On Mon, Dec 13, 2004 at 11:25:29PM +0100, Frederic Lehobey wrote: > After a few hours of X running, it seems actually that both removing > the xlibmesa-dri and disabling the DRI in the X server configuration > prevents the occurence of the lockup (I am still setting memory > limitation to 128 Mb as explained previously). I cannot be completely > sure as I have no reproducible way to trigger the problem (but I do > not recall the machine having run that long flawlessly). I will > perform the same tests with the -9 packages as soon as they hit my > mirror.
Michel, I have to admit I'm a little surprised by this. What precisely does disabling the "DRI" module in the XF86Config-4 accomplish that making the DRI modules unavailable by removing the xlibmesa-dri package doesn't? If I had to guess, I'd say that codepaths usually found in XFree86's driver/$FOO/FOO_dri.c don't get followed. And in that case, any bug which doesn't go away when xlibmesa-dri is removed but does when the "DRI" module is disables properly belongs in the xserver-xfree86 package. Does setting LIBGL_ALWAYS_INDIRECT=1 do anything distinguishable from removing the xlibmesa-dri package? If so, can you propose a debugging order for seemingly DRI-related problems? Maybe the following: 1) try to repro with LIBGL_ALWAYS_INDIRECT=1 2) try to repro with xlibmesa-dri removed 3) try to repro with 'Load "DRI"' commented out of XF86Config-4 ? Thanks for your help in understanding these pretty basic issues. -- G. Branden Robinson | When we call others dogmatic, what Debian GNU/Linux | we really object to is their [EMAIL PROTECTED] | holding dogmas that are different http://people.debian.org/~branden/ | from our own. -- Charles Issawi
signature.asc
Description: Digital signature