[PATCH 1/6] acpi: do not write ACPI tables on TARGET_QEMU_VIRT

[Heinrich Schuchardt]

On 11/16/23 19:33, Tom Rini wrote:
> On Wed, Nov 15, 2023 at 03:23:50PM +0100, Heinrich Schuchardt wrote:
> 
>> RISC-V QEMU provides the ACPI tables. We do not need to generate them
>> ourselves.
>>
>> Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt at canonical.com>
>> ---
>>   lib/acpi/Makefile      | 2 +-
>>   lib/acpi/acpi_writer.c | 2 +-
>>   2 files changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/lib/acpi/Makefile b/lib/acpi/Makefile
>> index c1c9675b5d..19fa6ac869 100644
>> --- a/lib/acpi/Makefile
>> +++ b/lib/acpi/Makefile
>> @@ -12,7 +12,7 @@ obj-$(CONFIG_$(SPL_)ACPIGEN) += acpi_table.o
>>   obj-y += acpi_writer.o
>>   
>>   # With QEMU the ACPI tables come from there, not from U-Boot
>> -ifndef CONFIG_QEMU
>> +ifeq ($(CONFIG_QEMU)$(CONFIG_TARGET_QEMU_VIRT),)
> 
> The goal here is to say that we have been passed ACPI tables, and not to
> generate them, yes? Lets just introduce a symbol which says that and
> enable it as needed in both QEMU targets and others as needed please.

These are the conflicting function implementations:

drivers/misc/qfw.c:158:ulong write_acpi_tables(ulong addr)
lib/acpi/acpi_writer.c:72:ulong write_acpi_tables(ulong start_addr)

drivers/misc/Makefile has:

ifdef CONFIG_QFW
obj-y += qfw.o

but inside the driver we find another constraint:

#if defined(CONFIG_GENERATE_ACPI_TABLE) && !defined(CONFIG_SANDBOX)

So probably we should probably have a symbol which is set for

CONFIG_QFW && CONFIG_GENERATE_ACPI_TABLE && !CONFIG_SANDBOX

Maybe call it CONFIG_QFW_ACPI, "Read ACPI tables from QEMU".

Best regards

Heinrich