*_find_end_of_memory Mach Dep?

*_find_end_of_memory Mach Dep?

[Grant Erickson]

Instead of having a mess of ifdef's and such in MMU_init for {apus, pmac,
m8xx, oak, chrp, etc}_find_end_of_memory, is there any compelling reason
not to make a ppc_md machine dependency entry for these?

If no one is in disagreement, I'd like to put a patch together for it.

Regards,

Grant


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: *_find_end_of_memory Mach Dep?

[Grant Erickson]

On Fri, 14 Jan 2000, Charles Lepple wrote:
> Grant Erickson wrote:
> > 
> > Instead of having a mess of ifdef's and such in MMU_init for {apus, pmac,
> > m8xx, oak, chrp, etc}_find_end_of_memory, is there any compelling reason
> > not to make a ppc_md machine dependency entry for these?
> 
> When I converted some code to the ppc_md interface, I seem to remember a
> few cases where I couldn't do it that way because the ppc_md structure
> was not completely filled out. It can get to be a catch-22 -- there has
> to be one messy function.

Fair enough. However, since identify_machine is called before MMU_init and
<platform>_init is called in identify_machine, then it's reasonable that
the ppc_md structure is filled out by the time we hit MMU_init.

So, it seems reasonable that "identify_machine" remains the messy
function and MMU_init could be cleaned up somewhat.

Grant


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: *_find_end_of_memory Mach Dep?

[Dan Malek]

Grant Erickson wrote:
> 
> Instead of having a mess of ifdef's and such in MMU_init for {apus, pmac,
> m8xx, oak, chrp, etc}_find_end_of_memory, is there any compelling reason
> not to make a ppc_md machine dependency entry for these?

I didn't see a mess of ifdef's.....in fact a ppc_md entry wouldn't
change the ifdef's, it would just remove a couple of lines of code
in MMU_init().

I put the ifdef's there originally to remove lots of code from the 8xx
port that isn't needed.  Perhaps with your 40x stuff we should consider
a more modular MMU_init() instead of the ifdef's, like creating multiple
mm/xxx_init.c files for the different MMU configurations (I can't believe
I said this, as I hate multiple files like this :-).

> If no one is in disagreement, I'd like to put a patch together for it.

Let's see what it looks like........


	-- Dan

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: *_find_end_of_memory Mach Dep?

[Dan Malek]

Grant Erickson wrote:

> On Fri, 14 Jan 2000, Charles Lepple wrote:
> > When I converted some code to the ppc_md interface, I seem to remember a
> > few cases where I couldn't do it that way because the ppc_md structure
> > was not completely filled out. It can get to be a catch-22 -- there has
> > to be one messy function.


I already checked this before I sent my reply, and it should work fine.


> So, it seems reasonable that "identify_machine" remains the messy
> function and MMU_init could be cleaned up somewhat.


It only looks messy because you haven't seen it grow over the years :-).
I think it is aging rather gracefully.....


	-- Dan

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/