From: "Bryan J. Smith b.j.smith@ieee.org"
Hmmm, maybe I haven't built from a stock kernel in a long time. ;-> All I know is that I've seen it in several other distros. I thought the whole purpose of the 4Gi/4Gi model was to give a support/performance point between the 1Gi/3Gi and full 64GiB model.
Okay, I did some elementary research and it appears it was never adopted in the stock 2.4 kernel. Although 2.6 seems to have been designed for it -- although I see the issues you were talking about, largely when user >2GiB. I'd be interested in finding out if they have the same issues with the 64GiB mode and formal PAE36 paging.
Anyhoo, my apologies on that _poor_ assumption. Given my lack of exposure to stock kernels, I should have kept my mouth shut. I don't know why I didn't put 2-2 together.
But it should be noted that it's no longer a Red Hat-only thing. I've seen several non-Red Hat (or long-gone Red Hat forked) distros who are using it. I should have made my responses in that context. Again, my apologies for that.
-- Bryan J. Smith mailto:b.j.smith@ieee.org
On Wed, 2005-07-20 at 10:46 -0400, Bryan J. Smith wrote:
Okay, I did some elementary research and it appears it was never adopted in the stock 2.4 kernel. Although 2.6 seems to have been designed for it -- although I see the issues you were talking about, largely when user >2GiB. I'd be interested in finding out if they have the same issues with the 64GiB mode and formal PAE36 paging.
Just FYI, I confirmed it is a 2.6 thing, and Red Hat backported it to 2.4. I also refreshed myself to find out that _both_ 1G/3G (aka HIGHMEM) and 4G/4G are PAE36 approaches.