Re: [PATCH 2/8] uprobes: revamp uprobe refcounting and lifetime management

[Oleg Nesterov <oleg@redhat.com>]

On 07/31, Andrii Nakryiko wrote:
>
> @@ -732,11 +776,13 @@ static struct uprobe *alloc_uprobe(struct inode *inode, loff_t offset,
>  	uprobe->ref_ctr_offset = ref_ctr_offset;
>  	init_rwsem(&uprobe->register_rwsem);
>  	init_rwsem(&uprobe->consumer_rwsem);
> +	RB_CLEAR_NODE(&uprobe->rb_node);

I guess RB_CLEAR_NODE() is not necessary?

> @@ -1286,15 +1296,19 @@ static void build_probe_list(struct inode *inode,
>  			u = rb_entry(t, struct uprobe, rb_node);
>  			if (u->inode != inode || u->offset < min)
>  				break;
> +			u = try_get_uprobe(u);
> +			if (!u) /* uprobe already went away, safe to ignore */
> +				continue;
>  			list_add(&u->pending_list, head);

cosmetic nit, feel to ignore, but to me

			if (try_get_uprobe(u))
				list_add(&u->pending_list, head);

looks more readable.

Other than the lack of kfree() in put_uprobe() and WARN() in _unregister()
the patch looks good to me.

Oleg.