Hi Original software has no fw_setenv or fw_printenv so only overwriting parts of nand is an option to set or reset env. WD's software do not use nand for storing setting - configs are stored on one of drives partitions and can be reset to defaults with reset button pressed for some time. One thing i can do more to confirm this, is to resetenv and check if firmware can set MAC and wriby osa - Debian
So, i verified reset button and i think i was wrong. It does not clear env when booting with button pressed (maybe sets other boot order?? can't confirm) But resetenv does and fills last 64K of nand with zeros (and clearing checksum) See log below - just after resetenv I use fox-exe kernel and initrd because it fits into 0xa00000/0xf00000 adresses BootROM - 1.51 Booting from SPby osa - Debian
It seems boot order, commands and adressess are hardcoded in uboot, but uses bootargs env If crc is corrupted default bootargs is used If i set manually root partition with bootargs (using mkenvimage) I can boot debian from HDD!!! Funny thing is that kernel overlaps initrd and it is disabled (like previously posted in other thread) and it allows kernel to boot I can't remember whatby osa - Debian
Hi CRC is an issue, because when if CRC corruption is detected u-boot uses default values. NAND env is cleared when board is started with button pressed. But i made some progress again ;-) We can use mkenvimage from u-boot to create valid but empty env and then overwrite last sector of NAND and ... now fw_setenv allows us to set env variables! Downside of this is no MAC set (and seriby osa - Debian
Hi all! We have working new box - WD My Cloud (Gen2) But there's also big issue with u-boot. There's Default CRC is messed and default values are always used, saveenv is not availiable. Maybe i'ts hard coded or maybe just bugged. For example reset button clears env so mac address is cleared causing network not usable. I think there are some ways to fix this without WD'sby osa - Debian
5.3.5-mvebu-tld-1 works for me The only issue i see is gigabit led (in the socket) but i can live with this ;-) DTS is from here with phy-mode changed and led triggers set There is another file but looks very similar (not tested) /* * Device Tree file for WD MyCloud based on Marvell Armada 375 * (DB-88F6720) * * Copyright (C) 2014 Marvell * * Gregory CLEMENT <gregory.clby osa - Debian
All DTBs are binary identical. 5.2 and 5.3 were not working for me either (i think they returned PHY error 22) I tried 5.5.0-rc7 with no luck, but now ethernet port led is turned off and no status change on cable plug/unplug (maybe this patch? https://github.com/torvalds/linux/commit/f3f2364ea14d1cf6bf966542f31eadcf178f1577#diff-38b28d4e5f70b4f5e7ec9b5e787a0e13) root@wdmc:~# ifdown eth0by osa - Debian
DTS were compiled (4.18 and 4.19) and are binary identical as fox-exe I will check "workaround" from bugzilla to confirm if this is related to mvpp2 driver change (RSS/multiqueue) and try other DTSesby osa - Debian
I tried vanilla kernel 4.18 and network works but gigabit (led blinks, no transmission). On 4.19.1 link is detected but no transmission at all (no matter if 100 or 1000 bit/s, but not throwing "could not attach PHY" error). MyCloud has only one network port, so multiqueue can be misconfigured somehow Are you sure your Mirabox is Armada 375 based, not 370?by osa - Debian
Hi I'm testing WD My Cloud Gen2. So far i tried kernels and dts from fox-exe - 4.15.0-rc6 from fox-exe - runs fine, but network crashed after 1 day (Edit: gigabit led not working, detects 1Gbps/Full detected when connecting cable, but not sending data) - 4.19 from debian buster but network card is not working ( mvpp2 f10f0000.ethernet eth0: could not attach PHY (-22)) - 5.2.9 Bohdi'by osa - Debian
Hi, do you still have this dump? Can you put it somewhere in the cloud and provide link?by osa - uBoot
@Bohdi Here's my original 3.2 kernel patch @@ -2777,6 +2777,23 @@ { struct phy_device *phy = mp->phy; + int reg; + + /*Leds link and activity */ + phy_write(mp->phy, 22, 0x3); + reg = phy_read(mp->phy, 16); + reg &=~0xf; + reg |=0xf; + phy_write(mp->phy, 16, reg); + phy_write(mp->phy, 22, 0x0); +by osa - Debian
Quotentpd is sync to the rtc, So.. we do not need rtc driver? Proper or just not stupid date/time is needed to get dhcp working (in my case) I think ht1382 driver is working fine now and we could trust it a little more. If there's no net connection - there's no ntpd working, right? @JohnnyUSA: please fix your rtc clock on 4.2 and we'll se if it was the reasonby osa - Debian
ht1382 was patched recently. Please set new date, set persistent with hwclock and. AFAIR i had some troubles with old rtc driver becouse it was setting year 1930 and dhcp was not able to renew.by osa - Debian
JohnnyUSA Wrote: ------------------------------------------------------- [ 12.478384] rtc rtc0: __rtc_set_alarm: > err=-22 What does hwclock write about stored time? RTC driver was patched and set to 24h mode by defauld Previously i could get year 1931 set and it caused dhcp problems for me Maybe it issue is related? Have you ever set alarm with zyxel software? Our driver do not supportby osa - Debian
I have a 320 box with 2 4tb drives spinning slowly and fixed it by reseting ide from uboot (my kernel/initrd is in nand)by osa - Debian
bobafetthotmail Wrote: ------------------------------------------------------- > Does that erase stock firmware? We have 2 copies of kernel/rootfs so it detects which was used to boot and overwrite second copy This is limitation of OMV. Did somebody filled such bug or feature request?by osa - Debian
Sorry It should be regs.hour = bin2bcd(tm->tm_hour) | HT1382_HOUR_1224; I copied it from my patch. I called it HT1328_24H_BIT because this bit sets 24h mode but HT1382_HOUR_1224 is commented clearly. Also, please test on your box ( it will always use 24h mode for simplicity and compatibility with zyxel's driver)by osa - Debian
I skipped GPT limit by putting kernel and initram into NAND if disk is > 2TB I remember i was able to create lvm by hand and was able to use it in OMVby osa - Debian
Hi bobafetthotmail Please review my scripts (based on arch's) It supports GPT, raids and nand installer https://www.dropbox.com/sh/w5zpp0utb88a1jt/AACkxs_PqH2IZdzvW4MSxt83a?dl=0&s=soby osa - Debian
@Bohdi: do you have source code for 310s/320/320s/325/325v2 ? I think they all have fancy phy_init like this #if defined(CONFIG_ZYXEL_NSA325) /* Set LED[2] pin is tristate; for communicating with MCU */ mvEthPhyRegWrite(mvBoardPhyAddrGet(ethPortNum),22,0x3); mvEthPhyRegRead(mvBoardPhyAddrGet(ethPortNum),17,®); reg &= ~(1 << 4); regby osa - Debian
I think Bohdi should write tutorial or something There's my way: - set up dev enviroment or cross compilation - download kernel from kernel.org (ie 4.1.3) - extract kernel - apply bohdi's patch - apply my patch - compile using make-kpkg and fakeroot so i finally get linux-image-4.1.3-tld-1_1.0_armel.deb You can try my build https://www.dropbox.com/s/n6n6b0tgwfei4yv/linux-imaby osa - Debian
Guys, i think i have working patch for 4.1. I'll post it here for testing [ 22.152970] NET: Registered protocol family 10 [ 22.414745] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready [....] Configuring network interfaces...Internet Systems Consortium DHCP Client 4.3.1 Copyright 2004-2014 Internet Systems Consortium. All rights reserved. For info, please visit https://www.isc.oby osa - Debian
> Quoteosa > I have alse "link not ready" fixed with this patch > > > Can I somehow try this with phypeek and phypoke? > Tried with > > phypoke 1 22 3 > > Gives error "illegal address"... Hi! 22 = 0x16 so try phypoke 1 16 3 I have tried on my box phypoke 1 16 3 phypoke 1 16 0 My network works after restart ( patch hacks regby osa - Debian