On Tue, Sep 15, 2026 at 06:06:05PM -0700, Kees Cook wrote:
> On Mon, Sep 14, 2026 at 10:22:20AM +0100, Lorenzo Stoakes (ARM) wrote:
> > diff --git a/Makefile b/Makefile
> > index 790ef23c5e8a..38c0cdc9f591 100644
> > --- a/Makefile
> > +++ b/Makefile
> > @@ -561,7 +561,7 @@ PERL = perl
> > PYTHON3 = python3
> > CHECK = sparse
> > BASH = bash
> > -KGZIP = gzip
> > +KGZIP := $(if $(shell command -v pigz 2>/dev/null),pigz,gzip)
> > KBZIP2 = bzip2
> > KLZOP = lzop
> > LZMA = lzma
>
> I had another idea that I think solves both my concerns (limits
> to parallelism, new surprise users) and your concerns (complexity,
> single-threaded when needed)... how about:
>
> Makefile:
>
> KPGZIP := $(if $(shell command -v pigz 2>/dev/null),$(PYTHON3)
> $(abs_srctree)/scripts/jobserver-exec
> $(abs_srctree)/scripts/parallel-gzip,gzip)
That's nice :) Keeps things simple, separates out the jobserver-exec execution
(which it turns out is no big deal also), and abstracts the idea of 'parallel
gzip'.
Let's do this.
>
> and the same scripts/parallel-gzip:
>
> #!/bin/sh
> exec pigz -p ${PARALLELISM:-1} "$@"
>
> And then replace the /boot/ KGZIP uses with KPGZIP. Then we don't need
> to change modules, config_data, new users, nor cmd_gzip, etc, which all
> continue to use the regular KGZIP and the "use python" part for the
> KPGZIP case becomes is a 1-time cost?
Yeah that's really nice.
This is clearly the way forward, will do that in v3, thanks! :)
[Oh and on precedent, it seems the arm64 FIT image builder,
scripts/make_fit.py, uses pigz already if available :>)]
>
> -Kees
>
> --
> Kees Cook
--
Cheers, Lorenzo
