ide: fix IDE ACPI regression breaking suspend

Commit 4dde449 ("ide: make drive->id
an union (take 2)") accidentally introduced an IDE ACPI regression which
at least on some machines confuses ACPI and breaks suspend. Fix it.

Fixes kernel.org bug #12279:
http://bugzilla.kernel.org/show_bug.cgi?id=12279

[ Sorry about the bug (IDE ACPI is not used by default so it managed
  to slip in).  Still it is kind of surprising that the bug breaks
  suspend as it shouldn't impact anything except IDE ACPI (it seems
  that it uncovered either some ACPI/PM deficiency or a BIOS bug). ]

Reported-and-bisected-by: David Roka <roka@dawid.hu>
Tested-by: David Roka <roka@dawid.hu>
Cc: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Len Brown <len.brown@intel.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

drivers/ide/ide-acpi.c

@@ -615,10 +615,10 @@ void ide_acpi_push_timing(ide_hwif_t *hwif)
 	in_params[0].buffer.length = sizeof(struct GTM_buffer);
 	in_params[0].buffer.pointer = (u8 *)&hwif->acpidata->gtm;
 	in_params[1].type = ACPI_TYPE_BUFFER;
-	in_params[1].buffer.length = sizeof(ATA_ID_WORDS * 2);
+	in_params[1].buffer.length = ATA_ID_WORDS * 2;
 	in_params[1].buffer.pointer = (u8 *)&master->idbuff;
 	in_params[2].type = ACPI_TYPE_BUFFER;
-	in_params[2].buffer.length = sizeof(ATA_ID_WORDS * 2);
+	in_params[2].buffer.length = ATA_ID_WORDS * 2;
 	in_params[2].buffer.pointer = (u8 *)&slave->idbuff;
 	/* Output buffer: _STM has no output */