Create a hole in high linear address space
PATCH 5/5 Create a hole in high linear address space
Allow compile time creation of a hole at the high end of linear address space. This makes accomodating a hypervisor a much more tractable problem by giving it ample playground to live in. Currently, the hole size is fixed at config time; I have experimented with dynamically sized holes, and have a later patch that developes this potential, but it becomes much more useful once the exact negotiation of linear address space with the hypervisor is defined.
The fixed compile time solution is sufficient for now.
arch/i386/Kconfig +8 -0
@@ -803,6 +803,14 @@ config ARCH_SELECT_MEMORY_MODEL def_bool y depends on ARCH_SPARSEMEM_ENABLE +config MEMORY_HOLE + int "Create hole at top of memory (0-512 MB)" + range 0 512 + default "0" + help + Useful for creating a hole in the top of memory when running + inside of a virtual machine monitor. + source "mm/Kconfig" config HAVE_ARCH_EARLY_PFN_TO_NID
include/asm-i386/fixmap.h +1 -1
@@ -20,7 +20,7 @@* Leave one empty page between vmalloc'ed areas and * the start of the fixmap. */ -#define __FIXADDR_TOP 0xfffff000 +#define __FIXADDR_TOP (0xfffff000-(CONFIG_MEMORY_HOLE << 20))#ifndef __ASSEMBLY__#include <linux/kernel.h>
Quoted 8 lines · * Zachary Amsden ([email protected]) wrote:
Allow compile time creation of a hole at the high end of linear address space. This makes accomodating a hypervisor a much more tractable problem by giving it ample playground to live in. Currently, the hole size is fixed at config time; I have experimented with dynamically sized holes, and have a later patch that developes this potential, but it becomes much more useful once the exact negotiation of linear address space with the hypervisor is defined.
The fixed compile time solution is sufficient for now.
Xen moves __FIXADDR_TOP like this:
#ifdef CONFIG_X86_PAE
# define HYPERVISOR_VIRT_START (0xF5800000UL) #else # define HYPERVISOR_VIRT_START (0xFC000000UL) #endif
and
#define __FIXADDR_TOP (HYPERVISOR_VIRT_START - 2 * PAGE_SIZE)
and also adds bits to fixmap.
So this proposed mechanism isn't quite good enough.
thanks,
-chris
Quoted 22 lines · Chris Wright
Quoted 10 lines · * Zachary Amsden ([email protected]) wrote:
Allow compile time creation of a hole at the high end of linear address space. This makes accomodating a hypervisor a much more tractable problem by giving it ample playground to live in. Currently, the hole size is fixed at config time; I have experimented with dynamically sized holes, and have a later patch that developes this potential, but it becomes much more useful once the exact negotiation of linear address space with the hypervisor is defined.
The fixed compile time solution is sufficient for now.
Xen moves
__FIXADDR_TOPlike this:#ifdef
CONFIG_X86_PAE# define HYPERVISOR_VIRT_START (0xF5800000UL) #else # define HYPERVISOR_VIRT_START (0xFC000000UL) #endif
Quoted 9 lines
and
#define
__FIXADDR_TOP(HYPERVISOR_VIRT_START- 2 *PAGE_SIZE)and also adds bits to fixmap.
So this proposed mechanism isn't quite good enough.
Hmm. I was thinking it would be compile time variable with defaults -- like
config MEMORY_HOLE
int "Create hole at top of memory (0-512 MB)"
range 0 512
default "0"
default 168 if (CONFIG_X86_PAE && CONFIG_X86_HYPERVISOR)
default 64 if (!CONFIG_X86_PAE && CONFIG_X86_HYPERVISOR)
help
Useful for creating a hole in the top of memory when running
inside of a virtual machine monitor.Adding things to the fixmap is a separate concept, thus a separate patch ;)
Zach
Quoted 8 lines · * Zachary Amsden ([email protected]) wrote:
Hmm. I was thinking it would be compile time variable with defaults -- like
config
MEMORY_HOLEint "Create hole at top of memory (0-512 MB)" range 0 512 default "0" default 168 if (CONFIG_X86_PAE && CONFIG_X86_HYPERVISOR) default 64 if (!CONFIG_X86_PAE && CONFIG_X86_HYPERVISOR)
That's fine, I had done some braindead math anyway ;-)
Quoted 5 lines
help
Useful for creating a hole in the top of memory when running inside of a virtual machine monitor.Adding things to the fixmap is a separate concept, thus a separate patch ;)
Sure, good point.
thanks,
-chris