Boston Linux & Unix (BLU) Home | Calendar | Mail Lists | List Archives | Desktop SIG | Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings
Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU

BLU Discuss list archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Discuss] Limiting amount of memory



Bill Bogstad wrote:
> It would still be nice to know where the memory is being used.

Some Googling about lead me to the PCI memory hole:
http://en.wikipedia.org/wiki/PCI_hole

Short version: in order to boot 32-bit operating systems the BIOS or EFI 
firmware needs to map all PCI devices to the first 4GB of address space. 
This leads to "holes" in the memory map between what's available in the 
lower 4GB and everything above 4GB. That's the "missing" RAM.

I checked five different boxes: two very different Dells, a white box 
server, and two VMware virtual machines. All of them are "missing" some 
RAM. How much varies widely: The white box is missing about 1GB; the big 
Dell about 2.5GB; the little Dell about 1/2GB, and the two virtual 
machines only about 150MB. Make that six: my little HP server is short 
about 160MB. The little Dell and one of the VMs are running 32-bit 
Linux; the others are 64-bit Linux.

So, not a Dell thing specifically and not a 32-bit thing specifically, 
but a PCI thing generally.

As to why mem=16G leaves the system with only 9GB detected? Dunnow. If 
mem=8G yields ~6GB useable then mem=16G should yield ~14GB usable on 
that box. There could be some kind of memory mapping weirdness between 
the PCI allocations and the kernel RAM constraints.

-- 
Rich P.



BLU is a member of BostonUserGroups
BLU is a member of BostonUserGroups
We also thank MIT for the use of their facilities.

Valid HTML 4.01! Valid CSS!



Boston Linux & Unix / webmaster@blu.org