[Openswan dev] building with gcc 3.4

[Nate Carlson]

On Wed, 15 Sep 2004, Michael Richardson wrote:
> We have been testing klips26 in HEAD. Using a UML based upon 2.6.6, it 
> works for kernel things with 2.6.6, but there is a UML bug that runs 
> things out of file descriptors, so it is hard to test the pluto items. I 
> have seen some successes though.
>
> Using a UML based upon 2.6.7, we get crashes (even without klips 
> loaded). There may be compile problems against 2.6.8.1.

I have done testing on a Debian-based UML session running 2.6.8.1 without 
any problems. I did have to apply a patch to the UML kernel to export the 
memcpy() symbol properly, and possibly fix a very minor compile problem. 
(Been awhile, I don't remember for sure.)

With current 2.6 UML's, be sure to disable 'HIGHMEM' in the kernel - if 
you have it enabled, it will not compile.

> Reports of problems against 2.6.6 are appreciated, as they can be 
> reproduced.

Do you have the scripts available to set up a 2.6-based UML testing 
environment for Openswan? If so, I'll see what I can do to get them 
working with 2.6.8.1 or one of the 2.6.9-pre's. Apparently, 2.6.9 is going 
to work in UML without any patches!

------------------------------------------------------------------------
| nate carlson | natecars at natecarlson.com | http://www.natecarlson.com |
|       depriving some poor village of its idiot since 1981            |
------------------------------------------------------------------------