Quiet

I've been quiet as the Linux machine has been running perfectly and I had no cause for complaints. I'm not running on the 4.4 kernel. I haven't tried a fix for the wireless mouse interference yet instead just being lazy and relocating the dongle to the USB port on the other side.

Acer V3-372 SSD Interference Wireless Mouse

I noticed that plugging the usb mouse receiver on the right side USB slots causes the following ATA errors:

[    9.998225] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300)[   10.001522] ata1.00: configured for UDMA/100[   10.002694] ata1: EH complete[   10.046132] ata1: limiting SATA link speed to 3.0 Gbps[   10.047274] ata1.00: exception Emask 0x10 SAct 0x700 SErr 0x280100 action 0x6 frozen[   10.048415] ata1.00: irq_stat 0x08000000, interface fatal error[   10.049557] ata1: SError: { UnrecovData 10B8B BadCRC }[   10.050707] ata1.00: failed command: READ FPDMA QUEUED[   10.051842] ata1.00: cmd 60/80:40:80:f8:3a/01:00:1c:00:00/40 tag 8 ncq 196608 in                        res 50/00:50:20:f8:3a/00:00:1c:00:00/40 Emask 0x10 (ATA bus error)[   10.054118] ata1.00: status: { DRDY }[   10.055250] ata1.00: failed command: READ FPDMA QUEUED[   10.056376] ata1.00: cmd 60/38:48:40:f8:3a/00:00:1c:00:00/40 tag 9 ncq 28672 in                        res 50/00:50:20:f8:3a/00:00:1c:00:00/40 Emask 0x10 (ATA bus error)[   10.058647] ata1.00: status: { DRDY }[   10.059759] ata1.00: failed command: READ FPDMA QUEUED[   10.060883] ata1.00: cmd 60/18:50:20:f8:3a/00:00:1c:00:00/40 tag 10 ncq 12288 in                        res 50/00:50:20:f8:3a/00:00:1c:00:00/40 Emask 0x10 (ATA bus error)[   10.063158] ata1.00: status: { DRDY }[   10.064277] ata1: hard resetting link[   10.382466] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 320)

Re-locating it to the left side solves it but ties up a USB3.0 slot. I will see if it is possible to shield the SSD.

Skylake bug confirmed

Mersenne Prime search has found a reproducible way to hang the CPU. Henk_NL writes:

The fatal combination seems to be:

- Skylake AND
- hyperthreading AND
- CpuSupportsFMA3=0. (only needed if you use the latest version of prime95, since previous version stem from before the introduction use FMA3 anyway)

There are people contemplating writing a piece of software based on prime95 that will simplify freezing the Skylake based on just a few instructions. Very scary indeed and further analysis of the downloadable code is required

https://communities.intel.com/mobile/mobile-access.jspa#jive-content?content=%2Fapi%2Fcore%2Fv3%2Fcontents%2F524553

Hopefully a microcode update will be issued soon.

Acer V3-372 Skylake Instability RC6

I have been able to consistently reproduce a GPU hang by playing Portal 2 under Linux. The game starts fine and you can play the first level until you sleep and then it crashes.

The fact that the crash occurs when the graphics are simple lead me to suspect that it could be a power-state issue and so I tried again this time disabling RC6 power-states.

Sure enough, I was able to play beyond this point. I guess it will be a case of waiting until the bugs are all fixed, which could take some time.

Although permanently disabling RC6 could be a workaround, it isn't appealing due to the reduced battery life. Also under the 4.4.0-rc6 kernel, the gpu hangs are more infrequent and no longer seem to lock up the entire system, so for the moment, I leave RC6 enabled.

Acer V3-372 with LInux Kernel 4.4 - stable?

I upgraded to kernel 4.4-rc4 and so far things are working great. I even re-enabled pstate and everything has been stable so far.

Also contributing to stability is that I have stopped using the buggy Asus wifi dongle and started using my Nexus 5 with USB tethering.

I determined a lot of the Steam crashes were due to crashed wifi.

Everything has been running great for 3 days now. I hope this works out. Tomorrow the Intel wifi card arrives which should mean the laptop will be finally fully functional!

Acer V3-372 WiFi Backup Plan with RTL8192cu

My $2 USB wifi adapter arrived a few days ago. At first I was very disappointed as the speed was excruciatingly slow.

I learned that this was due to poor hardware crypto performance. A simple swenc=1 option in module load up fixed it.

While I believe it is possible to compile a driver for the QCA6174 wireless module, I think this is likely to be more hassle than it is worth and so I ordered an Intel 7265n wifi card until the Qualcomm card gets mainline support.

Acer V3-372 SATA failed command: READ FPDMA QUEUED

Just when you think you got the laptop fixed and running well, I noticed these errors in dmesg:

[    9.998225] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
[   10.001522] ata1.00: configured for UDMA/100
[   10.002694] ata1: EH complete
[   10.046132] ata1: limiting SATA link speed to 3.0 Gbps
[   10.047274] ata1.00: exception Emask 0x10 SAct 0x700 SErr 0x280100 action 0x6 frozen
[   10.048415] ata1.00: irq_stat 0x08000000, interface fatal error
[   10.049557] ata1: SError: { UnrecovData 10B8B BadCRC }
[   10.050707] ata1.00: failed command: READ FPDMA QUEUED
[   10.051842] ata1.00: cmd 60/80:40:80:f8:3a/01:00:1c:00:00/40 tag 8 ncq 196608 in
                        res 50/00:50:20:f8:3a/00:00:1c:00:00/40 Emask 0x10 (ATA bus error)
[   10.054118] ata1.00: status: { DRDY }
[   10.055250] ata1.00: failed command: READ FPDMA QUEUED
[   10.056376] ata1.00: cmd 60/38:48:40:f8:3a/00:00:1c:00:00/40 tag 9 ncq 28672 in
                        res 50/00:50:20:f8:3a/00:00:1c:00:00/40 Emask 0x10 (ATA bus error)
[   10.058647] ata1.00: status: { DRDY }
[   10.059759] ata1.00: failed command: READ FPDMA QUEUED
[   10.060883] ata1.00: cmd 60/18:50:20:f8:3a/00:00:1c:00:00/40 tag 10 ncq 12288 in
                        res 50/00:50:20:f8:3a/00:00:1c:00:00/40 Emask 0x10 (ATA bus error)
[   10.063158] ata1.00: status: { DRDY }
[   10.064277] ata1: hard resetting link
[   10.382466] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 320)

Read errors leading to a down-rating of the link to 3Gbps. OK, not the end of the world, but potentially worrying if it portends future data integrity issues.

Having tested this further, I'm able to reproduce the errors when I put the Logitech mouse usb receiver on the usb port on the right of the machine (near the SSD). Putting the receiver on the left side away from the SSD eliminates this problem.

I will check whether it can be fixed with shielding.