On Mon, Oct 08, 2007 at 12:35:52PM -0500, David Young wrote:
quoted text
> 
> I believe the inherent flexibility of a microcontroller-based design,
> as compared with an ASIC or microcode, coupled with the peculiarly low
> quality and unnecessary complexity of WiFi firmwares, adds up to a worse
> threat than that posed by, say, ASIC bugs.

David,

You're aware that many of the "microcoded" devices we support are actually
instances of full general-purpose CPU architectures, and that their vendors
call their software load "microcode" simply as a matter of convention?

This has been the case for many, many years (look at "isp" as an example,
or "ti" or "bge" -- noting that the only real difference with bge is
that it loads a *default* software image at power-up, but the driver knows
how to load a new image).

Even the special-purpose CPUs like 53c8xx or the Adaptec SCSI "sequencers"
are certainly as flexible as many simple microcontrollers -- and we provide
toolchains for many of these in our own tree!

I do not believe that it is reasonable to draw a distinction between
programmable, DMA-capable devices as you are doing.

Thor