This bug is not very obvious but it seems to show up sooner or later no matter what I do. About the bug: Every now and then an app quits with the message, failure loading /lib/libdl.so.2: Unsupported version 243853 of Verdeef (or something like that) I guess the configure scripts are most vulnerable because lot's of small apps are started all the time. But this can happen to any process, gcc, sed, anything. Bootstrap from stage-1 fails too. What I've tried: Happens with kernel: gentoo, emachines, gentoo-dev-sources and mm-sources. With and without bootsplash. With 2004.0 and 2004.1 test. From chrooted environment from the livecd, and when the system boots itself. Actually, it's never happened during the first step of the installtion. It's allways been after I've started compiling the first few programs. Could be because you don't start many app's during the basic install. About my computer: Athlon64 3200+ Abit KV8-MAX3 mobo. 120 GB SATA. 512 MB DDR 400, tested with memtest86+ 1.11 from 2004.1 test. Connect 3D Radeon 9600XT Since I have so much trouble I'm running windows most of the time now, but any info you need from linux I could reboot and look up. Reproducible: Always Steps to Reproduce:
To clarify: It's not a glibc libdl compiled by me that crashes, it's the libdl from stage3 and stage1 in the case of failed bootstrap.
Got x86 up and running now, no problems. It's got to be something x86_64 related.
please provide output of "emerge info".
Since this happens with the installer stages I got the tip to add livecd@gentoo.org Steps to reproduce: Start my computer from A livecd. (2004.0, 2004.1, any kernel) Make new filesystem (any type), mount, extract stage 1. Chroot, and emerge sync. Do not edit anything, -O2 cflags, just keep everything at default. scripts/bootstrap.sh Then you'll get this after about a minute, maybe on sed, maybe on gcc.. /lib/libdl.so.2: unsupported version 25968 of Verneed record This is the exact number, and it's allmost allways that number.
Also tried with ext2 and ide=nodma. Ran memtest86 for 10 hours today, no problems. Got another number this time but it seems pretty random. Got the latest via-4-in-1 drivers and everthing latest from Abit's homepage.
Damn! :-) It works now! As I said, I tired my memory with memtest86 for 10 hours straight and it reported no problems. But changing the memory speed in bios to 333 made the "Verneed" problem go away. I guess my memory TwinMOS 512 DDR 400 wasn't listed as "AMD Certified" on http://www.amd.com/us-en/Processors/DevelopWithAMD/0,,30_2252_893_10125,00.html Anyway, now you can just refer to this bug of anyone else has similar problems, while I'll go buy some Samsung RAM!