On 2019/6/20 23:09, Igor Mammedov wrote:
> On Thu, 20 Jun 2019 22:04:01 +0800
> gengdongjiu <gengdongjiu@xxxxxxxxxx> wrote:
>
>> Hi Igor,
>> Thanks for your review.
>>
>> On 2019/6/20 20:10, Igor Mammedov wrote:
>>>> + */
>>>> +struct AcpiGenericErrorStatus {
>>>> + /* It is a bitmask composed of ACPI_GEBS_xxx macros */
>>>> + uint32_t block_status;
>>>> + uint32_t raw_data_offset;
>>>> + uint32_t raw_data_length;
>>>> + uint32_t data_length;
>>>> + uint32_t error_severity;
>>>> +} QEMU_PACKED;
>>>> +typedef struct AcpiGenericErrorStatus AcpiGenericErrorStatus;
>>> there shouldn't be packed structures,
>>> is it a leftover from previous version?
>>
>> I remember some people suggest to add QEMU_PACKED before, anyway I will remove it in my next version patch.
>
> Question is why it's there and where it is used?
sorry, it is my carelessness. it should be packed structures.
I used this structures to get its actual total size and member offset in [PATCH v17 10/10].
If it is not packed structures, the total size and member offset may be not right.
>
> BTW:
> series doesn't apply to master anymore.
> Do you have a repo somewhere available for testing?
Thanks, I appreciated that you can have a test.
I still do not upload repo, you can reset to below commit[1] in master and apply this series.
BTW：
If test series, you should make an guest memory hardware error, let guest access the error address, then it will happen RAS error.
I provide a software hard code method to test this series, you can refer to https://www.mail-archive.com/qemu-devel@xxxxxxxxxx/msg619771.html
[1]:
commit efb4f3b62c69383a7308d7b739a3193e7c0ccae8
Merge: 5f02262 e841257
Author: Peter Maydell <peter.maydell@xxxxxxxxxx>
Date: Fri May 10 14:49:36 2019 +0100
>
>>
>>>
>>>> +
>>>> +/*
>>>> + * Masks for block_status flags above
>>>> + */
>>>> +#define ACPI_GEBS_UNCORRECTABLE 1
>>>> +
>>>> +/*
>>
>
> .
>
