SEO
vBulletin Search Engine Optimization
| |||||||
| Register | FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
02-22-2008, 12:21 PM
| ||||
| ||||
 Re: i386 PAE support (64g physmem) On Wed, Mar 29, 2006 at 08:28:40PM +0200, mickey wrote: > On Fri, Jan 13, 2006 at 06:53:12PM +0100, mickey wrote: > > On Thu, Jan 12, 2006 at 03:40:17PM +0100, mickey wrote: > > > re > > > here is the diff that does it. > > > make sure option I686_PAE is in your generic. > > > > > > current design tries to force all dmable memory > > > allocations from the 32bit memory thus avoiding > > > bounce buffering. normal page allocations are > > > otherwise preferred to be in the higher memory. > > > seems to work just fine on averagely loaded machine. > > > last resort is bouncy-bouncy of course... > > > > > > the diff is in: > > > http://doorslam.net/pae.gz > > now is a good time to get down on this. > the diff has been updated to -current. > main concern is to test plain kernels w/o any > large memory support. > there are currently problems w/ >8g memory due > to some uvm constraints that can be addressed later. new diff same place. now code is always compiled into GENERICs but not RAMDISKs. to enable use the option or patch cpu_pae variable to 1. code growth reduced due to more restricted type size promotion. comments? testing? 10x cu -- paranoic mickey (my employers have changed but, the name has remained)      |
Linear Mode
