Re: [Flightgear-devel] scenery paging memory leak
Frederic Bouvier wrote : > Durk Talsma wrote: > > > The new scenery (the 0.9.5 release, downloaded using terrasync) > > > > Cheers, > > Durk > > > > On Sunday 18 July 2004 23:08, Frederic Bouvier wrote: > > > Durk Talsma wrote: > > > > > Ehm, to both of you, was this using the real-weather option turned > on > > > > > or anything other which might be related? > > > > > > > > Euhhm, yes, I always have turned the real-weather option on by > default. > > > > > > Are you using the old or the new Scenery. There was some changes in the > > > animation code to support towers and beacons. > > It is too late here so I can't have a look right now, but I am thinking > specifically of the class SGPersonalityBranch in personality.[ch]xx that > stores data for each individual tower. I can't remember exactly but it > occured to me that there could be a leak here, keeping states for unloaded > towers, but I still have to remember how it was designed. > > I will look at it tomorrow if nobody beat me tonight. False alert : there is one personality branch for each instance of models and they are deleted by the decimation thread when the tile is unloaded. We must find something else. -Fred ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] scenery paging memory leak
Durk Talsma wrote: > The new scenery (the 0.9.5 release, downloaded using terrasync) > > Cheers, > Durk > > On Sunday 18 July 2004 23:08, Frederic Bouvier wrote: > > Durk Talsma wrote: > > > > Ehm, to both of you, was this using the real-weather option turned on > > > > or anything other which might be related? > > > > > > Euhhm, yes, I always have turned the real-weather option on by default. > > > > Are you using the old or the new Scenery. There was some changes in the > > animation code to support towers and beacons. It is too late here so I can't have a look right now, but I am thinking specifically of the class SGPersonalityBranch in personality.[ch]xx that stores data for each individual tower. I can't remember exactly but it occured to me that there could be a leak here, keeping states for unloaded towers, but I still have to remember how it was designed. I will look at it tomorrow if nobody beat me tonight. -Fred ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] scenery paging memory leak
The new scenery (the 0.9.5 release, downloaded using terrasync) Cheers, Durk On Sunday 18 July 2004 23:08, Frederic Bouvier wrote: > Durk Talsma wrote: > > > Ehm, to both of you, was this using the real-weather option turned on > > > or anything other which might be related? > > > > Euhhm, yes, I always have turned the real-weather option on by default. > > Are you using the old or the new Scenery. There was some changes in the > animation code to support towers and beacons. > > -Fred > > > > ___ > Flightgear-devel mailing list > [EMAIL PROTECTED] > http://mail.flightgear.org/mailman/listinfo/flightgear-devel ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] scenery paging memory leak
Durk Talsma wrote: > > > Ehm, to both of you, was this using the real-weather option turned on or > > anything other which might be related? > > > > Euhhm, yes, I always have turned the real-weather option on by default. Are you using the old or the new Scenery. There was some changes in the animation code to support towers and beacons. -Fred ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] scenery paging memory leak
> Ehm, to both of you, was this using the real-weather option turned on or > anything other which might be related? > Euhhm, yes, I always have turned the real-weather option on by default. Cheers, Durk ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] scenery paging memory leak
Durk Talsma wrote: I do remember seeing something similar happening, on a long haul test flight between Tokyo and Sydney, which judging from the timestamp on the scenery directory, I did around the 22nd of June. Same problem: Huge memory leak, up to the point where the aircraft became uncontrollable. I reran the same route the next day, and nothing happened. Weird. Anyways, hope that the date of my problem flight may be of any help finding this bug. Actually, I'm testing long range performance right now, doing a 10 hour KSFO to EHAM trip. Memory use appears to be rock solid. Ehm, to both of you, was this using the real-weather option turned on or anything other which might be related? Erik On Sunday 18 July 2004 21:05, Curtis L. Olson wrote: Hey guys, Did we introduce a memory leak when paging scenery recently? I left FG running all night (with ATC and AI traffic disabled) and memory usage was stable. Then I went for 2-3 hour flight and just about filled up all my main RAM + Swap on my linux machine before I finally killed landed and exited. ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] scenery paging memory leak
Curtis L. Olson wrote: Hey guys, Did we introduce a memory leak when paging scenery recently? I left FG running all night (with ATC and AI traffic disabled) and memory usage was stable. Then I went for 2-3 hour flight and just about filled up all my main RAM + Swap on my linux machine before I finally killed landed and exited. I know we were very careful when we set this all up to make sure memory usage was stable during scenery paging and that we didn't leak any memory, but something seems to have crept in along the way with one of the changes. The only thing I can remember recenly is support for terrain/object separation in the directory structure and some sort of simplistic state presorting when loading terrain. Could one of those things be doing this? It's hard to believe that the terrain/object separation has something to do with this, because it only affects the location where the scenery related files can be found. It doesn't touch the loader and or deletion of them. There was also a change in the texture loading code to allow more than one texture to be defined per material, but taking a quick look at that code (defined in mat.cxx search for \"texture\") didn't show anything obvious. And there is the code change to allow a different texture path for a 3d model but again, I find that one very unlikely to cause a memory leak. So I can't think of anything obvious right now. Erik ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] scenery paging memory leak
Curt, I do remember seeing something similar happening, on a long haul test flight between Tokyo and Sydney, which judging from the timestamp on the scenery directory, I did around the 22nd of June. Same problem: Huge memory leak, up to the point where the aircraft became uncontrollable. I reran the same route the next day, and nothing happened. Weird. Anyways, hope that the date of my problem flight may be of any help finding this bug. Actually, I'm testing long range performance right now, doing a 10 hour KSFO to EHAM trip. Memory use appears to be rock solid. Cheers, Durk On Sunday 18 July 2004 21:05, Curtis L. Olson wrote: > Hey guys, > > Did we introduce a memory leak when paging scenery recently? I left FG > running all night (with ATC and AI traffic disabled) and memory usage > was stable. Then I went for 2-3 hour flight and just about filled up > all my main RAM + Swap on my linux machine before I finally killed > landed and exited. I know we were very careful when we set this all up > to make sure memory usage was stable during scenery paging and that we > didn't leak any memory, but something seems to have crept in along the > way with one of the changes. The only thing I can remember recenly is > support for terrain/object separation in the directory structure and > some sort of simplistic state presorting when loading terrain. Could > one of those things be doing this? Has anything else changed? This is > not good if someone wants to do long flights > > Thanks, > > Curt. ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
[Flightgear-devel] scenery paging memory leak
Hey guys, Did we introduce a memory leak when paging scenery recently? I left FG running all night (with ATC and AI traffic disabled) and memory usage was stable. Then I went for 2-3 hour flight and just about filled up all my main RAM + Swap on my linux machine before I finally killed landed and exited. I know we were very careful when we set this all up to make sure memory usage was stable during scenery paging and that we didn't leak any memory, but something seems to have crept in along the way with one of the changes. The only thing I can remember recenly is support for terrain/object separation in the directory structure and some sort of simplistic state presorting when loading terrain. Could one of those things be doing this? Has anything else changed? This is not good if someone wants to do long flights Thanks, Curt. -- Curtis Olsonhttp://www.flightgear.org/~curt HumanFIRST Program http://www.humanfirst.umn.edu/ FlightGear Project http://www.flightgear.org Unique text:2f585eeea02e2c79d7b1d8c4963bae2d ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel