Please add support for TOZED ZLT S10

or help me how to build one myself, i am willing to learn

link to product page seems to be broken fffff

probably beause of the extra char you added at the end of the URL.

https://openwrt.org/docs/guide-developer/add.new.device
https://openwrt.org/docs/guide-developer/adding_new_device

chipset ZX297520 doesn't look you'll get too far unfortunately...

@maurer I think this is the second post I saw on this forum that plainly states it like this. Why do you give the platform so little chance for success?

The hardware looks so promising, I would like this to happen. It is running Linux already, so there is the first shimmer of hope.

I got my hands on one of these by chance and I have been doing some research over the past few days.

Tozed ZLT S10 is actually the "real" name of the device. I found a few rebrands:

  • Tecno TR210 (Russia and surrounding countries),
  • Yeacomm S10 (America, at least north America)
  • Strong 4GROUTER300M (Europe)
    And it looks like some Iranian cell provider sold a locked version of it to their customers at some point in time.

In a PDF for another (but slightly resembling looking) device I found the information "CPU ARM Cortex A7,1G, SANECHIPS ZX297520 V3". The information that this chipset has a cortex-a7 CPU seems legit.

For the 1.x firmwares there was a hack to enable telnetd. Unfortunately, my device has a 2.x firmware and that hack doesn't work anymore, so I cannot validate all the other information I believe applies to this device.

Openwrt relies on mainline Linux kernel support to be able to support SOCs and platforms

This platform has no mainline support and is therefore not supportable - unless someone upstreams that support. As that is likely to be months of work, with no vendor support, that’s unlikely to happen.

Unless you are planning to do it yourself

For any device, 'someone' will have to do the development work - in most cases the necessary device support patches are provided by a motivated owner of the device (often by semi-regular contributors, but also as kind of a drive-by contribution). OpenWrt as a project doesn't buy a batch of devices and assigns the porting to the developers, everyone buys their own devices for their own needs. They tend to choose their devices based on their requirement, interests and how good the expected success rates are.

For a device based on a well known, supported SOC & wireless, this is a relatively reasonable task, something between a loong rainy weekend and several weeks to get it working.

As jdwl1o1 already implied, for unknown SOCs with no mainline support (probably not even complete GPL sources), you're looking at months++ (double digit) of work by very motivated and experienced developers, which (ideally) includes polishing up the code for mainline. It doesn't exactly help if the SOC in question isn't very common, meaning there'd be little opportunity to reuse the SOC support for other devices (half a dozen+).

No one will stop you or anyone else to work on it and do the necessary development, but chances to find someone else to do it for you are slim to none (it is just much easier to work with well-known SOCs). It is just a heap of work, but if you are willing to pause your other hobbies and most social interactions for the next ~two years to get it supported, by all means, speed on.

There are plenty of devices/ SOCs that will never be supported by OpenWrt, because it's just too much work for little reward (rather exotic, not representative for other devices, system specs not good enough to be worthwhile). One example would be the Huawei AX3, available all around the world, cheap (~30 EUR) and it was one of the first affordable 802.11ax routers (at a time when the competition started around 400 EUR) - but it also had in-house HiSilicon SOC and wireless, which has no mainline support, nor would be used by (m)any other devices. While the former would be ideal starting positions, the later ensured that no one did pick it up. It's just way more work (we're talking about years, if full GPL sources are available in some form - and is pretty much futile in case of proprietary wireless drivers), than getting mt7621/ mt7622/ filogic 8x0 or ipq807x/ ipq60xx to work, even if those only entered the market later and did cost more, but Mediatek and Qualcomm do spend more efforts on mainlining their SOC support (and ipq807x support took around 3 years to cook, ipq60xx support isn't quite these yet (and will easily require a couple more months (and a better wireless firmware from QCA)). The various Triductor or Realtek SOCs are in a similar boat - and Broadcom is a well deserved persona non grata, despite being very common on the market.

There are many ways for those hardware to run Linux, however most chipset manufacturers choose the way that only making non-upstream driver/patch and then call it a day, yes it's still a Linux system but not scalable nor maintainable, if you want to blame someone for "giving the platform so little chance for success", you should blame the manufacturer

2 Likes

I wasn't looking to blame anybody. What I meant was more like: Why do you consider the chance for success so low, if there was anybody willing to try and install openwrt on such a device?

I think I will, just to learn a few things. However, reading slh's reply, I don't think there will be greater results than me learning what you guys are talking about.

this issue (probably) isn't installing it, but getting the SoC supported in mainline Linux.
installation is the last 5m of your 10km journey.

Trying to install is easy, either success or fail, would be great if success (just like someone tried to install the Buffalo WXR-5950AX12 firmware on WXR-6000AX12 and it worked), how about failed? If you, or no one else is driving it forward, why would you think the success rate is not low?

From these short definite one sentence "doesn't look you'll get too far unfortunately..." answers I get the impression that the authors see the idea as so futile, that the chance for failure is obviously so paramount, that they don't bother to explain their assessment.

IIRC I should expect 3.3V on TTL TX and 3V on TTL RX. I get 33V and 36V :see_no_evil:. Maybe it's just a factor 10 error? I think I need better equipment first.

It's been done, multiple times.

SoC isn't supported by Linux, you need to get it supported there, before you can even start looking at OpenWRT.

What part requires additional explaining ?
As for how hard it is, and how long it takes, depends entirely on your skill set, and the time you're willing to put in.

This :point_up_2: is tremendously more information than " chipset ZX297520 doesn't look you'll get too far unfortunately...".

I did some more research on this router. On many variants

  • there is a UI exploit that easily allows you to run arbitrary commands on the device.
  • there is another UI exploit that lets you export system settings
  • you can simply hook up to the uart and watch uboot

The exploits even work on the most recent firmware revisions.

I have one of the "Strong" branded versions and my UI is not exploitable and the uart does not show me anything. I want to try again tomorrow, because I don't trust my ground pin, but it seems this Strong brand is indeed stronger than its other-branded-mates.

Perhaps something like this will assist you in understanding the complexity of porting Linux to a new system on a chip design

Possibly you’re misunderstanding us due to thinking routers are similar to computers - and can then have general purpose operating systems easily installed. This is not the case. A router, be it arm or mips or PowerPC are embedded systems, more similar to a phone or tablet than to a pc. Each system on a chip is unique, with integrated, custom, hardware that needs to be supported for the kernel and devices to operate

1 Like