Hiccups on Tronsmart R28

[lymy66]

I would like to use this tv box for something but I have a problem with hiccups (while moving the mouse cursor, it's stops from time to time) in entire system. This issue occurs also in the games. I've tried several roms (wasser, ugoos stage etc.) but there is still the same problem. I've tried to use app for the phone to control the mouse because I thought that it could be because of the air mouse but the issue still occur.
Is there any chance to fix this issue?

[mo123]

I would like to use this tv box for something but I have a problem with hiccups (while moving the mouse cursor, it's stops from time to time) in entire system. This issue occurs also in the games. I've tried several roms (wasser, ugoos stage etc.) but there is still the same problem. I've tried to use app for the phone to control the mouse because I thought that it could be because of the air mouse but the issue still occur.
Is there any chance to fix this issue?

It is a problem with Tronsmart & Beelink RK3288 devices.
Internal components like RAM, EMMC are lower speed & quality than more expensive boxes.

[lymy66]

It's a very sad information
I presume that it will not be fixed, so I need to think about buying something else...

[TheCritter]

You can test Wasser Firmware. Maybe it works, but without Remotecontrol:

http://freaktab.com/forum/tv-player-support/rk3288-devices/beelink-3288-devices/507181-new-lollipop-wasser-firmware-for-beelink-r89?p=548850#post548850

[Gormar]

You can test Wasser Firmware.

He did that. Read first post in this thread.

It's a very sad information
I presume that it will not be fixed, so I need to think about buying something else...

One and only complete solution for this problem is resoldering different eMMC in place of that PoS FreeSee (or something). It was done by HiSound from 4pda.ru forum.

I tried to find what cause those hiccups and discovered something very strange. MMC in Beelink R89 (identical to Tronsmart R28) after CMD13 returns always status equal to "900" which indicate none busy state and allows new data to be written constantly which is impossible without drive of light speed . It doesn't matter if I was copying big file (~500MB) from RAM drive (/mnt/obb). That might indicate that this eMMC (ForeSee) haven't got proper state register and it always returns the same answer: "ready to work". But that can end in congestion on eMMC bus and failed data read/send which are then resend multiple times. In normal situation after write operation to MMC kernel drive for MMC waits untill chosen drive comes out from transit state (writting) or there is a timeout.

You can check this on your device: "cat /sys/kernel/debug/mmc0/mmc0\:0001/status"

Even if MMC host driver (for RK3288 eMMC controller) properly maintains its own state register ("cat /sys/kernel/debug/mmc0/state") it doesn't matter in write operation when data are constantly send to MMC even when it can not properly write them because drive is busy with previous write operation but didn't return this crucial information to host MMC controller.

At the end a little trivia. RockChip kernel programmers made silly mistake and when you try to check content of MMC host primary registers from user space you will always get identical answer because they print registers "addresses" (offset from base address which is constant) and not what that register contain even if those data are available within kernel on that scope. Stupid mistake .
Check: cat /sys/kernel/debug/mmc0/regs. You will always get this:

root@rk3288_box:/mnt/obb # cat /sys/kernel/debug/mmc0/regs
STATUS: 0x00000048
RINTSTS:        0x00000044
CMD:    0x0000002c
CTRL:   0x00000000
INTMASK:        0x00000024
CLKENA: 0x00000010

No one use register to just hold constants .

[lymy66]

Wow. Thank you for your answer

[TheCritter]

@Gormar: ok for you: Please test the NEW Wasser Firmware 3.0.3
That I mean! On the first post the new 3.0.3. does not exist

[Gormar]

TheCritter. Ok. Sorry for my previous comment .

I now use mo123 LP LightBiz OS firmware mixed with Wasser's LP 3.0.3 (also LightBiz) kernel, resource and boot. Those hiccups aren't so much anoying like in the past but nevertheless they happen from time to time. They are especially evident if you are using some software to download files into internal memory (eMMC) in background then try to use other apps. That's why I download files onto external SD or internal RAM drive (/mnt/obb) if they are smaller than 1GB. Otherwise I would throw this box out of the window long ago. There might be no solution for them ;(.