Re: Newbie question, difference between offscreen and onscreen image formats?
On Sat, 7 Feb 2004, Suhaib Chishtie wrote: Sent: Saturday, February 07, 2004 6:29 PM Subject: Re: Newbie question, difference between offscreen and onscreen image formats? The image formats registered via xf86XVScreenInit are ones that are exported to the clients. The ones registered via xf86XVRegisterOffscreenImages are for internal use only. The idea being that they are the hardware's native overlay format or formats that can be exposed to other parts of the server such as a module which uses V4L. Reasons why some formats would be exposed as client XvImages while they wouldn't be exposed as OffscreenImages could be: * The XvImage formats aren't implemented via the overlay, but with some texture or blit mechanism. * The XvImage formats, though using the overlay, aren't the native hardware format and require CPU reformating on the copy. * There are hardware or software complications related to other devices bus mastering data into those overlay formats. * Simply an oversight, or somebody just didn't get around to adding them, or didn't have a way to adequetly test them. Thanks for the explaination. I know chiip and tech 69000/69030 supports both YUV and RGB overlays. So I guess, all the formats should also be registered as offscreen formats. Any idea what else needs to be changed? Like which functions get called to set/start/stop/capture overlays. Or should I contact one of the chips driver's developer? I think you should at least to verify that there's not a reason why it shouldn't be exposed. Mark. ___ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
Re: Newbie question, difference between offscreen and onscreen image formats?
- Original Message - From: Mark Vojkovich [EMAIL PROTECTED] To: Suhaib Chishtie [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Sent: Monday, February 09, 2004 5:36 PM Subject: Re: Newbie question, difference between offscreen and onscreen image formats? I know chiip and tech 69000/69030 supports both YUV and RGB overlays. So I guess, all the formats should also be registered as offscreen formats. Any idea what else needs to be changed? Like which functions get called to set/start/stop/capture overlays. Or should I contact one of the chips driver's developer? I think you should at least to verify that there's not a reason why it shouldn't be exposed. I'll modify the code and make it work first. I'm sure its doable, I have it running under DOS using debug code and windows using my frame grabber driver and stock chips windows driver. Suhaib ___ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
Re: Newbie question, difference between offscreen and onscreen image formats?
Sent: Saturday, February 07, 2004 6:29 PM Subject: Re: Newbie question, difference between offscreen and onscreen image formats? The image formats registered via xf86XVScreenInit are ones that are exported to the clients. The ones registered via xf86XVRegisterOffscreenImages are for internal use only. The idea being that they are the hardware's native overlay format or formats that can be exposed to other parts of the server such as a module which uses V4L. Reasons why some formats would be exposed as client XvImages while they wouldn't be exposed as OffscreenImages could be: * The XvImage formats aren't implemented via the overlay, but with some texture or blit mechanism. * The XvImage formats, though using the overlay, aren't the native hardware format and require CPU reformating on the copy. * There are hardware or software complications related to other devices bus mastering data into those overlay formats. * Simply an oversight, or somebody just didn't get around to adding them, or didn't have a way to adequetly test them. Thanks for the explaination. I know chiip and tech 69000/69030 supports both YUV and RGB overlays. So I guess, all the formats should also be registered as offscreen formats. Any idea what else needs to be changed? Like which functions get called to set/start/stop/capture overlays. Or should I contact one of the chips driver's developer? Suhaib ___ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
Newbie question, difference between offscreen and onscreen image formats?
I'm trying to make v4l driver to work with chips (Chips and tech 69030) x11 driver. The frame grabber can only write directly to offscreen video memory in YUV format. But the X11 v4l driver reports RGB as the only offscreen image format available. Althoough the chips does support both RGB and YUV for offscreen overlays. Further investigation in chips driver revealed that chips driver is registering only RGB as offscreen format and both RGB and YUV as supported image format by calling xf86XVListGenericAdaptors and xf86XVScreenInit functions. My question is, What is the difference between offscreen image format registered via xf86XVRegisterOffscreenImages and image formats registered via calls to xf86XVListGenericAdaptors and xf86XVScreenInit? Almost all of the video drivers register only one (usually YUV) format as Offscreen image format and more formats via the screeninit calls. If they support overlay in both RGB and YUV formats, should they not be registering all the formats as offscreen formats too? Any suggestion/advice is appreciated Cheers, Suhaib
Re: Newbie question, difference between offscreen and onscreen image formats?
On Sat, 7 Feb 2004, Suhaib Chishtie wrote: I'm trying to make v4l driver to work with chips (Chips and tech 69030) x11 driver. The frame grabber can only write directly to offscreen video memory in YUV format. But the X11 v4l driver reports RGB as the only offscreen image format available. Althoough the chips does support both RGB and YUV for offscreen overlays. Further investigation in chips driver revealed that chips driver is registering only RGB as offscreen format and both RGB and YUV as supported image format by calling xf86XVListGenericAdaptors and xf86XVScreenInit functions. My question is, What is the difference between offscreen image format registered via xf86XVRegisterOffscreenImages and image formats registered via calls to xf86XVListGenericAdaptors and xf86XVScreenInit? Almost all of the video drivers register only one (usually YUV) format as Offscreen image format and more formats via the screeninit calls. If they support overlay in both RGB and YUV formats, should they not be registering all the formats as offscreen formats too? The image formats registered via xf86XVScreenInit are ones that are exported to the clients. The ones registered via xf86XVRegisterOffscreenImages are for internal use only. The idea being that they are the hardware's native overlay format or formats that can be exposed to other parts of the server such as a module which uses V4L. Reasons why some formats would be exposed as client XvImages while they wouldn't be exposed as OffscreenImages could be: * The XvImage formats aren't implemented via the overlay, but with some texture or blit mechanism. * The XvImage formats, though using the overlay, aren't the native hardware format and require CPU reformating on the copy. * There are hardware or software complications related to other devices bus mastering data into those overlay formats. * Simply an oversight, or somebody just didn't get around to adding them, or didn't have a way to adequetly test them. Mark. ___ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
newbie question
Hello all, i have a question about Xfree internals. i would like to know how xfree capture the keyboard and mouse events. Is it with a simple read ? And does it rely on the OS? i know that in text mode, the characters comes from read(), which comes from somewhere in the tty driver... any direct answer, (accurate) source code pointers, or urls are welcome :) i hope this is ok to ask for that here. thx in advance :) btw, i haven't suscribed to the ml, plz reply by email -- David ___ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
Re: the relationship between framebuffer and starting speed----a newbie question
--Original Message Text--- From: Tao, Qian (Ý+ IES) Date: Sun, 17 Aug 2003 15:04:31 +0800 the relationship between framebuffer and starting speeda newbie question I want to make Xserver start more fast. Why? What is the point? The typical X server starts once and runs for days, weeks, or months. Shaving a second off of initialization time is nothing but a waste of optimization time. Concentrate on things that are done over and over: that's where you make your optimization wins. And I study the code in dix/main.c,and reduce some code in Keyboard and Mouse initialization.I find that it takes only 0.5 second from the start of main.c to the dispatch() function(dix/main.c). But when I type in XFree86,It seems that from the the typing-XFree86 to the bitmap comes out on screen takes me 2~3 second. I think that most time is spent in from RAM to Screen. What makes you think so? It is true that copying a screen-sized bitmap to your frame buffer will take time, but you would need to do some profiling to know for sure. It's probably not more than tens of milliseconds. Some drivers use I/O ports to write to the VGA-compatible registers on their chips. I/O ports on the x86 are notoriously slow. Some drivers actually have to delay at certain points to allow the hardware to initialize itself. It also takes time to load XFree86 and all of its drivers from disk: there's a lot of code to load and link. But I also hear that framebuffer doesn't support acclerration and it doesn't work if I play DVD. In fact I don't know the exact meaning of framebuffer. The word "frame buffer" refers to the area in the video card's memory that contains the screen image. The frame buffer driver is a basic driver that treats that memory as a simple, dumb bitmap with no acceleration. It is a way to support graphics chips that do not yet have accelerated drivers. Can you explain to me,the relationship between framebuffer and starting speed ,and If framebuffer can support accleration. No, the frame buffer driver does not do any acceleration. It is completely generic: all it gets is a memory pointer. It does not know what kind of graphics chip it has, so cannot know how to use the acceleration features of any single chip. -- -TimRoberts,[EMAIL PROTECTED] Providenza&Boekelheide,Inc.