RE: [RFC][PATCH 3/5] [PATCH 3/5] kvm-ept-idle: HVA indexed EPT read

["Peng, DongX" <dongx.peng@intel.com>]

kvm_get_kvm() kvm_put_kvm()

-----Original Message-----
From: Nikita Leshenko [mailto:nikita.leshchenko@oracle.com] 
Sent: Tuesday, September 4, 2018 3:57 PM
To: Wu, Fengguang <fengguang.wu@intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>; Linux Memory Management List <linux-mm@kvack.org>; Peng, DongX <dongx.peng@intel.com>; Liu, Jingqi <jingqi.liu@intel.com>; Dong, Eddie <eddie.dong@intel.com>; Hansen, Dave <dave.hansen@intel.com>; Huang, Ying <ying.huang@intel.com>; Brendan Gregg <bgregg@netflix.com>; kvm@vger.kernel.org; LKML <linux-kernel@vger.kernel.org>
Subject: Re: [RFC][PATCH 3/5] [PATCH 3/5] kvm-ept-idle: HVA indexed EPT read

On 1 Sep 2018, at 13:28, Fengguang Wu <fengguang.wu@intel.com> wrote:
> +static ssize_t ept_idle_read(struct file *file, char *buf,
> +			     size_t count, loff_t *ppos)
> +{
> +	struct task_struct *task = file->private_data;
> +	struct ept_idle_ctrl *eic;
> +	unsigned long hva_start = *ppos << BITMAP_BYTE2PVA_SHIFT;
> +	unsigned long hva_end = hva_start + (count << BITMAP_BYTE2PVA_SHIFT);
> +	int ret;
> +
> +	if (*ppos % IDLE_BITMAP_CHUNK_SIZE ||
> +	    count % IDLE_BITMAP_CHUNK_SIZE)
> +		return -EINVAL;
> +
> +	eic = kzalloc(sizeof(*eic), GFP_KERNEL);
> +	if (!eic)
> +		return -EBUSY;
> +
> +	eic->buf = buf;
> +	eic->buf_size = count;
> +	eic->kvm = task_kvm(task);
> +	if (!eic->kvm) {
> +		ret = -EINVAL;
> +		goto out_free;
> +	}
I think you need to increment the refcount while using kvm, otherwise kvm can be destroyed from another thread while you're walking it.

-Nikita
> +
> +	ret = ept_idle_walk_hva_range(eic, hva_start, hva_end);
> +	if (ret)
> +		goto out_free;
> +
> +	ret = eic->bytes_copied;
> +	*ppos += ret;
> +out_free:
> +	kfree(eic);
> +
> +	return ret;
> +}