On Thu, 28 Feb 2002, Mark Vojkovich wrote: >> > The "nv" driver doesn't have anything to do with RENDER. >> >It's completely out of the picture. I don't see how this >> >can be specific to the "nv" or "nvidia" drivers. Growing >> >memory usage in the server is almost always a client resource >> >leak. You'll notice that the "nv" driver doesn't allocate >> >any memory outside of initialization. >> >> Since then I have learned that this is an X library memory leak >> which the nv driver apparently triggers somehow. I've got an >> Xlib memleak patch I've just applied however i dont know if it >> has to do with this problem or not. > > I can't even think of a case where Xlib can behave differently >depending on an X driver. I don't think it's possible for this to >be a driver specific problem.
The thing is, this only seems to occur on Nvidia hardware, and only when using AA fonts under KDE aparently. I am not able to reproduce it personally since I have no Nvidia hardware, but Bero, and others here have confirmed the bug reports we've received. Initially it was thought to be a driver bug, but now it seems to be something else. Whatever it is, I think more data is needed from people who can reproduce it before it can be pinpointed. I wont be able to look into it for at least a week or so though, but if someone hasn't found it by then, I'll be glad to gather data together, and post it here for others to have a gander at as well. > The X-server growing by 100 meg/hour sounds like a client resource >leak to me. It would have to be done by a client who remained alive >that whole time (like the window manager). It is possible that a >library bug could cause a leak like that. That is what it sounded like to me as well, and it is possible that it is the problem still. If so though, very odd it only happens with certain hardware. It is probably not a driver bug, but possibly a problem that only shows up due to certain assumptions or somesuch in the driver or other code it interacts with. > I once talked about adding an XFree86 option where the server could >print out sum totals of particular resources left unfreed by clients >when they died. I should probably add that sooner rather than later. Sounds like a good idea. Jim Gettys also suggested doing something with NFS to help troubleshoot, however I'd have to hunt my IRC logs to pull up the details. Can you reproduce this on 4.1.0 with nv? -- ---------------------------------------------------------------------- Mike A. Harris Shipping/mailing address: OS Systems Engineer 190 Pittsburgh Ave., Sault Ste. Marie, XFree86 maintainer Ontario, Canada, P6C 5B3 Red Hat Inc. Phone: (705)949-2136 http://www.redhat.com ftp://people.redhat.com/mharris Red Hat XFree86 mailing list: [EMAIL PROTECTED] General open IRC discussion: #xfree86 on irc.openprojects.net ---------------------------------------------------------------------- _______________________________________________ Xpert mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/xpert
