[PATCH v4 03/13] arm/kexec: refactor for kernel/Kconfig.kexec

[Eric DeVolder]

On 7/5/23 10:05, Arnd Bergmann wrote:
> On Wed, Jul 5, 2023, at 16:19, Eric DeVolder wrote:
>> The kexec and crash kernel options are provided in the common
>> kernel/Kconfig.kexec. Utilize the common options and provide
>> the ARCH_SUPPORTS_ and ARCH_SELECTS_ entries to recreate the
>> equivalent set of KEXEC and CRASH options.
>>
>> Signed-off-by: Eric DeVolder <eric.devolder at oracle.com>
> 
>> +config ARCH_SUPPORTS_KEXEC
>> +	def_bool (!SMP || PM_SLEEP_SMP) && MMU
>>
>>   config ATAGS_PROC
>>   	bool "Export atags in procfs"
>> @@ -1668,17 +1656,8 @@ config ATAGS_PROC
>>   	  Should the atags used to boot the kernel be exported in an "atags"
>>   	  file in procfs. Useful with kexec.
>>
>> -config CRASH_DUMP
>> -	bool "Build kdump crash kernel (EXPERIMENTAL)"
>> -	help
>> -	  Generate crash dump after being started by kexec. This should
>> -	  be normally only set in special crash dump kernels which are
>> -	  loaded in the main kernel with kexec-tools into a specially
>> -	  reserved region and then later executed after a crash by
>> -	  kdump/kexec. The crash dump kernel must be compiled to a
>> -	  memory address not used by the main kernel
>> -
>> -	  For more details see Documentation/admin-guide/kdump/kdump.rst
>> +config ARCH_SUPPORTS_CRASH_DUMP
>> +	def_bool y
>>
> 
> I see this is now in linux-next, and it caused a few randconfig
> build issues, these never happened in the past:

Arnd,
Thanks for looking at this!

I received randconfig errors from Andrew Morton's machinery. When investigating I
found that randconfig was able to specify CRASH_DUMP without KEXEC, and that lead
to problems. I believe this situation existed prior to this series as well.
Specifically CRASH_DUMP does not have a dependency on KEXEC, or select (only s390
has this hole closed).

For CRASH_DUMP, this series now selects KEXEC to close this gap, which is what a
sane config would have (ie both CRASH_DUMP and KEXEC).

Do you think the changes outlined below are still needed?
eric


> 
> * The #ifdef CONFIG_KEXEC check in arch/arm/include/asm/kexec.h
>    needs to be changed to CONFIG_KEXEC_CORE:
> 
> include/linux/kexec.h:41:2: error: #error KEXEC_SOURCE_MEMORY_LIMIT not defined
> 
>    same thing on m68k
> 
> * ARCH_SUPPORTS_CRASH_DUMP needs the same dependency as ARCH_SUPPORTS_KEXEC,
>    otherwise we seem to run into an obscure assembler error building the kdump
>    core on architectures that do not support kdump:
> 
>    /tmp/ccpYl6w9.s:1546: Error: selected processor does not support requested special purpose register -- `mrs r2,cpsr'
> 
> * Most architectures build machine_kexec.o only when KEXEC is enabled,
>    this also needs to be changed to KEXEC_CORE:
> 
> --- a/arch/arm/kernel/Makefile
> +++ b/arch/arm/kernel/Makefile
> @@ -59,7 +59,7 @@ obj-$(CONFIG_FUNCTION_TRACER) += entry-ftrace.o
>   obj-$(CONFIG_DYNAMIC_FTRACE)   += ftrace.o insn.o patch.o
>   obj-$(CONFIG_FUNCTION_GRAPH_TRACER)    += ftrace.o insn.o patch.o
>   obj-$(CONFIG_JUMP_LABEL)       += jump_label.o insn.o patch.o
> -obj-$(CONFIG_KEXEC)            += machine_kexec.o relocate_kernel.o
> +obj-$(CONFIG_KEXEC_CORE)       += machine_kexec.o relocate_kernel.o
>   # Main staffs in KPROBES are in arch/arm/probes/ .
>   obj-$(CONFIG_KPROBES)          += patch.o insn.o
>   obj-$(CONFIG_OABI_COMPAT)      += sys_oabi-compat.o
> 
> 
>     Arnd