Re: Sierra Wireless EM9191 integration issues in mhi+wwan

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Aleksander,

On Mon, 2021-11-08 at 16:11 +0100, Aleksander Morgado wrote:
> Hey Thomas,
> 
> Reviving an old email :)
> 
> > On our setup, using i.MX6DL based board and a PCIe Sierra Wireless
> > EM9190 module, running Yocto and Linux 5.13, we don't have much
> > success
> > for the moment, qmi and mbim commands very often end in timeout.
> > 
> > Otherwise, when responses are received, we also can observe strange
> > things: unexpected messages, response to previous commands or queue
> > buffer issue.
> > 
> 
> Once all my boot reliability issues seem solved, I've also started to
> notice what you mean here. If I run a normal ModemManager build with
> both QMI and MBIM enabled, MM will try to probe both the QMI and MBIM
> ports. When that happens, I have no idea why, the modem gets in some
> weird state with commands timing out and what not. Maybe it's because
> we're using both ports at the same time, maybe it's because we run
> QMI
> on both the QMI and MBIM ports, no idea, the only thing I know is
> that
> if you choose to use either one or the other, the whole setup is
> fully
> stable.
> 
> E.g. I'm right now testing my build after compiling ModemManager
> using
> --without-mbim (so QMI only), and I have absolutely no error. 

That's very good news!

> Another
> option if you don't want to rebuild MM is to flag the MBIM or the QMI
> port as ID_MM_PORT_IGNORE with udev rules, which is very likely what
> I'll end up doing in upstream ModemManager to have a proper default.
> 
> I was thinking in preparing and sending for review the EM91xx entry
> for drivers/bus/mhi/pci_generic.c, but it's mostly based on what you
> suggested in the Sierra Wireless forum, so not sure if you'd like to
> send it yourself here? 

I can send it myself here, and I add you as developer if you'd like ?

Best regards,
Thomas

> The only changes I did w.r.t. what you
> suggested are setting sideband_wake to false, and listing the
> PCI_DEVICE_SUB() before the more generic PCI_DEVICE() one.
> 

-- 
Thomas Perrot, Bootlin
Embedded Linux and kernel engineering
https://bootlin.com

Attachment: signature.asc
Description: This is a digitally signed message part


[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [Linux for Sparc]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux MIPS]     [ECOS]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux