| authors | sponsor | state |
|---|---|---|
Peter Tribble <[email protected]> |
Joshua M. Clulow <[email protected]> |
published |
The illumos codebase contains support for a large variety of Sun desktops and servers. Given that some models are old, rare, and expensive, there's a likelihood that nobody running illumos will ever possess some of these hardware models, and some of the code we have may not only be useless, but also untestable.
What of this code is useful and worth keeping and fixing, and what should be dropped?
The aim of this project is twofold: to reduce the maintenance burden by removing code that has no utility, and enabling better code quality and support for the platforms that remain.
This project takes place in the context of significant changes within the illumos ecosystem, namely: the addition of gcc7 as a shadow compiler (and its potential promotion to the primary compiler), the replacement of lint by smatch, redefinition of NULL, and other potential modernizations of the toolchain (amongst which, on SPARC, we might include replacemnt of the old Sun assembler with the GNU assembler). Clearly, reducing the volume of code to be modernized would be a benefit.
The plan is to keep support largely as-is for the sun4v platform, but to limit sun4u support to those systems which current users either have or might easily be able to obtain. This essentially means that we will support desktop or volume server systems, but remove support for the specialist high-end server ranges.
An informal survey of SPARC models known to be currently running an illumos distribution, or that have run illumos in the past, generated no surprises:
- Ultra 5
- Ultra 60
- Sun Blade 1000
- Sun Blade 1500
- Sun Blade 2000
- Sun Blade 2500
- V210
- V240
- V245
- V490
- T1000
- T2000
- T5120
- T5220
- T5140
- T5240
The starfire range - the venerable Sun E10K. This has already been removed in 10318.
The V880z, a dedicated graphics variant of the V880 with an XVR-4000 graphics card, was removed in 6027.
The sunfire servers - E3000-E600, E3500-E6500. While not a terribly complex platform, there are no known users, it's 2 decades old, and ties us to sbus, sf, and socal.
The starcat range - The F15K and variants. Like the starfire, these were big expensive systems requiring dedicated controller hardware. See issue 10864.
The serengeti range, which are the newer Sun-Fire E2900-E6800 systems. Although more modern, there are no known users, and there's a big blob of complex code.
The Lightweight 8, or V1280, which is some serengeti boards in a volume server chassis.
Certain Netra systems, specifically the NetraCT compactPCI blade chassis systems. (Code names are montecarlo for the SUNW,UltraSPARC-IIi-Netract; makaha for SUNW,UltraSPARC-IIe-NetraCT-40; sputnik for SUNW,UltraSPARC-IIe-NetraCT-60; and snowbird for SUNW,Netra-CP2300.)
The B100s server blade.
Certain platforms bring in a certain amount of code complexity which would qualify them for removal, but there are good reasons for keeping them.
The opl, or Olympus platform, which is the Fujitsu-derived M-series. The reason here is that the M3000 model is readily available and cheap. It may be possible to thin out the opl support to exclude the complex domain and DR operations.
The Ultra-2, which would be the last remaining sbus system. The issue here is that the Ultra-2 is the base platform from which many of the smaller desktop systems are inherited.
The proposed plan for implementation is to work through the platforms, removing one at a time (similar to starfire removal). Breaking the work up like this makes each step more manageable, and also allows easier triage if anything does get broken.
It's likely that there will need to be final cleanup of unused modules, as there are some modules that are shared between platforms.
It is planned to defer other work, such as gcc7, NULL, and smatch, until this project has completed, to avoid having to clean up code that is then removed.