Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 13 Dec 2001 01:33:10 -0800
From:      Terry Lambert <tlambert2@mindspring.com>
To:        hiten@uk.FreeBSD.org
Cc:        chat@FreeBSD.org, grog@FreeBSD.org, phk@FreeBSD.org
Subject:   Re: IBM suing (was: RMS Suing was [SUGGESTION] - JFS for FreeBSD)
Message-ID:  <3C1875D6.5DE4F996@mindspring.com>
References:  <3C186381.6AB07090@yahoo.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Hiten Pandya wrote:
> IBM has release JFS code under the 'General Public License'. What
> we have to do is ask this question internally to IBM, which Greg
> might do that for us, if possible; and then we see what the JFS
> Team from IBM's views are about this.

This whole legal discussion stemmed from whether or not you can
have a root FS that was JFS, and still comply with the GPL.

There are several ways to do this with a JFS port; the last five
skate by on technicalities, while the first simply disregards the
license entirely:

1)	Statically link JFS into a distribution kernel, and then
	distribute it, in violation of the GPL.

	o	IBM would hate this; they hate people violating
		licenses.

2)	Statically link a kernel as part of the installation
	process, so that you are technically not distributing a
	kernel that has GPL'ed code linked into it.

	o	IBM would hate this, and RMS has sued over this
		before, and won a settlement (no binding case
		law in the U.S., yet).

		This is legally risky, if Greg is right, and the
		intent was to prevent commercial use of the code,
		since you defeat their intent.  Further, case law
		in the U.S. favors the plaintiff, due to historical
		interface copyright issues (RMS argued a crypto
		library was GPL'ed for being written to interface
		only to GPL'ed code, since there were no other
		instances, and thus this was an attempt to subvert
		the intent of the licensor).

3)	Provide the capability to build a kernel with a JFS, but
	provide no means of installing a FreeBSD system configured
	this way automatically.

	o	This holds the least risk; it is also the least
		generally useful implementation, since people would
		need to build their own release CDROMs locally to
		be able to successfully obtain native JFS installs;
		you will never be able to ship a product with such a
		FreeBSD preinstalled on it.

		This is potentially legally risky if the company is
		ever sold, most particularly if its assets are sold
		seperately (the company is a legal entity, so a bulk
		sale might be OK), since you can not legally transfer
		ownership of the binaries, since it is impossible to
		compy with the GPL on the JFS code.  FWIW: IBM made
		Whistle rip SQUID out _before_ the sale of Whistle
		to IBM was made final; however Whistle was privately
		held at the time.

4)	Build boot code that can at least read JFS, and load the
	real JFS as a kernel module.

	o	Since you have to deal with all the lookup and
		log version selection issues, this is more than
		half way to a reimplementation without the GPL.

		This is legally risky, if Greg is right, and the
		intent was to prevent commercial use of the code,
		since you defeat their intent.

5)	Use an alternate boot partition of a different FS type
	(e.g. FAT32 or FFS), and load the JFS module from there.

	o	This is harder than option #3, since it requires
		minimally that union mounts work, since mounts
		over mount points before the kernel is booted
		can't work, and therefore for your / to be JFS,
		it will have to be mounted over top of the boot /,
		which would be non-JFS.

		This is legally risky, if Greg is right, and the
		intent was to prevent commercial use of the code,
		since you defeat their intent.

6)	Write a JFS compatible FS, not using the IBM JFS code.

	o	It would be difficult to prove "clean room"
		techniques, and court cases in the U.S. recently
		have decided in increasingly high barriers to
		clean-room coding.

		This is legally risky, if Greg is right, and the
		intent was to prevent commercial use of the code,
		since you defeat their intent.


> If they think that porting JFS to FreeBSD is a waste of time and
> energy, because of the licensing issues, than we can take an
> alternate approach to 'crash recovery' problems and journalling,

As I said before: feel free to write the code.  Just don't expect
people who, philosophically, want the FreeBSD code to be usable
(and used!) as a source of reference implementations to participate
in this process.

-- Terry

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-chat" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3C1875D6.5DE4F996>