Hasivo switches

It looks like whatever uboot they gave you was rubbish.
It seems that it is 'valid' Uboot, but without any of the command line operations compiled in.

The only real solution from here is probably to use the SOIC clip and read out the firmware that you currently have, in the hopes that the 'EEPROM partition' (where MAC address etc is stored) is still ok.
Then find a proper flash image, and use the SOIC clip and flash writer to overwrite the Uboot partition.
That should give you an ok Uboot again. You could then check on what the linux partition is like...

Overall it would be pretty painful.

@olliver Hey, sorry to mention you on this thread, but you mentioned above that swapping out uboot is pretty trivial.
I've got a CH341a + SOIC clip ready n raring to go.
What's involved in flashing a new uboot onto a S600WP-5GT-2S_SE? I'm not sure if it's just a case of getting a copy of uboot that's g2g with a rtk9303 or if it's more involved.

I'm pretty sure that the only issue is a crappy linux partition write. I'll read off what's on it first.
Hasivo got back to me and suggested it's a bad flash chip, which wouldn't surprise me. Might end up just RMA-ing it.
Was really hoping I could use this as a learning experience about these things, but if the uboot is as garbage as it is...
Anyway, how would I go about determining the partitions/offsets etc once I pull an image? binwalk?

Binwalk is worth a try. But often the partitions aren't really partitions. But are just memory offsets (not actually written to any partition table or such), with blobs of data afterwards. i.e. the EEPROM type partitions are just generally a certain offset, and then a C struct dumped to FLASH with certain info, like device type, PHY config, MAC address(es), perhaps some calibration data..

So the best way is to boot a known working firmware, and see what it says that they are. Normally when they get into linux boot it will reveal them, since it will have been 'documented' in the Platform Config / Device Tree.

Whelp.
Using the SOIC clip I was able to pull the flash
Linux part looks like it starts @ 0x00300000.
Going to try and overwrite in the firmware there.

Doing a comparison between the backup firmware and one I hacked together where I overwrote the linux section with the original firmware file I was given, a big chunk ~ 0x04067784 was FF'd so I think 'there's my problem'.
Flashing now, lets see if I can get this dang thing to boot


Success!

3 Likes

Just a comment. At least on my Hasivo switch, the SCHS# prompt is not u-boot, from to the top of my mind this is something like diag application. From SCHS prompt you should type XXXX as a command to go to the actual u-boot prompt and try rtk network on. I guess the help command from the SCHS prompt is hshelp.

2 Likes

I'll have to try that next time! (hopefully, there won't be a next time tho)

Hi, I want to buy the S1100WP-8GT-SE and I understand it comes with 52V power adapter. I want to use a reliable high efficiency (level VI) brand powersupply with it. The issue is in EU the "legal" voltage for adapters tops at 50V so brands like Meanwell sell only 48V adapters.
How likely it is the switch will work with 48V power adapter?

I’m pretty sure EU legislation allows for up to 50V a.c. and 75V d.c. under the ELV categorisation.

However common industrial power supplies are generally 12V, 24V or 48V. Which is why the likes of Meanwell target these.

I doubt 48V will work very well. There is a reason the supplied power supply is 52V and not 48V.. and it is specifically to align with the PoE standards, which generally target >48V at the end device and a few volts of cable loss.

A quick Google shows a few Level VI efficiency 52V supplies…

Some of the Meanwell (and of course, also from other suppliers) industrial power supplies sold as 48V can be adjusted in the range of 41V to 56V.

Thanks, I did not know pulsar. I was looking what mouser sells and it is not 52V power supplies.

I asked on Aliexpress not to ship the adapter and to give me a discount to get below €150,- in order not to go through customs clearance process. (Make the order and do not pay for it, ask seller for adjusting the price for your order ID and pay after it is adjusted) I would recommend doing the same to anyone who lives in an EU country with an expensive customs clearance process.

I did not know that, this could be a way to go.
Apparently the POE+ 802.3at spec requires the switch to have minimum 50V on the port, so it can not be achieved with 48V power supply without internal DC-DC convertor.

Just found this thread, because I was googling for hardware that my usual supplier suggested. I'm already running Openwrt on a bunch of h/w including lots of xiaomis, some d-links and so on and I'm firm with compiling it myself and the casual one-line c patch for dirty stuff. So coming to my ignorant question because I just skipped over the thread: is the hardware basically usable for someone who knows how to build it himself?

It's still very early days for rtl93xx support, way beyond just building from source, you would need to be comfortable to debug- and develop issues as they arise. If you aren't comfortable with low-level hardware- and kernel development, don't count on it to become end user friendly any time soon (months++).

There's a 10g model available: S1100W-8XGT-SE

Here's a teardown: https://www.servethehome.com/finally-a-cheap-8-port-10gbase-t-managed-poe-switch-the-hasivo-s1100wp-8xgt-se/

Main switch is RTL9303. Appears to be two other PHYs: RTL8264

I'm not up to speed on these kinds of devices, no Realtek development. Think this would also be on the roadmap for support?

Another lite version of the above switch is available for even less, this one lists the storage and CPU details - S1108XGTP-SL

1 Like

Found these even cheaper on Alibaba, but you have to buy 2. I bought one of each to get OpenWrt running.

Please let me know how I can contribute.

Here’s the S1100W-8XGT-SE internals.




The PoE version arrived too, but I figure it'd be easier to get non-PoE working first.

1 Like