Message-ID: <1431541996.3625.55.camel@redhat.com>
Subject: Re: [PATCH 4/5] VFIO: platform: populate reset function according
 to compat
From: Alex Williamson <alex.williamson@redhat.com>
To: Eric Auger <eric.auger@linaro.org>
Cc: eric.auger@st.com, christoffer.dall@linaro.org,
        linux-arm-kernel@lists.infradead.org, kvmarm@lists.cs.columbia.edu,
        kvm@vger.kernel.org, b.reynal@virtualopensystems.com,
        linux-kernel@vger.kernel.org, patches@linaro.org, agraf@suse.de,
        Bharat.Bhushan@freescale.com
Date: Wed, 13 May 2015 12:33:16 -0600
In-Reply-To: <1431008843-28411-5-git-send-email-eric.auger@linaro.org>
References: <1431008843-28411-1-git-send-email-eric.auger@linaro.org>
	 <1431008843-28411-5-git-send-email-eric.auger@linaro.org>
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3902
Lines: 127

On Thu, 2015-05-07 at 16:27 +0200, Eric Auger wrote:
> Add the reset function lookup according to the device compat
> string. This lookup is added at different places:
> - on VFIO_DEVICE_GET_INFO
> - on VFIO_DEVICE_RESET
> - on device release
> 
> A reference to the module implementing the reset function is taken
> on first reset function lookup and released on vfio platform device
> release.
> 
> Signed-off-by: Eric Auger <eric.auger@linaro.org>
> ---
>  drivers/vfio/platform/vfio_platform_common.c | 50 ++++++++++++++++++++++++++++
>  1 file changed, 50 insertions(+)
> 
> diff --git a/drivers/vfio/platform/vfio_platform_common.c b/drivers/vfio/platform/vfio_platform_common.c
> index 0d10018..bd7e44c 100644
> --- a/drivers/vfio/platform/vfio_platform_common.c
> +++ b/drivers/vfio/platform/vfio_platform_common.c
> @@ -28,6 +28,52 @@ LIST_HEAD(reset_list);
>  
>  static DEFINE_MUTEX(driver_lock);
>  
> +static vfio_platform_reset_fn_t vfio_platform_lookup_reset(const char *compat)
> +{
> +	struct vfio_platform_reset_node *iter;
> +
> +	list_for_each_entry(iter, &reset_list, link) {

Racy

> +		if (!strcmp(iter->compat, compat) &&
> +			try_module_get(iter->owner))
> +			return iter->reset;
> +	}
> +
> +	return ERR_PTR(-ENODEV);

return NULL imo

> +}
> +
> +static vfio_platform_reset_fn_t vfio_platform_get_reset(
> +					struct vfio_platform_device *vdev)
> +{
> +	struct device *dev = vdev->get_device(vdev);
> +	const char *compat_str_array[2];
> +	vfio_platform_reset_fn_t reset;
> +	int ret;
> +
> +	if (!IS_ERR_OR_NULL(vdev->reset))
> +		return vdev->reset;
> +
> +	ret = device_property_read_string_array(dev, "compatible",
> +						compat_str_array, 2);
> +	if (!ret)
> +		return ERR_PTR(ret);
> +
> +	reset = vfio_platform_lookup_reset(compat_str_array[0]);
> +	return reset;

Something got allocated into compat_str_array and gets leaked here.

> +}
> +
> +static void vfio_platform_put_reset(struct vfio_platform_device *vdev)
> +{
> +	struct vfio_platform_reset_node *iter;
> +
> +	list_for_each_entry(iter, &reset_list, link) {

Racy

> +		if (iter->reset == vdev->reset) {
> +			module_put(iter->owner);
> +			vdev->reset = NULL;
> +			return;
> +		}
> +	}
> +}
> +
>  static int vfio_platform_regions_init(struct vfio_platform_device *vdev)
>  {
>  	int cnt = 0, i;
> @@ -103,10 +149,12 @@ static void vfio_platform_release(void *device_data)
>  	mutex_lock(&driver_lock);
>  
>  	if (!(--vdev->refcnt)) {
> +		vdev->reset = vfio_platform_get_reset(vdev);
>  		if (!IS_ERR_OR_NULL(vdev->reset))
>  			vdev->reset(vdev);
>  		vfio_platform_regions_cleanup(vdev);
>  		vfio_platform_irq_cleanup(vdev);
> +		vfio_platform_put_reset(vdev);
>  	}
>  
>  	mutex_unlock(&driver_lock);
> @@ -164,6 +212,7 @@ static long vfio_platform_ioctl(void *device_data,
>  		if (info.argsz < minsz)
>  			return -EINVAL;
>  
> +		vdev->reset = vfio_platform_get_reset(vdev);
>  		if (!IS_ERR_OR_NULL(vdev->reset))
>  			vdev->flags |= VFIO_DEVICE_FLAGS_RESET;
>  		info.flags = vdev->flags;
> @@ -260,6 +309,7 @@ static long vfio_platform_ioctl(void *device_data,
>  		return ret;
>  
>  	} else if (cmd == VFIO_DEVICE_RESET) {
> +		vdev->reset = vfio_platform_get_reset(vdev);
>  		if (!IS_ERR_OR_NULL(vdev->reset))
>  			return vdev->reset(vdev);
>  		else

I count 3 gets and 1 put, isn't the module reference count increase
showing that?  This looks like it hasn't been tested.  Why would we do a
get every time we want to do a reset?  Do one get when the device is
registered or opened and one put when the device is unregistered or
closed.  We don't want erratic userspace behavior that the reset
property of a device can come and go.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/