[Toybox] [PATCH v3] taskset: Add -c cpulist support

[Rob Landley]

On 3/8/26 15:45, Jesse Rosenstock wrote:
> On Wed, Mar 4, 2026 at 10:45 AM Jesse Rosenstock <jmr at google.com> wrote:
>> `taskset -c` is supported by util-linux as a user-friendly way of
>> specifying the affinity mask.
> 
> Hopefully correctly formatted patch attached.

Sorry for the delay, I'm trying to figure out if having to delete mkroot 
by the end of the year (due to Gavin "spyware" Newsom fining open source 
developers $7500/download if we don't suppress our speech because he 
wants to put a spotlight on children online to make them easier to find 
and follow around the net instead of blending in with the crowd)... 
anyway, does that mean I should just take toybox down off the net entirely?

https://mstdn.jp/@landley/116237441233152688

After all, mkroot had an /etc/os-release file which is pretty much all 
https://agelesslinux.org/ adds to debian...

This would be an easier decision if Android wasn't becoming a 
hermetically sealed closed platform in september (presumably in response 
to losing the epic megagames lawsuit, but let's be honest they've been 
inching this way since 2019) to the point I feel like vichy franch 
collaborating with you at this point. (How much did Google donate to the 
Epstein ballroom again? You _realize_ you're being publicly mocked in 
https://www.youtube.com/watch?v=DAg1-HFdaR4 and so on, right?)

I suppose Motorola shipping GrapheneOS gives me a glimmer of hope, but I 
dunno if they even use toybox, or what (if anything) they'll do about 
https://github.com/upper-up/meta-lobbying-and-other-findings and such. 
Honestly, 
https://www.theregister.com/2026/03/13/opinion_os_verification/ makes me 
root for taiwan to blow the fabs, that silicon mine in north carolina to 
flood, etc. Go ahead with your vibe coded rust. I already cancelled 
crunchyroll and started getting DVDs again.

But in the meantime:

Your help text doesn't say what the :2 is for, and checking debian's man 
page... sigh. Is "stride" actually needed? (I'm assuming this is a 
hyper-threading microoptimization thing? Because -c 1,3,5,7,9 wasn't 
good enough? Also... it autodetects it for output? If you do -c 5,10,15 
will it print 5-15:5? Is that an _improvement_?)

The #define gives me bad ASMR. (You consistently use 8*sizeof(long) but 
have a define for 8*sizeof(toybuf)? Constant evaluated at compile time 
either way...)

[two days later]

I'm just gonna hit "send".

Rob