If you can point me where this new site is? I have asked the support but the did not provide me with the firmware nor a pointer to the firmware.
networkingsupport.hpe.com but if you ask here someone may have a copy. The site is a user-hostile disaster.
That's a search query for the officeconnect 1920 series firmware. But during my attempt to login I was asked to reset my password, then asked to reset again as I had apparently used that random 20 character string before (I hadn't) then I got an error trying to request a password reset and finally a prompt to setup 2FA, in what felt less like a login and more like an admittance of defeat from the corporate overlord, before I was able to search for firmware. Utter disgrace.
I really want to see the User Journey diagram for this one.
1 Like
FYI, the buildbot got restarted this week and all new targets have snapshots now at https://downloads.openwrt.org/snapshots/targets/realtek/
Happy testing
2 Likes
Unfortunately, rtl83xx still does not build (presumably because of realtek-poe not being released for the new CMake version yet).
it does build now \o/
after https://github.com/Hurricos/realtek-poe/pull/65 and https://github.com/openwrt/packages/pull/27889 have been merged. Kudos to Hurricos and hauke
4 Likes
Is the current snapshot for rtl930x (specifically on the XikeStor SKS8300-8X) stable?
I'm running a recent main build on my XGS1210-12, and it's the backbone of my network basically. No complaints.
1 Like
Been running various versions of main on the identical Onti for about 6 months - stable as core switch in my LAN.
Twice had weird upgrade failures, who knows why but flashing from scratch worked both times.
1 Like
Since my ZYOPM 10G 80M doesn't work due to awful quirk, and doesn't work with my CAT6 cables (mysteriously, CAT5E cables work), I purchased another Xicom 10G 80M.
I asked the seller about the chip, they replied "Broadcom" first, then "Realtek" next, and finally "Marvell 3510" 
There is almost zero information about "Marvell 3510". The only thing I could find was a datasheet in this page labelled "Marvell CUX3510". I don't know if the new Xicom module really uses "Marvell 3510", I don't know this datasheet is correct.
What I can see is, both CAT5E and CAT6 cables work, multi-rate(10G/5G/2.5G/1G/100M/10M) works, slightly hotter than ZYOPM 10G 80M, but looks almost good to me.
root@OpenWrt:~# ethtool -m lan8
Identifier : 0x03 (SFP)
Extended identifier : 0x04 (GBIC/SFP defined by 2-wire interface ID)
Connector : 0x07 (LC)
Transceiver codes : 0x10 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
Transceiver type : 10G Ethernet: 10G Base-SR
Encoding : 0x06 (64B/66B)
BR Nominal : 10300MBd
Rate identifier : 0x00 (unspecified)
Length (SMF) : 0km
Length (OM2) : 80m
Length (OM1) : 20m
Length (Copper or Active cable) : 0m
Length (OM3) : 300m
Laser wavelength : 850nm
Vendor name : OEM
Vendor OUI : 00:1b:21
Vendor PN : SFP+-T80
Vendor rev : 1
Option values : 0x00 0x1a
Option : TX_DISABLE implemented
BR margin max : 0%
BR margin min : 0%
Vendor SN : C20251103****
Date code : 251105
Optical diagnostics support : Yes
Laser bias current : 6.000 mA
Laser output power : 0.5000 mW / -3.01 dBm
Receiver signal average optical power : 0.4000 mW / -3.98 dBm
Module temperature : 65.19 degrees C / 149.34 degrees F
Module voltage : 3.3400 V
Alarm/warning flags implemented : Yes
Laser bias current high alarm : Off
Laser bias current low alarm : Off
Laser bias current high warning : Off
Laser bias current low warning : Off
Laser output power high alarm : Off
Laser output power low alarm : Off
Laser output power high warning : Off
Laser output power low warning : Off
Module temperature high alarm : Off
Module temperature low alarm : Off
Module temperature high warning : Off
Module temperature low warning : Off
Module voltage high alarm : Off
Module voltage low alarm : Off
Module voltage high warning : Off
Module voltage low warning : Off
Laser rx power high alarm : Off
Laser rx power low alarm : Off
Laser rx power high warning : Off
Laser rx power low warning : Off
Laser bias current high alarm threshold : 15.000 mA
Laser bias current low alarm threshold : 1.000 mA
Laser bias current high warning threshold : 13.000 mA
Laser bias current low warning threshold : 2.000 mA
Laser output power high alarm threshold : 1.9952 mW / 3.00 dBm
Laser output power low alarm threshold : 0.1584 mW / -8.00 dBm
Laser output power high warning threshold : 1.5848 mW / 2.00 dBm
Laser output power low warning threshold : 0.1778 mW / -7.50 dBm
Module temperature high alarm threshold : 95.00 degrees C / 203.00 degrees F
Module temperature low alarm threshold : -50.00 degrees C / -58.00 degrees F
Module temperature high warning threshold : 90.00 degrees C / 194.00 degrees F
Module temperature low warning threshold : -45.00 degrees C / -49.00 degrees F
Module voltage high alarm threshold : 3.6000 V
Module voltage low alarm threshold : 3.0000 V
Module voltage high warning threshold : 3.5000 V
Module voltage low warning threshold : 3.1000 V
Laser rx power high alarm threshold : 1.1220 mW / 0.50 dBm
Laser rx power low alarm threshold : 0.0199 mW / -17.01 dBm
Laser rx power high warning threshold : 1.0000 mW / 0.00 dBm
Laser rx power low warning threshold : 0.0223 mW / -16.52 dBm
As long as you stick to 10 Gbps, I donāt think you need a PHY driver at all, at least in some cases. 
Thatās at least my experience with FlexOptix 100 m, ZYOPM 100 m and Xicom 80 m Broadcom based 10GBASE-T transceivers in the BananaPi BPI-R4 running OpenWrt. It claims to not find a PHY in the logs, but it works perfectly fine at 10 Gbps.
Well, what I said is that my new Marvell 3510 based Xicom 80M is working well at 10Gbps and other speeds(*) without any hack/quirk. So I will throw away troublesome ZYOPM 80M
(*) To be precise, I can achieve 10 Gbps in both directions when running iperf3 --bidir. However, at link speeds lower than 10 Gbps, I only get the full speed in one direction. I believe this behavior was also observed with other multi-rate RJ45 modules. It also works well at 5G/1G/100M/10Mbps in both directions.
1 Like
Yes, the 2.5G performance was poor, seems like it just pauses the 10G interface to reach 2.5G.
I mainly meant an USB device with SFP+ slots, yes, but Iād settle for a really low power/temperature RJ45 transceiver as well. My 10G switches are passively cooled and thus donāt like existing RJ45 transceivers.
I ordered two Xikestor SKN-U310GT from AliExpress (though the are now on Amazon.de as well, albeit almost twice as expensive as on AE) and so far they seem to work fine at the same level as their earlier 5Gb counterparts. One of the two computers does not have a 3.2/Gen2x2 port, so something like 6 Gb/s bidirectional is the maximum achievable (though I wonder how even that works over a 10 Gb USB port), but the dongles stay quite cool even under load, much cooler than the 5Gb dongles.
Iām still on the May 2025 Linux driver, so stability is not perfect (but it was not for the 5Gb dongles either), but Iāve heard good things of the new version released a few days ago.
Good to hear that the SKN-U310GT works and remains reasonably cool, mine is still in China
FWIW, the 80m Broadcom phy works in my passively cooled Onti switch (I do have a low speed fan blowing across it, though) but I would not want to use 8 of them :-). Still waiting for the RTL phy, in any case.
There appears to be an (intel based) dual SFP+ USB adapter on AE: https://de.aliexpress.com/item/1005009094867211.html no idea if its any good.
1 Like
Yeah, I know (and I am currently waiting for one), but those are not really USB adapters, but Thunderbolt 3 using PCIe tunneling. What Iāve read online, they do get hot because the chipset is pretty old and power-hungry (but thatās not a problem for the switch of course).
Considering that the RTL8127 apparently can do SFP+, we may yet get RTL8159 with SFP+...
Personally, I have no real need for it but likely it would be doable.
That section of the wiki is depreciated. I'm running current stable OpenWRT on this device and the LEDs function as expected with an empty rc.local
When I tried it about a month ago on 24.10.4 it did not work for me without the rc.local fix. Maybe you have the U-boot bootcmd alternative in place?
It has now arrived and I can confidently say it is not good. It gets seriously hot even with just single RJ45 transceiver (albeit a bad one for such an enclosure, I would not want to use that one for anything serious). However, the real problem is the weird way it splits the PCIe lanes tunneled via thunderbolt. It apparently connects as gen2 (5 GT/s) x 4 and then splits that into gen1 x 4 for each of the two SFP+ slots, for a total of 8 Gb/s for each of the two NICs 
.
Actual throughput with iperf is better than a 5 Gb Realtek USB dongle, but not by that much:
[ ID] Interval Transfer Bandwidth
[ 2] 0.00-8.63 sec 5.29 GBytes 5.26 Gbits/sec
[ 1] 0.00-10.00 sec 8.07 GBytes 6.93 Gbits/sec
[SUM] 0.00-10.00 sec 13.4 GBytes 11.5 Gbits/sec
compared to
[ ID] Interval Transfer Bandwidth
[ 2] 0.00-10.00 sec 3.27 GBytes 2.81 Gbits/sec
[ 1] 0.00-10.02 sec 5.31 GBytes 4.55 Gbits/sec
[SUM] 0.00-10.02 sec 8.58 GBytes 7.36 Gbits/sec
Not sure how they actually wired it looking at the specs of the TB3 module, but getting the dual SFP+ device (instead of the single slot version) looks like a mistake in hindsight.
2 Likes
There is a pending PR from @jonasj for device owners of the Engenius EWS2910P
Please test and report back in the PR https://github.com/openwrt/openwrt/pull/20687