[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#239952: Summary: binary firmware in the kernel

On Mon, Apr 05, 2004 at 02:16:41PM +0900, GOTO Masanori wrote:
> One idea is to separate or remove firmware part from both drivers
> rathar than to delete those entire files from kernel package.
> Look at drivers/net/tg3.c.  If NETIF_F_TSO (see
> include/linux/netdevice.h, offload implementation) is defined, then
> such firmware parts are needed.  So theoretically we can disable them
> and remove firmware code with keeping driver code.  One example of
> diff (not compiled) is at:
> 	http://gotom.jp/~gotom/patch/linux/patch.2.6.5_tg3.c-remove-firmware

No I will not accept patches that cause the driver to run in a
completely different way to the rest of the Linux community.
This is just a support nightmare.  (I know that 2.4 runs this
code-path, but the rest of the networking stack is totally different).

Of course, if you can get upstream to do this then it's perfectly
fine with me.  In fact, apparently they've just done exactly that
for e100.  So why not ask them about tg3?

> OTOH, unfortunatelly, firmware code of qla2xxx/* is coupled with
> hardware initialization code.  We need to have hardware specification
> to know how to remove these parts.
> I think it's also good idea that we contact to upstream author to
> separate firmware from files and to make them work without any
> firmware code.  If you already thought/did, this mail is just my noise
> about this issue.

I personally will not waste my time contacting upstream authors
about this since the most likely answer will be "don't waste my time".
Of course if you believe otherwise, please feel free to contact them.

IMHO, if you really want to keep the driver in main, the most
direct route to achieving that is to write the code to get it
to use the firmware loader.  This shouldn't be hard at all.

It'd be even better if you can get it integrated upstream.

> In addition, I wonder this issue is really well-discussed, though.
> Binary characters are marked as GPL by driver authors, so I still
> think there is no problem to exist in linux kernel.  To say radically,
> removing firmware code from debian is just escape to think about linux
> kernel GPL-ness.

This issue has been beaten to death on lkml so please read the archives
before raising it again.
Debian GNU/Linux 3.0 is out! ( http://www.debian.org/ )
Email:  Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

Reply to: