Re: [PATCH v2] kernel/module: Fix mem leak in module_add_modinfo_attrs

[Miroslav Benes]

On Thu, 30 May 2019, YueHaibing wrote:

> In module_add_modinfo_attrs if sysfs_create_file
> fails, we forget to free allocated modinfo_attrs
> and roll back the sysfs files.
> 
> Fixes: 03e88ae1b13d ("[PATCH] fix module sysfs files reference counting")
> Signed-off-by: YueHaibing <yuehaibing@xxxxxxxxxx>
> ---
> v2: free from '--i' instead of 'i--'  
> ---
>  kernel/module.c | 16 +++++++++++++++-
>  1 file changed, 15 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/module.c b/kernel/module.c
> index 6e6712b..78e21a7 100644
> --- a/kernel/module.c
> +++ b/kernel/module.c
> @@ -1723,15 +1723,29 @@ static int module_add_modinfo_attrs(struct module *mod)
>  		return -ENOMEM;
>  
>  	temp_attr = mod->modinfo_attrs;
> -	for (i = 0; (attr = modinfo_attrs[i]) && !error; i++) {
> +	for (i = 0; (attr = modinfo_attrs[i]); i++) {
>  		if (!attr->test || attr->test(mod)) {
>  			memcpy(temp_attr, attr, sizeof(*temp_attr));
>  			sysfs_attr_init(&temp_attr->attr);
>  			error = sysfs_create_file(&mod->mkobj.kobj,
>  					&temp_attr->attr);
> +			if (error)
> +				goto error_out;
>  			++temp_attr;
>  		}
>  	}
> +
> +	return 0;
> +
> +error_out:
> +	for (; (attr = &mod->modinfo_attrs[i]) && i >= 0; --i) {
> +		if (!attr->attr.name)
> +			break;
> +		sysfs_remove_file(&mod->mkobj.kobj, &attr->attr);
> +		if (attr->free)
> +			attr->free(mod);
> +	}
> +	kfree(mod->modinfo_attrs);
>  	return error;
>  }

Hi,

would not be better to reuse the existing code in 
module_remove_modinfo_attrs() instead of duplication? You could add a new 
parameter "limit" or something and call the function here. I suppose the 
order does not matter and if it does you could rename it "start" and go 
backwards like in your patch.

Btw, looking more into it, it would also be possible to let 
mod_sysfs_setup() go to out_unreg_modinfo_attrs in case of an error from 
module_add_modinfo_attrs() (and then clean the error handling in 
mod_sysfs_setup() a bit). module_remove_modinfo_attrs() almost does the 
right thing, because it checks attr->attr.name. The only problem is the
last failing attribute, because it would have attr.name set, but its 
sysfs_create_file() failed. So calling sysfs_remove_file() and the rest 
would not be correct in that case.

Miroslav