Summary: | xorgcfg fails to start | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | xorg | Reporter: | Tom Gall <tgall> | ||||||
Component: | App/xorgcfg | Assignee: | Adam Jackson <ajax> | ||||||
Status: | RESOLVED FIXED | QA Contact: | |||||||
Severity: | normal | ||||||||
Priority: | high | CC: | ajax, alan.coopersmith, dberkholz, henry.zhao, seemant, s.loeuillet | ||||||
Version: | 6.8.0 | ||||||||
Hardware: | PowerPC | ||||||||
OS: | Linux (All) | ||||||||
Whiteboard: | |||||||||
i915 platform: | i915 features: | ||||||||
Bug Depends on: | |||||||||
Bug Blocks: | 400 | ||||||||
Attachments: |
|
Description
Tom Gall
2004-09-04 05:29:16 UTC
and to add to it a bit, running a 2.6.7 ppc64 kernel > dlopen: /usr/X11R6/lib/modules/drivers/atimisc_drv.so: undefined symbol:
> ATIPublicOptionSize
dlloader issue. interesting, i'll take a look.
looks like the usual dlloader problem i have with binary nvidia drivers with 6.7.0 and 6.8.0 (gentoo too, but x86) as a workaround, re-emerge xorg like this for now USE="-dlloader" emerge xorg-x11 (temporary, waiting a real fix) Stephane, this was a build from a cvs co, emerge portage was not used in any way during the build. This is a tradition, make World ;-) so changing USE flags obviously is not appropriate here tom, i'm gonna go out on a limb here and say you have +hardened in your use flags, correct? Tom, feel like responding? *** Bug 2787 has been marked as a duplicate of this bug. *** not a hardened issue, it turns out. atimisc is referring to global data that's normally only present in the ati stub driver, which is bad juju. nv and cirrus may have similar braindamage. as alan mentioned in bug #1288, part of the solution is to add stubs to xorgcfg to get things to resolve. the other part (like with atimisc) is to not refer to globally visible data from other modules directly, which is standard dlloader practice. Created attachment 2982 [details]
Thaana Keyboard Map
This is the patch Henry Zhao wrote for Sun, which we're using in Solaris to
fix this bug. (Not sure how well this applies to current CVS, but hopefully
it helps ajax who said he was going to look into this soon.)
alan, that patch seems to have been mangled, can you re-attach it if you get the chance? Created attachment 3486 [details] [review] dlmodule support patch against Xorg 6.8.2 Reattached patch against 6.8.2 a quick review of this shows quite a bit of tweaking with the app-defaults and some varargs tuning that i don't think is strictly needed. i'll clean it up but the principle looks good. Created attachment 4053 [details] [review] cleaned-dlloader-fixes.patch cleaned version of the patch, testing... mmm, not quite right. as part of the fix for #4361 we have a hack in place now to sort the driver list coming off the filesystem. probably the best thing to do for now is to sort ati before atimisc, which should fix the ordering problem. it's still ugly, but that's the ati driver for you. sorted, ati loads before atimisc, which keeps the symbols resolving themselves. SOMEONE PLEASE KILL THE ATI WRAPPER FOR THE LOVE OF GOD |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.