jake1981 wrote:After all, I had to postpone ordering one with 2 weeks due to my financial situation, just cannot afford one immediately And as I am not likely going to be raised or something like that in years, it's very unlikely that atleast I could afford to donate one
Donation was just a proposal, open source is not behaving like churches even if it may sometimes look like a religion
I'll also try to give some money to b43 devs, but also not in the range of the price of the router, but if many of us are contributing, telling them we want 4322 support, it may help. Who said one guy should pay for all others ?
jake1981 wrote:I know, it would be better if we coud use b43 driver instead - but how I see it, currently other products using wireless by broadcom are using proprietary drivers, why couldn't we keep using one 'till b43 team can provide us a decent driver? I didn't quite understand what you said by "it's not working on MIPS, and even it was, it would require us to disable ssb which just happens to be required for tg3 to recognize the switch.."
I'll try to explain, at least what I understood... Corrections and contributions are welcome.
- About MIPS: the CPU running in embedded systems like this router are generally not intel x86 derivative. There many options on the market, and many (if not most) are built around the MIPS architecture. The problem is that these two types of CPUs do not use the same "language" (not the same assembler), so a program that you compile to run on a standard PC won't work on a MIPS CPU. If you have the source code, you can adapt it to support multiple architecture, and compile it for both architectures ( this will produce one binary per type of CPU). You certainly already understood, that if you don't have source code, you're out of luck.
- Broadcom has released a proprietary (ie: closed source), but only for x86 and x86_64 (no MIPS release). In order to protect their "intellectual property" (ie:owning the inventions created by other's brains), they hide most of the information and knowledge about their chips into some source code they will never distribute except perhaps to their direct clients (ie: linksys, asus, dlink, netgear....). These guys then build their final product and ship only the binary to public. Public can then use the driver but only for the specific uses intended by the manufacturers. Lets add one point of honesty: another reason for hiding code is compliance (have a look at http://linuxwireless.org/en/developers/Regulatory). Lets add one point of speculation: it's far more in the interest of routers manufacturers to lock the source code. I guess that the guys in these companies just get mad when they see the amount of features you can put on these boxes with openwrt. They can't sell boxes providing all these features at the price we get them, or they would have a hard time justifying the prices of their "enterprises" boxes, so they protect their advance by using secrecy...even if more than 90% of the software used in their boxes is open source. This is why everyone will correctly point out that they suck!
- SSB is an inter-chip communication bus (like PCI) that, IIRC, has specifically been developed by Broadcom. New drivers for managing this bus have been added to Linux 2.6, and both the tg3 driver for the ethernet card, and the b43 drivers are developed to use that ssb driver. However, the documentation shipped with broadcom's driver specifically exclude the activation of ssb in kernel, so we could not use tg3 and this driver at the same time (even there is a way to run the binary on MIPS, which IMHO is itself near to impossible)
jake1981 wrote:By a proprietary driver; do you mean a driver used on, for e.g. Linksys wrt54gs's and other older products? Why not use one from WRT610n? One used on different models most likely after all wouldn't work - wrt610n has dual radios and a/b/g/n connectivity.. There's gotta to be a source where earlier proprietary driver was from after all.. And as b43 team says: BCM 4322 802.11a/b/g/n (Has PCI-ID 0x432B) - This device has an N Phy. There is no support for any Draft 802.11n features.
Well, linksys are based on 2.4 kernels. The only "short term" solution seems to integrate the proprietary wlan drivers and the proprietary ethernet driver (as it looks from the latest firmware at least). As I don't care for wlan at the moment, I'm more willing to go for 2.6 for all the functionality it provides.
There is a source for recent or old proprietary drivers... The problem is that we don't and won't have access to it ;-)
jake1981 wrote:Therefore it's very unlikely that driver would be available any time soon (maybe in 2 years, and that propably ain't enough time even if they would get donations) as they are still working on previous versions of hardware and have got so less results in eyes of a normal user in years, that I guess maybe there's a proper driver for wrt54gs after 2 years and maybe then we can start expecting some results on draft-N nics.. Just trying to be realistic here..
Well, these guys are reverse-engineering the beast and that's a hard, long and uncertain process... So you're right, until someone open up the specs (don't hope), the process will certainly be long, and I would not for a full working release of openwrt on this platform anytime soon...
(Last edited by sentenza on 19 Nov 2008, 22:50)