Re: [APPL:DarwinKernel] Re: type sizes
Re: [APPL:DarwinKernel] Re: type sizes
- Subject: Re: [APPL:DarwinKernel] Re: type sizes
- From: Koryn Grant <email@hidden>
- Date: Thu, 28 Oct 2004 12:34:39 +1300 (NZDT)
On Thu, 28 Oct 2004, Quinn wrote:
> At 10:10 -0500 26/10/04, Jay A. Kreibich wrote:
> >The big question is how the kernel is going to run.
>
> The Tiger kernel will run 32 bit, even on G5. That is, sizeof(void
> *) == 4 for kernel code.
>
> Portions of the kernel have been modified to support 64 bit user
> address spaces but those portions represent user addresses using a 64
> bit integer type (that is, they don't use a C pointer).
So the Tiger release:
* has a 32-bit kernel address space regardless of host CPU;
* has some system libraries optimized for the host CPU, which includes
taking advantage of 64-bit operations when on G5;
* supports 32-bit user address spaces on any CPU, and additionally 64-bit
user address spaces (depending on whether the binary is 32/64-bit) on G5.
Is this an accurate summary?
Cheers,
Koryn
--
Dr Koryn Grant phone +64 7 8346729
Software Engineering Manager mobile +64 21 488386
Endace Technology Ltd email email@hidden
Hamilton, New Zealand web http://www.endace.com/
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden