Has someone a real experience on that? I mean, do you really think the vendor will try to read the flash on a bricked device?
And if not bricked, why not re-install the original firmware before requesting warranty?
It seems linksys (at least in Canada) only tell they do not warranty open source firmwares, not that this will void your waranty reference.
The issue is that if the power unit or the switch bridge fail - both common failures - there's no way to flash it back to OEM firmware. And even if with a lot of effort you manage to do, they can tell if the device has been flashed before. So, there's no point in exploring those kinds of tricks in a commercial setting.
Power unit fails -- Order a new one from Mouser, Digikey, or your favorite professional parts supplier. A Level VI unit from a upper-tier manufacturer is $7-15. Much cheaper than the time cost of someone creating the RMA, packaging the unit, and sending it back. More importantly, as you have one as a spare on hand, no down time.
Switch bridge fails -- If that falls into "common failures" maybe you should be looking at a different vendor. Even if you have an occasional failure, is it worth your time to RMA a $100-class item, then potentially get back a "refurb" unit, possibly inheriting someone else's problems?
Think from a LEDE perspective, we need to reach a similar point to where Linux or any open source OS is with servers, i.e. Dell, HP, IBM, Lenovo won't void their warranty if you put Linux on it. The contrary, they will see it as an opportunity to sell their hardware at an effectively lower price point. Any thoughts as to why this would be different in the embedded Linux case with Linksys, Netgear, Asus, TP-Link etc.? Is it the low retail price point of those routers that creates this cost sensitivity around the warranty? Would that sensitivity be alleviated with volume orders? Hypothetically, if you go to Linksys and order 1M units, would they be willing to support their hardware under any open source firmware you wish to load it with?
The idea is not to support the open source firmware but their hardware running open source firmware. As a parallel example, Dell is not required to support Linux but they do support the hardware running Linux or any other open source OS.
In EU flashing a custom firmware does not void warranty, in the sense that the seller of the device (EU laws state that the seller provides 2 years of warranty) can't say a thing for the first year or so, and then for the second year it must prove that the device was damaged by the custom firmware.
Since in most cases the cost to certify that the custom firmware damaged an embedded device is orders of magnitude higher than just sending a new one as a replacement, they don't usually bother.
Technically speaking, the reason vendors of x86 hardware is treated differently is because there the OS isn't running bare, but many of the "dangerous" things are done by a proprietary firmware (BIOS or UEFI) which should not allow your OS to damage the hardware. (theoretically anyway)
So if your motherboard fails the manufacturer does not care what you've been running on it. Because whatever was running on it was still using UEFI or BIOS to control dangerous hardware settings.
On an embedded device there is no such thing, the Linux kernel runs bare with no board firmware interfering, if it screws up initialization or something it can damage the hardware.
That said, even in x86 land only company-grade OEMs like Dell, HP, or SuperMicro actually care about officially supporting Linux (i.e. caring if you have Linux-only issues).
Consumer-grade OEMs (ASUS, Gigabyte, MSI and so on) usually support only Windows.
Really, I doubt most vendors even care about looking into why the board isn't booting anymore. Most service manuals I've seen only tell to do basic troubleshooting, and then if that fails it instructs to "change the motherboard". And this for laptops and PCs that cost 500-1000$. Who is going to care about a device that costs less than 100$? The time spent by the technician on it will cost more than just shipping a new device to you.
Also, in a commercial setting the cost of replacing the rare hardware failures (seriously, how many power units fail, and how many switch bridges fail? If you actually care about their numbers it's a sign you are using crappy hardware to begin with) is negligible.
I mean, you don't field gold-plated toys with 9 antennas and RGB leds costing 400$, but more normal units costing like 50-100$ tops, and in pretty much any case you're so much better off just not calling any vendor and replace the device with another one you pull from your stockpile.
We are not discussing morals, but the actual chances the vendor would care about that. With embedded devices the cost of a new device is so low (it costs them much less of the price you paid them to buy the device) that getting a technician to do forensics is not worth it.
If you send them stuff in RMA they can either send back something new and pay a relatively low cost, OR pay a technician more than a new device is worth to decide who is the fault.
Really if the device was worth less than 100$ new it's not even remotely worth it to care.
Then ok, there are many scumbag sellers that just just don't accept RMAs unless you sue them, but that's an issue for consumers buying 1 device, not for company buying a ton of devices.
Technically speaking, they do already. The final price of the product (any product really) has the estimated cost of RMAs and other assorted estimated future costs already calculated in its release price.
I'd like to point out that the situation here was not about someone trying to make a fraudolent RMA, but someone that flashed OpenWrt, then the device fails because of some normal issue that would be covered by warranty, and then the vendor refuses the RMA because it was using a custom firmware.
As I said it is 100% possible to damage the hardware by using a custom firmware so the vendor can't support a custom firmware, period.
But with OpenWrt you cannot do that casually as a user.
You really need to go looking for trouble, by overclocking, changing voltages and other hardware options in the board dtb files in the source code, and hacking around with the drivers, which are things that don't happen in normal operation of OpenWrt.
Think a good summary of this discussion is that for low-cost routers, you may or may not care about hardware warranty. If you are a businesses, you would care and retail branded OEMs are not going to cover it, even under their open source intended versions.
However, if you go with a white box vendor and order large quantities, they will do this.
The whole argument then concludes that it's just a matter of pricing for vendors to accept that risk.
Now, the question is, are there any other creative ways around this?
Is there any third party willing to underwrite an insurance on the hardware or provide after-market warranty?
Or, is the only solution a white box vendor, possibly losing a bit on economies of scale and paying a slight premium on the warranty?