Havic & GEN3 Project Status 2022-03-30

Andrew Sharp



Meeting w/ Kelly Spafford and Rick Klaus 3/28

Some important information regarding use cases for the two boxes. Havic is apparently the box used by the engineers for building of all the different sources: the firmware; the embedded software; and the host (Gen3 and Havic) software. That information has been put to good use as I have made some changes to the hardware spec for Havic to reflect that use. One important software implication is that I think the kernel on Havic should be 64-bit as this would help speed up the building significantly, as well as move things into the 2020's and beyond.

Meeting w/ Lance Spaulding 8:00am 3/30

Good meeting this morning, lots of questions answered. PCIe cards don't have the logic onboard themselves to talk the bus protocols, that is in the additional hardware that is attached for testing purposes, namely Landcruiser for ISA testing and ? for Havic. This explains why the cards don't show up on the bus scan and how the drivers can get loaded, which they currently don't on my hardware.

Lance gave me a pointer to the kernel driver source, which I have now found in some of the source heirarchies that are on the Havic machine. Not sure why I wasn't able to find them before, but possibly I ignored the `fw' directory under which they reside. The source directories that I have are dated 2020, so I'm not sure how up to date they are.

Lance mentioned that he didn't think there would be a lot of 32/64 bit issues involving Tron, the project name for the most current ASIC(?). He did mention that there are packets transferred between the host tools and the testing hardware and ASIC that might be affected, but possibly not when working on Tron.

There was some good information obtained about why Gentoo as the previous/current Linux distro, as well as some software dependencies of the previous (and possibly current) host kernel drivers and kernel boot parameters. Pointers were given for more specifics on most of those situations.

I recorded the meeting so I will be going over it again later to see what else I can absorb, as there was a lot of information disseminated.

A thank you to Lance for taking the time to help us out, it has helped me greatly.



Road ahead

I'll be going over the host kernel driver source to try and get an understanding of any 32/64 issues that might be looming. It appears that the embedded kernel driver source is also there, so I can look at that as well.

I'll be sending out a more informative spec for the proposed Havic hardware replacement, as sending the config from the HP configurator was pretty much a bust.

I will also home in on a spec for the Gen3 hardware replacement. This will be more generic as I suspect whatever sub HP uses to build them will choose specific cases, motherboards and the like.

It is now pretty clear that Rick was correct in that I do need the testing hardware that attaches to these two boxes, so we should be that ball rolling. I'm not sure what the lead time might be on that.

I will be investigating the distro question over the next couple of days. Lance made a good point regarding the systemd question, so I'll be taking a look at that. The only other distro I would consider is Debian, for it's longevity, long term support for security patches, and for it's [possible] technical superiority. Gentoo and Debian have different philosophies regarding ongoing adoption of new versions, so that will be considered as well.