[linux-elitists] I hate you, Greg.
Wed Oct 26 16:03:12 PDT 2005
Quoting Greg KH (email@example.com), writing to Eugen Leitl:
> And if you just want to whine because things change, and try to pick on
> someone else for you not following directions properly, well, bully for
> you. Feel better now?
As Nick said, Gene forwards stuff. And who are we to say it's not art?
> If you want to lament about devfs going away and I should have never
> taken such a wonderful thing out of the kernel, and how horrible I am
> for breaking your machine that had been working for the past eight
> years, then just do that.
Or you can just sit back and wait for Usenet's Lumber Cartel (TINLC) to
sue them for violation of the patent on methods for energetically
flaying deceased equines.
Any number of things could have saved that horse, including Gooch not
hiding in a cave for years. But, as it is, the filly's stone dead and
even my opinion that removal was inevitable irrespective of udev's
(considerable and growing) merits won't resurrect it.
> If you have a problem with me personally, fine get in line over there
> behind the others and I'll add you to my procmail filters (right now
> john gilmore is at the rear of that line for some misguided reason:
Although nobody could fault you for annoyance with whines-without-patch,
you might consider not taking it personally. Just a thought.
Proprietary inclusions in kernelspace strike me as a mess that just
happened without significant overall plan, but can be straightened
out and clarified among people of good will.
Not that you asked, but I have some half-baked thoughts that I might as
well air here, using Keyspan (or, well, the cited description of the
Keyspan donnybrook) as an example -- per my preference, not as a vehicle
for advocacy, but rather seeking clarity.
The code in question is described as "binary firmware". Such matters
have arisen before concerning, say, Prism54 802.11g wireless drivers (as
implemented in Linux until, I gather, recently). My ancient Lucent
Orinoco-based card has certain very low-level functions built into a
ROM. The Prism54 chip's design, by contrast, elects to save the cost of
EPROMs and shuffle off the _very same_ functions into a binary blob.
Intersil wrote that blob, apparently issued redistribution permission to
authorised distributors only, and then died. Globespanvirata inherited
the corpse, which in turn died and ended up being buried by Conexant,
which disavows all knowledge and would like us all to go away and buy
The blob can be flung into working memory at the time of hardware
initialisation, under any operating system, present or future, exactly
as if it had been loaded from ROM, by mechanisms such as Linux's
(userspace) hotplug subsystem. Effectively, it is part of hardware
initialisation -- exactly as my Lucent's ROM is.
My own view is that because the blob has exactly the same initialisation
function as my Lucent's ROM, and is entirely OS-neutral, at least in the
abstract it's no more of concern than that ROM is (other than the
regrettable omission of a public grant of permission to redistribute).
The Keyspan "binary firmware" might be of the same sort -- judging from
the description. If so, the authors might be asserting that it isn't
a derivative work of the kernel itself -- and good luck to them on that
(or whatever luck the legal gods bestow). It might be better to load it
via /sbin/hotplug -- but I can well understand your "Well, send a patch"
> If you have a problem with the way Debian handles udev, I'll point you
> to the proper place to complain about that. Debian is slowly sinking
> into the muck and it's fun watching it happen.
Such rancour! Did Software in the Public Interest steal your lunch money?
 Witness Torvalds's meandering (if consistently imperious) policy,
which I've tracked here: "Proprietary Kernel Modules" on
Cheers, Never anger a bard, for your name sounds funny and
Rick Moen scans to many popular songs.
firstname.lastname@example.org -- Stephen Savitzky
More information about the linux-elitists