ACPICA: Parser: Cleanup aml_offset in struct acpi_walk_state

ACPICA commit d254405814495058276c0c2f9d96794d15a6c91c

This patch converts aml_offset in struct acpi_walk_state to AML address.

AML offset is actually only used by the debugger, using AML address is more
direct and efficient during the parsing stage so that we don't need to
calculate it during the parsing stage.

On the other hand, we can see several issues in the current parser logic
around the aml_offset:
1. union acpi_operand_object.Common.aml_offset is redundantly assigned in
   acpi_ps_parse_loop().
2. aml_offset is not an indication of the offset from the table header but
   the offset from the entry of a list of objects. Sometimes, it indicates
   an entry for a Method/Package/Buffer, which makes it difficult to be
   reversely calculated to a table header offset.
3. When being used with method tracers (for example, Linux function trace),
   it's better to have AML address logged instead of the AML offset because
   the address is the only attribute that can uniquely identify the opcode.
This patch is required to solve the above issues. Lv Zheng.

Link: https://github.com/acpica/acpica/commit/d2544058
Signed-off-by: Lv Zheng <lv.zheng@intel.com>
Signed-off-by: Bob Moore <robert.moore@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

1 files changed, 1 insertions, 1 deletions

diff --git a/drivers/acpi/acpica/acstruct.h b/drivers/acpi/acpica/acstruct.h
index 44997ca..f9992dc 100644
--- a/drivers/acpi/acpica/acstruct.h
+++ b/drivers/acpi/acpica/acstruct.h
@@ -85,7 +85,7 @@ struct acpi_walk_state {
 	u8 namespace_override;	/* Override existing objects */
 	u8 result_size;		/* Total elements for the result stack */
 	u8 result_count;	/* Current number of occupied elements of result stack */
-	u32 aml_offset;
+	u8 *aml;
 	u32 arg_types;
 	u32 method_breakpoint;	/* For single stepping */
 	u32 user_breakpoint;	/* User AML breakpoint */