Kikat - Wasser Firmware for Beelink R89

[sillykeith]

Disable Journaling

I too would like to try this out. But, I'm unable to complete the necessary unmount or e2fsck commands on any of the found ext4 volumes. They are all in use. Is there a way to boot just to a console to accomplish this?

[ICU]

BTW. Of all the ROMs I have tested mo123 dual boot (only installed android) is the only one where 1080p YouTube app works. I know wasser will turn the bee link 111k4 into a nice ROM.

[no_spam_for_me]

...
For install i'can use Factory Tool procedure like point 3 of this thread?

http://www.freaktab.com/showthread.php?16490-Beelink-R89-UBOX-ROM-Tools-release

Thanks at all, expecially at wasser!!

Why Factory Tool?
Use AndroidTool_Release_v2.3

[vanzylst]

can someone please confirm "disable journal" fixes micro stutters?
i can add this fix in next firmware release if it really fixes micro stutters

Wasser

After doing some further testing and after removing the ext4 journal, I still get micro stutters from time to time. I had a high bitrate BluRay rip play for 2 hours without any frame drops or skips, but then another much lower bit-rate file did have some frame drops/skips.

However, my experience is that removing the journal does improve things quite a bit, but it would be good if someone else can confirm as well.

I did manage to remove the journal on all the mounted partitions. I found a different e2fsck which would allow me to force the fsck on all the mounted partitions, whereafter I could run tune2fs -O ^has_journal to remove the journal.

This e2fsck worked for me (I assume it's part of Rom Manager which one can install from the Play Store):
-rwxr-xr-x 1 app_65 app_65 428664 Dec 25 20:01 /data/data/com.koushikdutta.rommanager/files/e2fsck

This one didn't:
-rwxr-x--- 1 system system 289448 Jan 1 1970 /sbin/e2fsck

[Gilgamesh]

guys, these micro stutters, what is the cause of them? the not so "speedy" eMMC? or CPU?

[vanzylst]

@ICU. Hold your horses for a moment . I've just found something strange. There are writes to disk quite often because jbd2 process is updating EXT4 jurnal and sometime kworker is hiking along repeatedly every 5-6 seconds. These operations doesn't generate too much data because only 4kB (8 sectors) is written for each but if we have 30-40 similar events in one second and factor poor IO performance of this device eMMC then it isn't funny.

<7>[12447.539933] jbd2/mmcblk0p13(128): WRITE block 12891048 on mmcblk0p13 (8 sectors)
...
<7>[12448.654973] kworker/u8:3(13334): WRITE block 22871632 on mmcblk0p13 (72 sectors)

To log these operation in debug messages (dmesg) you should enable debug in VM for block_dumps, like that:
echo "1" > /proc/sys/vm/block_dump
Disable:
echo "0" > /proc/sys/vm/block_dump

Tomorow I will check if my old BOX have similar behavior. Maybe this device is salvageable

After removing the ext4 journal I get the following two processes writing fairly constantly to disk:

<7>[ 1989.823437] NetworkStats(530): WRITE block 4264 on mmcblk0p12 (8 sectors)
<7>[ 1990.541758] NetworkStats(530): WRITE block 1772776 on mmcblk0p12 (16 sectors)
<7>[ 1990.543191] NetworkStats(530): WRITE block 3944 on mmcblk0p12 (8 sectors)
<7>[ 1990.556987] NetworkStats(530): WRITE block 1772792 on mmcblk0p12 (8 sectors)
<7>[ 1990.558007] NetworkStats(530): WRITE block 3872 on mmcblk0p12 (8 sectors)
<7>[ 1991.179627] NetworkStats(530): WRITE block 1772800 on mmcblk0p12 (16 sectors)
<7>[ 1991.180402] NetworkStats(530): WRITE block 4264 on mmcblk0p12 (8 sectors)
<7>[ 1991.194462] NetworkStats(530): WRITE block 1772816 on mmcblk0p12 (8 sectors)
<7>[ 1991.195243] NetworkStats(530): WRITE block 3944 on mmcblk0p12 (8 sectors)
<7>[ 1991.290620] kworker/u8:1(6387): WRITE block 1088 on mmcblk0p12 (8 sectors)
<7>[ 1991.290652] kworker/u8:1(6387): WRITE block 1160 on mmcblk0p12 (8 sectors)
<7>[ 1991.290664] kworker/u8:1(6387): WRITE block 1416 on mmcblk0p12 (8 sectors)
<7>[ 1991.290674] kworker/u8:1(6387): WRITE block 3872 on mmcblk0p12 (8 sectors)
<7>[ 1991.290688] kworker/u8:1(6387): WRITE block 67968 on mmcblk0p12 (8 sectors)
<7>[ 1991.290706] kworker/u8:1(6387): WRITE block 8 on mmcblk0p12 (8 sectors)
<7>[ 1991.290716] kworker/u8:1(6387): WRITE block 1080 on mmcblk0p12 (8 sectors)

Interestingly enough, the NetworkStats process doesn't show up on ps but it is in /proc. Not really sure what it's used for...

[vanzylst]

guys, these micro stutters, what is the cause of them? the not so "speedy" eMMC? or CPU?

I believe the CPU is more than adequately powerful.

It looks to me that the slow eMMC coupled with background processes doing regular writes is a contributor to the problem, but there are probably other causes as well.

[Gilgamesh]

ok will see if the HPH NT-V6 4GB/32GB box I ordered will behave the same as they use a much faster Samsung eMMC, write speed about 20MB/s, read 55MB/s compared to R28 Meta 10/35MB/s... but the Samsung eMMc is supposed to be even faster than that when looking at the specs.. but will see.

Tapatalked from my GALAXY S5

[crazymark]

Why Factory Tool?
Use AndroidTool_Release_v2.3

thanks for your answer!!
Using Androidtool 2.3, tab "Upgrade frmware" and select "firmare" --> "wasser firmware".img and click upgrade?

[gffmac]

Re: NEW - Wasser Firmware V2.0.3 for Beelink R89/Tronsmart R28

thanks for your answer!!
Using Androidtool 2.3, tab "Upgrade frmware" and select "firmare" --> "wasser firmware".img and click upgrade?

That is correct!

Sent from my One using Tapatalk

[Hui91]

antutu video tester feedback

I ran Antutu video tester and here are the results of Wasser V2.0.3

Some audio problems occured vs. 110k4 roms had no indicated playback issue with this test.



I am still looking for a solution to solve Phillips TVs HDMI HDCP handshake issue: resulting in green/blue screens randomly for a short time. Any suggestion? 110k4 roms dont have this issue.

[maxxxy]

sorry but v. 2.0.1

[hueb_s]

NEW - Wasser Firmware V2.0.3 for Beelink R89/Tronsmart R28

sorry but v. 2.0.1

What are you trying to tell us?

[roolmapool]

BTW. Of all the ROMs I have tested mo123 dual boot (only installed android) is the only one where 1080p YouTube app works. I know wasser will turn the bee link 111k4 into a nice ROM.

Wasser rom is based on lastest RK SDK... same than 111K4.

[sillykeith]

This e2fsck worked for me (I assume it's part of Rom Manager which one can install from the Play Store):
-rwxr-xr-x 1 app_65 app_65 428664 Dec 25 20:01 /data/data/com.koushikdutta.rommanager/files/e2fsck

Thanks vanzylst! Using Koushik's e2fsck worked!

NOTE: Once you install "Rom Manager" you need to launch it so that the app can update and pull down the e2fsck file.

The only mount points that I had to do this to were the following:

/dev/block/mmcblk0p8
/dev/block/mmcblk0p10
/dev/block/mmcblk0p11
/dev/block/mmcblk0p12

I ran the following commands on each one:

tune2fs -l /dev/block/mmcblk0p8

("| grep features" didn't work for me so I omitted it and just scrolled up)
This will let you see if the line Filesystem features: has_journal exists

/data/data/com.koushikdutta.rommanager/files/e2fsck -f /dev/block/mmcblk0p8

(answer Y to all)

tune2fs -O ^has_journal /dev/block/mmcblk08

(this is a capital "O"!)
This command removes journaling.

Do this for all four and reboot.

Now to test for micro stutters...
*UPDATE*
stutters still exist. Especially bad with 24p content (SD or HD).
The only way I was able to solve this was to turn OFF "adjust display refresh rate to match video" and turn ON "Sync playback to display" from within XMBC System -> Video -> Playback. Oh well..