On my Chromebook: 4.8-rc1 doesn’t crash wifi, has audio over HDMI and USB3, and fixed USB with LPAE! YAY!.
…and ac adapter module (bq24735) is not loading when not plugged in, again (which causes sbs-battery to not load, and battery monitors to not work). UGH!
The patch I submitted should still be working… It’s reaching the initialization code that it shouldn’t touch le-sigh…
In other news, removing the bq24725-charger as a power-supply in the DT for sbs-battery (sludge hammer attempt to get it to load w/o bq24735 module present) now makes it visible to the Cinnamon power applet? This was something I figured was a Cinnamon problem (as xfce4-power-manager would detect the battery properly), but maybe it’s a combination?
Also this makes me happy: mount my external USB3 SSD,
hdpartm -Tt yieldes 203 Mb/s read time!!! Whoo!
Actually, I now think Cinnamon doesn’t sanitize or handle the ‘@’ in bq24735-charger’s path name (funny enough, this was also true in udev originally), as the power-manager doesn’t show the battery info when the bq24735 module is loaded at all (even when sbs-battery doesn’t have it listed as a dependent power-supply).
Also, wifi does crash on my home network… There are bug reports that mwifiex is unstable in high throughput 5GHz (which would match my experience). However, exposing a 2.4 GHz AP and using that still results in crashes… grumble…
Also, I’ve noticed weird behavior in my USB3 drive: when I was at work letting things compile, I would come back to the filesystem being marked read-only, and apparent communication errors. I’m using dm-crypt so the read-only and communication errors may be misleading, but appears the USB loses connection briefly and reassigns the drive as a new dummy drive under sdb (and hence why dm-crypt may freak out). I think this is power-saving related somehow? If I sit actively working I never witness this problem (and had it running much longer last night while I was actively using my chromebook).
I think the WiFi problem may be related to MMC/SDHCI… The errors are related to timeouts waiting for interrupts, then an interrupt occurring when not waiting. I can’t seem to pin down the errors to anything mwifiex related (well, aside from not getting an interrupt through causing the firmware to get into an irrecoverable state…) 4.6 is when the problem started, and that is when they disabled UHS tuning for tegra124. There are a handful of commits related to timing as well… I’m patching to reenable UHS for tegra124 and seeing if it mitigates the problem (the Chromebook has become almost unusable at home…).