On Fri, 7 Dec 2007 08:54:24 -0600, Kirk Wolf <[EMAIL PROTECTED]> wrote:
>The general discussion on IBM-MAIN of whether the z architecture is "open" >or not leads me to wonder a bit about Linux on system z. >Perhaps those more experienced with Linux on system z can help me understand >a couple of things. > >AFAIK: >1) Linux for system z is still able to run on raw LPARs, without z/VM >2) IBM contributes kernel patches and tool chain code to support the z >instruction set, under the GPL > >So: >Q1) Are any closed/proprietary instructions and hardware interfaces used? I >would guess that SIE would not be *executed* by Linux, but are others? If >so, does contribution of code under the GPL that "links" to >closed/proprietary interfaces imply anything? IBM has some history of contributing code to Linux that reveals previously undocumented aspects of the hardware architecture. The one that comes to mind is the Compare and Swap and Purge instruction, that was used in Linux, and then later added to an update to POPS. Along the same line, there are several cases where IBM has contributed an OCO module (typically a device driver) to the kernel, and then later released a source code version. The LCS driver comes to mind. Distribution under the GPL does indeed raise very interesting questions about patent licensing. Obviously the OCO modules are not under the GPL, but once source is made available, does it imply a licence to any patents necessary to run the code? Does that extend to patents on other than the code itself, e.g. patents on hardware or firmware? Well, I doubt that the intent of the GPL is to automatically license hardware, but of course "I am not a lawyer", etc. Certainly the GPL V3 contains much more detailed patent terms than earlier versions, and to my understanding contains automatic patent licensing rather than just the previous denial of distribution rights under the GPL. But again, IANAL, or even particularly well read on the GPL. >Q2) Might we expect that eventually Linux on system z will require z/VM, so >that "platform enablement" (for the kernel and device drivers) can be moved >into "closed" DIAG instructions so that IBM can further protect its IP? Presumably GPL'd code will necessarily expose interfaces like this, although not their implementations. Tony H. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html