Re: [git pull request] ACPI and driver patches for 2.6.33.merge

[Yinghai Lu <yinghai@kernel.org>]

Shaohua Li wrote:
> On Wed, Dec 23, 2009 at 10:09:04AM +0800, Yinghai Lu wrote:
>> On Tue, Dec 22, 2009 at 5:38 PM, Yinghai Lu <yinghai@kernel.org> wrote:
>>> Shaohua Li wrote:
>>>> On Tue, Dec 22, 2009 at 06:08:57PM +0800, Yinghai Lu wrote:
>>>>> Ingo Molnar wrote:
>>>>>> * Shaohua Li <shaohua.li@intel.com> wrote:
>>>>>>
>>>>>>> On Fri, Dec 18, 2009 at 09:28:50AM +0800, Yinghai Lu wrote:
>>>>>>>> On Wed, Dec 16, 2009 at 12:06 PM, Len Brown <lenb@kernel.org> wrote:
>>>>>>>>> Hi Linus,
>>>>>>>>>
>>>>>>>>> please pull from:
>>>>>>>>>
>>>>>>>>> git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6.git release
>>>>>>>>>
>>>>>>>> ;..
>>>>>>>>> Shaohua Li (3):
>>>>>>>>> ? ? ?ACPI: Add a generic API for _OSC -v2
>>>>>>>>> ? ? ?ACPI: cleanup pci_root _OSC code.
>>>>>>>>> ? ? ?ACPI: Add platform-wide _OSC support.
>>>>>>>> it seems these three patches broke the _OSC on my intel new systems.
>>>>>>>>
>>>>>>>> revert them fix the problem with AER and pciehp and etc
>>>>>>> can you give more details? I just cleaned up the _OSC code for AER and
>>>>>>> pciehp, no function changes.
>>>>>> Famous last words ;-)
>>>>>>
>>>>>> Yinghai, i suspect Shaohua needs the kind of info you'd need if you tried to
>>>>>> fix it: acpidump, before/after debug boot log, a description of what goes bad,
>>>>>> etc.
>>>>> the so called clean up, change the ret length checking.
>>>>>
>>>>> -       if (!output.length)
>>>>> -               return AE_NULL_OBJECT;
>>>>> -
>>>>>
>>>>> +       /* return buffer should have the same length as cap buffer */
>>>>> +       if (context->ret.length != context->cap.length)
>>>>> +               return AE_NULL_OBJECT;
>>>> Wield BIOS. ACPI spec does mention the return buffer have the same length.
>>>> Does changing the check back make the issue go away?
>>> change to
>>> if (context->ret.length < context->cap.length)
>>>
>>> make AER work, but pciehp still fail.
> Can you try below patch please? Looks the returned acpi buffer is a two-tiled buffer.
> Strange is it doesn't fail at my hand.
> 
> diff --git a/drivers/acpi/bus.c b/drivers/acpi/bus.c
> index 65f7e33..0c1ad31 100644
> --- a/drivers/acpi/bus.c
> +++ b/drivers/acpi/bus.c
> @@ -397,6 +397,7 @@ acpi_status acpi_run_osc(acpi_handle handle, struct acpi_osc_context *context)
>  	union acpi_object *out_obj;
>  	u8 uuid[16];
>  	u32 errors;
> +	struct acpi_buffer output = {ACPI_ALLOCATE_BUFFER, NULL};
>  
>  	if (!context)
>  		return AE_ERROR;
> @@ -419,16 +420,16 @@ acpi_status acpi_run_osc(acpi_handle handle, struct acpi_osc_context *context)
>  	in_params[3].buffer.length 	= context->cap.length;
>  	in_params[3].buffer.pointer 	= context->cap.pointer;
>  
> -	status = acpi_evaluate_object(handle, "_OSC", &input, &context->ret);
> +	status = acpi_evaluate_object(handle, "_OSC", &input, &output);
>  	if (ACPI_FAILURE(status))
>  		return status;
>  
> -	/* return buffer should have the same length as cap buffer */
> -	if (context->ret.length != context->cap.length)
> +	if (!output.length)
>  		return AE_NULL_OBJECT;
>  
> -	out_obj = context->ret.pointer;
> -	if (out_obj->type != ACPI_TYPE_BUFFER) {
> +	out_obj = output.pointer;
> +	if (out_obj->type != ACPI_TYPE_BUFFER
> +		|| out_obj->buffer.length != context->cap.length) {
>  		acpi_print_osc_error(handle, context,
>  			"_OSC evaluation returned wrong type");
>  		status = AE_TYPE;
> @@ -457,11 +458,20 @@ acpi_status acpi_run_osc(acpi_handle handle, struct acpi_osc_context *context)
>  		goto out_kfree;
>  	}
>  out_success:
> -	return AE_OK;
> +	context->ret.length = out_obj->buffer.length;
> +	context->ret.pointer = kmalloc(context->ret.length, GFP_KERNEL);
> +	if (!context->ret.pointer) {
> +		status =  AE_NO_MEMORY;
> +		goto out_kfree;
> +	}
> +	memcpy(context->ret.pointer, out_obj->buffer.pointer,
> +		context->ret.length);
> +	status =  AE_OK;
>  
>  out_kfree:
> -	kfree(context->ret.pointer);
> -	context->ret.pointer = NULL;
> +	kfree(output.pointer);
> +	if (status != AE_OK)
> +		context->ret.pointer = NULL;
>  	return status;
>  }
>  EXPORT_SYMBOL(acpi_run_osc);

aer and pciehp work again with this patch.

Thanks

Yinghai