Date: Fri, 19 Jan 1996 15:47:27 -0800 From: dmk@uask4it-223 (David Kahn) Subject: Item #307: PPC -- Real Base, Load Base comments Asign this stuff an item number and record it. -David ----- Begin Included Message ----- From owner-p1275-wg@risc.sps.mot.com Fri Jan 19 11:35:38 1996 To: p1275-wg@prombo.eng.sun.com Cc: kingman@austin.ibm.com Subject: Item #307: Real Base, Load Base, etc. Date: Fri, 19 Jan 96 10:27:58 -0600 From: "John A. Kingman" To: Mail list p1275-wg From: "John A. Kingman" Hi Jordan, Here's a note from John O'Quin which outlines his concerns about the way these variables are defined and used. John's references are to the PowerPC processor binding Version 1.7 which I passed out at the meeting. If you could help me come up with some verbiage to clarify or replace what's there, I'd appreciate it. Thanks, John Kingman ~ ~_~ ~ ~ ~_~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ | | ___ | |_ ._ _ | John A. Kingman _| |/ . \| . || ' | | Kingman@Austin.IBM.com \__/\___/|_|_||_|_| | IBM RSD, Austin, Texas ::::::::::::::::::::: referenced message :::::::::::::::::::::::::: Date: Thu, 18 Jan 96 11:36:02 -0600 From: "John O'Quin" To: "John Kingman" Cc: "Todd Moore" , "Bob Coffin" Subject: Item #307: PowerPC Processor Binding Comments/Concerns PowerPC processor binding to: IEEE Std 1275-1994 Standard for Boot (Initialization, Configuration) Firmware: Revision 1.7 dated January 11, 1996 I have the following comments/concerns about the above identified Open Firmware binding: 1) Page 9 lines 25-27: This is wrong. It also conflicts with page 16 lines 10-12. It should say something like OF may choose its base location if real-base is -1 and real-mode? is true or virt-base is -1 and real-mode? is false. 2) Page 9 lines 20-27: Please indicate what happens if real-size or virt-size are -1. I expect to set real-mode? to true and real-base to an address, but want to be able to set real-size to -1. Is this ok? 3) Page 9 lines 20-27, or page 21 lines 19-25: Need to indicate the relationship between real-base/virt-base and load-base. I recommend that we say that load-base is ignored if it conflicts with real-base/virt-base. Note that real-base/virt-base can be specified in the ELF header, but load-base can not so this conflict is very likely and OSes can not prevent. 4) Page 13, lines 6-23: Need to add max_number_cpus property that defines the maximum number of processors that this platform is designed for. This would indicate the difference between a uniprocessor platform and an SMP enabled platform with one processor. 5) Page 13, lines 6-23: Need optional slot-names property for external labeling. See the PCI binding. The bit mask should be based on cpu reg value. (Ron had indicated adding another field to the cpu name property for this.) -- Regards, John O'Quin International Business Machines Internet: oquin@austin.ibm.com 11400 Burnet Rd IBM LAN: oquin@austin.ibm.com Austin, TX 78758-3493 U.S.A. IBM VNET: OQUIN at AUSVM6 RISC System/6000 Division Phone: (512) 838-3904 T/L 678-3904 Bldg 905, Room 8A-021, ZIP 9582 FAX: (512) 838-0735 T/L 678-0735 ----- End Included Message ----- [ P1275 Item #307 -- Received: Fri Jan 19 15:47:11 PST 1996 ]