[dpdk-dev] [PATCH v2 9/9] examples/l3fwd-power: fix not stop and close device
Tan, Jianfeng
jianfeng.tan at intel.com
Fri Dec 30 07:56:48 CET 2016
> -----Original Message-----
> From: Yuanhan Liu [mailto:yuanhan.liu at linux.intel.com]
> Sent: Friday, December 30, 2016 2:45 PM
> To: Tan, Jianfeng
> Cc: dev at dpdk.org; stephen at networkplumber.org
> Subject: Re: [PATCH v2 9/9] examples/l3fwd-power: fix not stop and close
> device
>
> On Thu, Dec 29, 2016 at 07:30:43AM +0000, Jianfeng Tan wrote:
> > As it gets killed, in SIGINT signal handler, device is not stopped
> > and closed. In virtio's case, vector assignment in the KVM is not
> > deassigned.
>
> What wrong could happen then?
Actually, no strange things happen so far. My purpose for this patch is to verify that irqfd is deassigned.
And as calling dev_stop() and dev_close() is a good practice, so I use the word "fix" here.
>
> > This patch will invoke dev_stop() and dev_close() in signal handler.
>
> I will just say, it may workaround the bug you encountered, particulary,
> for this example only.
There's no bug to work around.
> If some people want to use the virtio interrupt
> somewhere at another app, he also has to do similar thing.
Vfio-pci is the best place to put device into original state, but is it really necessary to do that? If another program takes over that device, it will be re-initialized.
>
> Is there a more clean way to handle such case in the driver?
Let's do with the necessity firstly.
Thanks,
Jianfeng
>
> --yliu
More information about the dev
mailing list