[LITMUS^RT] New LITMUS-RT Version Based on Linux 5.4

[Andrea Bastoni]

Hi Joshua,

On 30/10/2020 20:29, Joshua Bakita wrote:
> Hello all,
> 
> Nathan Otterness, Leo Chen, and myself have been working on an effort to update
> LITMUS-RT to Linux 5.4 at UNC. We've been working on-and-off for a few months,
> but it's finally almost ready!

This is great news! Nice job! Is your rebasing also on github?

> 
> What is the preferred approach to upstreaming a major upgrade like this? Our
> current tree squashes all the historical LITMUS-RT commits. There's some
> internal debate as to whether we should convert our tree into a traditional
> rebase which preserves the LITMUS-RT commit history and modifies each commit as
> appropriate, or leave it as the squash that it is. Some input from the current
> LITMUS-RT maintainers would be appreciated.

I haven't been involved with the maintenance of LITMUS in a long time, but I've
seen major upgrades of several projects, and I strongly advise *against*
squashing for this.

Squashing removes history, it removes the locality of the commit message +
patch, and makes it impossible to understand any design / fix in the long run.

Also, squashing removes the attribution of patches, which I personally find a no-go.

Squashing is IMO a handy way to consolidate and rework commits from the same
authors during development, but major releases should be either rebased or merged.

Some general thoughts about git-branching / workflow models:
https://nvie.com/posts/a-successful-git-branching-model/

Just my two cents...

Best,
Andrea


> 
> We also found a couple bugs in upstream LITMUS-RT while testing our port:
> 1. GSN-EDF can complete jobs twice when budget enforcement is enabled
> 2. The TRACE buffer will fail to allocate when kernel memory is highly
> fragmented (solution: use kvmalloc)
> 
> Best,
> 
> Joshua Bakita
> University of North Carolina at Chapel Hill
> 
> _______________________________________________
> litmus-dev mailing list
> litmus-dev at lists.litmus-rt.org
> https://lists.litmus-rt.org/listinfo/litmus-dev
> 

-- 
Thanks,
Andrea Bastoni