Re: MHI driver issue with Sierra Wireless AirPrime EM919X

[Loic Poulain <loic.poulain@linaro.org>]

Hi Thomas,

On Fri, 30 Jul 2021 at 19:52, Thomas Perrot <thomas.perrot@bootlin.com> wrote:
>
> Hello,
>
> I’m trying to use the MHI driver with a Sierra Wireless AirPrime EM919X
> connected to an NXP i.MX6DL through a PCIe gen2 bus, on 5.13.6, but the
> device initialisation fails, as follows:
>
> [    3.029401] mhi-pci-generic 0000:01:00.0: BAR 0: assigned [mem
> 0x01100000-0x01100fff 64bit]
> [    3.041095] mhi-pci-generic 0000:01:00.0: enabling device (0140 ->
> 0142)
> [    3.118299] mhi mhi0: Requested to power ON
> [   11.370361] mhi mhi0: Power on setup success
> [   11.372503] mhi mhi0: Wait for device to enter SBL or Mission mode
> [   15.664068] mhi-pci-generic 0000:01:00.0: failed to suspend device:
> -16
> [   22.619948] mhi-pci-generic 0000:01:00.0: refused to change power
> state from D3hot to D0
> [   57.289998] mhi-pci-generic 0000:01:00.0: can't change power state
> from D3hot to D0 (config space inaccessible)
> [   73.769675] mhi-pci-generic 0000:01:00.0: can't change power state
> from D3cold to D0 (config space inaccessible)
> [   73.780188] mhi-pci-generic 0000:01:00.0: can't change power state
> from D3hot to D0 (config space inaccessible)
> [   73.836150] PC is at mhi_pci_read_reg+0xc/0x14
> [   73.840642] LR is at mhi_get_mhi_state+0x2c/0x5c
> [   74.121120] [<c05dcaf4>] (mhi_pci_read_reg) from [<c05d754c>]
> (mhi_get_mhi_state+0x2c/0x5c)
> [   74.129529] [<c05d754c>] (mhi_get_mhi_state) from [<c05da440>]
> (mhi_pm_resume+0x38/0x298)
> [   74.137754] [<c05da440>] (mhi_pm_resume) from [<c05dcbd0>]
> (mhi_pci_runtime_resume+0x70/0xe0)
> [   74.146326] [<c05dcbd0>] (mhi_pci_runtime_resume) from [<c0604da0>]
> (pci_pm_runtime_resume+0x84/0xa0)
>
> Do you have any idea where the issue might come from?
>
> MHI seems to enter in READY state and waiting to enter in SBL or
> Mission mode when the issue occurs, the host can start MHI
> initialization by programming MMIO registers and sets the device into
> MHI_M0 state.

runtime-pm is only enabled when the device is in mission mode, so the
device has likely been successfully initialized, It is just that there
is no info message indicating initialization completion and we should
probably add that. I think maybe about some PCIe suspend/resume
issues, could you try disabling runtime-pm for that device? or simply
comment `pm_runtime_allow` call?

Regards,
Loic