ShapeNotify
Hi all ;), Somebody knows what functions and in what conditions generate ShapeNotify events? I thought that XShape extension calls, like XShapeCombineRectangles will generate them, but experience told me that that's not the case. In the scarce XShape documentation there is no info about this. Just a testimony that ShapeNotify events exists and how to select them... Thanks, -- David Gómez The question of whether computers can think is just like the question of whether submarines can swim. -- Edsger W. Dijkstra ___ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
Re: X11 CVS with linux 2.6.0-test9?
On Tue, 2003-11-25 at 09:09, Fred Heitkamp wrote: Just to aggravate myself I am trying to use XFree86 with linux kernel 2.6.0-test9. I have a Radeon 8500 and a ASUS A7M266-D with two 1600MP processors. I have compiled and installed the radeon kernel modules from the X cvs sources into the kernel modules directory. Why? The one that comes with the kernel is at least as current. When I load the radeon module into the kernel and start X glxinfo shows I have no DRI (first problem) even though dri is enabled. Does LIBGL_DEBUG=verbose glxinfo give a hint? Second problem: X/Gnome 2.4 only runs for a couple minutes before the display freezes hard. The only way to clear the display is to reboot (CTL-ALT-BKSPC does not work; kill -9 X does not clear the display). kill -9 rids the X server of any chance to restore the display; not that it likely has much chance in this case (sounds like a chip lockup). Are you using any AGP or otherwise DRI related options? My question is what is the status of XFree86 and linux 2.6.x kernels? Has been working fine here for a while. :) Does anyone have DRI and Radeon working with 2.6.x on any x86 hardware? I'm running on PPC though. -- Earthling Michel Dnzer | Debian (powerpc), X and DRI developer Software libre enthusiast| http://svcs.affero.net/rm.php?r=daenzer ___ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
XvShmPutImage with XDraw commands
I am using XvShmPutImage to draw video via a YUV overlay on an nVidia Quadro 4 XGL card (PNY 580 XGL). I then do an XSync() to make sure that the video is completely drawn before continuing. I am having a problem where the video is sometimes not keeping up. I am drawing a 1280x960image andhave noticed that the first 14-16 draws take about 22 ms, and all subsequent ones take about 9-10 ms. The timing starts before the XvShmPutImage and ends after the XSync. Sometimes after switching video windows around a few times, I get into a state where the draws get slow again and stay that way indefinitely. The input rate of the data does not seem to be a factor - I tried lowering it from 30 FPS to 10 FPS and got the same results. Trying XvPutImage didn't help either. I get the same problem if I switch to the blitter, rather than the overlay port. I DO NOT get the problem however if I blit RGB data instead of YUV. Does anybody have any idea what is going on and if there is some way to get around this? I am kind of dead in the water right now. Thanks in advance, Steve Thrash P.S. On a related note, does anybody know what the XV_ITURBT_709 attribute does?
Re: must builds be done as root?
On Tue, Nov 25, 2003 at 07:02:29PM +0100, Alexander Stohr wrote: i tried a make World/make on a current CVS snapshot, using RH Linux 7.3 - and found that the drm kernel module makefiles will need root rights on /lib/modules/kernel-version. Set BuildXF86DRM to NO, and this won't be an issue. It is set that way by default for almost all cases. I'll change things so that it is always turned off by default. i know that prior versions of xf86 sources did not force me to be root for building them. and i think it is a bad habit to force users to run such a huge _build_ process as root whilst 99% of the code does not need it. The easiest solution is to build the kernel modules separately from the main build. That's what I do. Unfortunately, the combination of the mechanisms for portably building Linux kernel modules outside of the kernel source tree, and the ways various vendors supply the source for the running kernels requires write access in at least some cases. It has been a while since I played with it, but I think that if you go and do an actual kernel build in the kernel tree that RH installs, then it will be in the state required for the DRM module build to work without needing write access to it. Just checking on a RH7.2 system I have here, that works. Another way would be to have a separate copy of the kernel tree, that your build uid can write to, and point the DRM module build to that. Improvements to the current process are most welcome. What is there right now is the only way I've seen that works for a wide range of kernel versions and the default installations of the more popular linux distros. It's all much simpler on FreeBSD :-) btw, i am a bit in doubt if it is a good idea at all to let compile tasks modyfiy system wide settings at all. what if you run two of such processes concurrently? i would like to see a separation of any root rights requiring tasks into a special build call rather than letting all compile in the same run with the maximum risk factor... % echo '#define BuildXF86DRM NO' config/cf/host.def % make World % su # (cd programs/Xserver/hw/xfree86/os-support/linux/drm/kernel; \ make -f Makefile.linux) # whatever-is-needed-to-install-the-kernel-module (an install target would be nice, but I didn't want to push my luck :-) # make install install.man # ldconfig ProjectRoot/lib ... David -- David Dawes developer/release engineer The XFree86 Project www.XFree86.org/~dawes ___ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
