Consider backporting zero huge_page support for the 3.7 release?
From www.phoronix.com/scan.php?page=news_item&px=MTI1MTQ, the patch can save a lot of RAM for some situations on systems that have transparent_hugepage set to always.
A discussion on zero huge_page may be read at comments.gmane.org/gmane.linux.kernel.mm/85623. And the patch may be found at git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=f6e858a00af788bab0fd4c0b7f5cd788000edc18. But the patch will be merged for Linux 3.8. Damentz, would you consider backporting this patch when you release 3.7? It would be interesting to see how this change will affect memory usage, particularly in my system where having transparent_hugepages=always results in large RSS values for Firefox. Here\'s a bug report on that: bugzilla.mozilla.org/show_bug.cgi?id=770612. Back to top |
|||||
Hmm, that's interesting. I'll take a look and see how possible this is. Because this patch was merged before the RC release, it's probably based on 3.7, which should be easy to manage.
Back to top |
|||||
Good to see my post surviving the purge. Thanks techAdmin.
Although, it would have been more helpful if a message was displayed in case of a post being put on hold. I was left wondering why my post didn't appear on the board, so I tried posting again. And when that didn't work out, that was when I thought that it might have been put under review or filtered as spam. :-) Back to top |
|||||
3.8-rc1 is here
|
|||||
This feature is now in 3.7.0-2.dmz.1 using the latest patch set on LKML (v6): lkml.org/lkml/2012/11/15/466
Back to top |
|||||
All times are GMT - 8 Hours
|