Re: DRI on FreeBSD with a PCIe X800
So I thought I might try and debug this problem further... I was looking at the various options in the radeon man page and came across the BusType option. I had tried both leaving that option out, and setting the option to PCIE. This morning, I decided to try setting it to PCI though, according to the man page, PCIE simply falls back to PCI at the present time, so I figured it wouldn't make a difference... Except that it does make a difference. If I set the BusType to PCI dmesg shows: info: [drm] Initialized radeon 1.25.0 20060524 info: [drm] Setting GART location based on new memory map error: [drm:pid1311:radeon_do_init_cp] *ERROR* Cannot use PCI Express without GART in FB memory The only thing different that shows up in the Xorg log file is: (WW) RADEON(0): Direct rendering disabled This is after all the usual DRI setup including the opening of the /dev/dri/card0 device. I'm not sure if this is at all related to the problem I'm seeing when I use BusType PCIE and get the screen corruption... However, at the very least it shows that BusType PCIE does not, in fact, fall back to PCI and something is being setup differently. If anyone wants to look at the full Xorg from the PCI session, it's available at: http://www.visualtech.com/Xorg.0.log.PCI.txt.gz Dmesg: http://www.visualtech.com/dmesg.txt.gz If anyone has any tips or pointers on debugging either of these problems further, I'm all ears :-) Adam Adam K Kirchhoff wrote: For anyone interested in following this, I've opened up a problem report: http://www.freebsd.org/cgi/query-pr.cgi?pr=106370 Adam Hello all, I'm having a problem getting direct rendering working on one of my workstations. I'm running -CURRENT from November 17th with Xorg installed from the modular Xorg ports tree yesterday (though I first noticed this a couple weeks back when I built modular Xorg using jhbuild): [ [EMAIL PROTECTED] - ~ ]: Xorg -version X Window System Version 7.1.1 Release Date: 12 May 2006 X Protocol Version 11, Revision 0, Release 7.1.1 Build Operating System: FreeBSD 7.0-CURRENT i386 Current Operating System: FreeBSD sorrow.ashke.com 7.0-CURRENT FreeBSD 7.0-CURRENT #7: Tue Nov 14 08:33:41 EST 2006 [EMAIL PROTECTED]:/usr/obj/usr/src/sys/GENERIC i386 Build Date: 28 November 2006 Before reporting problems, check http://wiki.x.org to make sure that you have the latest version. Module Loader present If I boot up with DRI enabled in the config file, the server starts, but the very top of the screen shows some visual corruption. http://www.visualtech.com/screenshot.png I dropped the resolution of the image from 2304x864 to 1600x800, but you can still make out the corruption. What's particularly odd, though, is that the root window is never drawn. The background you see is actually the background from my previous X session (when I had DRI disabled), using windowmaker. This time I launched X and had fvwm2 in my .xinitrc file (you can see the outline of the fvwm pager in the screenshot, though that never finished drawing, either). After that nothing else gets drawn. I can move the mouse pointer, but that's about it. I can safely kill X and restart it, but the same thing happens unless I disable DRI. In comparison, I have another workstation with an AGP x700. -CURRENT from the same date, and modular Xorg from the ports tree from yesterday, too. It works just fine (start up fine, and the mesa demos run with acceleration). You can find the Xorg log file from the PCIe system at http://www.visualtech.com/Xorg.0.log.gz Any ideas? Thanks! Adam ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-x11 To unsubscribe, send any mail to [EMAIL PROTECTED] - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: DRI on FreeBSD with a PCIe X800
I was looking at the various options in the radeon man page and came across the BusType option. I had tried both leaving that option out, and setting the option to PCIE. This morning, I decided to try setting it to PCI though, according to the man page, PCIE simply falls back to PCI at the present time, so I figured it wouldn't make a difference... Except that it does make a difference. If I set the BusType to PCI dmesg shows: PCIE never falls back to PCI so the manpage is wrong, this used to be true before I implemented PCIE GART support, the PCIE cards don't have the PCI gart unlike the AGP cards.. Dave. - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: DRI on FreeBSD with a PCIe X800
Adam K Kirchhoff schreef: For anyone interested in following this, I've opened up a problem report: http://www.freebsd.org/cgi/query-pr.cgi?pr=106370 The screen corruption picture looks slightly familiar. I also get some red lines/pixels at the top of the screen when I switch to a console and back again. After a few seconds the screen is ok again. Dell Latitude D810, config at http://home.tiscali.nl/rladan/conf/ Adam Regards, Rene -- GPG fingerprint = E738 5471 D185 7013 0EE0 4FC8 3C1D 6F83 12E1 84F6 (subkeys.pgp.net) It won't fit on the line. -- me, 2001 - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: DRI on FreeBSD with a PCIe X800
On Sat, 2006-12-02 at 12:14 -0500, Adam K Kirchhoff wrote: So something occurred to me last night... I've seen these same symptoms before when trying to get DRI working on FreeBSD a long time ago... On another machine, if I accidentally had the AGPSize set to a value higher than was set in the BIOS, I saw the exact same problem in FreeBSD (but not in Linux, which didn't have any problem with that particular situation). I don't have any config options set for GARTSize, but could something similar be happening now? I think that's unlikely with PCIe, although my first guess would have been something related to GART as well given that the same configuration seems to run or not depending on the OS. Maybe the OSs set up something differently related to PCIe. I'm having a problem getting direct rendering working on one of my workstations. I'm running FreeBSD -CURRENT from November 17th with Xorg installed from the modular Xorg ports tree yesterday (though I first noticed this a couple weeks back when I built modular Xorg using jhbuild): If you're saying that some previous version worked with the same configuration, could you try isolating the regression with git-bisect? If I boot up with DRI enabled in the config file, the server starts, but the very top of the screen shows some visual corruption. http://www.visualtech.com/screenshot.png I dropped the resolution of the image from 2304x864 to 1600x800, but you can still make out the corruption. What's particularly odd, though, is that the root window is never drawn. The background you see is actually the background from my previous X session (when I had DRI disabled), using windowmaker. This time I launched X and had fvwm2 in my .xinitrc file (you can see the outline of the fvwm pager in the screenshot, though that never finished drawing, either). After that nothing else gets drawn. Sounds like a GPU hang/lockup. I can move the mouse pointer, but that's about it. I can safely kill X and restart it, but the same thing happens unless I disable DRI. That's a relatively graceful way for it to deal with the above though. :} This seems to be a FreeBSD specific problem as the same PCIe x800 works fine with the OSS drivers under Linux, but no on on the freebsd-x11 lists seems to have any ideas, and I'm running out of ideas. I thought some of the great minds on this list might be able to shed some light. Any interesting differences between the server log files, DRM related kernel output etc. between OSs? -- Earthling Michel Dänzer | http://tungstengraphics.com Libre software enthusiast | Debian, X and DRI developer - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: DRI on FreeBSD with a PCIe X800
On Mon, 2006-12-04 at 14:42 +0100, Michel Dänzer wrote: On Sat, 2006-12-02 at 12:14 -0500, Adam K Kirchhoff wrote: So something occurred to me last night... I've seen these same symptoms before when trying to get DRI working on FreeBSD a long time ago... On another machine, if I accidentally had the AGPSize set to a value higher than was set in the BIOS, I saw the exact same problem in FreeBSD (but not in Linux, which didn't have any problem with that particular situation). I don't have any config options set for GARTSize, but could something similar be happening now? I think that's unlikely with PCIe, although my first guess would have been something related to GART as well given that the same configuration seems to run or not depending on the OS. Maybe the OSs set up something differently related to PCIe. I'm having a problem getting direct rendering working on one of my workstations. I'm running FreeBSD -CURRENT from November 17th with Xorg installed from the modular Xorg ports tree yesterday (though I first noticed this a couple weeks back when I built modular Xorg using jhbuild): If you're saying that some previous version worked with the same configuration, could you try isolating the regression with git-bisect? If I boot up with DRI enabled in the config file, the server starts, but the very top of the screen shows some visual corruption. http://www.visualtech.com/screenshot.png I dropped the resolution of the image from 2304x864 to 1600x800, but you can still make out the corruption. What's particularly odd, though, is that the root window is never drawn. The background you see is actually the background from my previous X session (when I had DRI disabled), using windowmaker. This time I launched X and had fvwm2 in my .xinitrc file (you can see the outline of the fvwm pager in the screenshot, though that never finished drawing, either). After that nothing else gets drawn. Sounds like a GPU hang/lockup. I can move the mouse pointer, but that's about it. I can safely kill X and restart it, but the same thing happens unless I disable DRI. That's a relatively graceful way for it to deal with the above though. :} This seems to be a FreeBSD specific problem as the same PCIe x800 works fine with the OSS drivers under Linux, but no on on the freebsd-x11 lists seems to have any ideas, and I'm running out of ideas. I thought some of the great minds on this list might be able to shed some light. Any interesting differences between the server log files, DRM related kernel output etc. between OSs? The log file on FreeBSD doesn't show the same warning as the log file on linux about the support being experimental. Hmmm... FreeBSD shows: (II) RADEON(0): [drm] DRM interface version 1.2 (II) RADEON(0): [drm] created radeon driver at busid pci::07:00.0 (II) RADEON(0): [drm] added 8192 byte SAREA at 0xc56df000 (II) RADEON(0): [drm] mapped SAREA 0xc56df000 to 0x28557000 (II) RADEON(0): [drm] framebuffer handle = 0xc000 (II) RADEON(0): [drm] added 1 reserved context for kernel (II) RADEON(0): [pci] 8192 kB allocated with handle 0x (II) RADEON(0): [pci] ring handle = 0xc56e3000 (II) RADEON(0): [pci] Ring mapped at 0x38648000 (II) RADEON(0): [pci] Ring contents 0xfc181616 (II) RADEON(0): [pci] ring read ptr handle = 0xc57e4000 (II) RADEON(0): [pci] Ring read ptr mapped at 0x28559000 (II) RADEON(0): [pci] Ring read ptr contents 0xff3a492d (II) RADEON(0): [pci] vertex/indirect buffers handle = 0xc57e5000 (II) RADEON(0): [pci] Vertex/indirect buffers mapped at 0x38749000 (II) RADEON(0): [pci] Vertex/indirect buffers contents 0xff252d1e (II) RADEON(0): [pci] GART texture map handle = 0xc59e5000 (II) RADEON(0): [pci] GART Texture map mapped at 0x38949000 (II) RADEON(0): [drm] register handle = 0xdfbe (II) RADEON(0): [dri] Visual configs initialized And linux shows: (II) RADEON(0): [drm] DRM interface version 1.3 (II) RADEON(0): [drm] created radeon driver at busid pci::07:00.0 (II) RADEON(0): [drm] added 8192 byte SAREA at 0xf8dee000 (II) RADEON(0): [drm] mapped SAREA 0xf8dee000 to 0xb7bac000 (II) RADEON(0): [drm] framebuffer handle = 0xc000 (II) RADEON(0): [drm] added 1 reserved context for kernel (II) RADEON(0): [pci] 8192 kB allocated with handle 0xf8f2e000 (II) RADEON(0): [pci] ring handle = 0xf8f2e000 (II) RADEON(0): [pci] Ring mapped at 0xa797 (II) RADEON(0): [pci] Ring contents 0x (II) RADEON(0): [pci] ring read ptr handle = 0xf902f000 (II) RADEON(0): [pci] Ring read ptr mapped at 0xb7bab000 (II) RADEON(0): [pci] Ring read ptr contents 0x (II) RADEON(0): [pci] vertex/indirect buffers handle = 0xf903 (II) RADEON(0): [pci] Vertex/indirect buffers mapped at 0xa777 (II) RADEON(0): [pci] Vertex/indirect buffers contents 0x (II) RADEON(0): [pci] GART texture map handle = 0xf923 (II) RADEON(0): [pci]
Re: DRI on FreeBSD with a PCIe X800
On Mon, 2006-12-04 at 09:26 -0500, Adam K Kirchhoff wrote: The log file on FreeBSD doesn't show the same warning as the log file on linux about the support being experimental. Hmmm... FreeBSD shows: (II) RADEON(0): [drm] DRM interface version 1.2 [...] And linux shows: (II) RADEON(0): [drm] DRM interface version 1.3 So it looks like you're using a newer DRM on Linux but a newer xf86-video-ati on FreeBSD. Does making either of these equal between OSs change anything? -- Earthling Michel Dänzer | http://tungstengraphics.com Libre software enthusiast | Debian, X and DRI developer - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: DRI on FreeBSD with a PCIe X800
On Mon, 2006-12-04 at 15:42 +0100, Michel Dänzer wrote: On Mon, 2006-12-04 at 09:26 -0500, Adam K Kirchhoff wrote: The log file on FreeBSD doesn't show the same warning as the log file on linux about the support being experimental. Hmmm... FreeBSD shows: (II) RADEON(0): [drm] DRM interface version 1.2 [...] And linux shows: (II) RADEON(0): [drm] DRM interface version 1.3 So it looks like you're using a newer DRM on Linux but a newer xf86-video-ati on FreeBSD. Does making either of these equal between OSs change anything? Well, the DRM module itself is the same version as under linux (1.25.0). Are you referring to libdrm? I'll see if I can upgrade that on FreeBSD. Adam - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: DRI on FreeBSD with a PCIe X800
On Mon, 2006-12-04 at 12:49 -0500, Adam K Kirchhoff wrote: On Mon, 2006-12-04 at 15:42 +0100, Michel Dänzer wrote: On Mon, 2006-12-04 at 09:26 -0500, Adam K Kirchhoff wrote: The log file on FreeBSD doesn't show the same warning as the log file on linux about the support being experimental. Hmmm... FreeBSD shows: (II) RADEON(0): [drm] DRM interface version 1.2 [...] And linux shows: (II) RADEON(0): [drm] DRM interface version 1.3 So it looks like you're using a newer DRM on Linux but a newer xf86-video-ati on FreeBSD. Does making either of these equal between OSs change anything? Well, the DRM module itself is the same version as under linux (1.25.0). Are you referring to libdrm? No, see the different DRM interface versions (corresponding to the 'drm' kernel module) above. Maybe they're just different between OSs though. -- Earthling Michel Dänzer | http://tungstengraphics.com Libre software enthusiast | Debian, X and DRI developer - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: DRI on FreeBSD with a PCIe X800
On Mon, 2006-12-04 at 19:01 +0100, Michel Dänzer wrote: On Mon, 2006-12-04 at 12:49 -0500, Adam K Kirchhoff wrote: On Mon, 2006-12-04 at 15:42 +0100, Michel Dänzer wrote: On Mon, 2006-12-04 at 09:26 -0500, Adam K Kirchhoff wrote: The log file on FreeBSD doesn't show the same warning as the log file on linux about the support being experimental. Hmmm... FreeBSD shows: (II) RADEON(0): [drm] DRM interface version 1.2 [...] And linux shows: (II) RADEON(0): [drm] DRM interface version 1.3 So it looks like you're using a newer DRM on Linux but a newer xf86-video-ati on FreeBSD. Does making either of these equal between OSs change anything? Well, the DRM module itself is the same version as under linux (1.25.0). Are you referring to libdrm? No, see the different DRM interface versions (corresponding to the 'drm' kernel module) above. Maybe they're just different between OSs though. Yeah, I think that must be the case. I pulled the latest drm from the git repo and built both a newer libdrm and a newer drm.ko and radeon.ko kernel modules. I get the same results (and the same output in the Xorg log file about the interface version). I've also updated the xf86-video-ati driver on the linux installation to 6.6.3, so it's now the same version as the FreeBSD driver. DRI still works fine on the linux installation. Any other ideas? :-) Adam - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: DRI on FreeBSD with a PCIe X800
On Mon, 2006-12-04 at 13:17 -0500, Adam K Kirchhoff wrote: On Mon, 2006-12-04 at 19:01 +0100, Michel Dänzer wrote: On Mon, 2006-12-04 at 12:49 -0500, Adam K Kirchhoff wrote: On Mon, 2006-12-04 at 15:42 +0100, Michel Dänzer wrote: On Mon, 2006-12-04 at 09:26 -0500, Adam K Kirchhoff wrote: The log file on FreeBSD doesn't show the same warning as the log file on linux about the support being experimental. Hmmm... FreeBSD shows: (II) RADEON(0): [drm] DRM interface version 1.2 [...] And linux shows: (II) RADEON(0): [drm] DRM interface version 1.3 So it looks like you're using a newer DRM on Linux but a newer xf86-video-ati on FreeBSD. Does making either of these equal between OSs change anything? Well, the DRM module itself is the same version as under linux (1.25.0). Are you referring to libdrm? No, see the different DRM interface versions (corresponding to the 'drm' kernel module) above. Maybe they're just different between OSs though. Yeah, I think that must be the case. I pulled the latest drm from the git repo and built both a newer libdrm and a newer drm.ko and radeon.ko kernel modules. I get the same results (and the same output in the Xorg log file about the interface version). I've also updated the xf86-video-ati driver on the linux installation to 6.6.3, so it's now the same version as the FreeBSD driver. DRI still works fine on the linux installation. Any other ideas? :-) Adam For what it's worth, I've also heard from a user on the freebsd-x11 mailing list with a PCIe x600 mobility that has fully functional 2D with DRI enabled. Adam - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
Re: DRI on FreeBSD with a PCIe X800
So something occurred to me last night... I've seen these same symptoms before when trying to get DRI working on FreeBSD a long time ago... On another machine, if I accidentally had the AGPSize set to a value higher than was set in the BIOS, I saw the exact same problem in FreeBSD (but not in Linux, which didn't have any problem with that particular situation). I don't have any config options set for GARTSize, but could something similar be happening now? Adam On Fri, 2006-12-01 at 15:16 -0500, Adam K Kirchhoff wrote: Hello all, I'm having a problem getting direct rendering working on one of my workstations. I'm running FreeBSD -CURRENT from November 17th with Xorg installed from the modular Xorg ports tree yesterday (though I first noticed this a couple weeks back when I built modular Xorg using jhbuild): [ [EMAIL PROTECTED] - ~ ]: Xorg -version X Window System Version 7.1.1 Release Date: 12 May 2006 X Protocol Version 11, Revision 0, Release 7.1.1 Build Operating System: FreeBSD 7.0-CURRENT i386 Current Operating System: FreeBSD sorrow.ashke.com 7.0-CURRENT FreeBSD 7.0-CURRENT #7: Tue Nov 14 08:33:41 EST 2006 [EMAIL PROTECTED]:/usr/obj/usr/src/sys/GENERIC i386 Build Date: 28 November 2006 Before reporting problems, check http://wiki.x.org to make sure that you have the latest version. Module Loader present If I boot up with DRI enabled in the config file, the server starts, but the very top of the screen shows some visual corruption. http://www.visualtech.com/screenshot.png I dropped the resolution of the image from 2304x864 to 1600x800, but you can still make out the corruption. What's particularly odd, though, is that the root window is never drawn. The background you see is actually the background from my previous X session (when I had DRI disabled), using windowmaker. This time I launched X and had fvwm2 in my .xinitrc file (you can see the outline of the fvwm pager in the screenshot, though that never finished drawing, either). After that nothing else gets drawn. I can move the mouse pointer, but that's about it. I can safely kill X and restart it, but the same thing happens unless I disable DRI. In comparison, I have another workstation with an AGP x700. -CURRENT from the same date, and modular Xorg from the ports tree from yesterday, too. It works just fine (starts up fine, and the mesa demos run with acceleration). You can find the Xorg log file from the PCIe system at http://www.visualtech.com/Xorg.0.log.gz This seems to be a FreeBSD specific problem as the same PCIe x800 works fine with the OSS drivers under Linux, but no on on the freebsd-x11 lists seems to have any ideas, and I'm running out of ideas. I thought some of the great minds on this list might be able to shed some light. Any ideas? Thanks! Adam - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.phpp=sourceforgeCID=DEVDEV -- ___ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel