So, I worked for Broadcom for some years after this went down. This post is purely descriptive to give people some insight into the history from inside the company; I'm not commenting on who should have done what (although I was not directly involved, so if someone who was comments, take their word over mine).
Broadcom made an error of judgement here, but this incident fostered a deep distrust of open source, at senior levels, that persisted for more than a decade after; perhaps to this day.
Firstly at this point Cisco was, at the time, Broadcom's largest customer by a large margin. This caused huge tension in that relationship that was totally unforseen, and was very painful for a while.
Secondly, a at a certain point it dawned on Cisco and Broadcom that the GPL lawsuit was not like a normal business dispute , because businessmen after a certain point will settle for money even if they didn't get everything they want. Sure a few people will keep going to the detriment of their own business, but most aim to make profit, not expound a principle. Many companies in the position of the FSF would have settled for a cut of the revenue. But the FSF wanted the source code released, and they were prepared to kill the business to get it. So Cisco and Broadcom had to concede. The source code was released, and OpenWRT was born.
The fallout, though was that subsequently Broadcom router ICs were designed with hardware accelerators which were separate from the main CPU. They were driven by separate CPUs on the same SoC that did not run linux and whose drivers could not be demanded under the GPL. none of the open source firmwares can run these devices efficiently unless someone spends weeks reverse engineering them.
I'm not sure about the last point. I would think hardware dedicated accelerators were done because it was the cheapest way to achieve that performance not because it allowed to somehow bypass GPL. However, choosing to not run Linux but some proprietary OS could most certainly have something to do with that.
At the end of the day, was it a good thing? I would say it was. It opened many generations of home router hardware to being modded/replaced with user controlled software. It even created a market of its own where certain consumer router hardware is advertised as being designed to run custom/third-party software and where vendors themselves ship with some heavily modified software and release the sources for it from day 1 (which are the only wifi routers I shop for these days).
Indeed, hardware accelerators weren't introduced because of the GPL. What changed was that previously they were connected to the main cpu and driven by drivers that fell under the GPL; to avoid the GPL secondary CPUs were introduced not running linux at all.
I can't speak to Broadcom's motivations, but the end result has certainly been that they are the least open source-friendly WiFi vendor, behind Qualcomm-Atheros, Mediatek, Marvell and Intel (client only). When Linksys wanted to do a successor to the WRT54 series trading on its open source reputation, Broadcom wasn't an option because they've made themselves the NVIDIA of WiFi.
Broadcom made an error of judgement here, but this incident fostered a deep distrust of open source, at senior levels, that persisted for more than a decade after; perhaps to this day.
Firstly at this point Cisco was, at the time, Broadcom's largest customer by a large margin. This caused huge tension in that relationship that was totally unforseen, and was very painful for a while.
Secondly, a at a certain point it dawned on Cisco and Broadcom that the GPL lawsuit was not like a normal business dispute , because businessmen after a certain point will settle for money even if they didn't get everything they want. Sure a few people will keep going to the detriment of their own business, but most aim to make profit, not expound a principle. Many companies in the position of the FSF would have settled for a cut of the revenue. But the FSF wanted the source code released, and they were prepared to kill the business to get it. So Cisco and Broadcom had to concede. The source code was released, and OpenWRT was born.
The fallout, though was that subsequently Broadcom router ICs were designed with hardware accelerators which were separate from the main CPU. They were driven by separate CPUs on the same SoC that did not run linux and whose drivers could not be demanded under the GPL. none of the open source firmwares can run these devices efficiently unless someone spends weeks reverse engineering them.