Re: 4g4g-Patch for kernel 2.6.8
On 10/25/05 01:35:26PM +0200, Sven Luther wrote:
> On Tue, Oct 25, 2005 at 12:51:23PM +0200, Kai Militzer wrote:
> > Hello,
> >
> > do you think it would be possible to take the 4g4g-Patch from the
> > feodora/red hat kernel into the debian 2.6.8 kernel?
>
> As the 2.6.8 kernel is the stable/sarge kernel, this is highly unlikely, but
> there may be a chance for the etch/sid kernels (the upcoming 2.6.14 probably),
> and there will be backports of those to sarge.
>
> So, what is this patch good for, whatever is a 4g4g, and why has it not made
> it upstream ?
It changes the VM addressing on 32-bit kernels. I'm not sure, but it
probably also only applies to x86 kernels. The idea is to allow each
process access to the full 4G of VM available and to do this it also gives
the kernel it's own 4G VM space, hence the name 4G/4G. The problem is speed
because it causes the TLBs to be flushed out on every call into the kernel
in addition to each context switch.
There's a more complete description here: http://lwn.net/Articles/39283/
>
> Also, could you get either the patch itself or an url to it, and file a
> propoer bug report against the linux-2.6 package ?
I looked and didn't see the patch in the FC4 src.rpm, I can't imagine why
they would include it anyway. A little poking around on google produced the
above URL and the patch at http://people.redhat.com/mingo/4g-patches/ but
it doesn't look like it's maintained any more.
>
> Friendly,
>
> Sven Luther
>
Jim.
Reply to: