Re: [Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
On Thu, 18 Sep 2003, [rkulagow] wrote: --- Additional Comments From [EMAIL PROTECTED] 2003-18-09 21:58 --- It appears that something still isn't right: Thread 6 (Thread 65541 (LWP 25107)): #0 0x40be3006 in nanosleep () from /lib/i686/libc.so.6 No symbol table info available. #1 0x in ?? () No symbol table info available. Thread 5 (Thread 49156 (LWP 25106)): #0 0x40be3006 in nanosleep () from /lib/i686/libc.so.6 No symbol table info available. #1 0x in ?? () No symbol table info available. Can you please fill in a new bug report? I have not looked at it yet. This bug (SIG32 events) is fixed with new glibc 2.3.2-15mdk, available on Cooker. Once people have confirmed that, I will make it a 9.2 update.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-10-10 14:42 --- It works for me with glibc-2.3.2-15mdk. No SIG32 stops. Thanks, Jan -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: NEEDINFO creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-20-09 16:27 --- The main test is to check if info threads returns a list of running threads and that you can change from one thread to another by using the thread n (where n is the thread number) command. Try a bt on each thread to confirm that each thread has its own context. I fixed this on my Mandrake 9.0 by recompiling glibc from the .srpm then copying the unstripped pthread libraries over the top of my existing pthread libraries which I had backed up first. After a quick reboot cos this was quick and dirty, gdb worked fine on my threaded apps. I notice that the spec file for the glibc srpm did not have the pthread libraries in the no strip section - that should be the only required change. -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: NEW creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-18-09 21:58 --- It appears that something still isn't right: Thread 6 (Thread 65541 (LWP 25107)): #0 0x40be3006 in nanosleep () from /lib/i686/libc.so.6 No symbol table info available. #1 0x in ?? () No symbol table info available. Thread 5 (Thread 49156 (LWP 25106)): #0 0x40be3006 in nanosleep () from /lib/i686/libc.so.6 No symbol table info available. #1 0x in ?? () No symbol table info available. -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: NEW creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
Re: [Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
On Mon, 15 Sep 2003, [jan.ciger] wrote: I would really like to see this one fixed, because I saw similar problems since 9.1. It is a huge pain to debug threads when you are unable to use debugger :-( I would really like to see a testcase/sample program so that I can check this is actually fixed. Thanks, Gwenole.
Re: [Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
On Tue, 16 Sep 2003, [jan.ciger] wrote: Compile (first Threads, then Producer, finally OSG itself and dataset), install and try to run any of the demos (e.g. osgview). Ouch! Sorry, I won't. Alternatively, send me the updated glibc, I can test it here. http://people.mandrakesoft.com/~gbeauchesne/glibc/9.2/
Re: [Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
On Tue, 16 Sep 2003, [jan.ciger] wrote: YES !! It works, it seems that you fixed it. I am able to debug the application without any problems now. OK, thanks for testing. Note however, that both rpm those glibc packages are not in 9.2 final ISOs. Normal user will have to grab them from a usual updates mirror.
Re: [Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Gwenole Beauchesne wrote: | On Tue, 16 Sep 2003, [jan.ciger] wrote: | | |YES !! It works, it seems that you fixed it. I am able to debug the |application without any problems now. | | | OK, thanks for testing. Note however, that both rpm those glibc packages | are not in 9.2 final ISOs. Normal user will have to grab them from a usual | updates mirror. Yep, but better than nothing. Hopefully the updates on the mirrors will be available soon after release. Jan - -- Jan Ciger VRlab EPFL Switzerland GPG public key : http://www.keyserver.net/ -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.3 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQE/Zz0+n11XseNj94gRAv+xAKCAsKoLUSJ0FrrxoSXFNdjxV5Z+cACgyuKS SDUuMtKFhQZV3PZeiDho5iM= =XHzY -END PGP SIGNATURE-
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-16-09 17:25 --- Unfortunatelly I have no small (few lines) example causing this bug to manifest itself, but this product does it for me every time : http://openscenegraph.sourceforge.net/ Just download OpenSceneGraph, OpenProducer and OpenThreads (I use CVS version myself, because the download versions didn't work right, but you may try anyway) OpenSceneGraph http://sourceforge.net/project/showfiles.php?group_id=11278release_id=174030 Dataset https://sourceforge.net/project/showfiles.php?group_id=11278release_id=166150 Producer http://www.andesengineering.com/Producer/Download/Producer-0.8.4-1.tar.gz OpenThreads http://prdownloads.sourceforge.net/openthreads/OpenThreads-v1.2dev2-osg0.9.5.tar.gz?download Compile (first Threads, then Producer, finally OSG itself and dataset), install and try to run any of the demos (e.g. osgview). Alternatively, send me the updated glibc, I can test it here. BTW, we have the OSG packages in contribs, but they are obsolete and do not work. Jan -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: NEEDINFO creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-16-09 18:05 --- YES !! It works, it seems that you fixed it. I am able to debug the application without any problems now. Thanks, Jan -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: NEEDINFO creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-16-09 18:34 --- Which files need to be installed from Gwenole's directory to test this? -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: NEEDINFO creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-16-09 19:17 --- I was able to get gdb to run without the SIG32 interruptions, so at least that part works. I will see if I can get a test case on the not showing all thread stacks issue. -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: NEW creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
Re: [Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
Hi, The nostop merely hides the issue and causes problems later on. This severely impacts the ability to debug the multi-threaded application. The fix first implies to fix rpm then rebuild glibc. A workaround would be to manually handle EXCLUDE_FROM_STRIP stuff -debuginfo generation from within glibc package. In either case, this is an 9.2 update candidate. Bye, Gwenole.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-15-09 20:42 --- From what I've been told on #mythtv by the developer, doing the nostop doesn't fix the actual issue, which is that gdb should be working correctly in the first place. The nostop merely hides the issue and causes problems later on. This severely impacts the ability to debug the multi-threaded application. Debian (distro of MythTV developer) doesn't have this issue. -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: NEW creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-15-09 21:43 --- I would really like to see this one fixed, because I saw similar problems since 9.1. It is a huge pain to debug threads when you are unable to use debugger :-( Jan -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: NEEDINFO creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-13-09 09:40 --- [jan.ciger] wrote: I know a couple of folks who switched from Mandrake to RedHat because of the gdb and multi threaded debugging issues... -randy -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: UNCONFIRMED creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-13-09 20:40 --- I would hate to have to do that, but if this problem is not fixed, I may be forced to switch too. We are developing an application, which is heavily multithreaded, so this is quite a showstopper for me. Jan -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: UNCONFIRMED creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.
[Cooker] [Bug 5568] [gdb] gdb continuously stopping for SIG32 events
http://qa.mandrakesoft.com/show_bug.cgi?id=5568 --- Additional Comments From [EMAIL PROTECTED] 2003-11-09 23:36 --- SIG32 is caused by pthreads - it is used internally to notify a thread manager about some things inside of pthreads implementation. It is not a bug per se. This sets gdb to ignore it : handle SIG32 nostop noprint pass BTW, are you seeing also SIGTRAPs ? That is more interesting, since it breaks debugging of multi-threaded applications for me completely. I described the problems I have got here : http://qa.mandrakesoft.com/show_bug.cgi?id=4709 Usually I get SIG32, then set it to nostop,noprint, pass and immediately get SIGTRAP after typing continue. This usually hoses the debugger completely - no stacktrace, kill, quit do not work (hang) etc. -- Configure bugmail: http://qa.mandrakesoft.com/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug, or are watching someone who is. --- Reminder: --- assigned_to: [EMAIL PROTECTED] status: UNCONFIRMED creation_date: description: MythTV (http://www.mythtv.org) is a multi-threaded application. When debugging in gdb, gdb keeps raising SIG32 events, making it difficult to run / debug the application. Additionally, it appears that gdb does not always capture all thread stacks when a backtrace is performed. Not all versions of gdb created by Mandrake have had these issues when debugging multi-threaded applications, but because the segfaults are rare, it is difficult to remember exactly which gdb packages have had issues, and which don't. 5.3-25 definately has an issue with SIG32 events.