Askey RAC2V1K / RT4230W REV6 Support

[eganov]

yes just want them as option for install later
Thank you again

[lmore377]

@eganov I think I added that source right. Here's the image https://drive.google.com/drive/folders/1KZFzp82FH0Rp-9BygH_0EmxeG-AdC9u4?usp=sharing

[eganov]

Thank you very much
I would post the result soon

[eganov]

I have tried the bin
everything looks good Thank you.
at some point I can't save the config change Via Luci and at Interface packet traffic inticator TX/RX become 0 , but reboot the router fix the issue, I can't replicate the error.

The OpenVPN speed is not what I have expected, got only 50 -60 Mbps (DL) on 260 Mbps line.

edit

after apply

echo 35 > /sys/devices/system/cpu/cpufreq/ondemand/up_threshold
echo 10 > /sys/devices/system/cpu/cpufreq/ondemand/sampling_down_factor

the speed increase to 75-85Mbps (DL)

question for you if this build you already include NSS drivers
like mention at Askey RAC2V1K / RT4230W REV6 Support

who know it could get better performance on VPN,

Thank you again

[efsg]

NSS not work

[lmore377]

I tried to get it working but I couldn't get it to compile correctly. I'll take another shot at it soon.

[eganov]

I am doing openssl test, but have strange result where the test with -evp came lower or insignificant increase at some blocks

root@OpenWrt_Askey:/#  openssl speed -evp aes-128-cbc -elapsed
You have chosen to measure elapsed time instead of user CPU time.
Doing aes-128-cbc for 3s on 16 size blocks: 9160604 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 64 size blocks: 3003726 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 256 size blocks: 825778 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 1024 size blocks: 211126 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 8192 size blocks: 26430 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 16384 size blocks: 13267 aes-128-cbc's in 3.01s
OpenSSL 1.1.1g  21 Apr 2020
built on: Tue Jun 30 04:21:08 2020 UTC
options:bn(64,32) rc4(char) des(long) aes(partial) blowfish(ptr)
compiler: arm-openwrt-linux-muslgnueabi-gcc -fPIC -pthread -Wa,--noexecstack 
The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes  16384 bytes
aes-128-cbc      48856.55k    64079.49k    70466.39k    72064.34k    72171.52k    72214.79k
root@OpenWrt_Askey:/# openssl speed aes-128-cbc
Doing aes-128 cbc for 3s on 16 size blocks: 10289168 aes-128 cbc's in 2.99s
Doing aes-128 cbc for 3s on 64 size blocks: 2864147 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 256 size blocks: 743459 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 1024 size blocks: 187694 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 8192 size blocks: 23481 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 16384 size blocks: 11743 aes-128 cbc's in 3.00s
OpenSSL 1.1.1g  21 Apr 2020
built on: Tue Jun 30 04:21:08 2020 UTC
options:bn(64,32) rc4(char) des(long) aes(partial) blowfish(ptr)
compiler: arm-openwrt-linux-muslgnueabi-gcc -fPIC -pthread -Wa,--noexecstack 
The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes  16384 bytes
aes-128 cbc      55059.09k    61101.80k    63441.83k    64066.22k    64118.78k    64132.44k
root@OpenWrt_Askey:/#

I compare the test with NanoPi R2S (RK3328) from

NanoPI R2S is a great OpenWrt device

Over the past few weeks I have been fortunate enough to do some tests with early OpenWRT builds on the NanoPI R2S provided to me by @jayanta525. This is a dual Gigabit Ethernet board with a 4 core Rockchip SOC (RK3328) on it, manufactured by FriendlyARM. Performance of the board is great, especially considering its price point (€32 incl shipping to Europe in my case). In summary (more details below): AES h/w accelerated openssl, up to 12x openssl performance compared to RPi4 WAN-to-LAN/ingre…

Where the openssl test with -evp the performance result is way higher than w/o -evp

I also tested with atomic pi ( Intel Atom x5- Z8350), result below

BusyBox v1.31.1 () built-in shell (ash)

  _______                     ________        __
 |       |.-----.-----.-----.|  |  |  |.----.|  |_
 |   -   ||  _  |  -__|     ||  |  |  ||   _||   _|
 |_______||   __|_____|__|__||________||__|  |____|
          |__| W I R E L E S S   F R E E D O M
 -----------------------------------------------------
 Atomic-Pi SNAPSHOT, r13245-78bebe680f
 -----------------------------------------------------
root@OpenWrt:~# openssl speed -evp aes-128-cbc -elapsed
You have chosen to measure elapsed time instead of user CPU time.
Doing aes-128-cbc for 3s on 16 size blocks: 21984147 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 64 size blocks: 9518017 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 256 size blocks: 3060657 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 1024 size blocks: 824183 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 8192 size blocks: 105300 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 16384 size blocks: 52729 aes-128-cbc's in 3.00s
OpenSSL 1.1.1g  21 Apr 2020
built on: Sat May 16 02:37:55 2020 UTC
options:bn(64,64) rc4(16x,int) des(int) aes(partial) blowfish(ptr)
compiler: x86_64-openwrt-linux-musl-gcc -fPIC -pthread -m64 -Wa,--noexecstack 
The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes  16384 bytes
aes-128-cbc     117248.78k   203051.03k   261176.06k   281321.13k   287539.20k   287970.65k
root@OpenWrt:~# openssl speed aes-128-cbc
Doing aes-128 cbc for 3s on 16 size blocks: 8842149 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 64 size blocks: 2464777 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 256 size blocks: 638758 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 1024 size blocks: 161500 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 8192 size blocks: 20257 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 16384 size blocks: 10112 aes-128 cbc's in 3.00s
OpenSSL 1.1.1g  21 Apr 2020
built on: Sat May 16 02:37:55 2020 UTC
options:bn(64,64) rc4(16x,int) des(int) aes(partial) blowfish(ptr)
compiler: x86_64-openwrt-linux-musl-gcc -fPIC -pthread -m64 -Wa,--noexecstack
The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes  16384 bytes
aes-128 cbc      47158.13k    52581.91k    54507.35k    55125.33k    55315.11k   55225.00k
root@OpenWrt:~#

Surprisingly the test result w/o evp (hardware crypto) IPQ8065 is the higher than Intel Z8350 or RK3328

Just wondering if the kernel if QCE driver is included (?) it could be better outcome,
found from google but don't have idea whether it is related.

https://patchwork.ozlabs.org/project/openwrt/patch/20190920204804.10183-1-cotequeiroz@gmail.com/

and I am expecting that NSS driver also will have an impact on boosting performance for openssl

[tmomas]

Please use the "Preformatted text </>" button for logs, scripts, configs and general console output.

Please edit your post accordingly. Thank you!

[quarky]

@eganov

Just wondering if the kernel if QCE driver is included (?) it could be better outcome,
found from google but don't have idea whether it is related.

ipq806x does not have the QCE crypto engine. It has the NSS crypto engine.

@eganov

and I am expecting that NSS driver also will have an impact on boosting performance for openssl

It performed a lot worst. You can refer to the results I posted here:

IPQ806x NSS Drivers

@Ansuel, I got the multi segment patch done and did some benchmark. As expected, performance is worst compared to software. Small payload performance is very much worst compared to larger payloads. The only consolation would be that the Krait CPU utilisation is low when ciphering is performed. Results as shown below: With NSS crypto engine: cmd: openssl speed -multi <#> -elapsed -evp aes-128-cbc type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 16384…

Could be the driver not optimised, but at the moment, I have no clue how to improve the performance further. Hopefully I can find a solution in future.

[eganov]

Ouch, Fixed it , sorry

[eganov]

@IPQ806x NSS Drivers For Developers

I got the multi segment patch done and did some benchmark.

Thanks, can you explain how to apply multi-segment it seems with a software-based crypto engine, your result better than mine

I am switching to Wireguard now, heard that hardware Crypto is not necessary

[quarky]

The multi-segment approach in my post is meant for the NSS crypto drivers. It doesn't apply to OpenSSL per se.

For the better performance from my results, it's most likely due to the OPTIMIZE_FOR_SPEED option in my OpenWrt build. You can turn it on using 'make menuconfig'.

I think this script also helps:

quarkysg/openwrt/blob/master-ipq806x-nss-qsdk11/target/linux/ipq806x/base-files/etc/init.d/powerctl

Of course you can force the Krait CPU to it's max clock speed to get the best results.

As for WireGuard, yeah, I switched all my site-to-site VPN to wireguard now. It allows for better latency and thruput as it resides in kernel, so the performance will be much better compared to OpenVPN. Too bad WireGuard only support L3 tunnels. I still rely on OpenVPN when I need a L2 tunnel.

HTH.

[frustro]

I was just delivered this device and it is a miserable castrated thing. I just finished reading this thread beginning to end. Amazing work.

[efsg]

@lmore377 Can I repost your guide?

[eganov]

can you share here your Router info (Factory ID) close to USB and AC adapter port (picture if possible) . I am learning that several version of this router 256MiB / 512MiB Nand
related to Factory ID
Thanks

[frustro]

[frustro]

Strange, the Box says RAC2V1K and the status page says SAC2V1K

Refurb?

[eganov]

You picture kind of blurry There is wording Factory ID just right on top USB , if I am guessing you have ID:3
I have same thing but by original firmware RAC you might have the same unit with @lmore377
if you need to flash here the link for saving your time.to set the partition

Askey RAC2V1K / RT4230W REV6 Support

With my image you need to run setenv mtdids nand0=nand0 && set mtdparts mtdparts=nand0:0x1A000000@0x2400000(firmware) && ubi part firmware && ubi read 0x44000000 kernel 0x6e0000 && bootm saveenv after flashing then reboot That image doesn't have luci or anything extra installed. I'm rebuilding a new one with some extras

[eganov]

@lmore377 if you have spare time, can we try to include this?

IPQ806x NSS Drivers

My branch have different package You need to select nss drv nss gmac nss ecm For basic accelleration

someone has a better result there

IPQ806x NSS Drivers

That is much easier! wifi seems to be acting up so I'll have to troubleshoot that (using default drivers...I must of messed something up - 5 ghz not working, 2.4ghz not connecting with devices). Running your NSS build on the router (wired only) and a hynman build on a r7800 AP I’m getting 20-30mbps better on the wifi download on average. [image] I loaded those three packages and got great results via wired (I've never seen numbers this high before, appreciate the work!): [Screen Shot 2020…

Thanks

[lmore377]

@efsg sure, just credit my GitHub or something

@frustro don't flash openwrt onto a device being rented from an ISP. There's no way to go back to stock on these routers because trying to restore the stock firmware just breaks /overlay which makes it enter a boot loop. Also the routers that have sac2v1k are identical to rac2v1k except for the fact that they have a smaller flash since the sac routers have simpler firmware(mainly no web interface for changing settings and a lot less packages)

@eganov I've been busy with stuff but I'll look into it soon