Description of problem:
I've bricked a box when testing glibc, by removing glibc. Because I'm a smart guy, I had glibc rpms downloaded, and busybox isntalled for this purpose, just in case. So I've tried to fix things easily:

# busybox rpm -i glibc-2.5-24.i686.rpm

which kabooms like this:
*** glibc detected *** busybox: malloc(): memory corruption: 0x085c5d20 ***
======= Backtrace: =========
[0x8146e18]
[0x81483cb]
[0x80febb4]
======= Memory map: ========
00800000-00801000 r-xp 00800000 00:00 0          [vdso]
08048000-0823b000 r-xp 00000000 fd:00 10584066   /sbin/busybox
0823b000-0823e000 rw-p 001f2000 fd:00 10584066   /sbin/busybox
0823e000-0830e000 rw-p 0823e000 00:00 0
085c4000-085e6000 rw-p 085c4000 00:00 0
b7c00000-b7c24000 rw-p b7c00000 00:00 0
b7c24000-b7d00000 ---p b7c24000 00:00 0
b7d25000-b7d4b000 r--p 00000000 00:18 83198326   /mnt/qa/scratch/i386-5c-m1/2009:8182/tps/glibc-2.5-24.i686.rpm
b7d4b000-b7f4b000 r--p 00000000 fd:00 463506     /usr/lib/locale/locale-archive
bfec1000-bfed7000 rw-p bfec1000 00:00 0          [stack]
Aborted

Version-Release number of selected component (if applicable):
# rpm -q busybox
busybox-1.2.0-4.el5.i386

How reproducible:
always, at least on i386.
on x86, I see 'Segmentation fault' wih glibc packages in cca 50% cases

Steps to Reproduce:
1. busybox rpm -i glibc-2.5-24.i686.rpm (can be even rhel5.3's 2.5-29 glibc)
2. see the abort
  
Actual results:
abort, segfaults & friends

Expected results:
my box breathing fresh glibc again

Additional info:
It seems like only glibc packages aborts busybox. I've tried few more, not seeing the problems