Re: [PATCH] sched/isolation: Allow "isolcpus=" to skip unknown sub-parameters

[Ming Lei]

On Tue, Feb 04, 2020 at 11:16:39AM -0500, Peter Xu wrote:
> The "isolcpus=" parameter allows sub-parameters to exist before the
> cpulist is specified, and if it sees unknown sub-parameters the whole
> parameter will be ignored.  This design is incompatible with itself
> when we add more sub-parameters to "isolcpus=", because the old
> kernels will not recognize the new "isolcpus=" sub-parameters, then it
> will invalidate the whole parameter so the CPU isolation will not
> really take effect if we start to use the new sub-parameters while
> later we reboot into an old kernel. Instead we will see this when
> booting the old kernel:
> 
>     isolcpus: Error, unknown flag
> 
> The better and compatible way is to allow "isolcpus=" to skip unknown
> sub-parameters, so that even if we add new sub-parameters to it the
> old kernel will still be able to behave as usual even if with the new
> sub-parameter is specified.
> 
> Ideally this patch should be there when we introduce the first
> sub-parameter for "isolcpus=", so it's already a bit late.  However
> late is better than nothing.
> 
> CC: Ming Lei <ming.lei@xxxxxxxxxx>
> CC: Ingo Molnar <mingo@xxxxxxxxxx>
> CC: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> CC: Juri Lelli <juri.lelli@xxxxxxxxxx>
> CC: Luiz Capitulino <lcapitulino@xxxxxxxxxx>
> Suggested-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> Signed-off-by: Peter Xu <peterx@xxxxxxxxxx>
> ---
>  kernel/sched/isolation.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/sched/isolation.c b/kernel/sched/isolation.c
> index 008d6ac2342b..d5defb667bbc 100644
> --- a/kernel/sched/isolation.c
> +++ b/kernel/sched/isolation.c
> @@ -169,8 +169,12 @@ static int __init housekeeping_isolcpus_setup(char *str)
>  			continue;
>  		}
>  
> -		pr_warn("isolcpus: Error, unknown flag\n");
> -		return 0;
> +		str = strchr(str, ',');
> +		if (str)
> +			/* Skip unknown sub-parameter */
> +			str++;
> +		else
> +			return 0;

Looks fine, even though the 'old' kernel has to apply this patch. 

Reviewed-by: Ming Lei <ming.lei@xxxxxxxxxx>


thanks,
Ming