Re: Etcupdate runtime has more than doubled....Why?

From: bob prohaska <fbsd_at_www.zefox.net>
Date: Wed, 01 Apr 2026 17:54:37 UTC 
On Wed, Apr 01, 2026 at 01:26:02PM -0400, Charlie Li wrote:
> bob prohaska wrote:
> > On Wed, Apr 01, 2026 at 07:14:54PM +0200, Dimitry Andric wrote:
> > > On 1 Apr 2026, at 19:13, bob prohaska <fbsd@www.zefox.net> wrote:
> > > > 
> > > > For some time now I've been using etcupdate on a self-hosting
> > > > armv7 (RPi2) system. Lately, the runtime for etcupdate after
> > > > a world/kernel update from source has seemingly more than doubled.
> > > 
> > > It's unfortunately not an April Fools joke, but I think it's due to the new LLVM_BINUTILS default, which can cause some stuff to get rebuilt during etcupdate. I am unsure if there is any PR or pending fix.
> > > 
> This is exactly it. After emaste@'s commit, LLVM binutils is getting built
> directly in the etcupdate tree because the built artefacts do not exist
> there (but rather in /usr/obj).
> > If a workaround emerges is would be very helpful on a Pi2 !
> > 
> The workaround is to revert a follow-up typo fix and the original commit in
> order:
> 
> e8dac3944dca - main - Makefile.inc1: Remove extraneous }
> 17494c6e6b7d - main - build: Boostrap LLVM_BINUTILS for cross-tools
>
 
I'm a trifle confused; is the present situation considered a bug
to be corrected in a later revision, or a feature to be preserved
for general use? If it's going to be preserved I'll attempt the 
two reversions. 

Thanks for writing!

bob prohaska