I did the GCC 4.1 migration on stella last week after it got marked stable and the 2006.1 profile was pushed out. In retrospect, I jumped the gun way too early. There was a bit of package breakage, but most of all, X and my nvidia drivers stopped working and I haven't been able to get them working since. In the meantime, I'm using the nv driver with X. Read on.
I followed the
migration guide to a t.
emerge -eav system finished fine with no breakage, much to my surprise. Things started falling apart when I recompiled world with
emerge -eav world.
These are the packages that failed to compile. I am going to list them if someone happens to stumble upon this post and find this information useful.
=dev-lisp/cmucl-19a-r1
-19c worked for me
games-fps/doomsday
I just unistalled it
=app-arch/unshield-0.2
=gnome-base/gconf-1.0.9
=gnome-extra/guppi-0.4.0.3-r2
-r3 works
=gnome-extra/gtkhtml - I got an access violation. I gave me a path to some html documentation. Deleting all the documentation and recompiling worked
=dev-libs/pwlib-1.6.3-r3
I've read that versions above 1.8.7 works but openh323 wants a lower version. Bugs have been filed
net-libs/openh323
=net-analyzer/iptraf-2.7.0-r1
-3.0.0-r2 works
=media-libs/libdv-0.102
0.104-r1 works
That's all the broken packages.
X.org and nVidia
After I finished all the compilation, I tried starting X, crossed my fingers, but nothing happened. I first realized I had to recompile my kernel (gentoo-sources-2.6.15-r1) with the new toolchain (this made sense since the nvidia module was originally compiled with gcc3 and my version of X that I was using was compiled with gcc4). I did that but no luck. I did get this error:
(EE) NVIDIA(0): The requested configuration of display devices is not
(EE) NVIDIA(0): supported in the hardware.
Fatal server error:
AddScreen/ScreenInit failed for driver 0
*** If unresolved symbols were reported above, they might not
*** be the reason for the server aborting.
Backtrace:
0: X(xf86SigHandler+0x87) [0x80b4dda]
1: [0xffffe420]
2: X(AbortServer+0x23) [0x81a4b72]
3: X(FatalError+0x66) [0x81a505e]
4: X(InitOutput+0x85d) [0x809f846]
5: X(main+0x277) [0x806ffcb]
6: /lib/libc.so.6(__libc_start_main+0xdc) [0xb7d23864]
7: X(FontFileCompleteXLFD+0xad) [0x806f531]
FatalError re-entered, aborting
Caught signal 11. Server aborting
Now, I've already recompiled glibc since all this started, and I'm still getting this error. /lib/libc.so.6 belongs to glibc which is why I bring it up. The only other thing that I find interesting is line 7. I can't really put the pieces together.
The only other idea I have right now is to upgrade my kernel but the nvidia docs say that their drivers work with kernels 2.6.7+ (give or take 2)