Using the MC7354 LTE Modem with QMI


I have a email out to my rep to get the Sierra Wireless Linux QMI SDK, but it is taking some time. I have several questions about QMI, since I have only worked with PPP connections with AT commands.

Will the MC7354 also work with the cdc-wdm and qmi_wwan drivers? Then I can use libqml to talk with it? If so what kernel should I be using?

Also what should the AT!UDUSBCOMP? command be set to in order to use QMI?

Currently it is set to 7, which is:


Should this be set to 6, which is:


Not sure what the difference is between RMNET1 and QMI in the AT!UDUSBCOMP? command.


qmi_wwan/libqml is not developed by SierraWireless and I think it appears since kernel 3.1.
However, it is suggested to use the QMI SDK/driver developed by Sierra as which was developed and tested for MC73xx.
Also reminded that qmi_wwan cannot work together with Sierra’s QMI driver, and Sierra’s driver was developed for kernel 2.6.31+.

Both 6 and 7 configuration in !UDUSBCOMP exposes the QMI interface, 7 comes with multi-PDN support.

Hope this helps.

As late as 3.4 actually.

And you’d need the latest stable v3.10.x, v3.12.x, v3.14.x or v3.15 version to have support for MC73xx, which was added just recently. As a community effort, new device additions will always lag behind because they depend on end user reports and testing.

I still wonder about the RMNET3 function… I note that this is disabled in the Sierra driver, and I assume that’s the main outcome of your testing :wink:

This is now also the case in the qmi_wwan driver. At least it will be when this commit hits mainline: … c59539725b

But would it be possible have some official statement explaining the state of RMNET3 support? I believe a note in the “USB Driver Developer’s Guide” would be appropriate. I have wondered about this question ever since I first laid my hands on a MC7710, which AFAICS has the exact same problem. It was interesting to see that the function now was disabled in Sierra’s own driver for the MC73xx.

BTW, I found it hilarious that your driver now includes some of my code, working around one of the more annoying Qualcomm firmware bugs. Thanks for that. It’s good to see that you have the same problems I’ve had trying to support these things :slight_smile: But I would have appreciated a little more respect for my copyright on that code… It is GPLv2 licensed so you are of course free to copy like everyone else. But that doesn’t mean that there aren’t rules to follow. A line mentioning the copyrights would be nice. And with or without that: It’s really not OK to just relicense GPLv2 code owned by someone else under a dual BSD/GPL license. You would need to have the copyright holder’s consent to do that. Which I’m pretty sure would be possible to get in this case, if you just asked for it…