In EL6.2, there are a handful of rpms that use lua for rpm %post/%postun scripts.  When these rpms are upgraded manually using 'rpm -Uv', rpm complains that it cannot find the rest of the rpms after processing the first rpm that uses a lua script.  Using yum works but it's not always an option.

I've attached 2 proof of concept rpm spec files which will show the problem. To reproduce, build the rpms and run: 
     cd /path/to/rpms && rpm -Uvh test{foo,bar}*.rpm

The output looks like this:

[root@optimus ~]# ls
anaconda-ks.cfg  install.log.syslog  testbar-1.0-1.sc00.x86_64.rpm
install.log      ks-post.log         testfoo-1.0-1.sc00.x86_64.rpm
[root@optimus ~]# rpm -Uvh test*.rpm
Preparing...                ########################################### [100%]
   1:testfoo                ########################################### [ 50%]
Tue 10 Apr 2012 11:52:46 AM PDT
error: open of testbar-1.0-1.sc00.x86_64.rpm failed: No such file or directory


runLuaScript() doesn't properly reset the current working directory after running the lua script.  The attached patch fixes the problem for me.

[root@optimus ~]# rpm -e testfoo
[root@optimus ~]# rpm -Uvh test*.rpm
Preparing...                ########################################### [100%]
   1:testfoo                ########################################### [ 50%]
Tue 10 Apr 2012 12:33:55 PM PDT
   2:testbar                ########################################### [100%]