Re: tunables overridability patch vs release

On 06/07/26 17:46, DJ Delorie wrote:
> 
> Re: https://patchwork.sourceware.org/project/glibc/patch/[email protected]/
> 
> So I originally left this bit out intentionally because (1) I thought
> it would be hard, (2) it wasn't needed by most of the folks waiting
> for the tunables work, and (3) I didn't expect the patch set to land
> so close to a release.
> 
> However, tunables is in and it's already hit the press[*], including
> the bit about overridability, because I implemented the parser side of
> this already (needed by the AT_SECURE parts).  The patch turned out to
> be fairly simple but still needs a reviewed-by.
> 
> Thus, I would like to consider this patch as "desirable" for this
> release.  It was posted before the freeze, one could consider it a
> bugfix and/or continuation of the main tunables patches, or we could
> delay it and backport it later.
> 
> [*] https://www.phoronix.com/news/Glibc-System-Tunables
> 

I think this should be a release blocker because it does fix some issues.
Andreas Schwab has already approved, and I only have a small comment.

As a side-note, I am plan send some additional fixes to the system-wide
tunables that I missed on my previous reviews (I was about to re-start
it and saw that Arjun has already approved them).  Some of them are
harderning, one it a performance optimization, and another is semantic
fix for the old environment variables.

Since this is a new features, so I think we should *not* release with some
missing fixes or proper defined semantic.