Re: [PATCH] ACPICA: Log Exceptions and Errors as warning while loading extra tables

[Hans de Goede <hdegoede@redhat.com>]

Hi,

On 14-03-17 09:56, Hans de Goede wrote:
> Hi,
>
> On 14-03-17 09:15, Zheng, Lv wrote:
>> Hi, Hans
>>
>>> From: Hans de Goede [mailto:hdegoede@redhat.com]
>>> Subject: Re: [PATCH] ACPICA: Log Exceptions and Errors as warning while loading extra tables
>>>
>>> Hi,
>>>
>>> On 13-03-17 10:52, Zheng, Lv wrote:
>>>> Hi, Hans
>>>>
>>>> For log level issue, is this fix useful for you?
>>>> https://github.com/acpica/acpica/pull/121/commits/a505d3942
>>>
>>> That fixes reloading already loaded tables, the problem I'm
>>> getting errors about its loading a different table with identical
>>> contents.
>>>
>>> I will look into your suggestion to do something similar to
>>> AcpiTbInstallStandardTable using AcpiTbCompareTables for the
>>> SSDT tables. I will send a new patch when I can make some time
>>> to look into this.
>>
>> I just completed a prototype here:
>> https://github.com/acpica/acpica/pull/121
>>
>> I guess the original "duplicate table check" couldn't cover static SSDTs.
>> Actually the duplicate table check should be a sanity check of table load.
>> And for table install, we should have a different sanity check like:
>> https://github.com/acpica/acpica/pull/121/commits/6e825cae5e5
>> I'm not sure if this is what you want.
>
> This checks for having 2 table_descriptors pointing to the same table
> (address in memory). But in my case there are 2 identical copies of
> the table at 2 different addresses in memory, so this won't work.
>
> After looking at this a bit myself, I think fixing this is actually
> quite easy (now that you've pointed me to acpi_tb_install_standard_table()
>
> I've come up with the attached patch to fix this. I will give this a test
> spin and then submit it officially (assuming it works).

Ok the approach of doing the check during acpi_tb_install_standard_table
does not work because then acpi_gbl_verify_table_checksum is false so
we are only loading the header of the table, that and we are not supposed
to load more data / use more mem this early, which the call to
acpi_tb_acquire_table() will do for the table being compared against.

So it looks like we will need to go with some version of my patch which
does the check later when acpi_gbl_verify_table_checksum is true.

Regards,

Hans