Date: Fri, 2 Aug 1996 18:13:28 -0400 From: milendor@chicklet.East.Sun.COM (Michael Milendorf (Godin) - Sun BOS Firmware) Subject: Item #383: clarifications on MEMORY-TEST-SUITE and MASK [ dmk: reformatted text to 80 char lines ] P1275 Openboot Working Group Proposal -- Proposal #:383 Ver Title: Clarification on MASK and MEMORY-TEST-SUITE fcodes Author: Michael Milendorf Date: August 2, 1996 Ed/Tech: Technical Synopsis: Clarification Doc & Version: IEEE Std P1275-1994 Problem: MEMORY-TEST-SUITE and MASK definitions need clarification The P1275 definition of MEMORY-TEST-SUITE says "perform tests of memory, starting at addr for len bytes". It says that "the actual tests performed are machine specific" however it doesn't clarify on what kinds of memories the test could be applied to, and what kinds of accesses it is using. The test can't be applied to both system and device memories and always provide a correct error reporting in case of failure. The same types of accesses presumably used in MEMORY-TEST-SUITE wouldn't always work the same way on system and device memories. Proposal: The definition for MEMORY-TEST-SUITE should clearly state that the test is intended for device registers only and shall implement device type reads and writes. Otherwise MEMORY-TEST-SUITE is not guaranteed to work on plug-in cards. In conjunction with clarifications on MEMORY-TEST-SUITE the definiton for MASK should be changed from "which bits out of every cell will be tested" to "which bits will be tested with MEMORY-TEST-SUITE". Note: To make MEMORY-TEST-SUITE more usefull TWO other FCODEs could be introduced: REG@ and REG! MEMORY-TEST-SUITE shall use REG@ and REG! types of accesses. Each particular FCODE driver or FCODE selftest then shall set the required cell-read and cell-write words along with MASK value prior to calling MEMORY-TEST-SUITE: ['] rl@ to reg@ ['] rl! to reg! ffff.0000 mask ! addr len MEMORY-TEST-SUITE [ P1275 Item #383 -- Received: Fri Aug 2 15:19:03 PDT 1996 ]