Root [3.4.107] [Linaro] Schwartz v1.0 4/24/15

[spock1104]

Actually I'm not sure if this works on stock or any variant (i.e. DSF), I've only used it with sawbones' rom.

 

[H_Man]

I installed this and it seems that the phone got unstable. Buying another microSD card so I can try with a blank slate on everything to see if the kernel is at fault (I liked the improved performance). Does the kernel ignore the grade of chip indicated by the fuses and run the lowest voltage table?

 

[spock1104]

I installed this and it seems that the phone got unstable. Buying another microSD card so I can try with a blank slate on everything to see if the kernel is at fault (I liked the improved performance). Does the kernel ignore the grade of chip indicated by the fuses and run the lowest voltage table?

No, I kept the voltage tables, expanded each one to allow overclocking but the best bins will overclock the furthest. Instability comes from the way it's compiled; improved performance but more risk of crashing. Stability will be ymmv, completely dependent on which apps are in use and how they interact with the kernel. My phone is stable but I really don't use it much, it's a backup for me. That said, after any crashes if you post a last_kmsg I will be happy to analyze and see if there's anything I can fix.

 

[H_Man]

Good. I found it odd that my phone had the best voltage binning ( best gives 384 MHz 850mV, next best gives 384 MHz 900mV, worst gives 384 MHz 950 mV) considering it was a $25 referb unit.
It will be a bit until I can post a last_kmsg since I'm waiting on a new microSD card. I may just have to stick to the stock kernel on the sawbones rom since this phone is my main smartphone. I need the phone to run waze for my 70 mile one way commute. (I have a tracfone that costs me $6 a month via piggybacking off of a faimly plan for rock solid reliable text and voice; then the ZTE force on freedompop costs me $0 a month for questionably reliable bulk text, voice, and data. The setup is a bit cumbersome but it works and I can't argue with a setup that has cost me less than $10 a month even when counting the upfront hardware cost in that number.)

I tried applying an amplify config I've ran before on the stock kernal and it gave similar issues as what I had with this excellent kernel. I need to approach this scientifically I think, I may have broken something in an overly aggressive effort to get idle power draw below 1%/hour. I guess this may be a case of 'Fix it until it is broke.' Good think I took a backup of the phone before I broke it.


Any hope to get USB OTG working? I'd like to be able to do backups to a USB HDD some day, or maybe even play emulators with a wired controller.

 

[spock1104]

Good. I found it odd that my phone had the best voltage binning ( best gives 384 MHz 850mV, next best gives 384 MHz 900mV, worst gives 384 MHz 950 mV) considering it was a $25 referb unit.
It will be a bit until I can post a last_kmsg since I'm waiting on a new microSD card. I may just have to stick to the stock kernel on the sawbones rom since this phone is my main smartphone. I need the phone to run waze for my 70 mile one way commute. (I have a tracfone that costs me $6 a month via piggybacking off of a faimly plan for rock solid reliable text and voice; then the ZTE force on freedompop costs me $0 a month for questionably reliable bulk text, voice, and data. The setup is a bit cumbersome but it works and I can't argue with a setup that has cost me less than $10 a month even when counting the upfront hardware cost in that number.)

I tried applying an amplify config I've ran before on the stock kernal and it gave similar issues as what I had with this excellent kernel. I need to approach this scientifically I think, I may have broken something in an overly aggressive effort to get idle power draw below 1%/hour. I guess this may be a case of 'Fix it until it is broke.' Good think I took a backup of the phone before I broke it.


Any hope to get USB OTG working? I'd like to be able to do backups to a USB HDD some day, or maybe even play emulators with a wired controller.

I don't think OTG is going to work, or at least not well. The phone can't supply power to its USB port, so using OTG would require a Y adapter.

 

[H_Man]

I don't think OTG is going to work, or at least not well. The phone can't supply power to its USB port, so using OTG would require a Y adapter.

Drive is powered by a wall wart. It is a 4 TB external. Why can't the phone supply power? Did ZTE just not put down the traces for it? The power management chip supports it.
On another note, this kernel does seem to improve internal flash performance by 50%.

 

[spock1104]

Drive is powered by a wall wart. It is a 4 TB external. Why can't the phone supply power? Did ZTE just not put down the traces for it? The power management chip supports it.
On another note, this kernel does seem to improve internal flash performance by 50%.

Probably the traces. Nexus 4 notoriously had the same issue.

 

[H_Man]

I wonder who made the choice to save the penny and omit that.
Anyway, is there any USB-OTG hope for devices that don't need the phone to supply power? dmesg seems to detect the cable correctly, it just won't detect anything else and the cable is good.

 

[H_Man]

Okay, I tried this kernel again and although it isn't outright crashing on me, it is laggy at times. Rolling back to the sawbones kernel made issue go away, I'm making a backup so that when stuff breaks, I can unmess it up. With a samsung pro+ 64 GB micro SD card in it, I get this (About 95% improvement in read performance vs the sawbones kernel):

However I can't seem to mount it if I'm using exFAT.


Still trying to chase down issues with the linaro kernel. Phone did forget how wake up, but that may have been a rouge wakelock draining the battery on me because the handset would only vibrate, no lights, when I pressed the power button. Trying to reproduce the issue so I can post logs. I want to use your kernel since on the sawbones kernel, the phone is spending 1.3% of the battery an hour (after putting wooden stakes through a bunch of wakelock sources and bloat services) by having the CPU awake 40% of the time the screen is off due to qcom_rx_wakelock and mmc1_detect, compared to the sub 10% with your kernel.