Inception: 2005/112 sunv4 io API Submitter: david.kahn@sun.com Owner: Hitendra Zhangada Intern: Stephen Ehring DMK gave an overview about the project. The following is summary of his overview: * This case defines the I/O API for sun4v, and specifically for PCI* style buses. Therefore, I/O API can be referred as PCI I/O API. Issues ------ Hitendra Zhangada 1. 18Q, Q# 15, last bullet refers to "some measurable effect.." Have measure the boot time effects? Similar question for the core API case (2005/116) states that there are no measurable boot time effects. So, is this true that IO APIs have boot time effects but the core APIs don't? Further, for the same answer, it states that the systems that implement sun4v tend to be larger systems. I don't think this is true. Sun4v is implemented on Ontario and Niagara-2 based smaller systems. DMK: Will work offline with Hitu on this issue. Following comments/issues are for IO API specification version 1.29. 2. Section 1, middle of third paragraph seems to have a typo. "Results 0 - 5 are returned in %o0 - %o4" seems to be incorrect. I think the correct sentence is "Results 0 - 4 are ..." DMK: Will fix this. 3. Section 5.1, we may want to have registry for MSI function numbers. Is there a reason why we skipped 0xcf? DMK: Yes, there is a reason to skip 0xcf because MSI and MSIQ have different range. It does not matter which number to use, it can be random number. Status: CLOSED. 4. Section 5.3.13, typo, the last word should be "msiqid". 5. Section 5.3.13.1, typo, "msium" should be "msinum". 6. Section 5.3.17, typo, the last word should be "msiqid". The above issues are accepted by DMK and will fix the specs. Tony - note in the opinion that this case is dependent upon 2005/116 and there is no issue in 2005/116 that affects this case. Redman: The registry can affect this case too. Question: Converting this case into a fast-track case ?? ** VOTE to APPROVE this case as a fast-track: Redman, Hitu, Tony, DMK: Yes Status: This case is approved contingent upon updated spec from DMK. Then, Hitu will mark this case closed approved. It's also agreed that the registry that is created by case FWARC 2005/116 will include APIs from this case. ===========================================================================