Thank you for confirming the original proposal for the ASL format , Hanjun.
I will refine this patchset, implement the clock definition, and then resubmit to the list for review.
Brandon
-----Original Message----- From: Hanjun Guo [mailto:hanjun.guo@linaro.org] Sent: Wednesday, October 23, 2013 2:38 AM To: Anderson, Brandon; linaro-acpi@lists.linaro.org; linux-acpi@vger.kernel.org Cc: lenb@kernel.org; rjw@sisk.pl; naresh.bhat@linaro.org; graeme.gregory@linaro.org; Suthikulpanit, Suravee Subject: Re: [RFC PATCH 0/3] ACPI: ARM AMBA bus connector resource
On 2013-10-23 8:13, Anderson, Brandon wrote:
Is the proposed ASL format appropriate? I recognize that the ASL
It is ok for me. it represents the device topology very clear also.
definitions for these devices must be standardized across all ARM ACPI platform definitions if we want the software to work consistently, so this is an important aspect to agree upon.
How about contrasting it with an alternative where each device has _HID of "AMBA0000" and there's no top-level 'Device (AMBA)'? This would mean that all the devices would be labeled by ACPI subsystem as "AMBA0000:NN" instead of "device:NN". However, with this solution there would be no easy way to define a default clock. See example below.
I think a top-level 'Device (AMBA)' is needed. there is a good example for PCI host bridge, its HID is PNP0A08 or PNP0A03, and there are a top level device for PCI host bridge and child devices with different HIDs.
if there's no top-level 'Device (AMBA)', AMBA0000:NN will confuse people if there are multi AMBAs in the system, people will think that AMBA0000:01 is the second AMBA bus in the system but not some devices under AMBA bus.
Thanks Hanjun