a lot of people are wrong
mt7622rfb1 is wrong
the exact files is the file the write E8450ubi ....
2 Likes
Seems I'm stuck in a recovery loop even if I flash the old working
openwrt-22.03-snapshot-r19338-ae64d0624c-mediatek-mt7622-linksys_e8450-ubi-squashfs-sysupgrade.itb
from
I assume there's a crashlog or something keeping me in recovery from the wrong flash... checking now
Update:
yep, /sys/fs/pstore had a oops file
1 Like
i has a same problem if i remember before... i has search while 1hours but is done finaly . try after install recovery to install the snaphot of dango
then the rc2 but not sure
1 Like
Removing the
/sys/fs/pstore/*
oops files stopped the endless recovery loop (this is by design)
I just flashed rc2 cleanly with config unchecked now and all is well. Thanks for putting up with my mistake
1 Like
Hi. I just got this router and installed OpenWrt on it. I've paid for around 250 Mbps download and upload speed with my ISP. Anyone got any suggestions for the sqm settings that i need to use? And should I enable software offloading?
Also I've got irqbalance installed, is that good for sqm? and do I use irqbalance or packet steering?
I'd like to help to fix this but I am a bit of a newbie when it comes to locating the dts files for a device (although I've stumbled upon them for my DIR-878 in the past).
I am, however, and electronic engineer and from the FCC photographs of the Belkin cross-referenced with the BananaPi R64's schematic and MT7622A datasheet, it would be fairly trivial to measure the CPU core rail provided by the MT6380N PMIC as most/all of the circuit around the MT6380N follows the BananaPi. It looks as though The MT6380N should have it's output voltage controlled by I2C commands issued by the Belkin/Linksys' MT7622B and I can identify the MT6380N CPU core output capacitors from the FCC photos of the Belkin/Linksys PCB.
What I can't find is a definition of voltage versus frequency. I'm currently in too much of a hurry to leaf through the datasheet, but is there a seperate microcontroller in the 7622 which controls the PMIC? I'd guess not if the DTS file has anything to do with it.
My RT3200 has just arrived on my doorstep so I'm not going to open it just yet... At least not today!
Please could you show me the patches and evidence of the voltage and frequency specifications for the MT7622A or ideally the MT7622B if possible?
1 Like
Well, when the cpu has not enough voltage it will crash, and you will notice it 
It works for me since about 1 year. I've created new patches for the 5.15 kernel.
It seems i am not allowed to attach files, so get the file here: https://easyupload.io/ruhnbi
MD5: 601908983b77703b812b56490c44e5b6
Unpack the tar, and then apply the 2 patches into the main directory of your checkout. The one needs p1 the other p0.
I recommend to install statistics package to see temperature and mhz changes
4 Likes
Thank you. I will try to understand how to use this patch in due course, but right now it's a bit above my skill level! Sorry.
Does anyone know if release v0.6.5 definately works with latest Belkin signed firmware 1.2.x ?
Just to be more confident I'm not likely to encounter an early life failure of the hardware, I am going to test stock firmware for a few days before taking the plunge and installing OpenWRT but am a bit worried that if I upgrade from 1.0.01.101415 to 1.2.00.273012 I might have OpenWRT installation issues.
I'm guessing that as 1.2.x is so new, few have tried. Anyone?
@quarky @Ansuel
Today I've contacted Belkin support and they couldn't help much but escalated the issue and I'm waiting for a call to explain the details to the tech support.
Do you have any suggestions what specifically I can told or ask them that may help to resolve this performance issue.
1 Like
Any reason you're not using 'schedutil' for the gov?
Also, here is the DTS patch inside your tar file:
8450-515-add125_use1V.patch:
--- /dev/null
+++ target/linux/mediatek/patches-5.15/722-dts-mt7622-add125-use1V.patch
@@ -0,0 +1,19 @@
+--- a/arch/arm64/boot/dts/mediatek/mt7622.dtsi
++++ b/arch/arm64/boot/dts/mediatek/mt7622.dtsi
+@@ -23,9 +23,15 @@
+ cpu_opp_table: opp-table {
+ compatible = "operating-points-v2";
+ opp-shared;
++
++ opp-125000000 {
++ opp-hz = /bits/ 64 <125000000>;
++ opp-microvolt = <1000000>;
++ };
++
+ opp-300000000 {
+ opp-hz = /bits/ 64 <300000000>;
+- opp-microvolt = <950000>;
++ opp-microvolt = <1000000>;
+ };
+
+ opp-437500000 {
[removed my misread]
How would you improve the DTS overall if they are messy? I only ask because I'm curious and want to learn.
Here is your second file:
8450-515-governor_ondemand.patch:
--- target/linux/mediatek/mt7622/config-5.15
+++ target/linux/mediatek/mt7622/config-5.15
@@ -94,8 +94,8 @@ CONFIG_COMPAT_OLD_SIGACTION=y
CONFIG_CONSOLE_LOGLEVEL_DEFAULT=15
# CONFIG_CPUFREQ_DT is not set
CONFIG_CPU_FREQ=y
+CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y
# CONFIG_CPU_FREQ_DEFAULT_GOV_PERFORMANCE is not set
-CONFIG_CPU_FREQ_DEFAULT_GOV_USERSPACE=y
CONFIG_CPU_FREQ_GOV_ATTR_SET=y
CONFIG_CPU_FREQ_GOV_COMMON=y
CONFIG_CPU_FREQ_GOV_CONSERVATIVE=y
Hosted files tend to disappear over time, and this saves people a download to look. 
Because ondemand works well °-°
Pasted .patch files tend to no fit as spaces and tabs are not correct.
So blame the forum-mods to accept only pictures and not txt/patche
Read this thead why voltage below 1.0v does not work with ubi-uboot (the cpu does)
Is this just adding a 1,25ghz
No, please count the "0"s.
This is the same counting failure exists in the original dts which is yet fixed in openwrt
Whoops - ok, just trying to help out.
I know there are some PRs out there somewhere to fix the "30"mhz error and to modify the thermal points that trigger early. I think there was some debate over the right way to go about that.
[removed bad info]
With git is only 1,25ghz is used, so there is no reason to remove.
And with this patch 300 and lower are working
remove the 300mhz because of the u-boot reboot issue.
With openwrt problem resolving is funny
1 Like
I'm not sure if this ever got applied. It has many duplicates, for the 300mhz fix:
I see it was a voltage issue for 300mhz - my bad - you did fix it.
Looks like this is about to be merged, which fixes the thermal trigger:
I keep putting my foot in my own mouth - thanks for being patient.
This is the same discussion since 1 year... From https://github.com/openwrt/openwrt/pull/4440 only 1 commit was merged. The other 3 are in my "attached" file above, changed for new kernel version. Merged to simple use local
1 Like
I think you pretty much have a handle on the issue at hand. I'll be interested to know what Belkin finds, tho. I think their firmware images most probably are using a very old Linux kernel. But it looks like what is causing the issue for the old kernel still exists in the newer kernels.
1 Like
I realised that my 5 ghz wifi range is very poor. Its enough but the signal keeps dropping and increasing back, anyway to fix this bug?
I doubt this is very helpful since you've probably already seen it, but just in case, on Daniels git it mentions 6.5 supports stock firmware v1.1 (not 1.2), and 6.5 was released after stock v1.2, so probably not worth the try.
edit - (deleting that edit since Daniels git mentions signed stock 1.1 and not 1.2)
1 Like
Thanks for looking into this mike. Good spot on noticing that 1.2.00.273012 and 1.1.01.272918 were released on the same day, too.
To quote Daniel "now supports signed stock firmware >=v1.1". I interpreted this as a common shorthand/pseudocode syntax meaning greater than or equal to 1.1, hence my question about whether anyone has tested 1.2 as the implication of that statement is that it's compatible. So I am afraid I'm still looking for the answer to that question!
1 Like