@svanheule I'm seeing error messages about the initramfs being too big for my GS1900 builds now, is the 6976k cap still there out of caution or are some ZyXEL u-boot versions known to break with bigger ramdisks?
WARNING: Image file openwrt-realtek-rtl838x-zyxel_gs1900-10hp-initramfs-kernel.bin is too big: 7971711 > 7143424
WARNING: Image file openwrt-realtek-rtl838x-zyxel_gs1900-8hp-v1-initramfs-kernel.bin is too big: 7825735 > 7143424
WARNING: Image file openwrt-realtek-rtl838x-zyxel_gs1900-8hp-v2-initramfs-kernel.bin is too big: 7825736 > 7143424
The cap is there because the initramfs is also the factory install image.
If we could make a size-limited initramfs-based factory image, that would also work. Then the normal (without the Zyxel device info) initramfs maybe wouldn't have to be size limited anymore.
Okay. Further inspection shows the 24.10 pre-release snapshots are lacking the same initramfs images (8HP v1 / 8HP v2 / 10HP). The non PoE GS1900-8 and -16 e.g. do have them, weirdly enough.
Nothing weird about it, the size is right on the cusp - and since intramfs images are fixed to evaluate DEVICE_PACKAGES, including PoE support can push them over the limit.
Maybe we could cull the default packages to only those that are really necessary? I think we can drop tc-bpf (+ libbpf) and ip-full and still cover the basic needs in a fresh install. See target/linux/realtek/Makefile.
I have managed to add support for Sodola SL-SWTGW2C8F (8x10gbe SFP+) which is based on RTL9303. I have also added support for jumbo frames (max mtu 10000) for this chipset.
I have only been able to test it with Sfp+ 10G optics (no DAC cables) and able to achieve near wire speed. We still need u-boot to initialize the network (rtk network on), however there is no access to serial console on this switch. However, there is an unpopulated max23x header that I was able to tap into to get serial console working
Hi all,
I've recently purchased a Netgear GS310TP switch and flashed it with Openwrt (both 23.05.5 and snapshot build); everything is working as on stock firmware minus a couple details:
I was able to make the port LEDs work but for some reason the Power status LED stays off.
POE seems to be working fine, but when I try to query POE status with ubus call poe info all ports are showing status 'unknown'.
I've got a few SFP modules laying around so I wanted to test them with both OpenWrt and stock firmware. I was able to make it work on both with some SFP fiber modules, but for some reason I am unable to make it work with my 2 SFP copper modules on Openwrt. Those same 2 modules are working properly when I switch to stock firmware.
Any idea on the above ? I'm willing to help in fixing this but not too sure where to start.
Thanks !
Are there any Realtek based switches known, that have components on the underside?
What would you expect to be "available" on the UART header in the suspected "no-memory-onboard" case? A really bare-bones cli configuration tool, or "just' some telemetry/stats? A no-function vestigial remnant from a more advanced product series?
Yeah, my assumption has been that these super-cheap, unmanaged devices are built with an ASIC instead of a CPU/SOC, with just enough KBs of on-chip RAM to hold a FIB.
Hi!
I've just bought a similiar device with 4x2.5GBE and 2x10GBE SFP+. It attached a serial console and got this:
==========Loader start V0.2===========
Press any key to start the normal procedure.
To run SPI flash viewer, press [v]
To enforce the download of the runtime kernel, press [ESC] .....
cmd -1
Check Runtime Image.....
Chksum Correct!
RunTime Kernel Starting....
Ver8372N=2
Ver8373N=2
LINE 1462, RL6818C_pwr_on_patch_phy_v007 , patch 0xf0 finished!
So looks like RTL8372, which is not supported by openwrt ATM.
