What would the ramifications be of increasing VM_AREA size in
include/asm-i386/page.h from 128mb to 256mb. What would be the proper way to
increase this if the above isn't?
Regards
Mark
On Wed, 14 Jan 2004, Mark Hounschell wrote:
> What would the ramifications be of increasing VM_AREA size in
> include/asm-i386/page.h from 128mb to 256mb. What would be the proper way to
> increase this if the above isn't?
I think you mean __VMALLOC_RESERVE? For the most part it's straightforward
to bump it up. _However_, that breaks boot loader assumptions about where
to load initrd, causing mayhem in that case (and initramfs?). That's
second hand info: if I'm wrong or out-of-date, Peter is the authority
and will correct me; or try Google VMALLOC_RESERVE boot.
Hugh
Hugh Dickins wrote:
> On Wed, 14 Jan 2004, Mark Hounschell wrote:
>
>>What would the ramifications be of increasing VM_AREA size in
>>include/asm-i386/page.h from 128mb to 256mb. What would be the proper way to
>>increase this if the above isn't?
>
> I think you mean __VMALLOC_RESERVE? For the most part it's straightforward
> to bump it up. _However_, that breaks boot loader assumptions about where
> to load initrd, causing mayhem in that case (and initramfs?). That's
> second hand info: if I'm wrong or out-of-date, Peter is the authority
> and will correct me; or try Google VMALLOC_RESERVE boot.
>
I think it affects initramfs too.
It only affects boot loaders which don't support version 2.03 of the
boot protocol; unfortunately that includes GRUB last I checked.
We really need a better dialog with the GRUB people, unfortunately at
least I have unsuccessful in starting such a dialog.
GRUB, in particular, needs to report the boot loader ID they're using,
plus support version 2.03 of the protocol.
-hpa
"H. Peter Anvin" wrote:
>
> Hugh Dickins wrote:
> > On Wed, 14 Jan 2004, Mark Hounschell wrote:
> >
> >>What would the ramifications be of increasing VM_AREA size in
> >>include/asm-i386/page.h from 128mb to 256mb. What would be the proper way to
> >>increase this if the above isn't?
> >
> > I think you mean __VMALLOC_RESERVE? For the most part it's straightforward
> > to bump it up. _However_, that breaks boot loader assumptions about where
> > to load initrd, causing mayhem in that case (and initramfs?). That's
> > second hand info: if I'm wrong or out-of-date, Peter is the authority
> > and will correct me; or try Google VMALLOC_RESERVE boot.
> >
>
> I think it affects initramfs too.
>
> It only affects boot loaders which don't support version 2.03 of the
> boot protocol; unfortunately that includes GRUB last I checked.
>
> We really need a better dialog with the GRUB people, unfortunately at
> least I have unsuccessful in starting such a dialog.
>
> GRUB, in particular, needs to report the boot loader ID they're using,
> plus support version 2.03 of the protocol.
>
> -hpa
Thanks guys. I "was" using grub I guess it's now back to lilo.
Mark