Re: [cocci] Cocci MUCH more slow after 6.11

On Mon, 5 Jan 2026, Ricardo Ribalda wrote:

> Hi Julia
>
>
>
> On Sun, 4 Jan 2026 at 22:29, Julia Lawall <[email protected]> wrote:
> >
> >
> >
> > On Tue, 16 Dec 2025, Ricardo Ribalda wrote:
> >
> > > Hi Laura
> > >
> > > Has there been any progress on this?
> > >
> > > I am stuck with an old Debian version in our CI and I would like to
> > > update to debian stable.
> >
> > Hello,
> >
> > I made some improvements to the minmax semantic patch that made it go 4x
> > faster on my test machine.  Maybe it will help.
> >
>
> Thanks for the file. Testing with 6.19-rc4.
>
> You should be easily be able to repro with docker/podman. If you need
> help reproducing, let me know
>
> With bookworm and old minmax.cocci (baseline)
> real    3m37.891s
> user    77m19.816s
> sys     2m12.033s
>
> With trixie and old minmax.cocci
> eal    9m8.442s
> user    195m42.341s
> sys     5m53.336s
>
> With trixie and new minmax.cocci
> real    8m19.803s
> user    165m8.025s
> sys     5m52.308s

Interesting.  I had 32 minutes for old minmax and 8 minutes for the new
one.

> With these extra warnings:
> warning: rmaxif: metavariable func not used in the - or context code
> warning: errcode: metavariable func not used in the - or context code
> warning: rmin: metavariable func not used in the - or context code
> warning: rminif: metavariable func not used in the - or context code
> warning: rmaxif: metavariable func not used in the - or context code
> warning: errcode: metavariable func not used in the - or context code
> warning: rmin: metavariable func not used in the - or context code
> warning: rminif: metavariable func not used in the - or context code
> warning: rmaxif: metavariable func not used in the - or context code
> warning: errcode: metavariable func not used in the - or context code
> warning: rmin: metavariable func not used in the - or context code
> warning: rminif: metavariable func not used in the - or context code

This is noise.  I will remove these variables.

> It is definitely better, but it is still ~2.4 times slower

Interesting.  It's not clear to me what would have ever caused the old
version to be efficient...  Thanks for the report.

julia

> Best regards!
>
>
>
>
> --
> Ricardo Ribalda
>