On Tue, 9 May 2006, Ludovic Courtès wrote:
I should note that the above remark applied to 2.6.15, probably _before_ running X. With 2.6.16-1 (from sid), I can observe the following funny thing: # file /usr/lib/xorg/modules/libxaa.so /usr/lib/xorg/modules/libxaa.so: Sun disk label 'ST39111A cyl 17660 alt 2 hd 16 sec 63' 14285 phys cys, 51 alts/cyl, 0 blocks, boot block present With 2.6.15, it's now `libramdac.so' that gets broken: # file /usr/lib/xorg/modules/libramdac.so /usr/lib/xorg/modules/libramdac.so: sparc executable This randomness may explain why we do not experience the same problems.
Ugh. The only thing which comes to mind is that your kernel got miscompiled somehow. Dave Miller mentioned on sparclinux list that gcc 4.1 causes problems, perhaps you are using it to build the kernels?
Have people been running X-free (no pun here) workloads, like builds, with recent kernels on sparc64? Do they trigger any such problem?
Yeah, I'm regularly building test kernels on a headless Ultra-60, running the latest available kernel (2.6.16, 2.6.17-rc), and it has been working quite nicely for me.
Best regards, Jurij Smakov jurij@wooyd.org Key: http://www.wooyd.org/pgpkey/ KeyID: C99E03CC