[PATCH][RFC] acpi: method: refine _EVT test
Colin Ian King
colin.king at canonical.com
Fri Jul 31 14:38:48 UTC 2015
On 31/07/15 15:25, Alex Hung wrote:
> There is a side-effect on this patch that I'd like to discuss (therefore
> RFC in topic).
>
> When I run "sudo make check", it shows an failure as below:
>
> < method _EVT.
> ---
>> method _AEI.
>
> As it is the result of below:
>
> method Test 2 of 180: Test _AEI.
> method SKIPPED: Test 2, Skipping test for non-existant object
> method _AEI.
> method
> method Test 3 of 180: Test _EVT (Event Method).
> method SKIPPED: Test 3, Skipping test for non-existant object
> method _AEI.
>
> From function perspective, the patch works fine but it may not look
> quite the same as other methods. Does anyone have any comments?
The skipping message is a bit confusing. One way around this is to add
an extra bit to the test_type flag in method_evaluate_method() so we can
decided to print messages or not, e.g.
#define METHOD_SILENT 8
...
and in method_evaluate_method() do:
if (!(test_type & METHOD_SILENT)) {
/* Manditory not-found test are a failure */
if (test_type & METHOD_MANDITORY) {
fwts_failed(fw, LOG_LEVEL_MEDIUM, "MethodNotExist",
"Object %s did not exist.", name);
}
/* Mobile specific tests on non-mobile platform? */
if ((test_type & METHOD_MOBILE) &&
(!fadt_mobile_platform)) {
fwts_skipped(fw,
"Machine is not a mobile platform,
skipping "
"test for non-existant mobile platform "
"related object %s.", name);
} else {
fwts_skipped(fw,
"Skipping test for non-existant object %s.",
name);
}
}
return FWTS_NOT_EXIST;
and in then:
static int method_test_EVT(fwts_framework *fw)
{
int ret;
/* Only test the _EVT method with pins defined in AEI. */
ret = method_evaluate_method(fw, METHOD_OPTIONAL |
METHOD_SILENT, "_AEI", NULL, 0, method_test_EVT_return, NULL);
if (ret == FWTS_NOT_EXIST)
fwts_skipped(fw,
"Skipping test for non-existant object %s.",name);
return ret;
}
Perhaps that's the simplest way forward.
>
>
> On 07/31/2015 09:44 PM, Alex Hung wrote:
>> According to ACPI spec, _EVT's events should be in adjacent _AEI
>> control method. This patch also reduces lengthy testing time.
>>
>> This was modified from a patch from Fan Wu <wufan at codeaurora.org>
>> Most changes are styles, comments and some small code refactoring.
>>
>> Signed-off-by: Alex Hung <alex.hung at canonical.com>
>> ---
>> src/acpi/method/method.c | 86 ++++++++++++++++++++++++++++++++++++++++++------
>> 1 file changed, 76 insertions(+), 10 deletions(-)
>>
>> diff --git a/src/acpi/method/method.c b/src/acpi/method/method.c
>> index 638e937..732facc 100644
>> --- a/src/acpi/method/method.c
>> +++ b/src/acpi/method/method.c
>> @@ -27,6 +27,7 @@
>> #include <inttypes.h>
>> #include "fwts_acpi_object_eval.h"
>>
>> +
>> /*
>> * ACPI methods + objects used in Linux ACPI driver:
>> *
>> @@ -958,21 +959,86 @@ static int method_test_AEI(fwts_framework *fw)
>> "_AEI", NULL, 0, method_test_AEI_return, NULL);
>> }
>>
>> -static int method_test_EVT(fwts_framework *fw)
>> +static void check_evt_event (
>> + fwts_framework *fw,
>> + ACPI_RESOURCE_GPIO *gpio)
>> {
>> ACPI_OBJECT arg[1];
>> - int ret, i;
>> + ACPI_HANDLE evtMethodHandle;
>> + ACPI_STATUS Status;
>> + char Path[256];
>> + uint16_t i;
>> +
>> + /* Skip the leading spaces in ResourceSource. */
>> + for (i = 0; i < gpio->ResourceSource.StringLength; i++) {
>> + if (gpio->ResourceSource.StringPtr[i] != ' ')
>> + break;
>> + }
>>
>> - arg[0].Type = ACPI_TYPE_INTEGER;
>> - for (i = 0; i < 65535; i++) {
>> - arg[0].Integer.Value = i;
>> - ret = method_evaluate_method(fw, METHOD_OPTIONAL,
>> - "_EVT", arg, 1, method_test_NULL_return, NULL);
>> + if (i == gpio->ResourceSource.StringLength ) {
>> + fwts_log_warning(fw, "Invalid ResourceSource");
>> + return;
>> + }
>>
>> - if (ret != FWTS_OK)
>> - break;
>> + /* Get the handle of return;the _EVT method. */
>> + sprintf (Path, "%s._EVT", &gpio->ResourceSource.StringPtr[i]);
>> +
>> + Status = AcpiGetHandle (NULL, Path, &evtMethodHandle);
>> + if (ACPI_FAILURE(Status)) {
>> + fwts_log_warning(fw, "Failed to find valid handle for EVT method (0x%x), %s", Status, Path);
>> + return;
>> }
>> - return ret;
>> +
>> + /* Call the _EVT method with all the pins defined for the GpioInt */
>> + for (i = 0; i < gpio->PinTableLength; i++) {
>> + ACPI_OBJECT_LIST arg_list;
>> +
>> + arg[0].Type = ACPI_TYPE_INTEGER;
>> + arg[0].Integer.Value = gpio->PinTable[i];
>> + arg_list.Count = 1;
>> + arg_list.Pointer = arg;
>> +
>> + method_evaluate_found_method(fw, Path, method_test_NULL_return, NULL, &arg_list);
>> + }
>> +}
>> +
>> +static void method_test_EVT_return (
>> + fwts_framework *fw,
>> + char *name,
>> + ACPI_BUFFER *buf,
>> + ACPI_OBJECT *obj,
>> + void *private)
>> +{
>> + ACPI_RESOURCE *resource;
>> + ACPI_STATUS Status;
>> +
>> + FWTS_UNUSED(buf);
>> + FWTS_UNUSED(private);
>> +
>> + if (method_check_type(fw, name, buf, ACPI_TYPE_BUFFER) != FWTS_OK)
>> + return;
>> +
>> + Status = AcpiBufferToResource( obj->Buffer.Pointer, obj->Buffer.Length, &resource );
>> + if (ACPI_FAILURE(Status))
>> + return;
>> +
>> + do {
>> + if (!resource->Length) {
>> + fwts_log_warning(fw, "Invalid zero length descriptor in resource list\n");
>> + break;
>> + }
>> +
>> + if (resource->Type == ACPI_RESOURCE_TYPE_GPIO && resource->Data.Gpio.ConnectionType == ACPI_RESOURCE_GPIO_TYPE_INT)
>> + check_evt_event(fw, &resource->Data.Gpio);
>> +
>> + resource = ACPI_NEXT_RESOURCE(resource);
>> + } while (resource->Type != ACPI_RESOURCE_TYPE_END_TAG);
>> +}
>> +static int method_test_EVT(fwts_framework *fw)
>> +{
>> + /* Only test the _EVT method with pins defined in AEI. */
>> + return method_evaluate_method(fw, METHOD_OPTIONAL,
>> + "_AEI", NULL, 0, method_test_EVT_return, NULL);
>> }
>>
>> /*
>>
>
>
More information about the fwts-devel
mailing list