[GIT PULL v2] interconnect changes for 5.5

[GIT PULL v2] interconnect changes for 5.5

[Georgi Djakov]

Hi Greg,

This is the updated pull request with interconnect patches for the 5.5 merge
window. The details are in the signed tag. Please pull into char-misc-next.

Thanks,
Georgi

The following changes since commit 4f5cafb5cb8471e54afdc9054d973535614f7675:

  Linux 5.4-rc3 (2019-10-13 16:37:36 -0700)

are available in the Git repository at:

  https://git.linaro.org/people/georgi.djakov/linux.git tags/icc-5.5-rc1

for you to fetch changes up to 83561a721dc23330d36e0ce499e716cd65f436fe:

  interconnect: Move interconnect drivers to core_initcall (2019-11-07 19:26:19
+0200)

----------------------------------------------------------------
interconnect patches for 5.5

Here are the interconnect updates for the 5.5-rc1 merge window.

- New interconnect driver for msm8974 platforms.
- Change the initcall level of a driver, so it is available earlier to
other dependent drivers.

Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>

----------------------------------------------------------------
Brian Masney (2):
      dt-bindings: interconnect: qcom: add msm8974 bindings
      interconnect: qcom: add msm8974 driver

Georgi Djakov (1):
      interconnect: Add locking in icc_set_tag()

Jordan Crouse (1):
      interconnect: Move interconnect drivers to core_initcall

Leonard Crestez (1):
      interconnect: qcom: Fix icc_onecell_data allocation

 Documentation/devicetree/bindings/interconnect/qcom,msm8974.yaml |  62 +
 drivers/interconnect/core.c                                      |   4 +
 drivers/interconnect/qcom/Kconfig                                |   9 +
 drivers/interconnect/qcom/Makefile                               |   2 +
 drivers/interconnect/qcom/msm8974.c                              | 796 ++++++++
 drivers/interconnect/qcom/qcs404.c                               |  17 +-
 drivers/interconnect/qcom/sdm845.c                               |  16 +-
 include/dt-bindings/interconnect/qcom,msm8974.h                  | 146 ++
 8 files changed, 1048 insertions(+), 4 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/interconnect/qcom,msm8974.yaml
 create mode 100644 drivers/interconnect/qcom/msm8974.c
 create mode 100644 include/dt-bindings/interconnect/qcom,msm8974.h

Re: [GIT PULL v2] interconnect changes for 5.5

[Greg Kroah-Hartman]

On Fri, Nov 08, 2019 at 10:59:11AM +0200, Georgi Djakov wrote:
> Hi Greg,
> 
> This is the updated pull request with interconnect patches for the 5.5 merge
> window. The details are in the signed tag. Please pull into char-misc-next.

Ugh, ok, can we just go back to sending patches instead?

The first two patches have "Fixes:" tags for older kernels, yet no
stable tags.  That's not ok.

For patch 5, you are playing "chicken" with what module gets loaded
first, and in the end you're not going to win, only delay the
inevitable.  We now have the driver core framework to support
dependencies like this, please use that instead.

So, can you drop patch 5, fix up patches 1 and 2, and go back to sending
patches instead of pull requests as if these were patches, I could have
fixed up 1 and 2 myself.

thanks,

greg k-h

Re: [GIT PULL v2] interconnect changes for 5.5

[Georgi Djakov]

On 8.11.19 г. 12:53 ч., Greg Kroah-Hartman wrote:
> On Fri, Nov 08, 2019 at 10:59:11AM +0200, Georgi Djakov wrote:
>> Hi Greg,
>>
>> This is the updated pull request with interconnect patches for the 5.5 merge
>> window. The details are in the signed tag. Please pull into char-misc-next.
> 
> Ugh, ok, can we just go back to sending patches instead?
> 
> The first two patches have "Fixes:" tags for older kernels, yet no
> stable tags.  That's not ok.

These two patches are for the current 5.4 release and you have them in your
char-misc-linus branch already. I assume that you will get them in -next later
via back merge anyway, so will drop them.
Should we use Fixes and stable tags always together and only when we fix
something in older kernels and not in the current one?

> For patch 5, you are playing "chicken" with what module gets loaded
> first, and in the end you're not going to win, only delay the
> inevitable.  We now have the driver core framework to support
> dependencies like this, please use that instead.

Yes, i keep an eye on this and am currently testing it.

> So, can you drop patch 5, fix up patches 1 and 2, and go back to sending
> patches instead of pull requests as if these were patches, I could have
> fixed up 1 and 2 myself.

Will do it!

Thanks,
Georgi

Re: [GIT PULL v2] interconnect changes for 5.5

[Greg Kroah-Hartman]

On Fri, Nov 08, 2019 at 01:47:51PM +0200, Georgi Djakov wrote:
> On 8.11.19 г. 12:53 ч., Greg Kroah-Hartman wrote:
> > On Fri, Nov 08, 2019 at 10:59:11AM +0200, Georgi Djakov wrote:
> >> Hi Greg,
> >>
> >> This is the updated pull request with interconnect patches for the 5.5 merge
> >> window. The details are in the signed tag. Please pull into char-misc-next.
> > 
> > Ugh, ok, can we just go back to sending patches instead?
> > 
> > The first two patches have "Fixes:" tags for older kernels, yet no
> > stable tags.  That's not ok.
> 
> These two patches are for the current 5.4 release and you have them in your
> char-misc-linus branch already. I assume that you will get them in -next later
> via back merge anyway, so will drop them.

Ah, yeah, that's odd, don't send me stuff like that unless there is a
build dependency on them.

> Should we use Fixes and stable tags always together and only when we fix
> something in older kernels and not in the current one?

stable, no, fixes is fine.  It wasn't obvious that these were already in
another branch of mine.

Because of all of this, let's just stick to patches in emails please.

thanks,

greg k-h