Re: [PATCH v2] ACPI / x86: Work around broken XSDT on Advantech DAC-BJ01 board

[Hans de Goede]

Hi,

On 3/2/22 05:08, Mark Cilissen wrote:
> On this board the ACPI RSDP structure points to both a RSDT and an XSDT,
> but the XSDT points to a truncated FADT. This causes all sorts of trouble
> and usually a complete failure to boot after the following error occurs:
> 
>   ACPI Error: Unsupported address space: 0x20 (*/hwregs-*)
>   ACPI Error: AE_SUPPORT, Unable to initialize fixed events (*/evevent-*)
>   ACPI: Unable to start ACPI Interpreter
> 
> This leaves the ACPI implementation in such a broken state that subsequent
> kernel subsystem initialisations go wrong, resulting in among others
> mismapped PCI memory, SATA and USB enumeration failures, and freezes.
> 
> As this is an older embedded platform that will likely never see any BIOS
> updates to address this issue and its default shipping OS only complies to
> ACPI 1.0, work around this by forcing `acpi=rsdt`. This patch, applied on
> top of Linux 5.10.102, was confirmed on real hardware to fix the issue.
> 
> Signed-off-by: Mark Cilissen <mark@xxxxxxxxxx>
> Cc: stable@xxxxxxxxxxxxxxx
> Reviewed-by: Hans de Goede <hdegoede@xxxxxxxxxx>
> ---
> v2:
> - Reduce DMI match count to 4 to not overflow dmi_system_id structure
> Reported-by: kernel test robot <lkp@xxxxxxxxx>
> - Change board ident to correct name
> - Fix small style issue
> - Fix up subject as per Rafael's changes
> 
> As this patch is CC'd to stable, it seemed wiser to submit a V2 rather
> than an additional fixup patch to process.
> ---
>  arch/x86/kernel/acpi/boot.c | 24 ++++++++++++++++++++++++
>  1 file changed, 24 insertions(+)
> 
> diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
> index 5b6d1a95776f..b47338cd579d 100644
> --- a/arch/x86/kernel/acpi/boot.c
> +++ b/arch/x86/kernel/acpi/boot.c
> @@ -1328,6 +1328,17 @@ static int __init disable_acpi_pci(const struct dmi_system_id *d)
>  	return 0;
>  }
>  
> +static int __init disable_acpi_xsdt(const struct dmi_system_id *d)
> +{
> +	if (!acpi_force) {
> +		pr_notice("%s detected: force use of acpi=rsdt\n", d->ident);
> +		acpi_gbl_do_not_use_xsdt = TRUE;
> +	} else {
> +		pr_notice("Warning: DMI blacklist says broken, but acpi XSDT forced\n");
> +	}
> +	return 0;
> +}
> +
>  static int __init dmi_disable_acpi(const struct dmi_system_id *d)
>  {
>  	if (!acpi_force) {
> @@ -1451,6 +1462,19 @@ static const struct dmi_system_id acpi_dmi_table[] __initconst = {
>  		     DMI_MATCH(DMI_PRODUCT_NAME, "TravelMate 360"),
>  		     },
>  	 },
> +	/*
> +	 * Boxes that need ACPI XSDT use disabled due to corrupted tables
> +	 */
> +	{
> +	 .callback = disable_acpi_xsdt,
> +	 .ident = "Advantech DAC-BJ01",
> +	 .matches = {
> +		     DMI_MATCH(DMI_SYS_VENDOR, "NEC"),
> +		     DMI_MATCH(DMI_PRODUCT_NAME, "Bearlake CRB Board"),
> +		     DMI_MATCH(DMI_BIOS_VENDOR, "Phoenix Technologies LTD"),
> +		     DMI_MATCH(DMI_BIOS_VERSION, "V1.12"),
> +		     },
> +	 },

Heh, I should have noticed this new version before replying. I see that
you've dropped the BIOS-date match. But that actually is often more useful
then the BIOS_VERSION, sometimes vendors don't bump the version when
doing a new BIOS build.

If you only want to match the exact BIOS you tested against I would
drop the BIOS_VENDOR check instead.

Regards,

Hans



>  	{}
>  };
>  
> 
> base-commit: 038101e6b2cd5c55f888f85db42ea2ad3aecb4b6