sh4 fails on check-textrel in glibc-2.3.2 even though there are norelocations?

[Dan Kegel <dank at kegel dot com>]

I'm cross-building glibc for sh4 using gcc-3.3 and binutils-2.13.90.0.18.
My scripts are online at http://kegel.com/crosstool
and are partly working with glibc-2.2.5 on sh4, with a few glibc
test execution failures (see http://www.m17n.org/linux-sh/ml/linux-sh/2003-07/msg00050.html).

Now I'm trying glibc-2.3.2, but "make tests" fails at build time with error
make[2]: *** [/gcc-3.3-glibc-2.3.2/build-glibc/elf/check-textrel.out] Error 1

Looking inside check-textrel.out, I see the errors
/gcc-3.3-glibc-2.3.2/build-glibc/libc.so: text relocations used
/gcc-3.3-glibc-2.3.2/build-glibc/linuxthreads/libc.so: text relocations used
/gcc-3.3-glibc-2.3.2/build-glibc/linuxthreads/libpthread.so: text relocations used
/gcc-3.3-glibc-2.3.2/build-glibc/math/libm.so: text relocations used
/gcc-3.3-glibc-2.3.2/build-glibc/resolv/libresolv.so: text relocations used

Kaz Kojima kindly provided a patch,
http://www.m17n.org/linux-sh/ml/linux-sh/2003-07/msg00085.html
which got rid of the relocations he knew about.  Unfortunately,
check-textrel still fails.  Oddly, when you use readelf -a to look
at the .so's that supposedly have relocations, there aren't any,
but there is a size-zero TEXTREL:

Dynamic segment at offset 0xd760 contains 25 entries:
  Tag        Type                         Name/Value
...
 0x00000016 (TEXTREL)                    0x0

The .so's in question are at http://kegel.com/crosstool/badso.tar.gz
if you want to see for yourself.  I tried updating to binutils-2.14.90.0.5
but the problem remains.

Anyone have any idea?
I'm jazzed that I have ppc405 and ppc750 gcc-3.3/glibc-3.2.2 toolchains
working, it'd be nice to be able to run on sh4, too.  (And it'd make my
boss happy.  I spent two months figuring out and documenting how to
build and test cross-toolchains (see http://kegel.com/crosstool), partly
to better support our sh4-based cards, and it's a bit discouraging
to keep running into problems on sh4...)

Thanks,
Dan

--
Dan Kegel
http://www.kegel.com
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=78045