Bug#467199: [Debian GNUstep maintainers] Bug#467199: RE : Bug#467199: Mpdcon.app: Segfault
PICCA Frédéric-Emmanuel wrote: I attached the xdpyinfo file. Thanks. Could you also please try to run the app like this: $ MPDCon --GNU-Debug=dflt Same thing Crash That's expected, it won't magically fix the bug :-) Can you post the messages that are printed on the console (no need to start it under gdb)? Thanks.
Bug#467199: [Debian GNUstep maintainers] Bug#467199: Mpdcon.app: Segfault
Frédéric-Emmanuel, Could you please rebuild gnustep-base with the attached upstream patch and see if you can reproduce the bug? Thanks. screen-object.diff Description: Binary data
Bug#467199: [Debian GNUstep maintainers] Bug#467199: Bug#467199: Mpdcon.app: Segfault
On Mon, Feb 25, 2008 at 10:43:10PM +0200, Yavor Doganov wrote: Could you please rebuild gnustep-base with the attached upstream patch and see if you can reproduce the bug? Sorry, I messed this up. Here's the patch. 2008-02-25 Fred Kiefer [EMAIL PROTECTED] * Source/x11/XGServer.m (-_screenContextForScreen:): Remove wrong local variable declaration and reorder code. * Source/x11/XGServer.m (-_initXContext:): Don't create default screen. Index: Source/x11/XGServer.m === --- Source/x11/XGServer.m (revision 26069) +++ Source/x11/XGServer.m (working copy) @@ -374,7 +374,6 @@ { int screen_number, display_number; NSString *display_name; - XGScreenContext *screen; display_name = [server_info objectForKey: GSDisplayName]; if (display_name == nil) @@ -437,9 +436,6 @@ screenList = NSCreateMapTable(NSIntMapKeyCallBacks, NSObjectMapValueCallBacks, 20); - screen = [[XGScreenContext alloc] initForDisplay: dpy screen: screen_number]; - AUTORELEASE(screen); - NSMapInsert(screenList, (void *)(uintptr_t)screen_number, (void *)screen); defScreen = screen_number; XSetErrorHandler(XGErrorHandler); @@ -504,12 +500,12 @@ screen = NSMapGet(screenList, (void *)(uintptr_t)screen_number); if (screen == NULL) { - XGScreenContext *screen; screen = [[XGScreenContext alloc] initForDisplay: dpy screen: screen_number]; - AUTORELEASE(screen); NSMapInsert(screenList, (void *)(uintptr_t)screen_number, (void *)screen); + RELEASE(screen); } + return screen; }
Bug#467199: RE : [Debian GNUstep maintainers] Bug#467199: RE : Bug#467199: Mpdcon.app: Segfault
Hello That's expected, it won't magically fix the bug :-) Can you post the messages that are printed on the console (no need to start it under gdb)? Thanks. Here the message: [EMAIL PROTECTED]:~$ MPDCon --GNU-Debug=dflt 2008-02-25 20:50:45.952 MPDCon[3948] File NSData.m: 167. In readContentsOfFile Open (/usr/share/GNUstep/Libraries/gnustep-base/Versions/1.14/Resources/NSTimeZones/localtime) attempt failed - Aucun fichier ou répertoire de ce type 2008-02-25 21:50:45.938 MPDCon[3948] Loading Backend from /usr/lib/GNUstep/Bundles/libgnustep-back-012.bundle 2008-02-25 21:50:46.449 MPDCon[3948] Initializing GNUstep x11 backend. 2008-02-25 21:50:46.751 MPDCon[3948] Loading 1 user defined AppKit bundles 2008-02-25 21:50:46.775 MPDCon[3948] Loaded '/usr/lib/GNUstep/Bundles/Camaelon.themeEngine' 2008-02-25 21:50:47.109 MPDCon[3948] Opened display :0.1, display 0 screen 1 Erreur de segmentation
Bug#467199: RE : [Debian GNUstep maintainers] Bug#467199: Bug#467199: Mpdcon.app: Segfault
Sorry, I messed this up. Here's the patch. I tryed it and it works, but before all windows appeared on the screen, a rectangular shape blink a few times the first time I start MPD. after a warm start no more blink. I do not observe that comportment with Gorm. Nice. See you Frederic here the message: [EMAIL PROTECTED]:~$ MPDCon --GNU-Debug=dflt 2008-02-25 22:43:14.853 MPDCon[19352] File NSData.m: 167. In readContentsOfFile Open (/usr/share/GNUstep/Libraries/gnustep-base/Versions/1.14/Resources/NSTimeZones/localtime) attempt failed - Aucun fichier ou répertoire de ce type 2008-02-25 23:43:14.851 MPDCon[19352] Loading Backend from /usr/lib/GNUstep/Bundles/libgnustep-back-012.bundle 2008-02-25 23:43:14.877 MPDCon[19352] Initializing GNUstep x11 backend. 2008-02-25 23:43:14.998 MPDCon[19352] Loading 1 user defined AppKit bundles 2008-02-25 23:43:14.999 MPDCon[19352] Loaded '/usr/lib/GNUstep/Bundles/Camaelon.themeEngine' 2008-02-25 23:43:15.101 MPDCon[19352] Opened display :0.1, display 0 screen 1 2008-02-25 23:43:15.446 MPDCon[19352] Loading Gorm `/usr/lib/GNUstep/Applications/MPDCon.app/Resources/Player.gorm'... 2008-02-25 23:43:15.446 MPDCon[19352] Loaded data from /usr/lib/GNUstep/Applications/MPDCon.app/Resources/Player.gorm/objects.gorm... 2008-02-25 23:43:15.447 MPDCon[19352] Invoking unarchiver 2008-02-25 23:43:15.561 MPDCon[19352] Calling awakeWithContext 2008-02-25 23:43:15.629 MPDCon[19352] Loading Gorm `/usr/lib/GNUstep/Applications/MPDCon.app/Resources/Preferences.gorm'... 2008-02-25 23:43:15.629 MPDCon[19352] Loaded data from /usr/lib/GNUstep/Applications/MPDCon.app/Resources/Preferences.gorm/objects.gorm... 2008-02-25 23:43:15.629 MPDCon[19352] Invoking unarchiver 2008-02-25 23:43:15.699 MPDCon[19352] Calling awakeWithContext 2008-02-25 23:43:15.804 MPDCon[19352] File NSView.m: 442. In [GSBackendWindowDecorationView -initWithFrame:] given negative height 2008-02-25 23:43:15.804 MPDCon[19352] File NSView.m: 442. In [NSView -initWithFrame:] given negative height 2008-02-25 23:43:15.804 MPDCon[19352] File NSView.m: 1024. In [NSView -setFrame:] given negative height 2008-02-25 23:43:15.805 MPDCon[19352] File NSView.m: 1024. In [GSBackendWindowDecorationView -setFrame:] given negative height
Bug#467199: Mpdcon.app: Segfault
On Mon, 25 Feb 2008 23:43:53 +0100, PICCA Frédéric-Emmanuel [EMAIL PROTECTED] said: Sorry, I messed this up. Here's the patch. I tryed it and it works, Good. I guess I'll roll that patch into a new package and upload soon. but before all windows appeared on the screen, a rectangular shape blink a few times the first time I start MPD. after a warm start no more blink. This seems to be an annoying, but normal, bug. I'm not sure exactly what it is, but I see it too, and I remember some discussion on gnustep-bug saying that it would be too much of an effort to get rid of it. -- Hubert Chathi [EMAIL PROTECTED] -- Jabber: [EMAIL PROTECTED] PGP/GnuPG key: 1024D/124B61FA http://www.uhoreg.ca/ Fingerprint: 96C5 012F 5F74 A5F7 1FF7 5291 AF29 C719 124B 61FA
Bug#467199: RE : RE : Bug#467199: Mpdcon.app: Segfault
We would need further help from you to investigate what's going on. In case you're willing to, please rebuild gnustep-base, gnustep-gui and gnustep-back with debugging symbols and send the backtrace. Ok so, what do I have to add in the debian/rules to build thoses packages in debug mode ? see you
Bug#467199: RE : RE : Bug#467199: Mpdcon.app: Segfault
On Sun, Feb 24, 2008 at 10:09:29AM +0100, PICCA Frйdйric-Emmanuel wrote: Ok so, what do I have to add in the debian/rules to build thoses packages in debug mode ? Nothing, just do: DEB_BUILD_OPTIONS=nostrip dpkg-buildpackage -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#467199: RE : RE : RE : Bug#467199: Mpdcon.app: Segfault
Hello Here the backtrace but no information from the libgnustep-art-012 Maybe we must change the makefile to add the debug information for the libgnustep-art-012 [Switching to Thread 0xb72418d0 (LWP 29525)] 0xb6e3fafc in ?? () from /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/./libgnustep-art-012 (gdb) bt #0 0xb6e3fafc in ?? () from /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/./libgnustep-art-012 #1 0xb6e42321 in ?? () from /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/./libgnustep-art-012 #2 0xb6e370a8 in ?? () from /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/./libgnustep-art-012 #3 0xb6e371da in ?? () from /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/./libgnustep-art-012 #4 0xb7da8d2f in +[GSDisplayServer serverWithAttributes:] (self=0xb7f0e680, _cmd=0xb7e7cb28, attributes=0x0) at GSDisplayServer.m:185 #5 0xb7c1ff94 in -[NSApplication _init] (self=0x81a7378, _cmd=0xb7e7cb78) at NSApplication.m:805 #6 0xb7915929 in -[NSObject performSelector:withObject:] (self=0x81a7378, _cmd=0xb7b4be40, aSelector=0xb7e7cb78, anObject=0x81a7378) at NSObject.m:1947 #7 0xb7975980 in -[NSObject(NSMainThreadPerformAdditions) performSelectorOnMainThread:withObject:waitUntilDone:modes:] ( self=0x81a7378, _cmd=0xb7b4be60, aSelector=0xb7e7cb78, anObject=0x81a7378, aFlag=1 '\001', anArray=0x81a9648) at NSThread.m:1046 #8 0xb7975bba in -[NSObject(NSMainThreadPerformAdditions) performSelectorOnMainThread:withObject:waitUntilDone:] (self=0x81a7378, _cmd=0xb7e7cb80, aSelector=0xb7e7cb78, anObject=0x81a7378, aFlag=1 '\001') at NSThread.m:1093 #9 0xb7c20467 in -[NSApplication init] (self=0x81a7378, _cmd=0xb7e7c918) at NSApplication.m:892 #10 0xb7c1feec in +[NSApplication sharedApplication] (self=0xb7e7c280, _cmd=0xb7e727e0) at NSApplication.m:771 #11 0xb7c06fbc in NSApplicationMain (argc=1, argv=0xbfd3a504) at Functions.m:72 #12 0x0805d5a2 in main ()
Bug#467199: RE : RE : RE : Bug#467199: Mpdcon.app: Segfault
PICCA Frédéric-Emmanuel wrote: Maybe we must change the makefile to add the debug information for the libgnustep-art-012 Yes, please comment line 244 in debian/rules: # strip \ Have you been able to reproduce this with another GNUstep app?
Bug#467199: Mpdcon.app: Segfault
Have you been able to reproduce this with another GNUstep app? Yes with another application gorm but I think it is true with all gnustep applications. here the backtrace: [Switching to Thread 0xb72c88d0 (LWP 16512)] 0xb6ec6afc in -[XGServer(WindowOps) _checkStyle:] (self=0x8217470, _cmd=0xb6f290f0, style=1) at ../../../Source/x11/XGServerWindow.m:571 571 window-depth = context-depth; Current language: auto; currently minimal (gdb) bt #0 0xb6ec6afc in -[XGServer(WindowOps) _checkStyle:] (self=0x8217470, _cmd=0xb6f290f0, style=1) at ../../../Source/x11/XGServerWindow.m:571 #1 0xb6ec9321 in -[XGServer(WindowOps) _setupRootWindow] (self=0x8217470, _cmd=0xb6f27900) at ../../../Source/x11/XGServerWindow.m:1424 #2 0xb6ebe0a8 in -[XGServer _initXContext] (self=0x8217470, _cmd=0xb6f27928) at ../../../Source/x11/XGServer.m:447 #3 0xb6ebe1da in -[XGServer initWithAttributes:] (self=0x8217470, _cmd=0xb7f95740, info=0x0) at ../../../Source/x11/XGServer.m:460 #4 0xb7e2fd2f in +[GSDisplayServer serverWithAttributes:] (self=0xb7f95680, _cmd=0xb7f03b28, attributes=0x0) at GSDisplayServer.m:185 #5 0xb7ca6f94 in -[NSApplication _init] (self=0x81a7378, _cmd=0xb7f03b78) at NSApplication.m:805 #6 0xb799c929 in -[NSObject performSelector:withObject:] (self=0x81a7378, _cmd=0xb7bd2e40, aSelector=0xb7f03b78, anObject=0x81a7378) at NSObject.m:1947 #7 0xb79fc980 in -[NSObject(NSMainThreadPerformAdditions) performSelectorOnMainThread:withObject:waitUntilDone:modes:] ( self=0x81a7378, _cmd=0xb7bd2e60, aSelector=0xb7f03b78, anObject=0x81a7378, aFlag=1 '\001', anArray=0x81a9648) at NSThread.m:1046 #8 0xb79fcbba in -[NSObject(NSMainThreadPerformAdditions) performSelectorOnMainThread:withObject:waitUntilDone:] (self=0x81a7378, _cmd=0xb7f03b80, aSelector=0xb7f03b78, anObject=0x81a7378, aFlag=1 '\001') at NSThread.m:1093 #9 0xb7ca7467 in -[NSApplication init] (self=0x81a7378, _cmd=0xb7f03918) at NSApplication.m:892 #10 0xb7ca6eec in +[NSApplication sharedApplication] (self=0xb7f03280, _cmd=0xb7ef97e0) at NSApplication.m:771 #11 0xb7c8dfbc in NSApplicationMain (argc=1, argv=0xbfab1a84) at Functions.m:72 #12 0x0805d5a2 in main () (gdb)
Bug#467199: [Debian GNUstep maintainers] Bug#467199: Mpdcon.app: Segfault
reassign 467199 gnustep-back0.12-art retitle 467199 Apps crashing on a second display in a dual screen setup thanks Many thanks for the backtrace. What is your DefaultDepth? Please try with a different value. Also, I am curious to know how what are the results with the cairo backend. Could you also please try to run the app like this: $ MPDCon --GNU-Debug=dflt Hubert, do you have any clues? Should we forward this upstream? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#467199: RE : [Debian GNUstep maintainers] Bug#467199: Mpdcon.app: Segfault
Many thanks for the backtrace. What is your DefaultDepth? Please try with a different value. Also, I am curious to know how what are the results with the cairo backend. I attached the xdpyinfo file. Could you also please try to run the app like this: $ MPDCon --GNU-Debug=dflt Same thing Crash xdpyinfo Description: xdpyinfo
Bug#467199: Bug#467199: Mpdcon.app: Segfault
On Sun, 24 Feb 2008 14:47:32 +0200, Yavor Doganov [EMAIL PROTECTED] said: [...] Hubert, do you have any clues? Should we forward this upstream? No, I don't have any clues. I've never encountered this, since I only have one display. Yes, please forward it upstream. Thanks. By the way, I've been planning on adding debugging packages for the libraries. I've already done it for GNUMail. The only thing holding me back right now is trying to decide what to do with gnustep-base and ffcall vs. libffi, which means trying to figure out why things are crashing on amd64 with libffi. I guess I can add the debugging packages to gnustep-gui and gnustep-back, though... -- Hubert Chathi [EMAIL PROTECTED] -- Jabber: [EMAIL PROTECTED] PGP/GnuPG key: 1024D/124B61FA http://www.uhoreg.ca/ Fingerprint: 96C5 012F 5F74 A5F7 1FF7 5291 AF29 C719 124B 61FA -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#467199: Mpdcon.app: Segfault
Package: Mpdcon.app Version: 1.1.99-3 Severity: important Hello, here is the message of gdb on MPDCon. Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0xb72986c0 (LWP 5013)] 0xb6e90cc5 in ?? () from /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/./libgnustep-art-012 (gdb) quit here the valgrind message. ==5089== Invalid read of size 4 ==5089==at 0x58B6CC5: (within /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/libgnustep-art-012) ==5089==by 0x58B950A: (within /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/libgnustep-art-012) ==5089==by 0x58AE1FE: (within /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/libgnustep-art-012) ==5089==by 0x58AE330: (within /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/libgnustep-art-012) ==5089==by 0x426FD2E: (within /usr/lib/libgnustep-gui.so.0.12.0) ==5089==by 0x40E6F93: (within /usr/lib/libgnustep-gui.so.0.12.0) ==5089==by 0x4555B78: (within /usr/lib/libgnustep-base.so.1.14.1) ==5089==by 0x45B5BCF: (within /usr/lib/libgnustep-base.so.1.14.1) ==5089==by 0x45B5E09: (within /usr/lib/libgnustep-base.so.1.14.1) ==5089==by 0x40E7466: (within /usr/lib/libgnustep-gui.so.0.12.0) ==5089==by 0x40E6EEB: (within /usr/lib/libgnustep-gui.so.0.12.0) ==5089==by 0x40CDFBB: NSApplicationMain (in /usr/lib/libgnustep-gui.so.0.12.0) ==5089== Address 0x18 is not stack'd, malloc'd or (recently) free'd ==5089== ==5089== Process terminating with default action of signal 11 (SIGSEGV) ==5089== Access not within mapped region at address 0x18 ==5089==at 0x58B6CC5: (within /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/libgnustep-art-012) ==5089==by 0x58B950A: (within /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/libgnustep-art-012) ==5089==by 0x58AE1FE: (within /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/libgnustep-art-012) ==5089==by 0x58AE330: (within /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/libgnustep-art-012) ==5089==by 0x426FD2E: (within /usr/lib/libgnustep-gui.so.0.12.0) ==5089==by 0x40E6F93: (within /usr/lib/libgnustep-gui.so.0.12.0) ==5089==by 0x4555B78: (within /usr/lib/libgnustep-base.so.1.14.1) ==5089==by 0x45B5BCF: (within /usr/lib/libgnustep-base.so.1.14.1) ==5089==by 0x45B5E09: (within /usr/lib/libgnustep-base.so.1.14.1) ==5089==by 0x40E7466: (within /usr/lib/libgnustep-gui.so.0.12.0) ==5089==by 0x40E6EEB: (within /usr/lib/libgnustep-gui.so.0.12.0) ==5089==by 0x40CDFBB: NSApplicationMain (in /usr/lib/libgnustep-gui.so.0.12.0) ==5089== ==5089== ERROR SUMMARY: 2 errors from 2 contexts (suppressed: 161 from 1) ==5089== malloc/free: in use at exit: 1,679,871 bytes in 31,354 blocks. ==5089== malloc/free: 59,955 allocs, 28,601 frees, 4,185,380 bytes allocated. ==5089== For counts of detected errors, rerun with: -v ==5089== searching for pointers to 31,354 not-freed blocks. ==5089== checked 3,488,232 bytes. ==5089== ==5089== LEAK SUMMARY: ==5089==definitely lost: 5,460 bytes in 674 blocks. ==5089== possibly lost: 338,165 bytes in 8,797 blocks. ==5089==still reachable: 1,336,246 bytes in 21,883 blocks. ==5089== suppressed: 0 bytes in 0 blocks. ==5089== Rerun with --leak-check=full to see details of leaked memory. Erreur de segmentation hope it can help. -- System Information: Debian Release: lenny/sid APT prefers testing APT policy: (500, 'testing'), (500, 'stable') Architecture: i386 (i686) Kernel: Linux 2.6.22-3-686 (SMP w/1 CPU core) Locale: LANG=fr_FR, LC_CTYPE=fr_FR (charmap=ISO-8859-1) Shell: /bin/sh linked to /bin/bash Versions of packages Mpdcon.app depends on: ii gnustep-back0.12 0.12.0-1 The GNUstep GUI Backend ii gnustep-base-runtime 1.14.1-2 GNUstep Base library ii gnustep-gpbs 0.12.0-1 The GNUstep PasteBoard Server ii gnustep-gui-runtime 0.12.0-2 GNUstep GUI Library - runtime file ii libc6 2.7-6
Bug#467199: Mpdcon.app: Segfault
Thanks for the report, but unfortunately I can't reproduce it neither on a sid nor on a lenny machine. picca frederic wrote: here is the message of gdb on MPDCon. Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0xb72986c0 (LWP 5013)] 0xb6e90cc5 in ?? () from /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/./libgnustep-art-012 (gdb) quit At this point, could you please type bt instead of quit? Most probably the backtrace will be meaningless, so it is better to rebuild the package with debugging symbols. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#467199: RE : Bug#467199: Mpdcon.app: Segfault
At this point, could you please type bt instead of quit? Most probably the backtrace will be meaningless, so it is better to rebuild the package with debugging symbols. here the backtrace [Switching to Thread 0xb727f6c0 (LWP 5905)] 0xb6e77cc5 in ?? () from /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/./libgnustep-art-012 (gdb) bt #0 0xb6e77cc5 in ?? () from /usr/lib/GNUstep/Bundles/libgnustep-art-012.bundle/./libgnustep-art-012 #1 0x08255e98 in ?? () #2 0x in ?? () not really interesting. In fact I am using two screens on my laptop, and the program do not crash on the main screen of the laptop (DISPLAY=:0.0). It crashed on the second display :0.1
Bug#467199: RE : Bug#467199: Mpdcon.app: Segfault
PICCA Frédéric-Emmanuel wrote: In fact I am using two screens on my laptop, and the program do not crash on the main screen of the laptop (DISPLAY=:0.0). IIRC from old discussions GNUstep does not support well dual screens, but I may be wrong. Either way, the application shouldn't crash. We would need further help from you to investigate what's going on. In case you're willing to, please rebuild gnustep-base, gnustep-gui and gnustep-back with debugging symbols and send the backtrace. Thanks.
Bug#467199: [Debian GNUstep maintainers] Bug#467199: RE : Bug#467199: Mpdcon.app: Segfault
On Sun, Feb 24, 2008 at 01:03:51AM +0200, Yavor Doganov wrote: We would need further help from you to investigate what's going on. Also, if it's not a hassle, it would be helpful to try other GNUstep apps with this setup. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]