.04 with all the updates on i386. This means firefox 3.02 (or 3.03), flash-plugin-9.0.124, and the plugin is run natively. What can I do to track this down? Thanks, Alex From mtrainer at central-data.net Thu Nov 6 02:04:43 2008 From: mtrainer at central-data.net (Murray Trainer) Date: Thu, 6 Nov 2008 10:04:43 +0800 (GMT+08:00) Subject: Failure on some streams (no crash) In-Reply-To: <560429.27986.qm@web51703.mail.re2.yahoo.com> Message-ID: <17749567.862551225937083587.JavaMail.root@mailstore01.gopc.net> ----- Original Message ----- From: "Alex Eskin" To: nspluginwrapper-devel-list at redhat.com Sent: Thursday, 6 November, 2008 10:54:09 AM GMT +08:00 Perth Subject: Failure on some streams (no crash) (Sorry if this gets posted twice.) On some sites the flash video stream does not play, even though there is no crash. The same stream works with a native flash plugin. The only test case I have at the moment is www.cbs.com/video (which is AFAIK not accessible outside the US). What does not work: fedora 9 with all the latest updates on x86_64, i.e. nspluginwrapper-1.1.2-2.fc9.x86_64 nspluginwrapper-1.1.2-2.fc9.i386 firefox-3.0.2-1.fc9.x86_64 flash-plugin-9.0.124.0-release.i386 libflashsupport-000-0.5.svn20070904.x86_64 libflashsupport-000-0.5.svn20070904.i386 If I install flash-plugin-10 and remove libflashsupport, I get exactly the same symptoms. What does work: ubuntu 8.04 with all the updates on i386. This means firefox 3.02 (or 3.03), flash-plugin-9.0.124, and the plugin is run natively. What can I do to track this down? Thanks, Alex ----------- Hi Alex, The info below worked pretty well for me ... http://ubuntuforums.org/showthread.php?t=954100&highlight=10.0.12.36-1hardy2+flash Regards Murray From alexeskin at yahoo.com Thu Nov 6 03:32:25 2008 From: alexeskin at yahoo.com (Alex Eskin) Date: Wed, 5 Nov 2008 19:32:25 -0800 (PST) Subject: Failure on some streams (no crash) In-Reply-To: <17749567.862551225937083587.JavaMail.root@mailstore01.gopc.net> Message-ID: <583445.36683.qm@web51706.mail.re2.yahoo.com> --- On Wed, 11/5/08, Murray Trainer wrote: > > Hi Alex, > > The info below worked pretty well for me ... > > http://ubuntuforums.org/showthread.php?t=954100&highlight=10.0.12.36-1hardy2+flash > I am curious: can you play streams from www.cbs.com/video? At the moment most sites work for me (e.g. youtube, www.hulu.com, etc). There is something special about www.cbs.com/video which causes problems for me with the wrapped plugin. Thanks, Alex From mtrainer at central-data.net Thu Nov 6 04:33:58 2008 From: mtrainer at central-data.net (Murray Trainer) Date: Thu, 6 Nov 2008 12:33:58 +0800 (GMT+08:00) Subject: Failure on some streams (no crash) In-Reply-To: <583445.36683.qm@web51706.mail.re2.yahoo.com> Message-ID: <27397619.865361225946038765.JavaMail.root@mailstore01.gopc.net> > > Hi Alex, > > The info below worked pretty well for me ... > > http://ubuntuforums.org/showthread.php?t=954100&highlight=10.0.12.36-1hardy2+flash > I am curious: can you play streams from www.cbs.com/video? At the moment most sites work for me (e.g. youtube, www.hulu.com, etc). There is something special about www.cbs.com/video which causes problems for me with the wrapped plugin. Thanks, Alex ----------- Hi Alex, It appeared to load flash OK but the videos just won't play. I just get a black box. Murray _______________________________________________ Nspluginwrapper-devel-list mailing list Nspluginwrapper-devel-list at redhat.com https://www.redhat.com/mailman/listinfo/nspluginwrapper-devel-list From alexeskin at yahoo.com Thu Nov 6 05:26:42 2008 From: alexeskin at yahoo.com (Alex Eskin) Date: Wed, 5 Nov 2008 21:26:42 -0800 (PST) Subject: Failure on some streams (no crash) In-Reply-To: <27397619.865361225946038765.JavaMail.root@mailstore01.gopc.net> Message-ID: <737157.29433.qm@web51712.mail.re2.yahoo.com> --- On Wed, 11/5/08, Murray Trainer wrote: > > Hi Alex, > > It appeared to load flash OK but the videos just won't > play. I just get a black box. > > Murray > > Yes, this is exactly what I am getting on x86_64 with the plugin wrapped. On x86 with the flash plugin native, the videos play with no problem. Thanks, Alex > > > _______________________________________________ > Nspluginwrapper-devel-list mailing list > Nspluginwrapper-devel-list at redhat.com > https://www.redhat.com/mailman/listinfo/nspluginwrapper-devel-list From gb.public at free.fr Thu Nov 6 08:16:32 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Thu, 6 Nov 2008 09:16:32 +0100 (CET) Subject: Failure on some streams (no crash) In-Reply-To: <3838990.1336171225959346072.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <31597003.1336351225959392986.JavaMail.root@spooler5-g27.priv.proxad.net> Hi, ----- "Alex Eskin" a ?crit : > On some sites the flash video stream does not play, even though there > is no crash. The same stream works with a native flash plugin. > The only test case I have at the moment is www.cbs.com/video Since you are using Firefox 3, could you do either of these? (i) Try with Firefox 2 (ii) Try with current SVN tree, which will be 1.1.4 tonight or tomorrow: (don't forget to re-wrap the plugin) Thanks, Gwenole. From alexeskin at yahoo.com Thu Nov 6 23:29:41 2008 From: alexeskin at yahoo.com (Alex Eskin) Date: Thu, 6 Nov 2008 15:29:41 -0800 (PST) Subject: Failure on some streams (no crash) Message-ID: <473576.26005.qm@web51704.mail.re2.yahoo.com> I got today's svn version to build (after mucking around with the fedora patches a bit), but I just realized that I cannot easily build the i386 parts. I think I will have to wait for the binaries. Also in a week or so I will again have access to an i386 machine, so I will try again. Sorry, Alex --- On Thu, 11/6/08, Gwenole Beauchesne wrote: > From: Gwenole Beauchesne > Subject: Re: Failure on some streams (no crash) > To: alexeskin at yahoo.com, "nspluginwrapper development discussion list" > Date: Thursday, November 6, 2008, 2:16 AM > Hi, > > ----- "Alex Eskin" a > ?crit : > > > On some sites the flash video stream does not play, > even though there > > is no crash. The same stream works with a native flash > plugin. > > The only test case I have at the moment is > www.cbs.com/video > > Since you are using Firefox 3, could you do either of > these? > (i) Try with Firefox 2 > (ii) Try with current SVN tree, which will be 1.1.4 tonight > or tomorrow: > > (don't forget to re-wrap the plugin) > > Thanks, > Gwenole. From gb.public at free.fr Fri Nov 7 07:01:17 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Fri, 7 Nov 2008 08:01:17 +0100 Subject: [ANN] nspluginwrapper 1.1.4 Message-ID: <5D59E5B2-9C29-447D-9575-B701DA0BBC24@free.fr> Hi, I have just released nspluginwrapper 1.1.4. Here is a short list of changes: * Fix memory leaks in NPRuntime bridge * Fix XEMBED support (workaround Gtk2 and Firefox bugs) * Fix DiamondX plugin with Konqueror4 * Fix NPP_URLNotify() (Bennet Yee) * Fix NPAPI version that is exposed to the plugin Some further details are available here: Regards, Gwenol?. From fhimpe at telenet.be Tue Nov 11 20:44:53 2008 From: fhimpe at telenet.be (Frederik Himpe) Date: Tue, 11 Nov 2008 21:44:53 +0100 Subject: nspluginwrapper, Flash 10 and flashblock 1.5.7: very unstable Message-ID: <1226436293.11801.37.camel@Anastacia.local> I'm using Flash 10 and nspluginwrapper 1.1.4 and Firefox 3.0.3 on Mandriva 2009.0 x86_64. nspluginwrapper (different previous versions up to the current 1.1.4) and Flash 10, have never worked stable for me. Different things I'm seeing: - npviewer.bin starts using 100% of cpu time, blocking my web browser while this happens. strace and ltrace don't give any information about what it's doing when this happens, but I could grab a backtrace with gdb: $ strace -p 4501 Process 4501 attached - interrupt to quit ^C Process 4501 detached $ ltrace -p 4501 --- SIGSTOP (Stopped (signal)) --- --- SIGSTOP (Stopped (signal)) --- --- SIGSTOP (Stopped (signal)) --- $ gdb -p 4501 [...] (gdb) thread apply all bt Thread 2 (Thread 0xf5168b90 (LWP 4516)): #0 0xffffe425 in __kernel_vsyscall () #1 0xf7865025 in pthread_cond_wait@@GLIBC_2.3.2 () from /lib/libpthread.so.0 #2 0xf67af6cf in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #3 0xf68e129f in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #4 0xf67afb8d in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #5 0xf7861315 in start_thread () from /lib/libpthread.so.0 #6 0xf77d4f0e in clone () from /lib/libc.so.6 Thread 1 (Thread 0xf71d46d0 (LWP 4501)): #0 0xf7862a18 in pthread_mutex_lock () from /lib/libpthread.so.0 #1 0xf6c9c412 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #2 0xf6c9f917 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #3 0xf6b04eeb in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #4 0xf67a61e4 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #5 0xf67a00f0 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #6 0xf6798091 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #7 0xf679cd24 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #8 0x08808220 in ?? () #9 0xffbab474 in ?? () #10 0xffbab488 in ?? () #11 0x0804d188 in handle_NPP_DestroyStream (connection=0x8808220) at ../src/npw-viewer.c:2818 #12 0x0804e712 in handle_NPP_Destroy (connection=0x8807470) at ../src/npw-viewer.c:2474 #13 0x0805515f in _rpc_dispatch (connection=0x8807470, message=0xffbab4d0) at ../src/rpc.c:1506 #14 0x08055239 in rpc_dispatch (connection=0x8807470) at ../src/rpc.c:1547 #15 0x0804aff1 in rpc_event_dispatch (source=0x8807ac8, callback=0x80551c0 , connection=0x8807470) at ../src/npw-viewer.c:3182 #16 0xf7a177ca in IA__g_main_context_dispatch (context=0x87f7b20) at gmain.c:2142 #17 0xf7a1aee8 in g_main_context_iterate (context=0x87f7b20, block=1, dispatch=1, self=0x87a8c80) at gmain.c:2776 #18 0xf7a1b3fb in IA__g_main_loop_run (loop=0x88076d8) at gmain.c:2984 #19 0xf7ca249f in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0 #20 0x0805038e in main (argc=1, argv=0xffbad764) at ../src/npw-viewer.c:3300 - A flash movie disappears as soon as another Flash movie on the same page loads - bug-buddy comes up and complains of a crash in npviewer.bin - The whole web browser just crashes What is probably a bit less common to my setup, is that I'm using the flashblock 1.5.7 extension in Firefox, so maybe that is what triggers my problems. Also what I just saw when running Firefox in a console and opening a movie on youtube (the movie played fine in spite of this error though): *** glibc detected *** /usr/lib/nspluginwrapper/i386/linux/npviewer.bin: double free or corruption (!prev): 0x093a5c00 *** (the movie played fine in spite of this error though) ======= Backtrace: ========= /lib/libc.so.6[0xf76d4784] /lib/libc.so.6(cfree+0x96)[0xf76d6866] /usr/lib/libglib-2.0.so.0(g_free+0x36)[0xf798ef96] /usr/lib/gtk-2.0/modules/libgnomebreakpad.so[0xf5267665] [0xffffe400] /usr/lib/mozilla/plugins/libflashplayer.so[0xf6a6e0ca] /usr/lib/mozilla/plugins/libflashplayer.so[0xf6873e88] /usr/lib/mozilla/plugins/libflashplayer.so[0xf6a6b6e8] /usr/lib/mozilla/plugins/libflashplayer.so[0xf6981691] /usr/lib/mozilla/plugins/libflashplayer.so[0xf6a6e0ca] /usr/lib/mozilla/plugins/libflashplayer.so[0xf698388e] /usr/lib/mozilla/plugins/libflashplayer.so[0xf6a8bdc4] /usr/lib/mozilla/plugins/libflashplayer.so[0xf6724681] /usr/lib/mozilla/plugins/libflashplayer.so[0xf670ea23] /usr/lib/libglib-2.0.so.0[0xf7986f37] ======= Memory map: ======== 08048000-08062000 r-xp 00000000 fc:03 254164 /usr/lib/nspluginwrapper/i386/linux/npviewer.bin 08062000-08063000 rw-p 0001a000 fc:03 254164 /usr/lib/nspluginwrapper/i386/linux/npviewer.bin 08063000-08068000 rw-p 08063000 00:00 0 09340000-0944d000 rw-p 09340000 00:00 0 [heap] f3700000-f3721000 rw-p f3700000 00:00 0 f3721000-f3800000 ---p f3721000 00:00 0 f38ab000-f38ac000 ---p f38ab000 00:00 0 f38ac000-f40ac000 rwxp f38ac000 00:00 0 f4166000-f416c000 r--s 00000000 fc:04 201326708 /home/frederik/.fontconfig/20b58f14c9b581391d79ea335a81488a-x86.cache-2 f416c000-f416f000 r--s 00000000 fc:04 201326707 /home/frederik/.fontconfig/87f5e051180a7a75f16eb6fe7dbd3749-x86.cache-2 f416f000-f4175000 r--s 00000000 fc:04 201326703 /home/frederik/.fontconfig/b79f3aaa7d385a141ab53ec885cc22a8-x86.cache-2 f4175000-f4178000 r--s 00000000 fc:04 201326702 /home/frederik/.fontconfig/5d999c1bbe32f61af008974facb58b71-x86.cache-2 f4178000-f417a000 r--s 00000000 fc:04 201326700 /home/frederik/.fontconfig/54c1e18aebbedcfe0b7e9af239200de0-x86.cache-2 f417a000-f4181000 r--s 00000000 fc:04 201326698 /home/frederik/.fontconfig/79aeb4e90a401e55ec91db207072ba77-x86.cache-2 f4181000-f418e000 r--s 00000000 fc:04 201326697 /home/frederik/.fontconfig/8d4af663993b81a124ee82e610bb31f9-x86.cache-2 f418e000-f4194000 r--s 00000000 fc:04 201326692 /home/frederik/.fontconfig/f6b893a7224233d96cb72fd88691c0b4-x86.cache-2 f4194000-f41be000 r--s 00000000 fc:04 201326689 /home/frederik/.fontconfig/17090aa38d5c6f09fb8c5c354938f1d7-x86.cache-2 f4273000-f42bc000 r-xp 00000000 fc:03 213318 /usr/lib/libfreebl3.so f42bc000-f42bd000 rw-p 00049000 fc:03 213318 /usr/lib/libfreebl3.so f42bd000-f42e4000 r-xp 00000000 fc:03 213337 /usr/lib/libnssdbm3.so f42e4000-f42e5000 rw-p 00026000 fc:03 213337 /usr/lib/libnssdbm3.so f42e5000-f4359000 r-xp 00000000 fc:03 213311 /usr/lib/libsqlite3.so.0.8.6 f4359000-f435a000 rw-p 00074000 fc:03 213311 /usr/lib/libsqlite3.so.0.8.6 f435a000-f4395000 r-xp 00000000 fc:03 213343 /usr/lib/libsoftokn3.so f4395000-f4396000 rw-p 0003a000 fc:03 213343 /usr/lib/libsoftokn3.so f4396000-f439e000 r-xp 00000000 fc:03 213412 /usr/lib/libkrb5support.so.0.1 f439e000-f439f000 rw-p 00007000 fc:03 213412 /usr/lib/libkrb5support.so.0.1 f439f000-f43c5000 r-xp 00000000 fc:03 213378 /usr/lib/libk5crypto.so.3.1 f43c5000-f43c6000 rw-p 00026000 fc:03 213378 /usr/lib/libk5crypto.so.3.1 f43c6000-f446a000 r-xp 00000000 fc:03 213411 /usr/lib/libkrb5.so.3.3 f446a000-f446c000 rw-p 000a3000 fc:03 213411 /usr/lib/libkrb5.so.3.3 f446c000-f4486000 r-xp 00000000 fc:03 213460 /usr/lib/libsasl2.so.2.0.22 f4486000-f4487000 rw-p 00019000 fc:03 213460 /usr/lib/libsasl2.so.2.0.22 f4487000-f44b5000 r-xp 00000000 fc:03 213341 /usr/lib/libgssapi_krb5.so.2.2 f44b5000-f44b6000 rw-p 0002e000 fc:03 213341 /usr/lib/libgssapi_krb5.so.2.2 f44b6000-f44c5000 r-xp 00000000 fc:03 213940 /usr/lib/liblber-2.4.so.2.1.0 f44c5000-f44c6000 rw-p 0000e000 fc:03 213940 /usr/lib/liblber-2.4.so.2.1.0 f44c6000-f450b000 r-xp 00000000 fc:03 213088 /usr/lib/libldap-2.4.so.2.1.0 f450b000-f450d000 rw-p 00044000 fc:03 213088 /usr/lib/libldap-2.4.so.2.1.0 f450d000-f4530000 r-xp 00000000 fc:03 213286 /usr/lib/libssh2.so.1.0.0 f4530000-f4531000 rw-p 00022000 fc:03 213286 /usr/lib/libssh2.so.1.0.0 f4531000-f4562000 r-xp 00000000 fc:03 213292 /usr/lib/libidn.so.11.5.39 f4562000-f4563000 rw-p 00030000 fc:03 213292 /usr/lib/libidn.so.11.5.39 f4563000-f45a9000 r-xp 00000000 fc:03 213297 /usr/lib/libcurl.so.4.1.0 f45a9000-f45ab000 rw-p 00045000 fc:03 213297 /usr/lib/libcurl.so.4.1.0 f45ab000-f468a000 r-xp 00000000 fc:03 213281 /usr/lib/libasound.so.2.0.0 f468a000-f468f000 rw-p 000de000 fc:03 213281 /usr/lib/libasound.so.2.0.0 f468f000-f469e000 r-xp 00000000 fc:02 65570 /lib/libresolv-2.8.so f469e000-f469f000 r--p 0000e000 fc:02 65570 /lib/libresolv-2.8.so f469f000-f46a0000 rw-p 0000f000 fc:02 65570 /lib/libresolv-2.8.so f46a0000-f46a2000 rw-p f46a0000 00:00 0 f46a2000-f4812000 r-xp 00000000 fc:03 213648 /usr/lib/libcrypto.so.0.9.8 f4812000-f4827000 rw-p 00170000 fc:03 213648 /usr/lib/libcrypto.so.0.9.8 f4827000-f482a000 rw-p f4827000 00:00 0 f482a000-f482f000 r-xp 00000000 fc:03 218753 /usr/lib/libasyncns.so.0.2.0 f482f000-f4830000 rw-p 00004000 fc:03 218753 /usr/lib/libasyncns.so.0.2.0 f4830000-f4879000 r-xp 00000000 fc:03 213724 /usr/lib/libssl.so.0.9.8 f4879000-f487d000 rw-p 00048000 fc:03 213724 /usr/lib/libssl.so.0.9.8 f487d000-f48d2000 r-xp 00000000 fc:03 218748 /usr/lib/libpulse.so.0.4.1 f48d2000-f48d3000 rw-p 00055000 fc:03 218748 /usr/lib/libpulse.so.0.4.1 f48d3000-f48d6000 r-xp 00000000 fc:03 218760 /usr/lib/libflashsupport.so f48d6000-f48d7000 rw-p 00002000 fc:03 218760 /usr/lib/libflashsupport.so f48d7000-f48d8000 ---p f48d7000 00:00 0 f48d8000-f50d8000 rwxp f48d8000 00:00 0 f50d8000-f51d9000 rw-p f50d8000 00:00 0 f51d9000-f51ed000 r-xp 00000000 fc:03 214186 /usr/lib/libelf-0.137.so f51ed000-f51ee000 r--p 00014000 fc:03 214186 /usr/lib/libelf-0.137.so f51ee000-f51ef000 rw-p 00015000 fc:03 214186 /usr/lib/libelf-0.137.so f51f1000-f5210000 r--s 00000000 fc:04 201326688 /home/frederik/.fontconfig/df311e82a1a24c41a75c2c930223552e-x86.cache-2 f5210000-f5212000 r--s 00000000 fc:04 201530899 /home/frederik/.fontconfig/6d55269ca890cf128802a9e1c76fc996-x86.cache-2 f5212000-f5262000 r-xp 00000000 fc:03 218763 /usr/lib/bug-buddy/libbreakpad.so.0.0.0 f5262000-f5263000 rw-p 00050000 fc:03 218763 /usr/lib/bug-buddy/libbreakpad.so.0.0.0 f5263000-f5271000 r-xp 00000000 fc:03 483412 /usr/lib/gtk-2.0/modules/libgnomebreakpad.so f5271000-f5272000 rw-p 0000d000 fc:03 483412 /usr/lib/gtk-2.0/modules/libgnomebreakpad.so f5272000-f527c000 r-xp 00000000 fc:02 65564 /lib/libnss_files-2.8.so f527c000-f527d000 r--p 00009000 fc:02 65564 /lib/libnss_files-2.8.so f527d000-f527e000 rw-p 0000a000 fc:02 65564 /lib/libnss_files-2.8.so f527e000-f5286000 r-xp 00000000 fc:02 65566 /lib/libnss_nis-2.8.so f5286000-f5287000 r--p 00007000 fc:02 65566 /lib/libnss_nis-2.8.so f5287000-f5288000 rw-p 00008000 fc:02 65566 /lib/libnss_nis-2.8.so f5288000-f528f000 r-xp 00000000 fc:02 65562 /lib/libnss_compat-2.8.so f528f000-f5290000 r--p 00006000 fc:02 65562 /lib/libnss_comMultiple segmentation faults occurred; can't display error dialog ERROR: ld.so: object '/usr/$LIB/libaoss.so.0' from LD_PRELOAD cannot be preloaded: ignored. ERROR: ld.so: object '/usr/$LIB/libaoss.so.0' from LD_PRELOAD cannot be preloaded: ignored. Gtk-Message: Failed to load module "canberra-gtk-module": libcanberra-gtk-module.so: cannot open shared object file: No such file or directory (npviewer.bin:5867): Gtk-WARNING **: Unable to locate theme engine in module_path: "clearlooks", ERROR: ld.so: object '/usr/$LIB/libaoss.so.0' from LD_PRELOAD cannot be preloaded: ignored. Gtk-Message: Failed to load module "canberra-gtk-module": libcanberra-gtk-module.so: cannot open shared object file: No such file or directory (npviewer.bin:5883): Gtk-WARNING **: Unable to locate theme engine in module_path: "clearlooks", What would be the best way to properly help debugging these problems? -- Frederik Himpe From gb.public at free.fr Tue Nov 11 21:59:20 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Tue, 11 Nov 2008 22:59:20 +0100 (CET) Subject: nspluginwrapper, Flash 10 and flashblock 1.5.7: very unstable In-Reply-To: <5324691.41651226440705275.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <25374847.41671226440760087.JavaMail.root@spooler5-g27.priv.proxad.net> Hi, ----- "Frederik Himpe" a ?crit : > I'm using Flash 10 and nspluginwrapper 1.1.4 and Firefox 3.0.3 on > Mandriva 2009.0 x86_64. nspluginwrapper (different previous versions > up > to the current 1.1.4) and Flash 10, have never worked stable for me. What site? I tried fully loaded things like TF1, Gaumont, Freenews, Youtube without crash in the plugin nor the browser. > Different things I'm seeing: > - npviewer.bin starts using 100% of cpu time, blocking my web browser > while this happens. strace and ltrace don't give any information > about I don't see anything here but an 'f' instead of the normal content, which I assume is the expected behaviour. Flash then normally starts when I click on the plugin area. I tested with Firefox 3.0.3 (from openSUSE 11.0 updates) + Flash Block 1.5.7 + Flash 10.0.12.36 + original nspluginwrapper 1.1.4 (fresh checkout from: http://svn.beauchesne.info/svn/gwenole/projects/nspluginwrapper/tags/release-1.1.4). > (npviewer.bin:5867): Gtk-WARNING **: Unable to locate theme engine in > module_path: "clearlooks", BTW, as a side|sad note: Mandriva used to have correct 32-bit run-time by default, including the default Gtk2 theme for both 64-bit and 32-bit applications. Please don't forget to ship with the 32-bit Gtk2 engines... Regards, Gwenole. From fhimpe at telenet.be Thu Nov 13 20:27:55 2008 From: fhimpe at telenet.be (Frederik Himpe) Date: Thu, 13 Nov 2008 21:27:55 +0100 Subject: nspluginwrapper, Flash 10 and flashblock 1.5.7: very unstable In-Reply-To: <25374847.41671226440760087.JavaMail.root@spooler5-g27.priv.proxad.net> References: <25374847.41671226440760087.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <1226608076.29576.5.camel@Anastacia.local> On di, 2008-11-11 at 22:59 +0100, Gwenole Beauchesne wrote: > Hi, > > ----- "Frederik Himpe" a ?crit : > > > I'm using Flash 10 and nspluginwrapper 1.1.4 and Firefox 3.0.3 on > > Mandriva 2009.0 x86_64. nspluginwrapper (different previous versions > > up > > to the current 1.1.4) and Flash 10, have never worked stable for me. > > What site? I tried fully loaded things like TF1, Gaumont, Freenews, Youtube without crash in the plugin nor the browser. I'll start with this very easily reproducible test case which I just found: just visit http://www.deredactie.be . Immediately bug-buddy pops-up here, before the site has even loaded completely and before I could activate on any of the flashblocked animations. Bug-buddy gives this information: System: Linux 2.6.27.4-desktop-2mnb #1 SMP Thu Nov 6 13:36:39 EST 2008 i686 X Vendor: The X.Org Foundation X Vendor Release: 10402000 Selinux: No Accessibility: Disabled GTK+ Theme: Clearlooks-Quicksilver Icon Theme: Tango Memory status: size: 65138688 vsize: 65138688 resident: 14741504 share: 10149888 rss: 14741504 rss_rlim: 18446744073709551615 CPU usage: start_time: 1226607576 rtime: 15 utime: 10 stime: 5 cutime:0 cstime: 1 timeout: 0 it_real_value: 0 frequency: 100 Backtrace was generated from '/usr/lib64/npviewer.bin' [?1034h[Thread debugging using libthread_db enabled] [New Thread 0xf72196d0 (LWP 28920)] 0xffffe425 in __kernel_vsyscall () #0 0xffffe425 in __kernel_vsyscall () #1 0xf78ae00b in waitpid () from /lib/libpthread.so.0 #2 0xf7a9635b in IA__g_spawn_sync (working_directory=0x0, argv=0x85e6e50, envp=0x0, flags=, child_setup=0, user_data=0x0, standard_output=0x0, standard_error=0x0, exit_status=0x0, error=0xffef1368) at gspawn.c:382 #3 0xf7a966b4 in IA__g_spawn_command_line_sync (command_line=0x86799c0 "bug-buddy --appname=\"npviewer.bin\" --pid=28920", standard_output=0x0, standard_error=0x0, exit_status=0x0, error=0xffef1368) at gspawn.c:694 #4 0xf533d003 in ?? () from /usr/lib/gtk-2.0/modules/libgnomebreakpad.so #5 0xf533d590 in ?? () from /usr/lib/gtk-2.0/modules/libgnomebreakpad.so #6 0xf533d665 in ?? () from /usr/lib/gtk-2.0/modules/libgnomebreakpad.so #7 #8 0x313d6575 in ?? () #9 0xf67f0b28 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #10 0xf6b576ef in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #11 0xf6b57ffe in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #12 0xf6b5b9d2 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #13 0xf6b5cd9c in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #14 0xf6b5cff8 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #15 0xf6a61262 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #16 0xf6b5e92a in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #17 0xf6b5ff45 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #18 0xf6b62128 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #19 0xf67fa681 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #20 0xf67e4a23 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so #21 0xf7a5cf37 in g_timeout_dispatch (source=0x8649568, callback=0xf54fb230, user_data=0xf54c1000) at gmain.c:3587 #22 0xf7a5c7ca in IA__g_main_context_dispatch (context=0x8639918) at gmain.c:2142 #23 0xf7a5fee8 in g_main_context_iterate (context=0x8639918, block=1, dispatch=1, self=0x85eac80) at gmain.c:2776 #24 0xf7a603fb in IA__g_main_loop_run (loop=0x8648fb8) at gmain.c:2984 #25 0xf7ce749f in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0 #26 0x0805038e in main (argc=1, argv=0xffef2434) at ../src/npw-viewer.c:3300 Thread 1 (Thread 0xf72196d0 (LWP 28920)): #0 0xffffe425 in __kernel_vsyscall () No symbol table info available. #1 0xf78ae00b in waitpid () from /lib/libpthread.so.0 No symbol table info available. #2 0xf7a9635b in IA__g_spawn_sync (working_directory=0x0, argv=0x85e6e50, envp=0x0, flags=, child_setup=0, user_data=0x0, standard_output=0x0, standard_error=0x0, exit_status=0x0, error=0xffef1368) at gspawn.c:382 outpipe = -1 errpipe = -1 pid = 28936 fds = {__fds_bits = {8, 140485816, 0, 0, -140095594, 160, 0, 8, -139566536, 0, 140406364, 140803360, -139997905, 8, 140803360, 4, -140010768, 140803360, -140095648, 0, -139983505, 0, -1, 68, 11, 577024580, -1109288, -1109292, 140406352, 3, 0, 807572212}} ret = outstr = (GString *) 0x0 errstr = (GString *) 0x0 failed = 0 status = __PRETTY_FUNCTION__ = "IA__g_spawn_sync" #3 0xf7a966b4 in IA__g_spawn_command_line_sync (command_line=0x86799c0 "bug-buddy --appname=\"npviewer.bin\" --pid=28920", standard_output=0x0, standard_error=0x0, exit_status=0x0, error=0xffef1368) at gspawn.c:694 retval = 0 argv = (gchar **) 0x85e6e50 __PRETTY_FUNCTION__ = "IA__g_spawn_command_line_sync" #4 0xf533d003 in ?? () from /usr/lib/gtk-2.0/modules/libgnomebreakpad.so No symbol table info available. #5 0xf533d590 in ?? () from /usr/lib/gtk-2.0/modules/libgnomebreakpad.so No symbol table info available. #6 0xf533d665 in ?? () from /usr/lib/gtk-2.0/modules/libgnomebreakpad.so No symbol table info available. #7 No symbol table info available. #8 0x313d6575 in ?? () No symbol table info available. #9 0xf67f0b28 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #10 0xf6b576ef in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #11 0xf6b57ffe in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #12 0xf6b5b9d2 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #13 0xf6b5cd9c in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #14 0xf6b5cff8 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #15 0xf6a61262 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #16 0xf6b5e92a in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #17 0xf6b5ff45 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #18 0xf6b62128 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #19 0xf67fa681 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #20 0xf67e4a23 in ?? () from /usr/lib/mozilla/plugins/libflashplayer.so No symbol table info available. #21 0xf7a5cf37 in g_timeout_dispatch (source=0x8649568, callback=0xf54fb230, user_data=0xf54c1000) at gmain.c:3587 No locals. #22 0xf7a5c7ca in IA__g_main_context_dispatch (context=0x8639918) at gmain.c:2142 No locals. #23 0xf7a5fee8 in g_main_context_iterate (context=0x8639918, block=1, dispatch=1, self=0x85eac80) at gmain.c:2776 max_priority = 0 timeout = 0 some_ready = 1 nfds = 4 allocated_nfds = fds = (GPollFD *) 0x86493f8 __PRETTY_FUNCTION__ = "g_main_context_iterate" #24 0xf7a603fb in IA__g_main_loop_run (loop=0x8648fb8) at gmain.c:2984 self = (GThread *) 0x85eac80 __PRETTY_FUNCTION__ = "IA__g_main_loop_run" #25 0xf7ce749f in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0 No symbol table info available. #26 0x0805038e in main (argc=1, argv=0xffef2434) at ../src/npw-viewer.c:3300 plugin_path = 0xffef3fa1 "/usr/lib/mozilla/plugins/libflashplayer.so" connection_path = 0xffef3fd9 "/org/wrapper/NSPlugins/libflashplayer.so/28744-2" cmd = 0 handle = (void *) 0x85e4050 ret = The program is running. Quit anyway (and detach it)? (y or n) [answered Y; input not from terminal] ----------- .xsession-errors (31 sec old) --------------------- Gtk-Message: Failed to load module "canberra-gtk-module": libcanberra-gtk-module.so: cannot open shared object file: No such file or directory (npviewer.bin:28179): Gtk-WARNING **: Unable to locate theme engine in module_path: "clearlooks", *** NSPlugin Viewer *** ERROR: NPN_GetURLNotify() invoke: Connection closed /usr/lib64/npviewer.bin: No such file or directory. ERROR: ld.so: object '/usr/$LIB/libaoss.so.0' from LD_PRELOAD cannot be preloaded: ignored. ERROR: ld.so: object '/usr/$LIB/libaoss.so.0' from LD_PRELOAD cannot be preloaded: ignored. Gtk-Message: Failed to load module "canberra-gtk-module": libcanberra-gtk-module.so: cannot open shared object file: No such file or directory (npviewer.bin:28513): Gtk-WARNING **: Unable to locate theme engine in module_path: "clearlooks", _IceTransSocketUNIXConnect: Cannot connect to non-local host Anastacia _IceTransSocketUNIXConnect: Cannot connect to non-local host Anastacia (firefox:28531): GnomeUI-WARNING **: While connecting to session manager: Could not open network socket. -------------------------------------------------- -- Frederik Himpe From gb.public at free.fr Thu Nov 13 22:10:21 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Thu, 13 Nov 2008 23:10:21 +0100 (CET) Subject: nspluginwrapper, Flash 10 and flashblock 1.5.7: very unstable In-Reply-To: <3566990.321961226614044974.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <33414040.322121226614221965.JavaMail.root@spooler5-g27.priv.proxad.net> Hi, ----- "Frederik Himpe" a ?crit : > I'll start with this very easily reproducible test case which I just > found: just visit http://www.deredactie.be . Immediately bug-buddy > pops-up here, before the site has even loaded completely and before I > could activate on any of the flashblocked animations. Well, it still works here. Something must be broken on your system. I also tried with firefox 3.1-beta-something. However, the only unusual message this site causes is: (npviewer.bin:9311): Gdk-CRITICAL **: _gdk_x11_window_tmp_unset_bg: assertion `GDK_IS_WINDOW (window)' failed Nothing else. Flash videos normally play afterwards when they are activated. > ----------- .xsession-errors (31 sec old) --------------------- > Gtk-Message: Failed to load module "canberra-gtk-module": > libcanberra-gtk-module.so: cannot open shared object file: No such > file > or directory > (npviewer.bin:28179): Gtk-WARNING **: Unable to locate theme engine > in > module_path: "clearlooks", I don't think those errors should be ignored, please make sure to fix them first. > *** NSPlugin Viewer *** ERROR: NPN_GetURLNotify() invoke: Connection > closed > /usr/lib64/npviewer.bin: No such file or directory. BTW, why a 32-bit binary would lay into */lib64/ ? Regards, Gwenole. From gb.public at free.fr Thu Nov 13 22:44:15 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Thu, 13 Nov 2008 23:44:15 +0100 (CET) Subject: nspluginwrapper, Flash 10 and flashblock 1.5.7: very unstable In-Reply-To: <33414040.322121226614221965.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <11029498.324361226616255701.JavaMail.root@spooler5-g27.priv.proxad.net> Him > Nothing else. Flash videos normally play afterwards when they are > activated. I have just tried on Fedora 9 + updates and could not see any problem either. firefox-3.0.2-1.fc9.x86_64 flash-plugin-10.0.12.36-release.i386 xulrunner-1.9.0.2-1.fc9.x86_64 + nspluginwrapper 1.1.4 From wtogami at redhat.com Fri Nov 14 01:25:01 2008 From: wtogami at redhat.com (Warren Togami) Date: Thu, 13 Nov 2008 20:25:01 -0500 Subject: Upstreaming Fedora's patches? Message-ID: <491CD36D.9050304@redhat.com> http://cvs.fedoraproject.org/viewvc/rpms/nspluginwrapper/F-10/ I don't know all the details because Stransky did all the work, but is there any chance that we can upstream some or all of the patches that we use against nspluginwrapper? What will it take to bring us closer to upstream? If there are reservations about technical issues please discuss. Thanks, Warren Togami wtogami at redhat.com From wtogami at redhat.com Fri Nov 14 01:45:34 2008 From: wtogami at redhat.com (Warren Togami) Date: Thu, 13 Nov 2008 20:45:34 -0500 Subject: How to use npplayer? Message-ID: <491CD83E.9070108@redhat.com> How do I use npplayer? I cannot figure it out with many different combinations of command line options. I would like it to play a simple SWF. Then I would like to see if I can make it play a .flv video that I captured from Youtube. Has anyone got this to work in npplayer? Warren Togami wtogami at redhat.com From stransky at redhat.com Fri Nov 14 08:21:06 2008 From: stransky at redhat.com (Martin Stransky) Date: Fri, 14 Nov 2008 09:21:06 +0100 Subject: Upstreaming Fedora's patches? In-Reply-To: <491CD36D.9050304@redhat.com> References: <491CD36D.9050304@redhat.com> Message-ID: <491D34F2.1050402@redhat.com> I sent reasonable patches to Gwenole (I can re-sent them here of course), other ones are fedora specific (install/build). ma. Warren Togami wrote: > http://cvs.fedoraproject.org/viewvc/rpms/nspluginwrapper/F-10/ > I don't know all the details because Stransky did all the work, but is > there any chance that we can upstream some or all of the patches that we > use against nspluginwrapper? > > What will it take to bring us closer to upstream? > > If there are reservations about technical issues please discuss. > > Thanks, > Warren Togami > wtogami at redhat.com > > _______________________________________________ > Nspluginwrapper-devel-list mailing list > Nspluginwrapper-devel-list at redhat.com > https://www.redhat.com/mailman/listinfo/nspluginwrapper-devel-list From gb.public at free.fr Fri Nov 14 08:47:48 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Fri, 14 Nov 2008 09:47:48 +0100 (CET) Subject: How to use npplayer? In-Reply-To: <25994877.341531226652411508.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <18156953.341701226652468335.JavaMail.root@spooler5-g27.priv.proxad.net> Hi, ----- "Warren Togami" a ?crit : > How do I use npplayer? I cannot figure it out with many different > combinations of command line options. OK, there is a bug that was exposed by the NPP_URLNotify() fix. I have attached a fix. The important part is to move down the "consider the stream in error if it was never started" check since the stream was manually started after in the "javascript:" condition. > I would like it to play a simple SWF. Then I would like to see if I > can make it play a .flv video that I captured from Youtube. Has anyone > got this to work in npplayer? It's generally enough to copy and paste the args. e.g. say I have downloaded a Flash game (bloxorz.swf): $ ./npplayer src=file:///home/gb/NPAPI/flash/bloxorz.swf If the player can't determine the source stream, you can help it with the type= arg to specify the mime type: $ ./npplayer src=file://home/gb/downloads/diamondx-0.1/test/a.diamondx width=400 height=300 type=application/x-diamondx If you want to play a Youtube video, it's not a direct process: - Go to the page of the desired video. e.g. - Observe the args in the right-hand side box. i.e. - Browse to the src arg. i.e. http://www.youtube.com/v/EuhU6jJjpAQ&hl=fr&fs=1 the URL will be expanded, use that one for npplayer $ ./npplayer type=application/x-shockwave-flash allowscriptaccess=always allowfullscreen=true width=425 height=344 src="http://www.youtube.com/swf/l.swf?swf=http%3A//s.ytimg.com/yt/swf/cps-vfl64504.swf&video_id=EuhU6jJjpAQ&rel=1&showsearch=1&eurl=&iurl=http%3A//i2.ytimg.com/vi/EuhU6jJjpAQ/hqdefault.jpg&sk=V7Po1QzD5cCenH9cB2LvAea08_tD7UTpC&use_get_video_info=1&load_modules=1&fs=1&hl=fr" Now, in order to play an .flv file, you will need the SWF player and know how to call it with a local file. I have never tested this case yet. -------------- next part -------------- A non-text attachment was scrubbed... Name: npw.fix.npplayer.diff Type: text/x-diff Size: 2140 bytes Desc: not available URL: From gb.public at free.fr Fri Nov 14 09:00:19 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Fri, 14 Nov 2008 10:00:19 +0100 (CET) Subject: Upstreaming Fedora's patches? In-Reply-To: <491CD36D.9050304@redhat.com> Message-ID: <25537033.343341226653219544.JavaMail.root@spooler5-g27.priv.proxad.net> Hi, ----- "Warren Togami" a ?crit : > What will it take to bring us closer to upstream? Yes, this is still planned for 1.2.0 which should be next release actually. The goal is to lower integration work because I'd like to reindent and reorganize the tree for 2.0. ;-) However, I am mostly interested in changes in the code (e.g. checking whether the plugin is still alive). Less in the configure changes because most of them were Fedora specific IIRC. e.g. I want to keep it buildable through the LSB headers because some distributions are not capable enough to build 32-bit binaries within a 64-bit environment (most notably Debian-based distributions and Gentoo). Future changes also include autotools and cleaner use of a TARGET compiler (e.g. build Linux viewer from *BSD with a cross compiler, or Windows viewer from winegcc and the like). Regards, Gwenole. From gb.public at free.fr Fri Nov 14 09:14:02 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Fri, 14 Nov 2008 10:14:02 +0100 (CET) Subject: [PATCH] Fix npw_get_stream_instance (Was: How to use npplayer?) In-Reply-To: <27906075.346271226654034997.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <24946457.346321226654042443.JavaMail.root@spooler5-g27.priv.proxad.net> Hi, ----- "Gwenole Beauchesne" a ?crit : > $ ./npplayer type=application/x-shockwave-flash > allowscriptaccess=always allowfullscreen=true width=425 height=344 > src="http://www.youtube.com/swf/l.swf?swf=http%3A//s.ytimg.com/yt/swf/cps-vfl64504.swf&video_id=EuhU6jJjpAQ&rel=1&showsearch=1&eurl=&iurl=http%3A//i2.ytimg.com/vi/EuhU6jJjpAQ/hqdefault.jpg&sk=V7Po1QzD5cCenH9cB2LvAea08_tD7UTpC&use_get_video_info=1&load_modules=1&fs=1&hl=fr" And here is the patch that will actually make this command work. In addition to the previous one. -------------- next part -------------- A non-text attachment was scrubbed... Name: npw.fix.streaminstance.diff Type: text/x-diff Size: 931 bytes Desc: not available URL: From stransky at redhat.com Fri Nov 14 10:03:19 2008 From: stransky at redhat.com (Martin Stransky) Date: Fri, 14 Nov 2008 11:03:19 +0100 Subject: Upstreaming Fedora's patches? In-Reply-To: <25537033.343341226653219544.JavaMail.root@spooler5-g27.priv.proxad.net> References: <25537033.343341226653219544.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <491D4CE7.40801@redhat.com> Gwenole Beauchesne wrote: > Hi, > > ----- "Warren Togami" a ?crit : > >> What will it take to bring us closer to upstream? > > Yes, this is still planned for 1.2.0 which should be next release > actually. The goal is to lower integration work because I'd like to > reindent and reorganize the tree for 2.0. ;-) Hi, please let me know if you like to comment any fedora patch, I hope I sent you all of them... ma. From alexeskin at yahoo.com Sat Nov 15 11:26:41 2008 From: alexeskin at yahoo.com (Alex Eskin) Date: Sat, 15 Nov 2008 03:26:41 -0800 (PST) Subject: Failure on some streams (no crash) In-Reply-To: <31597003.1336351225959392986.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <360174.16418.qm@web51710.mail.re2.yahoo.com> It also does not work with firefox 2 (fedora 8, firefox-2.0.0.18-1.fc8, nspluginwrapper-0.9.91.5-18.fc8, flash-plugin-10.0.12.36-release) Will try 1.1.4 in a few days. --- On Thu, 11/6/08, Gwenole Beauchesne wrote: > From: Gwenole Beauchesne > Subject: Re: Failure on some streams (no crash) > To: alexeskin at yahoo.com, "nspluginwrapper development discussion list" > Date: Thursday, November 6, 2008, 2:16 AM > Hi, > > ----- "Alex Eskin" a > ?crit : > > > On some sites the flash video stream does not play, > even though there > > is no crash. The same stream works with a native flash > plugin. > > The only test case I have at the moment is > www.cbs.com/video > > Since you are using Firefox 3, could you do either of > these? > (i) Try with Firefox 2 > (ii) Try with current SVN tree, which will be 1.1.4 tonight > or tomorrow: > > (don't forget to re-wrap the plugin) > > Thanks, > Gwenole. From wtogami at redhat.com Mon Nov 17 20:46:48 2008 From: wtogami at redhat.com (Warren Togami) Date: Mon, 17 Nov 2008 15:46:48 -0500 Subject: Leaked file descriptors from browser to npviewer.bin Message-ID: <4921D838.6000904@redhat.com> It seems that upon fork() file descriptors from the browser leak to the npviewer.bin process. There are potential security concerns whenever file descriptors leak to a forked process. Are any of these file descriptors actually expected and needed by the npviewer.bin process? Warren Togami wtogami at redhat.com From gb.public at free.fr Mon Nov 17 23:30:11 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Tue, 18 Nov 2008 00:30:11 +0100 Subject: Leaked file descriptors from browser to npviewer.bin In-Reply-To: <4921D838.6000904@redhat.com> References: <4921D838.6000904@redhat.com> Message-ID: <0ED4D9BD-EE16-4247-86F9-A0DD01AABA48@free.fr> Hi, Le 17 nov. 08 ? 21:46, Warren Togami a ?crit : > It seems that upon fork() file descriptors from the browser leak to > the npviewer.bin process. There are potential security concerns > whenever file descriptors leak to a forked process. Are any of > these file descriptors actually expected and needed by the > npviewer.bin process? Do you mean any file descriptor opened by the browser-side part of nspluginwrapper? Because, for file descriptors originating from the browser, I don't see what we could do. Isn't it the responsibility of the browser to O_CLOEXEC (resp. SOCK_CLOEXEC) them through fnctl() or "now" directly through creation functions? e.g. open() as I understood through Ulrich's article. nspluginwrapper only needs its sockets. I normally close the descriptor from the popen() and I don't see other descriptor nspluginwrapper may have opened. Could you be more specific? Thanks, Gwenole. From wtogami at redhat.com Wed Nov 19 03:25:28 2008 From: wtogami at redhat.com (Warren Togami) Date: Tue, 18 Nov 2008 22:25:28 -0500 Subject: Leaked file descriptors from browser to npviewer.bin In-Reply-To: <0ED4D9BD-EE16-4247-86F9-A0DD01AABA48@free.fr> References: <4921D838.6000904@redhat.com> <0ED4D9BD-EE16-4247-86F9-A0DD01AABA48@free.fr> Message-ID: <49238728.3040507@redhat.com> Gwenole Beauchesne wrote: > Hi, > > Le 17 nov. 08 ? 21:46, Warren Togami a ?crit : > >> It seems that upon fork() file descriptors from the browser leak to >> the npviewer.bin process. There are potential security concerns >> whenever file descriptors leak to a forked process. Are any of these >> file descriptors actually expected and needed by the npviewer.bin >> process? > > Do you mean any file descriptor opened by the browser-side part of > nspluginwrapper? Because, for file descriptors originating from the > browser, I don't see what we could do. Isn't it the responsibility of > the browser to O_CLOEXEC (resp. SOCK_CLOEXEC) them through fnctl() or > "now" directly through creation functions? e.g. open() as I understood > through Ulrich's article. > > nspluginwrapper only needs its sockets. I normally close the descriptor > from the popen() and I don't see other descriptor nspluginwrapper may > have opened. Could you be more specific? 1) Run firefox. 2) Go somewhere that uses flash. 3) Find the process ID of npviewer.bin 4) [root at newcaprica tmp]# ls /proc/2488/fd -l total 0 lr-x------ 1 warren warren 64 2008-11-18 22:22 0 -> /dev/null lrwx------ 1 warren warren 64 2008-11-18 22:22 1 -> /dev/null lrwx------ 1 warren warren 64 2008-11-18 22:22 10 -> socket:[2495578] l-wx------ 1 warren warren 64 2008-11-18 22:22 11 -> /home/warren/.mozilla/firefox/default.xzo/.parentlock lr-x------ 1 warren warren 64 2008-11-18 22:22 12 -> pipe:[2494840] l-wx------ 1 warren warren 64 2008-11-18 22:22 13 -> pipe:[2494840] lr-x------ 1 warren warren 64 2008-11-18 22:22 14 -> /home/warren/.mozilla/firefox/default.xzo/cert8.db lr-x------ 1 warren warren 64 2008-11-18 22:22 15 -> pipe:[2494863] l-wx------ 1 warren warren 64 2008-11-18 22:22 16 -> pipe:[2494863] lr-x------ 1 warren warren 64 2008-11-18 22:22 17 -> /dev/urandom lr-x------ 1 warren warren 64 2008-11-18 22:22 18 -> /home/warren/.mozilla/firefox/default.xzo/XPC.mfasl lr-x------ 1 warren warren 64 2008-11-18 22:22 19 -> /home/warren/.mozilla/firefox/default.xzo/key3.db lrwx------ 1 warren warren 64 2008-11-18 22:22 2 -> /dev/null lrwx------ 1 warren warren 64 2008-11-18 22:22 20 -> /tmp/FlashLn3ECB lr-x------ 1 warren warren 64 2008-11-18 22:22 21 -> pipe:[2497185] l-wx------ 1 warren warren 64 2008-11-18 22:22 22 -> pipe:[2497185] lr-x------ 1 warren warren 64 2008-11-18 22:22 23 -> /usr/lib64/firefox-3.0.4/chrome/en-US.jar lr-x------ 1 warren warren 64 2008-11-18 22:22 24 -> pipe:[2497186] l-wx------ 1 warren warren 64 2008-11-18 22:22 25 -> pipe:[2497186] lr-x------ 1 warren warren 64 2008-11-18 22:22 26 -> /usr/lib64/firefox-3.0.4/chrome/browser.jar lrwx------ 1 warren warren 64 2008-11-18 22:22 27 -> socket:[2497190] lr-x------ 1 warren warren 64 2008-11-18 22:22 28 -> /usr/lib64/xulrunner-1.9/chrome/en-US.jar lr-x------ 1 warren warren 64 2008-11-18 22:22 29 -> /usr/lib64/xulrunner-1.9/chrome/classic.jar lrwx------ 1 warren warren 64 2008-11-18 22:22 3 -> socket:[2495570] lr-x------ 1 warren warren 64 2008-11-18 22:22 30 -> /usr/lib64/xulrunner-1.9/chrome/toolkit.jar lr-x------ 1 warren warren 64 2008-11-18 22:22 32 -> pipe:[2494877] l-wx------ 1 warren warren 64 2008-11-18 22:22 33 -> pipe:[2494877] lr-x------ 1 warren warren 64 2008-11-18 22:22 36 -> /usr/lib64/firefox-3.0.4/chrome/classic.jar lr-x------ 1 warren warren 64 2008-11-18 22:22 37 -> /home/warren/.mozilla/firefox/default.xzo/extensions/max at subfighter.com/chrome/flvdownloader.jar lr-x------ 1 warren warren 64 2008-11-18 22:22 38 -> /home/warren/.mozilla/firefox/default.xzo/extensions/{c1309325-5574-41bc-ab8a-abae2acee24b}/chrome/fullerscreen.jar lrwx------ 1 warren warren 64 2008-11-18 22:22 4 -> socket:[2495574] lrwx------ 1 warren warren 64 2008-11-18 22:22 45 -> /home/warren/.mozilla/firefox/default.xzo/Cache/_CACHE_MAP_ lrwx------ 1 warren warren 64 2008-11-18 22:22 46 -> /home/warren/.mozilla/firefox/default.xzo/Cache/_CACHE_001_ lrwx------ 1 warren warren 64 2008-11-18 22:22 47 -> /home/warren/.mozilla/firefox/default.xzo/Cache/_CACHE_002_ lrwx------ 1 warren warren 64 2008-11-18 22:22 48 -> /home/warren/.mozilla/firefox/default.xzo/Cache/_CACHE_003_ lr-x------ 1 warren warren 64 2008-11-18 22:22 5 -> pipe:[2495576] l-wx------ 1 warren warren 64 2008-11-18 22:22 6 -> pipe:[2495576] lrwx------ 1 warren warren 64 2008-11-18 22:22 7 -> socket:[2495577] lr-x------ 1 warren warren 64 2008-11-18 22:22 8 -> pipe:[2494831] l-wx------ 1 warren warren 64 2008-11-18 22:22 9 -> pipe:[2494831] Are all of these file descriptors from npviewer.bin and Flash? It appears that some are leaked from the fork() from firefox. According to dwalsh, SELinux automatically closes some leaked file descriptors that are not allowed after the transition, but others remain? Warren Togami wtogami at redhat.com From wtogami at redhat.com Wed Nov 19 04:37:50 2008 From: wtogami at redhat.com (Warren Togami) Date: Tue, 18 Nov 2008 23:37:50 -0500 Subject: Crash with Flash 64bit and nspluginwrapper-1.1.4 Message-ID: <4923981E.5070908@redhat.com> http://finance.google.com/finance?q=gol Other URL's seem to work with nspluginwrapper-1.1.4 and flash 10.0.20.7 64bit, but it crashes npviewer.bin on this Google Finance page. npviewer.bin[4157]: segfault at 7f669823d058 ip 0000003cc5008a4a sp 00007fffa5e65970 error 4 in libpthread-2.9.so[3cc5000000+17000] Warren Togami wtogami at redhat.com From gb.public at free.fr Sat Nov 22 15:50:13 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Sat, 22 Nov 2008 16:50:13 +0100 Subject: [PATCH] Re: Crash with Flash 64bit and nspluginwrapper-1.1.4 In-Reply-To: <4923981E.5070908@redhat.com> References: <4923981E.5070908@redhat.com> Message-ID: <7FC56CD2-A531-450B-B1FC-67634E325699@free.fr> Hi Warren, Le 19 nov. 08 ? 05:37, Warren Togami a ?crit : > http://finance.google.com/finance?q=gol > Other URL's seem to work with nspluginwrapper-1.1.4 and flash > 10.0.20.7 64bit, but it crashes npviewer.bin on this Google Finance > page. I could not reproduce the crash per se but this page indeed exhibited problems, even with the 32-bit Flash plugin. I have attached some patches I committed to SVN, let me know which one (if any) fixes your problem. Thanks. * Fix for NPClass::Invoke() args release (could be your problem, though it's not a satisfying fix yet): nspluginwrapper-fix-NPClass::Invoke.patch * Fix NPN_ReleaseObject() whereby the npobj was still dereferenced even when deallocated: nspluginwrapper-fix-NPN_ReleaseObject.patch * Fix string_of_NPVariant() that did not check for NULL arguments... nspluginwrapper-fix-string_of_NPVariant.patch Regards, Gwenole. -------------- next part -------------- A non-text attachment was scrubbed... Name: nspluginwrapper-fix-NPClass::Invoke.patch Type: application/octet-stream Size: 1391 bytes Desc: not available URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: nspluginwrapper-fix-NPN_ReleaseObject.patch Type: application/octet-stream Size: 990 bytes Desc: not available URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: nspluginwrapper-fix-string_of_NPVariant.patch Type: application/octet-stream Size: 462 bytes Desc: not available URL: From wtogami at redhat.com Sun Nov 23 20:46:20 2008 From: wtogami at redhat.com (Warren Togami) Date: Sun, 23 Nov 2008 15:46:20 -0500 Subject: [PATCH] Re: Crash with Flash 64bit and nspluginwrapper-1.1.4 In-Reply-To: <7FC56CD2-A531-450B-B1FC-67634E325699@free.fr> References: <4923981E.5070908@redhat.com> <7FC56CD2-A531-450B-B1FC-67634E325699@free.fr> Message-ID: <4929C11C.4090402@redhat.com> Gwenole Beauchesne wrote: > Hi Warren, > > Le 19 nov. 08 ? 05:37, Warren Togami a ?crit : > >> http://finance.google.com/finance?q=gol >> Other URL's seem to work with nspluginwrapper-1.1.4 and flash >> 10.0.20.7 64bit, but it crashes npviewer.bin on this Google Finance page. > > I could not reproduce the crash per se but this page indeed exhibited > problems, even with the 32-bit Flash plugin. > > I have attached some patches I committed to SVN, let me know which one > (if any) fixes your problem. Thanks. > > * Fix for NPClass::Invoke() args release (could be your problem, though > it's not a satisfying fix yet): > nspluginwrapper-fix-NPClass::Invoke.patch > > * Fix NPN_ReleaseObject() whereby the npobj was still dereferenced even > when deallocated: > nspluginwrapper-fix-NPN_ReleaseObject.patch > > * Fix string_of_NPVariant() that did not check for NULL arguments... > nspluginwrapper-fix-string_of_NPVariant.patch > > Regards, > Gwenole. Great work! That fixed it. Hmm, other pages that would cause it to crash seem to work now too. Time for 1.1.5 soon? Warren From gb.public at free.fr Sun Nov 23 23:41:21 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Mon, 24 Nov 2008 00:41:21 +0100 Subject: Performance analysis Message-ID: Hi, After 1.2.0 is released, I would like to implement some ideas that should improve performance (reduce RPC trafic and timers). But how could this be evaluated in the first place? Since I never really carried out performance analysis and people expressed concerns with nspluginwrapper performance, here is my initial attempt. The idea is to collect websites that make (intensive) use of Flash. e.g. the following websites could be interesting: - 4 Flash areas: : - 2 Flash areas, including one with some constant animation that could trigger NPN_InvalidateRect() in windowless mode for example: - 1 Flash video that is auto-playing for a couple of minutes: Then, after a certain amount of time (e.g. 2 minutes), we check how much CPU both the viewer and the browser are using (ps with %CPU). I am not sure the latter is a reliable indicator though. I have attached a small script I ran for several hours along with initial results for both windowed and windowless modes. Note: the script was quick-and-dirty code, better run it on a spare machine, if you really want to test it. For example, I use sudo rm and sudo cp because opera requires the plugins to lay in specific directories. Yes, I also wanted to compare with Opera. ;-) BTW, if you know how to make Opera accept plugins from a user directory, please tell (thus avoiding sudo...). Thanks. The results are strange because I didn't expect nspluginwrapper overhead < 5% in windowless mode too. I know there are numerous NPP_HandleEvent() and NPN_InvalidateRect() calls normally. And, I actually got a case where npviewer.bin was taking up to 50% more CPU than in windowed case. So, there may be room to improve the benchmark process. ;-) Hmm, now that I remember, the manual testing was done with a Firefox 3.1 beta and the automated testing was done with Firefox 3.0.3 from openSUSE 11.0 updates. So, Firefox 3.1b was probably causing the excessive amount of events to pass through RPC. I will have to check another day. -------------- next part -------------- A non-text attachment was scrubbed... Name: bench-browsers.sh Type: application/octet-stream Size: 6468 bytes Desc: not available URL: -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: bench-browsers.txt URL: -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: bench-browsers.windowless.txt URL: From gb.public at free.fr Mon Nov 24 07:52:23 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Mon, 24 Nov 2008 08:52:23 +0100 Subject: [ANN] nspluginwrapper 1.1.6 Message-ID: Hi, I am sorry Martin, I hadn't the time to apply your patches yet (PID check + RPC error notification) and that release contains enough fixes already. That's also why this version is 1.1.6 and not 1.2.0. So, what's new since 1.1.4? * Enable glib memory hooks by default * Lower priority of RPC events so that timeouts are triggered first * Fix string_of_NPVariant() that could make some plugins crash * Fix NPClass::Invoke(|Default)() args release * Fix memory leak in NPN_GetStringIdentifiers() * Fix NPN_ReleaseObject() that could dereference a deallocated NPObject * Fix (sync) NPObject referenceCount when the object is passed to the plugin * Fix plugin window resize in XEMBED hack mode * Fix "javascript:" streams requests in standalone player * Fix NPP_Write() and propagate negative lengths too (DiamondX plugin) Announcement (no details this time): Sources: Regards, Gwenole. From gb.public at free.fr Sun Nov 30 09:37:10 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Sun, 30 Nov 2008 10:37:10 +0100 (CET) Subject: [PATCH] Re: Leaked file descriptors from browser to npviewer.bin In-Reply-To: <29969844.1973161228037814520.JavaMail.root@spooler5-g27.priv.proxad.net> Message-ID: <19336972.1973181228037830746.JavaMail.root@spooler5-g27.priv.proxad.net> Hi, ----- "Warren Togami" a ?crit : > Are all of these file descriptors from npviewer.bin and Flash? It > appears that some are leaked from the fork() from firefox. According > to > dwalsh, SELinux automatically closes some leaked file descriptors that > > are not allowed after the transition, but others remain? I have committed the following patch. 1) Close all open files. I used different strategies to determine OPEN_MAX. call to getrlimit() could be removed on contemporary systems though. 2) Add SOCK_CLOEXEC to socket(). I have not tested this one yet (e.g. mplayerplug-in) Regards, Gwenole. -------------- next part -------------- A non-text attachment was scrubbed... Name: nspluginwrapper-cloexec.patch Type: text/x-patch Size: 4497 bytes Desc: not available URL: From gb.public at free.fr Sun Nov 30 16:29:11 2008 From: gb.public at free.fr (Gwenole Beauchesne) Date: Sun, 30 Nov 2008 17:29:11 +0100 Subject: Upstreaming Fedora's patches? In-Reply-To: <491CD36D.9050304@redhat.com> References: <491CD36D.9050304@redhat.com> Message-ID: Hi, Le 14 nov. 08 ? 02:25, Warren Togami a ?crit : > http://cvs.fedoraproject.org/viewvc/rpms/nspluginwrapper/F-10/ > I don't know all the details because Stransky did all the work, but > is there any chance that we can upstream some or all of the patches > that we use against nspluginwrapper? I have committed the PID check and restart patches, and fixed a couple of issues at the same time. In particular, the NPObject hash is now kept for the lifetime of the browser. Otherwise, there were situations where NPobjects were lost since the hash table was renewed, thus losing the correct mapping. Concerning the PID check, is there a real situation that needed that patch? I mean, what plugin fork()'es the browser and does not exec() something afterwards? That check is currently enabled by default, though it can be disabled at compile time with USE_PID_CHECK = 0. Note that the npruntime bridge is not patched yet. Thanks, Gwenole.