Re: Windows XP 0xA5 BSoD at IASL_VERSION Address

There's more usage of ACPI QWORD data types here [1] in the Sandybridge 
code.

Regards,
Felix

[1] 
https://github.com/coreboot/coreboot/blob/main/src/northbridge/intel/sandybridge/acpi/hostbridge.asl#L372

On 03/04/2024 23:03, Felix Held wrote:
> Hi,
> 
> I think there was some comment in the code that Windows XP doesn't 
> support the ACPI QWORD data type. coreboot uses QWORD ACPI type in the 
> DSDT. Try if locally reverting this patch [1] will change anything. 
> Since coreboot uses a much newer ACPI version than 1, the resulting DSDT 
> should be correct regarding this and it's likely an OS bug if that local 
> revert fixes things.
> 
> Regards,
> Felix
> 
> [1] https://review.coreboot.org/c/coreboot/+/79096
> 
> On 03/04/2024 22:06, Rudolf Marek wrote:
>> Hi again,
>>
>> Dne 02. 04. 24 v 22:37 Rudolf Marek napsal(a):
>>> Hi
>>>
>>> Dne 02. 04. 24 v 21:12 Grayson Guarino napsal(a):
>>>> Do you have any suggestions for how I can further attempt to debug 
>>>> this? I'm struggling to find the release where ACPI v1.0 
>>>> compatibility was officially dropped, so if it is not 09eb8d0c9b, 
>>>> let me know and I will attempt to rebuild and flash again.
>>>
>>> Likely using windbg debugging windows kernel over serial line and 
>>> some checked build of windows would help to debug that. The coreboot 
>>> booted windows XP first time december 2008 (well that is a blast from 
>>> the past!) using some of my patches/email attached below. Please try 
>>> to check what were the tables versions (you will need coreboot SVN)
>>>
>>> I'm not a windows expert. I think windows often checks if 
>>> sizeof(acpitable) matches expected sizeof() defined in the ACPI 
>>> specs. So, if you  trying with ACPI 1.0 make sure to actually change 
>>> sizes of the table structures, otherwise it won't work. Also make 
>>> sure you don't get caught in the mess of RSDP and 1.0 
>>> https://uefi.org/specs/ACPI/6.5/05_ACPI_Software_Programming_Model.html#root-system-description-pointer-rsdp-structure
>>>
>>> I think I had to fix FACS table size to make windows happy... (see 
>>> rest of email/patches).
>>
>> So according to 
>> https://github.com/MicrosoftDocs/windows-driver-docs/blob/staging/windows-driver-docs-pr/debugger/bug-check-0xa5--acpi-bios-error.md
>>
>> Your stop code means that windows failed to load DSDT -> "ACPI failed 
>> to load DDB." You need to make sure that you generate AML bytecode 
>> which is OK for windows. I remember there was something about 32-bit 
>> vs 64-bit integers or QWORDS.... I would suggest to try to see how the 
>> DSDT mentioned in the previous email looked like.
>>
>> Maybe some 64-bit arithmetic is a problem... also try to find some 
>> microsoft document on  what windows xp actually expects - I suspect it 
>> wants ACPI 1.0b.
>>
>> Thanks,
>> Rudolf
>>
>> _______________________________________________
>> coreboot mailing list -- [email protected]
>> To unsubscribe send an email to [email protected]
> _______________________________________________
> coreboot mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
_______________________________________________
coreboot mailing list -- [email protected]
To unsubscribe send an email to [email protected]