Hi Arnd,
On Sat, Aug 3, 2024 at 4:01 PM Arnd Bergmann <arnd@xxxxxxxxxx> wrote:
From: Arnd Bergmann <arnd@xxxxxxxx>
It is possible to select an m68k MMU build but not actually
enable any of the three MMU options, which then results in a
build failure:
arch/m68k/include/asm/page.h:10:25: error: 'CONFIG_PAGE_SHIFT' undeclared here (not in a function); did you mean 'CONFIG_LOG_BUF_SHIFT'?
Change the Kconfig selection to ensure that exactly one of the
three options is always enabled whenever an MMU-enabled kernel
is built, but moving CONFIG_SUN3 into a top-level option next
to M68KCLASSIC and COLDFIRE.
All defconfig files should keep working without changes,
but alldefconfig now builds support for the classic MMU.
Reported-by: kernel test robot <lkp@xxxxxxxxx>
Closes: https://lore.kernel.org/oe-kbuild-all/202408032138.P7sBvIns-lkp@xxxxxxxxx/
Signed-off-by: Arnd Bergmann <arnd@xxxxxxxx>
---
I saw another copy of the bug report today.
Geert, any opinions on this approach? I posted this before but
it looks like you missed it.
I was still thinking about it, and then holidays kicked in...
--- a/arch/m68k/Kconfig.cpu
+++ b/arch/m68k/Kconfig.cpu
@@ -32,13 +32,23 @@ config COLDFIRE
select HAVE_LEGACY_CLK
select HAVE_PAGE_SIZE_8KB if !MMU
-endchoice
+config SUN3
+ bool "Sun3 support"
+ depends on MMU
+ select HAVE_ARCH_PFN_VALID
+ select LEGACY_TIMER_TICK
+ select NO_DMA
+ select M68020
+ help
+ This option enables support for the Sun 3 series of workstations
+ (3/50, 3/60, 3/1xx, 3/2xx systems). These use a classic 68020 CPU
+ but the custom memory management unit makes them incompatible with
+ all other classic m68k machines, including Sun 3x.
Yes, it's a good idea to factor out at the top level machines that
need special handling.
However, the name of the choice now sounds a bit odd. Perhaps it should
be changed to "CPU/machine family support"?
Likewise for M68KCLASSIC.
-if M68KCLASSIC
+endchoice
config M68000
- def_bool y
- depends on !MMU
+ def_bool M68KCLASSIC && !MMU
I think this is unrelated, and should be spun-off into a separate patch
to fix the "no CPU type selected" issue?
select CPU_HAS_NO_BITFIELDS
select CPU_HAS_NO_CAS
select CPU_HAS_NO_MULDIV64
@@ -56,7 +66,8 @@ config M68000
a paging MMU.
config M68020
- bool "68020 support"
+ bool "68020 support" if M68KCLASSIC
+ default !(M68030 || M68040 || M68060)
Part of the "no CPU type selected" fix?
depends on MMU
select FPU
select CPU_HAS_ADDRESS_SPACES
You also need:
config M68KFPU_EMU
bool "Math emulation support"
- depends on M68KCLASSIC && FPU
+ depends on (M68KCLASSIC || SUN3) && FPU
--- a/arch/m68k/kernel/Makefile
+++ b/arch/m68k/kernel/Makefile
@@ -5,16 +5,8 @@
extra-y += vmlinux.lds
-obj-$(CONFIG_AMIGA) := head.o
-obj-$(CONFIG_ATARI) := head.o
-obj-$(CONFIG_MAC) := head.o
-obj-$(CONFIG_APOLLO) := head.o
-obj-$(CONFIG_VME) := head.o
-obj-$(CONFIG_HP300) := head.o
-obj-$(CONFIG_Q40) := head.o
-obj-$(CONFIG_SUN3X) := head.o
-obj-$(CONFIG_VIRT) := head.o
-obj-$(CONFIG_SUN3) := sun3-head.o
+obj-$(CONFIG_M68KCLASSIC) := head.o
+obj-$(CONFIG_SUN3) := sun3-head.o
Nice cleanup ;-)
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
