Hi Bjorn,

On 3/9/2017 5:27 PM, Bjorn Helgaas wrote:

This is an aspect of the ASPM design that I don't like:

  - pcie_aspm_init_link_state() is called on a *bridge* after we've
    enumerated any devices below the bridge, and we allocate the
    link_state.

  - pcie_aspm_exit_link_state() is called on an *endpoint*, and if
    we're removing the last endpoint below a bridge, we release the
    *parent's* link_state.

I just posted V4 a minute ago. I divided ASPM init into two so that the
ASPM registers are captured in device_add path. 

There is a section of the ASPM init code that needs to walk all the devices
under the bridge. 

This section of the code is not working in device_add path as the link list
for child devices has not been set up yet. 

Unless you tell me there is a better way to handle this, I don't see how
we can remove the call from scan_slot. 

After further review, I think I can split PATCH V4 3/3 into two. Before
diving into too much clean up, I wanted to get some feedback as it is
obvious from the messed up cover letter.

Sinan


-- 
Sinan Kaya
Qualcomm Datacenter Technologies, Inc. as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project.