Re: [PATCH] sched: fix double kfree

[Mike Travis <travis@sgi.com>]

Li Zefan wrote:
> Impact: fix double kfree in failure path
> 
> It's not the responsibility of init_rootdomain() to free root_domain
> allocated by alloc_rootdomain().
> 
> Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
> ---
>  kernel/sched.c |    5 ++---
>  1 files changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/sched.c b/kernel/sched.c
> index 545c6fc..2bad712 100644
> --- a/kernel/sched.c
> +++ b/kernel/sched.c
> @@ -6970,7 +6970,7 @@ static int init_rootdomain(struct root_domain *rd, bool bootmem)
>  	}
>  
>  	if (!alloc_cpumask_var(&rd->span, GFP_KERNEL))
> -		goto free_rd;
> +		goto out;
>  	if (!alloc_cpumask_var(&rd->online, GFP_KERNEL))
>  		goto free_span;
>  	if (!alloc_cpumask_var(&rd->rto_mask, GFP_KERNEL))
> @@ -6986,8 +6986,7 @@ free_online:
>  	free_cpumask_var(rd->online);
>  free_span:
>  	free_cpumask_var(rd->span);
> -free_rd:
> -	kfree(rd);
> +out:
>  	return -ENOMEM;
>  }
>  

Thanks Li!

Another thing I noticed but didn't deal with is:

static int init_rootdomain(struct root_domain *rd, bool bootmem)
{
        memset(rd, 0, sizeof(*rd));

        if (bootmem) {
                alloc_bootmem_cpumask_var(&def_root_domain.span);
                alloc_bootmem_cpumask_var(&def_root_domain.online);
                alloc_bootmem_cpumask_var(&def_root_domain.rto_mask);
                cpupri_init(&rd->cpupri, true);
                return 0;
        }
        if (!alloc_cpumask_var(&rd->span, GFP_KERNEL))
		...

Note that under the if (bootmem) case, it specifies the addresses
directly instead of using (&rd->span) as the other alloc's do.

Not a big deal, just an inconsistency.

Thanks,
Mike