On Sun, 2 Jun 2002, Felix Kühling wrote: > Hi, > > I submitted another message about this yesterday. It didn't show up in > the SourceForge archive and you didn't refer to it, so I'm submitting it > again: > > ---------------------------------------------------------------------- > > [Michel Dänzer <[EMAIL PROTECTED]>] > > These messages are harmless unless one of the unresolved symbols is > > referenced, which shouldn't happen when DRI is disabled. And even if one > > of them was referenced, that would cause a server crash and not a > > lockup. (Unless the crash causes a lockup...) > > I had a closer look at this. I get a lockup when I run the Xserver > without DRI after switching from a 2D accelerated XFree86 4.1 server to > the text console. If DRI is loaded there is no problem. If I start the > Mach64-Xserver after a reboot and without DRI there is no problem > either. > > I guess this means that the 3D driver puts the card in some state that > the 2D driver relies on and the old Xserver messes it up.
Since the dri branch driver works on it's own without dri enabled, I don't see how it could be a 3D driver dependency. Unless I'm not understanding you, the problem is with two 2D drivers without dri enabled on either. Maybe it's a bad interaction between xfree 4.1 and 4.2? I just tried starting a 4.2 based gatos driver, then switching to a text console and starting the dri branch driver with dri disabled. I can switch back and forth between them and to a text console without a lockup. I don't have a 4.1 server anymore so I can't test that. > ---------------------------------------------------------------------- > > Maybe it's again a problem with the FIFO size. Just a guess. This shouldn't be a problem when dri isn't enabled. > Anyway, I think I'll switch to the new accelerated server as you > proposed. It seems that I'm just asking for trouble with my current > config :) ok. I'm still curious about what the problem is, though. How recent is the dri branch build you're using? Is the 4.1 server a vanilla xfree release or a gatos driver? > > On Sat, 1 Jun 2002 22:02:14 -0400 (EDT) > Leif Delgass <[EMAIL PROTECTED]> wrote: > > > On 31 May 2002, Michel Dänzer wrote: > > > > > On Fri, 2002-05-31 at 02:30, Felix Kühling wrote: > > > > > > > > I reported lockups shortly after starting the Xserver with a gcc 3.0 > > > > compiled drm module that triggered the strange permission problem. Now I > > > > tested it without loading the dri Xserver module in XF86Config and got > > > > the same lockup. The following messages in XFree86.1.log indicate that > > > > the problem is that the 2D driver currently doesn't work without drm: > > > > > > > > (II) ATI(0): Direct rendering disabled > > > > Symbol DRILock from module >/usr/X11R6-mach64004/lib/modules/drivers/atimisc_drv.o is unresolved! > > > > Symbol drmMach64WaitForIdle from module >/usr/X11R6-mach64004/lib/modules/drivers/atimisc_drv.o is unresolved! > > > > [ yadda yadda yadda ... ] > > > > > > Symbol DRICloseScreen from module >/usr/X11R6-mach64004/lib/modules/drivers/atimisc_drv.o is unresolved! > > > > Symbol DRIDestroyInfoRec from module >/usr/X11R6-mach64004/lib/modules/drivers/atimisc_drv.o is unresolved! > > > > > > > > I will recompile my entire kernel with gcc-3.0 tomorrow to see if the > > > > permission goes away, but this is still a problem, I think. The 2D > > > > Xserver should work without DRI loaded. > > > > > > These messages are harmless unless one of the unresolved symbols is > > > referenced, which shouldn't happen when DRI is disabled. And even if one > > > of them was referenced, that would cause a server crash and not a > > > lockup. (Unless the crash causes a lockup...) > > > > > > I think the messages can be silenced by adding the symbols to the > > > xf86LoaderRefSymLists() call. > > > > Thanks for the tip. I commited a fix for this that quiets these messages. > > I used xf86LoaderRefSymLists() in ATIPreInit (atipreinit.c) a la the r128 > > and Radeon drivers. > > > > Felix, 2D accel should work when dri is disabled with the current branch, > > so you might try running two Xservers based on the branch, one with dri > > enabled and one with it disabled and see if you still have the lockup. > > A couple of weeks ago, I also made sure that the DDX in the DRI branch > > explicitly sets the FIFO size to 128 entries, rather than reading the > > regsiter and saving that value (which is restored on mode/vt switches), in > > order to avoid the problem you mentioned before. IIRC, you had started > > the DRI enabled server from a console with the xfree 4.1 server already > > running. In that case, the DDX in the newly started server saved the > > value set by the 4.1 server, which would cause a lockup with dri. In this > > case though, without dri enabled, I don't how that could be the problem. > > > > -- > > Leif Delgass > > http://www.retinalburn.net > > > > > > > __\|/__ ___ ___ ___ > __Tschüß_______\_6 6_/___/__ \___/__ \___/___\___You can do anything,___ > _____Felix_______\Ä/\ \_____\ \_____\ \______U___just not everything____ > [EMAIL PROTECTED] >o<__/ \___/ \___/ at the same time! > -- Leif Delgass http://www.retinalburn.net _______________________________________________________________ Don't miss the 2002 Sprint PCS Application Developer's Conference August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel