Re: are GPLed sourceless firmwares legal to distribute ?
Frank -
> the ftpmasters and many others want to give those drives the
> benefit of doubt and assume that they aren't sourceless, but are, e.g.,
> just dumps of unnamed registers and therefore "the preferred form for
> modification". After all, they were what was given to the kernel people
> when the driver was released as .c and .h files under the GPL.
Commenting on the word "given" in that last sentence,
there are following basic classes of files:
1. originally written and released by the hardware manufacturer
2. written by someone else, with support from/cooperation with/payment from
the manufacturer, who also provided the binary blobs
3. written by someone else, who managed to cajole a "legitimate" copy of
the firmware blob from the manufacturer, but with unclear
or self-contradictory licensing arrangements
4. written by someone else, who just snarfed a copy of the needed
firmware from a driver CD or USB sniffer, with no consideration
to ownership or licensing
My analysis does not try to distinguish between these cases, although
in some cases I have suspicions.
No matter the path, the firmware provided is in binary form, and its
rendition in .c, .h, or sometimes hex-record form is an irrelevant
detail. I can convert between them at will, with no loss of information.
> So the real question is whether we want to do that, whether in the
> particular cases there's in fact any doubt, etc.
In making the list, I left off all cases where I had any doubt.
I am not perfect, but I have plenty of experience using and writing
firmware of many kinds. I would be very surprised if any of the
listed firmware is not derived from a human-legible design file of
one form or another.
So while it is perhaps a polite excuse that "we don't know for sure
if these thousands of bytes of hex code were ever compiled from source",
no sane person would bet against it.
Sven says:
> A quick survey based on the size of the firmware blobs suggests 1/3 of
> them may be register dumps, while 2/3 are most probably code.
but I disagree. Sven, can you contact me privately, so I can research
and/or drop some of your potential register dump files?
- Larry
Reply to: