On Fri, 7 Oct 2005 18:00:20 +0200, Blaisorblade wrote
> On Wednesday 05 October 2005 10:04, Vincent guffens wrote:
> > cheikhrouhou wrote:
> > > Hello,
> > >
> > > i compiled a 2.4.24 kernel with gcc.3.3.3 (which i was installed because
> > > the gcc-4.0 which comes with FC4 don't compile well uml kernel) and he
> > > generates this message :
>
> > > and when i try lo lance this kernel he shows :
> > > segmantation fault
>
> > > Any idea about the solution
>
> > I had time to investiguate your problem on a FC3. I found out that the
> > segmentation fault which occurs with the binaries when inserting the
> > module is due to :
>
> > _dl_lookup_versioned_symbol ()
>
> > So this is a library mismatch and the problem should disappear when
> > recompiling uml on the host that you will be using to run uml. In order
> > to do so, you have to download the sources I put on my web site and I
> > have added the patch to run uml 2.4 guest on a 2.6 host. This is just
> > tow lines and you have to modify these lines by hand (the patch doesn't
> > apply on this version). So the instructions are:
>
> > 1) download and untar the kernel.
> > 2) patch it with uml-patch-2.4.24-1
> > 3) modify the two lines as directed by the patch uml2.6host.patch
> > 4) apply the netbus patch
> > 5) compile the modified kernel with gcc 3.3
>
> > And that's it, no more segmentation fault !
> Vincent, an important question: since 2.4.24-1um is outdated enough,
> and various bugs result (including, but not limited to the 2.6 host
> problem), have you any idea if the whole thing would apply on a more
> recent kernel? I'm especially interested in 2.4.28-bs2 from my
> homepage, which is the latest 2.4 UML release, and the more trustworthy.
>
> Especially given that 2.4 has been quite stable and got little
> changes in recent times, at least starting with 2.4.26 - in fact any
> UML patch for
> >=2.4.26 applies without changes on any kernel >= 2.4.26...
I will try next week (I don't have a convenient access to the internet now)
but it should not be very difficult to make it work on any 2.4 kernel. The
netbus patch is not very intrusive and should apply with no problem. I'm now
working on a 2.6 version which would not be based on an existing driver
anymore, i.e more like the other UML network drivers. The problem with the
2.4 approach is that I used a "netbus" bus instead of writting a code that
would export all the pci_* functions. As a results, the driver itself must be
modified. Now, I'm not too sure if it worths the trouble to write this pci
code given that for each driver faked hardware has to be implemented.
--
Vincent Guffens
UCL/CESAME +32 10 47 80 30
Euler Building A017
-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
_______________________________________________
User-mode-linux-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user
