Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
haha, I can imagine your virtually slapping your hands. It happens on everyone so don't worry. The patch that Ron posted solved the problem, and now all JSBSim aircraft works fine as long as I've tested here. By the way, what kind of tools you are going to use? I'm more interested in these. I know Valgrind, but it doesn't work on Mac OS X. I also want to know some open source static analysis tools that tell us possible bugs like Prevent SQS does. Best, Tat I wanted to try kcachegrind and lcov (code coverage tool), as well as gprof. Jon -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
[Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
Hi there, I found the crash with Concorde is caused by in JSBSim update on 12/01. I made a reverse patch of JSBSim update on that date and applied to the fgfs source on 12/07, then Concorde worked just fine. So I suspect there are some inconsistencies between the update and Concorde's configuration file. I'm not so sure why this update causes crash on a sound related thread but somehow these are related. After the JSBSim update, the log shows tons of errors and warning like: CullVisitor::apply(Geode) detected NaN, depth=nan, center=(0 0 0), matrix={ nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan } Warning:: Picked up error in TriangleIntersect (-109.359 98.4674 -330.521, -55.9012 50.3336 -353.894, -139.955 45.4741 -330.521) (nan,nan,nan) Warning: invalid line segment passed to IntersectVisitor::addLineSegment(..) nan nan nan nan nan nan segment ignored.. Then finally I got OpenAL error (AL_INVALID_VALUE) and fgfs crashed. I'll investigate deeper on this, but I'm very happy if anyone tells me how to fix this. Tat On Dec 4, 2008, at 10:58 AM, Tatsuhiro Nishioka wrote: (2) Concorde crashes fgfs (SIGSEGV) when splash screen goes off. log shows: osgDB ac3d reader: could not find texture concorde-warning.rgb snip concorde systems started, version 2.5 snip 46: GEAR_CONTACT: LEFT_FWD 0 47: GEAR_CONTACT: LEFT_REAR 0 48: GEAR_CONTACT: RIGHT_FWD 0 49: GEAR_CONTACT: RIGHT_REAR 0 Updating adjusted fog parameters. Stopping audio after 0 sec: engine 1 shutdown Stopping audio after 0 sec: engine 2 shutdown Stopping audio after 0 sec: engine 3 shutdown Stopping audio after 0 sec: engine 4 shutdown AICarrier: Inside Operating Box AICarrier: Inside Operating Box -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
On 7 Dec 2008, at 15:37, Tatsuhiro Nishioka wrote: Then finally I got OpenAL error (AL_INVALID_VALUE) and fgfs crashed. I'll investigate deeper on this, but I'm very happy if anyone tells me how to fix this. I see this 'somethimes' - not every day, always with JSBsim aircraft, and I have no explanation, sorry. It doesn't always crash, but I've also seen the OpenAL error (AL_INVALID_VALUE and the the crash without the 'nans' flooding. My feeling is that there is still a memory-corruption issue lurking somewhere. James -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
On dimanche 07 décembre 2008, James Turner wrote: On 7 Dec 2008, at 15:37, Tatsuhiro Nishioka wrote: Then finally I got OpenAL error (AL_INVALID_VALUE) and fgfs crashed. I'll investigate deeper on this, but I'm very happy if anyone tells me how to fix this. I see this 'somethimes' - not every day, always with JSBsim aircraft, and I have no explanation, sorry. It doesn't always crash, but I've also seen the OpenAL error (AL_INVALID_VALUE and the the crash without the 'nans' flooding. My feeling is that there is still a memory-corruption issue lurking somewhere. James Just a way to look for , not to solve , when working on some old aircraft (not in cvs, not yet ..) i noticed that additional tanks which are not declared feeding an engine gave these tons of message and a black screen. If i remove the tanks everything is coming back, right. However, since i split the flight_control in several systems components , (the flight_control , becoming empty in the main branch) i get again the Aircraft loaded and working with the additional tanks. Probably it is a misfit during the initialization of the FDM within JSBSim, which does not come up, with flight_control replaced by systems files Cheers -- Gérard http://pagesperso-orange.fr/GRTux/ J'ai décidé d'être heureux parce que c'est bon pour la santé. Voltaire -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
On dimanche 07 décembre 2008, gerard robin wrote: On dimanche 07 décembre 2008, James Turner wrote: On 7 Dec 2008, at 15:37, Tatsuhiro Nishioka wrote: Then finally I got OpenAL error (AL_INVALID_VALUE) and fgfs crashed. I'll investigate deeper on this, but I'm very happy if anyone tells me how to fix this. I see this 'somethimes' - not every day, always with JSBsim aircraft, and I have no explanation, sorry. It doesn't always crash, but I've also seen the OpenAL error (AL_INVALID_VALUE and the the crash without the 'nans' flooding. My feeling is that there is still a memory-corruption issue lurking somewhere. James Just a way to look for , not to solve , when working on some old aircraft (not in cvs, not yet ..) i noticed that additional tanks which are not declared feeding an engine gave these tons of message and a black screen. If i remove the tanks everything is coming back, right. However, since i split the flight_control in several systems components , (the flight_control , becoming empty in the main branch) i get again the Aircraft loaded and working with the additional tanks. Probably it is a misfit during the initialization of the FDM within JSBSim, which does not come up, with flight_control replaced by systems files Cheers An other complementary information, that problem, to me, is not new . i noticed the bug before the last update ( which is for piston engine, ). 2008-11-30 Modified Files: FGAtmosphere.cpp FGAtmosphere.h FGAuxiliary.cpp FGFCS.cpp FGMassBalance.cpp FGMassBalance.h FGOutput.cpp FGPropulsion.cpp Log Message: Sync. with JSBSim CVS I was unable to explain the error ( it could have been from me ) , and the files i am working on are not available on cvs, so , i shut up :) May be an addition of errors did make that Concorde becoming suddenly not usable. Cheers -- Gérard http://pagesperso-orange.fr/GRTux/ J'ai décidé d'être heureux parce que c'est bon pour la santé. Voltaire -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
Hi Gérard, and James, Thanks for the info, I tried some other JSBSim aircraft like A6M2-jsbsim, f18, f15, etc, and found many fall into either black out or white out, which never happens on fgfs without the JSBSim update. As this update is included in Durk's tarball, we definitely need to either cut out the JSBSim update on 11/30 or fix this issue before the release. I think latter is better but I don't know if we can fix this easily need many helps from you guys. I now feel very much that I wish we had a release branch for separating two things, concentrating on bug-hunting for release and actively committing new features that might be buggy but very attractive. Best, Tat On Dec 8, 2008, at 2:30 AM, gerard robin wrote: On dimanche 07 décembre 2008, gerard robin wrote: On dimanche 07 décembre 2008, James Turner wrote: On 7 Dec 2008, at 15:37, Tatsuhiro Nishioka wrote: Then finally I got OpenAL error (AL_INVALID_VALUE) and fgfs crashed. I'll investigate deeper on this, but I'm very happy if anyone tells me how to fix this. I see this 'somethimes' - not every day, always with JSBsim aircraft, and I have no explanation, sorry. It doesn't always crash, but I've also seen the OpenAL error (AL_INVALID_VALUE and the the crash without the 'nans' flooding. My feeling is that there is still a memory-corruption issue lurking somewhere. James Just a way to look for , not to solve , when working on some old aircraft (not in cvs, not yet ..) i noticed that additional tanks which are not declared feeding an engine gave these tons of message and a black screen. If i remove the tanks everything is coming back, right. However, since i split the flight_control in several systems components , (the flight_control , becoming empty in the main branch) i get again the Aircraft loaded and working with the additional tanks. Probably it is a misfit during the initialization of the FDM within JSBSim, which does not come up, with flight_control replaced by systems files Cheers An other complementary information, that problem, to me, is not new . i noticed the bug before the last update ( which is for piston engine, ). 2008-11-30 Modified Files: FGAtmosphere.cpp FGAtmosphere.h FGAuxiliary.cpp FGFCS.cpp FGMassBalance.cpp FGMassBalance.h FGOutput.cpp FGPropulsion.cpp Log Message: Sync. with JSBSim CVS I was unable to explain the error ( it could have been from me ) , and the files i am working on are not available on cvs, so , i shut up :) May be an addition of errors did make that Concorde becoming suddenly not usable. Cheers -- Gérard http://pagesperso-orange.fr/GRTux/ J'ai décidé d'être heureux parce que c'est bon pour la santé. Voltaire -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
On Mon, 2008-12-08 at 03:44 +0900, Tatsuhiro Nishioka wrote: Hi Gérard, and James, Thanks for the info, I tried some other JSBSim aircraft like A6M2-jsbsim, f18, f15, etc, and found many fall into either black out or white out, which never happens on fgfs without the JSBSim update. As this update is included in Durk's tarball, we definitely need to either cut out the JSBSim update on 11/30 or fix this issue before the release. I think latter is better but I don't know if we can fix this easily need many helps from you guys. I now feel very much that I wish we had a release branch for separating two things, concentrating on bug-hunting for release and actively committing new features that might be buggy but very attractive. Best, Tat Anders found this bug. It was uninitialized variables in FGTank. The change is already in CVS, but here's the patch anyway: Index: src/FDM/JSBSim/models/propulsion/FGTank.cpp === RCS file: /var/cvs/FlightGear-0.9/source/src/FDM/JSBSim/models/propulsion/FGTank.cpp,v retrieving revision 1.6 diff -u -p -r1.6 FGTank.cpp --- src/FDM/JSBSim/models/propulsion/FGTank.cpp 30 Nov 2008 10:44:30 - 1.6 +++ src/FDM/JSBSim/models/propulsion/FGTank.cpp 7 Dec 2008 19:41:38 - @@ -59,6 +59,7 @@ FGTank::FGTank(FGFDMExec* exec, Element* Element* element_Grain; Area = 1.0; Temperature = -.0; + Ixx = Iyy = Izz = 0.0; Auxiliary = exec-GetAuxiliary(); Radius = Capacity = Contents = Standpipe = Length = InnerRadius = 0.0; PropertyManager = exec-GetPropertyManager(); -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
From: Ron Jensen [mailto:[EMAIL PROTECTED] Anders found this bug. It was uninitialized variables in FGTank. The change is already in CVS, but here's the patch anyway: ... I'll consider my hands virtually slapped. My mistake. Thanks, Anders, for fixing that. I'm working on getting a Linux box at home where I can use several good memory/validation tools to make sure that code changes don't have memory leaks in them. Jon -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
Hi Ron, Thanks for this wonderful news. I was also looking for the Leaky/uninititialized variables but was not found these yet, and am very glad to hear this. Great job Anders!, as always. Tat - Tatsuhiro Nishioka On Dec 8, 2008, at 5:38 AM, Ron Jensen [EMAIL PROTECTED] wrote: On Mon, 2008-12-08 at 03:44 +0900, Tatsuhiro Nishioka wrote: Hi Gérard, and James, Thanks for the info, I tried some other JSBSim aircraft like A6M2-jsbsim, f18, f15, etc, and found many fall into either black out or white out, which never happens on fgfs without the JSBSim update. As this update is included in Durk's tarball, we definitely need to either cut out the JSBSim update on 11/30 or fix this issue before the release. I think latter is better but I don't know if we can fix this easily need many helps from you guys. I now feel very much that I wish we had a release branch for separating two things, concentrating on bug-hunting for release and actively committing new features that might be buggy but very attractive. Best, Tat Anders found this bug. It was uninitialized variables in FGTank. The change is already in CVS, but here's the patch anyway: Index: src/FDM/JSBSim/models/propulsion/FGTank.cpp === RCS file: /var/cvs/FlightGear-0.9/source/src/FDM/JSBSim/models/ propulsion/FGTank.cpp,v retrieving revision 1.6 diff -u -p -r1.6 FGTank.cpp --- src/FDM/JSBSim/models/propulsion/FGTank.cpp30 Nov 2008 10:44:30 - 1.6 +++ src/FDM/JSBSim/models/propulsion/FGTank.cpp7 Dec 2008 19:41:38 - @@ -59,6 +59,7 @@ FGTank::FGTank(FGFDMExec* exec, Element* Element* element_Grain; Area = 1.0; Temperature = -.0; + Ixx = Iyy = Izz = 0.0; Auxiliary = exec-GetAuxiliary(); Radius = Capacity = Contents = Standpipe = Length = InnerRadius = 0.0; PropertyManager = exec-GetPropertyManager(); --- --- --- - SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
Hi, On Dec 8, 2008, at 7:55 AM, Jon S. Berndt wrote: From: Ron Jensen [mailto:[EMAIL PROTECTED] Anders found this bug. It was uninitialized variables in FGTank. The change is already in CVS, but here's the patch anyway: ... I'll consider my hands virtually slapped. My mistake. Thanks, Anders, for fixing that. I'm working on getting a Linux box at home where I can use several good memory/validation tools to make sure that code changes don't have memory leaks in them. haha, I can imagine your virtually slapping your hands. It happens on everyone so don't worry. The patch that Ron posted solved the problem, and now all JSBSim aircraft works fine as long as I've tested here. By the way, what kind of tools you are going to use? I'm more interested in these. I know Valgrind, but it doesn't work on Mac OS X. I also want to know some open source static analysis tools that tell us possible bugs like Prevent SQS does. Best, Tat -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Concorde crash on 1.9.0-pre1; (was Re: FlightGear 1.9.0-pre1 release - for Mac OS X)
Hi, On Dec 8, 2008, at 10:11 AM, gerard robin wrote: Anders found this bug. It was uninitialized variables in FGTank. The change is already in CVS, but here's the patch anyway: (snip) I hope that solve the problem for you. Here i have rebuilt FG with that patch and recovered one of my old JSBSim Aircraft FDM which did not worked, i am longer getting the same error , black screen and CullVisitor::apply(Geode) detected NaN, depth=nan, center=(-3.09272 -0.110464 -0.924417), matrix={ nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan } Anyhow, as i told before i have solved it in a different way. As I wrote in my last post, it solved the problem. And I'll try your way when I encounter these errors. Thanks a lot! Tat -- SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel