from the quill of Jeff Garzik <[EMAIL PROTECTED]> on scroll
<[EMAIL PROTECTED]>
> Chmouel Boudjnah wrote:
>
> Yep... if a binary driver works for you it is basically luck (and
> maybe
> good testing by the vendor).
And they always have with vmware's, thus far. Their binary drivers are
a bit different than most mind you. You actually build the drivers on
the host they will be loaded on. From my very quick peek at things it
looks like they ship a bunch of .o's and actually link them together on
the box they will be loaded on.
> In order to completely support a _single_
> distro, a VMWare (for example) needs to provide a binary module for
>
> * kernel-linus
> * kernel-fb
> * kernel
> * kernel-smp
> * kernel-smp-fb
> * kernel-your-flavor-of-the-month
But like I said, you link the driver on the kernel they will be loaded
on.
> For MandrakeSoft, I think it is MOSTLY the same situation. But since
> we
> are a vendor, we can make "quick hacks" if they are safe and get some
> binary modules working.
>
> So... I think exporting "best_memset" from all our kernels is a safe
> and good patch...
Funny thing is that from what I can tell (and I am not a seasoned kernel
hacker by any stretch of the immagination) best_memset is supposed to be
exported. See arch/i386/kernel/i386_ksyms.c. The .config for the
framebuffer version (the only one I checked) of the kernel sure seems to
have CONFIG_X86_CPU_OPTIMIZATIONS=y so I don't know why it is not being
exported.
Is there a way to examine the actual kernel image to determine if a
symbol is exported?
b.
--
Brian J. Murrell InterLinx Support Services, Inc.
North Vancouver, B.C. 604 983 UNIX
Platform and Brand Independent UNIX Support - R3.2 - R4 - BSD
