I have plans to start ASAP a fundraiser of 5,000.00 thousand dollars for you dos. I will be the first one to donate $500 to the fundraiser. We need to work together to push Librem 5 development.
My main wish is to quick release newer Redpine driver v2.6.1 for Librem 5 via downstream and upstream.
There is an MR by Angus to update it to 2.5.1 in our tree: https://source.puri.sm/Librem5/linux/-/merge_requests/578 that wasnāt pushed any further because of licensing issues. Seems like this got resolved in 2.6.1, so updating it in the downstream tree shouldnāt be a big problem.
Upstreaming is, however, a completely different matter. I donāt think Iād be able to do that, at least not without spending a lot of time getting familiar with mac80211 drivers first, and then Iām pretty sure there would be plenty of reworking needed to put that driver into upstreamable state. I can work on upstreaming some of our changes to drivers such as tps6598x, max170xx, bq25890, wm8962ā¦ or the thing I talked about at FOSDEMā¦ but Redpine driver is out of my competence.
Not at all. This issue was about USB resets happening occasionally when the modem attempted to wake itself up from USB suspend. You would see it disappearing from the bus and appearing back a moment later.
Was the talk not clear enough? Let me hear some feedback
The āhalf hangā (the problem that can result all sorts of erratic behavior, such as āsudoā hanging, the phone not suspending anymore etc.) is a result of use-after-free thatās happening in CDC-NCM gadget that implements the network interface over USB-C port. I havenāt really traced the whole execution path yet, but I found a recent patch on Linux mailing lists that was described to fix such an issue: Draft: usb: gadget: f_ncm: Fix Kernel Panic due to access of invalid gadget ptr (!769) Ā· Merge requests Ā· Librem5 / linux Ā· GitLab. This is a USB gadget, but ultimately the problem is not really related to USB - the same bug with the same consequences could end up in any other driver.
Thereās also the issue with cellular network connection getting stale (not receiving any incoming packets). This is actually related to USB power management (as in: this is what can trigger it), but so far it looks like the problem actually happens on the modem side rather than the host (but I donāt have any conclusive proof yet).
I moved on May 30 from one L5 to another. The new L5 has the SparkWLAN card, the older Redpine. Both are otherwise identical in kernel and below /home/purism. Since this, I faced no hangup when returning from suspend.
Another observation (as feeling) is, that the new L5 with this card draws more battery power when the L5 acts as Internet AccessPoint and is otherwise unused (display off, no apps running). I will do an exact measure when the device returns from display swap.
Senior leadership at Purism greatly appreciates @dosās work and is eager to ramp up billable hours as soon as financially viable. I scheduled a meeting with leadership for Thursday to sync up and kick off planning R&D efforts (more from a financial planning and high-level project management perspective at this time).
The half hang thing which I thought had something to do with the USB port issues. When I connect my Nexdock it will do this when I unplug or replug it.
Tested and it really work amazing. ( ! ) s2r.data bug it gone.
Fixed too.
If u want testing downloads packages below then install all: āsudo dpkg -i *.debā on Librem 5 then Reboot, check gsm.data box before and after s2r.
Gnu Byzantium download only.
All packages were processed from Petitboot System.
Gnu | Purism
Libre Software Sebastian Krzyszkowiak