Out of all the embedded network devices that had GPL issues, the Texas Instruments AR7 based devices probably have the worst GPL compliance history I've ever seen. The time has come to properly rant about this.
It's yet unclear whether this is TI's own fault, or just the fault of their OEM/ODM manufacturers. But I'm more than determined to find out.
Anyway, the list of problems with TI AR7 based devices is so incredibly long, that I don't even know where to start.
First of all, re-engineering their devices (for GPL compliance audits and legal action following up to such an audit) is incredibly difficult because they've added LZMA compression to both the kernel image (vmlinux) and squashfs.
Now what's so difficult about this? You might argue that the LZMA algorithm is (L)GPL licensed and publicly available. As is the original kernel source code, and the squashfs code. Also, you might know that numerous individuals have already released patches to add LZMA to kernel boot, initrd and squashfs.
However, there are various methods (with/without LZMA header, with/without p7zip header, etc.), and there simply is no standard on how to build a system from the algorithm.
Getting to the actual infringements. So far I've seen devices that
- remove the "(C) Netfilter Core Team" message that is usually printed during boot-up
- modify existing netfilter/iptables code, like add HTTP reply support to ipt_REJECT
- add binary-only new netfilter/iptables targets, like ipt_PNAT
- add new binary kernel modules that have "MODULE_LICENSE(GPL)" without providing source code
There are many other potential issues, on whose GPL compatibility (or lack thereof) I do not want to comment at this time, such as their binary only drivers for the DSL chipset, the WLAN driver.
Interestingly, all of the Vendors of TI AR7 based devices with whom I had contact on the GPL issues showed equally little interest into bringing their products into compliance. Now this could all just be a coincidence. But my personal guess is that they just forward whatever questionable policy they get from their upstream chipset and reference software development kit provider: TI.
You might wander about the device manufacturers in question? I'm still a bit hesitant in disclosing names. One of the first companies running into GPL trouble with TI AR7 was D-Link. Another company with anything but the cleanest GPL history on TI AR7 based devices is AVM, who produce the overly popular and widely branded FritzBox devices.
There is another brand that is sold in significant quantities, at least in the German market. We're on the brink of applying for the next gpl-violations.org preliminary injunction, so I won't be able to say any names.
[and now, after some five hours of gpl-violations related device re-engineering before getting up, I'll finally try to find some time go get some breakfast.]