Subject: Re: VM_TOPDOWN
To: Quentin Garnier <email@example.com>
From: Andrew Brown <firstname.lastname@example.org>
Date: 01/02/2006 14:22:34
On Mon, Jan 02, 2006 at 03:00:24PM +0100, Quentin Garnier wrote:
>I've just noticed that VM_TOPDOWN isn't used on amd64. Is it an
>overlook or is there a compelling reason for it?
no specific reason for it -- it's mainly just up to the portmaster.
>In any case, it seems to jsut work. The patch I'm including also adds
>code to make netbsd32 binaries use the topdown layout, and sync a few
>VM parameters with their i386 counterparts.
you should check with the portmaster. i don't know anything about the
vm layour on amd64, so i can't say if it's a win or not. on the alpha
(another 64 bit arch), for example, it's not, since the stack sits
roughly in the middle of the user's vm space, and doing topdown would
incur more tlb entries or something. i forget what.
>I'd like to commit it unless someone objects. Notice the huge available
>VM space for netbsd32 binaries on an amd64 system:
>0000000008048000 104K read/exec /mnt/netbsd32 -?-
>0000000008062000 4K read/write [ anon ]
>0000000008063000 24K read/write [ anon ]
>00000000FBEFF000 4K read/write [ anon ]
>00000000FBF00000 724K read/exec /mnt/netbsd32 -?-
much like the regular i386 space. :)
>amd64 binaries look like this:
>0000000000400000 116K read/exec / -?-
>000000000051C000 16K read/write / -?-
>0000000000520000 32K read/write [ anon ]
>00007F7FFD800000 800K read/exec / -?-
>00007F7FFD8C8000 1024K / -?-
>00007F7FFD9C8000 116K read/write / -?-
>00007F7FFD9E5000 68K read/write [ anon ]
looks fine to me, but i'm not the expert. :)
|-----< "CODE WARRIOR" >-----|
email@example.com * "ah! i see you have the internet
firstname.lastname@example.org (Andrew Brown) that goes *ping*!"
email@example.com * "information is power -- share the wealth."