Re: [RFC][PATCH 3/8] 4xx MMU

On Wed, 2007-07-11 at 22:56 +0200, Arnd Bergmann wrote:

On Wednesday 11 July 2007, Josh Boyer wrote:

quoted

+#elif defined(CONFIG_40x)
+/* 40x-style software loaded TLB */
+#  include <asm/mmu-4xx.h>
 #elif defined(CONFIG_44x)
 /* 44x-style software loaded TLB */
 #  include <asm/mmu-44x.h>

If you call it mmu-4xx, shouldn't it be used
for 44x as well? I would think this either
should be

No.  I was following the established convention that's been there for
years.  Which sucks I suppose for people not familiar with 40x/44x.
That would be a lot of stuff to change though...  e.g.:

arch/powerpc/platforms/4xx -> arch/powerpc/platforms/40x
arch/powerpc/kernel/head_4xx.S -> arch/powerpc/platforms/head_40x.S

etc.  I'd need a git tree to do that cleanly.  I've been toying with
creating one for powerpc4xx anyway so I might do that.

quoted

+#elif defined(CONFIG_4xx)
+/* 40x-style software loaded TLB */
+#  include <asm/mmu-4xx.h>
-#elif defined(CONFIG_44x)
-/* 44x-style software loaded TLB */
-#  include <asm/mmu-44x.h>

or 

quoted

+#elif defined(CONFIG_40x)
+/* 40x-style software loaded TLB */
+#  include <asm/mmu-40x.h>
 #elif defined(CONFIG_44x)
 /* 44x-style software loaded TLB */
 #  include <asm/mmu-44x.h>

I'll probably change it to this.  Will look a bit odd, given that C file
is arch/powerpc/mm/4xx_mmu.c.

Is it actually feasible to get to a point where
you can build a kernel that boots on both
40x and 44x, or is it just too different?

The MMUs are entirely different.  40x has real-mode and is 32-bit.  44x
is always virtual and is 36-bit.

josh