platform/x86: intel-hid: fix _DSM function index handling

According to the ACPI spec 9.1.1 _DSM (Device Specific Method),
intel_hid_dsm_fn_mask, acquired from function index 0, is "a buffer
containing one bit for each function index". When validitaing fn_index,
it should be compared with corresponding bit.

This buffer is usually longer than a byte. Depending on whether
INTEL_HID_DSM_HEBC_V2_FN exist, it could be either
"Buffer (0x02) { 0xFF, 0x01 }" or "Buffer (0x02) { 0xFF, 0x03 }".
Probably it won't grow larger according to the description. On older
platforms, available functions could be fewer or not supported at all,
i.e., "Buffer (One) { 0x00 }".

Signed-off-by: Zhen Gong <zhengong@usc.edu>
Link: https://lore.kernel.org/r/CAJCLVRCyp0ASdWTx-PxsrDC9zFBPw0U2AtPip+_Hpj2r5gUPwA@mail.gmail.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
This commit is contained in:
Zhen Gong 2020-11-08 01:23:19 -08:00 committed by Hans de Goede
parent 3be3955315
commit 97ab451620

View File

@ -141,7 +141,7 @@ static bool intel_hid_execute_method(acpi_handle handle,
method_name = (char *)intel_hid_dsm_fn_to_method[fn_index];
if (!(intel_hid_dsm_fn_mask & fn_index))
if (!(intel_hid_dsm_fn_mask & BIT(fn_index)))
goto skip_dsm_exec;
/* All methods expects a package with one integer element */
@ -214,7 +214,19 @@ static void intel_hid_init_dsm(acpi_handle handle)
obj = acpi_evaluate_dsm_typed(handle, &intel_dsm_guid, 1, 0, NULL,
ACPI_TYPE_BUFFER);
if (obj) {
switch (obj->buffer.length) {
default:
case 2:
intel_hid_dsm_fn_mask = *(u16 *)obj->buffer.pointer;
break;
case 1:
intel_hid_dsm_fn_mask = *obj->buffer.pointer;
break;
case 0:
acpi_handle_warn(handle, "intel_hid_dsm_fn_mask length is zero\n");
intel_hid_dsm_fn_mask = 0;
break;
}
ACPI_FREE(obj);
}