On Sat, Feb 04, 2017 at 03:02:15PM +0800, Hanjun Guo wrote:

Hi Lorenzo,

On 2017/2/4 2:36, Lorenzo Pieralisi wrote:

quoted

Hanjun, Sinan,

On Wed, Jan 18, 2017 at 08:54:50PM +0800, Hanjun Guo wrote:

quoted

From: Hanjun Guo <redacted>

With platform msi support landed in the kernel, and the introduction
of IORT for GICv3 ITS (PCI MSI) and SMMU, the framework for platform msi
is ready, this patch set add few patches to enable the ACPI platform
msi support.

For platform device connecting to ITS on arm platform, we have IORT
table with the named componant node to describe the mappings of paltform
device and ITS, so we can retrieve the dev id and find its parent
irqdomain (ITS) from IORT table (simlar with the ACPI ITS support).

Depending on how things go, I prepared a branch with the first 12
patches (I basically updated some logs and added some cosmetics changes)
for testing (Hanjun please have a look in details since I may have misread
some logs), whether or not I will send a pull request for it we shall see
next week.

Here:

git://git.kernel.org/pub/scm/linux/kernel/git/lpieralisi/linux.git acpi/platform-msi

Thanks a lot for putting them together, I fetched your git tree and
took a detail look, there is one issue in patch "msi: platform: make
platform_msi_create_device_domain() ACPI aware" which has two
"Cc: Greg KH [off-list ref]" in the commit log
(it's in the original patch from me, my bad), others are pretty good
to me (to make sure it works I retested those patches and patches
in your branch work as before).

BTW, patches in Lorenzo's branch have no conflicts with Agustin's
patch set + my mbigen one, so after a pull request to Rafael, could
them (Agustin's patch set + my mbigen one) go via one tree such as
Marc's one for 4.11?  I know it's a little bit late but those patches
are quite self-contained, which Agustin's changes are conditional
on the ACPI_GENERIC_GSI config which is only available with
ARM64 ACPI and others (interrupt combiner and mbigen) are specific
to QC and Hisilicon platforms.

Marc, Lorenzo, could you give some comments that how can we
proceed those patches (Agustin's patch set + my mbigen one)?
It's really critical for us, thank you very much.

Ok, given that:

- We have decided that from now onwards ACPI IORT patches should go
  via the ARM64 tree
- This series does not yet handle ARM SMMU MSIs (but it has to and I
  want to see how this will work - waiting for spec updates)
- It depends on fixes that will get merged via the v4.11 arm64 queue
- We are at -rc7 and I do not think it is fair at all to ask Will
  and Catalin to pull this code now given that it comes out of thin
  air for them
- Last but not least: it is not bad to take time for the dust to
  settle given that we merged lots of IORT code last two cycles

I have decided that the whole series will be considered v4.12
material, I do not expect it to be a major disruption given that
we should have a stable code base for it come v4.11-rc1 (inclusive
of Agustin's key probe deferral series).

Thanks,
Lorenzo