Re: [PULL 3/3] hw/loongarch: Modify flash block size to 256K

["Philippe Mathieu-Daudé" <philmd@linaro.org>]

On 19/7/24 10:18, Daniel P. Berrangé wrote:
> On Fri, Jul 19, 2024 at 10:12:20AM +0200, Philippe Mathieu-Daudé wrote:
>> On 19/7/24 04:26, Song Gao wrote:
>>> From: Xianglai Li <lixianglai@loongson.cn>
>>>
>>> loongarch added a common library for edk2 to
>>> parse flash base addresses through fdt.
>>> For compatibility with other architectures,
>>> the flash block size in qemu is now changed to 256k.
>>>
>>> Signed-off-by: Xianglai Li <lixianglai@loongson.cn>
>>> Reviewed-by: Song Gao <gaosong@loongson.cn>
>>> Message-Id: <20240624033319.999631-1-lixianglai@loongson.cn>
>>> Signed-off-by: Song Gao <gaosong@loongson.cn>
>>> ---
>>>    include/hw/loongarch/virt.h | 2 +-
>>>    1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/include/hw/loongarch/virt.h b/include/hw/loongarch/virt.h
>>> index 8fdfacf268..603c1cebdb 100644
>>> --- a/include/hw/loongarch/virt.h
>>> +++ b/include/hw/loongarch/virt.h
>>> @@ -20,7 +20,7 @@
>>>    #define VIRT_FWCFG_BASE         0x1e020000UL
>>>    #define VIRT_BIOS_BASE          0x1c000000UL
>>>    #define VIRT_BIOS_SIZE          (16 * MiB)
>>> -#define VIRT_FLASH_SECTOR_SIZE  (128 * KiB)
>>> +#define VIRT_FLASH_SECTOR_SIZE  (256 * KiB)
>>
>> Again, I believe this breaks machine migration. See the recent
>> example Daniel explained to me:
>> https://lore.kernel.org/qemu-devel/Zn6EQ39Q57KtmKPU@redhat.com/
> 
> Yes, changing flash size breaks migration compat, but note that loongarch
> does not have any versioned machine types, so it has zero migration compat
> right now regardles of this change. IOW, now is the right time to make the
> change, before someone asks for versioned machines with loongarch.

Oh I see now, thanks for clarifying!

It would have been nice to mention that in the commit description,
but too late so I don't mind.

Regards,

Phil.