I have found the problem, maybe I made some mistakes while build bootstrap, and the gcc-machdep did not change the linux-dynlinker to lsb-dynlinker, the intermediate need to use bootstrap, so ...
Shou Hua Wang
IBM China Development Laboratory
"Wichmann, Mats D" <firstname.lastname@example.org>
To: Shou Hua Wang/China/IBM@IBMCN, email@example.com
Subject: RE: One question about change linux-dynlinker to lsb-dynlinker
Sean is the best to answer, but the bootstrap process should be such that
things that will be linked against ls-lsb.so.1 will only be built after the
glibc build has put it in place. I haven't completed a build against the
new CVS stuff yet. I started one yesterday morning before i had to run
off, I notice it has died doing "at" during lsb-intermediate but I won't have
access to the console until Thursday (for some reason the logs that
nALFS leaves behind don't record any actual informaiton, just "failed").
Is there any chance of changing that, so the logs would be useful?
From: Shou Hua Wang [mailto:firstname.lastname@example.org]
Sent: Tuesday, August 13, 2002 3:48 AM
Subject: One question about change linux-dynlinker to lsb-dynlinker
While I try to build the phase2 and phase3, glibc.xml will search "linux-dynlinker"(it is ld.so.1 in PPC) to "lsb-dynlinker"(ld-lsb.so.1). But most packages such as gcc, bison, etc will compile after this, and they all try to find the /lib/ld.so.1, but this file doesn't exist, the exist file is ld-lsb.so.1. So the build process will failed.
Could somebody else tell me why LSB SI need to change this, and can I do this link manullay after the system was built(Both ld.so.1 and ld-lsb.so.1 are link to ld-2.2.5.so)?