[dpdk-dev,v2,09/12] ethdev: support attach vdev in secondary process
Checks
Commit Message
When vdev driver requests an ethdev entry in secondary process,
we will identify the correct entry in rte_eth_dev_data array
and return the correct entry in the rte_eth_devices arrays.
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
---
lib/librte_ether/rte_ethdev_vdev.h | 26 +++++++++++++++-----------
1 file changed, 15 insertions(+), 11 deletions(-)
Comments
On Thu, Sep 28, 2017 at 3:55 PM, Jianfeng Tan <jianfeng.tan@intel.com> wrote:
> When vdev driver requests an ethdev entry in secondary process,
> we will identify the correct entry in rte_eth_dev_data array
> and return the correct entry in the rte_eth_devices arrays.
>
> Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
> ---
> lib/librte_ether/rte_ethdev_vdev.h | 26 +++++++++++++++-----------
> 1 file changed, 15 insertions(+), 11 deletions(-)
>
> diff --git a/lib/librte_ether/rte_ethdev_vdev.h b/lib/librte_ether/rte_ethdev_vdev.h
> index 4d2c3e2..460749b 100644
> --- a/lib/librte_ether/rte_ethdev_vdev.h
> +++ b/lib/librte_ether/rte_ethdev_vdev.h
> @@ -58,25 +58,29 @@ rte_eth_vdev_allocate(struct rte_vdev_device *dev, size_t private_data_size)
> struct rte_eth_dev *eth_dev;
> const char *name = rte_vdev_device_name(dev);
>
> - eth_dev = rte_eth_dev_allocate(name);
> - if (!eth_dev)
> - return NULL;
> -
> - if (private_data_size) {
> - eth_dev->data->dev_private = rte_zmalloc_socket(name,
> - private_data_size, RTE_CACHE_LINE_SIZE,
> - dev->device.numa_node);
> - if (!eth_dev->data->dev_private) {
> - rte_eth_dev_release_port(eth_dev);
> + if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
> + eth_dev = rte_eth_dev_allocate(name);
> + if (!eth_dev)
> return NULL;
> +
> + if (private_data_size) {
> + eth_dev->data->dev_private = rte_zmalloc_socket(name,
> + private_data_size, RTE_CACHE_LINE_SIZE,
> + dev->device.numa_node);
> + if (!eth_dev->data->dev_private) {
> + rte_eth_dev_release_port(eth_dev);
> + return NULL;
> + }
> }
> + } else {
> + eth_dev = rte_eth_dev_attach_secondary(name);
I don't see the point why the secondary process should call
rte_eth_vdev_allocate() in the first place. The driver needs to setup
the IPC anyway so it should just call rte_eth_dev_attach_secondary()
instead.
> }
>
> eth_dev->device = &dev->device;
> eth_dev->intr_handle = NULL;
> -
> eth_dev->data->kdrv = RTE_KDRV_NONE;
> eth_dev->data->numa_node = dev->device.numa_node;
> +
> return eth_dev;
> }
>
> --
> 2.7.4
>
Hi Jan,
> -----Original Message-----
> From: jblunck@gmail.com [mailto:jblunck@gmail.com] On Behalf Of Jan
> Blunck
> Sent: Thursday, October 5, 2017 10:27 PM
> To: Tan, Jianfeng
> Cc: dev; Richardson, Bruce; Ananyev, Konstantin; De Lara Guarch, Pablo;
> Thomas Monjalon; yliu@fridaylinux.org; Maxime Coquelin; Tetsuya Mukawa;
> Yigit, Ferruh
> Subject: Re: [dpdk-dev] [PATCH v2 09/12] ethdev: support attach vdev in
> secondary process
>
> On Thu, Sep 28, 2017 at 3:55 PM, Jianfeng Tan <jianfeng.tan@intel.com>
> wrote:
> > When vdev driver requests an ethdev entry in secondary process,
> > we will identify the correct entry in rte_eth_dev_data array
> > and return the correct entry in the rte_eth_devices arrays.
> >
> > Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
> > ---
> > lib/librte_ether/rte_ethdev_vdev.h | 26 +++++++++++++++-----------
> > 1 file changed, 15 insertions(+), 11 deletions(-)
> >
> > diff --git a/lib/librte_ether/rte_ethdev_vdev.h
> b/lib/librte_ether/rte_ethdev_vdev.h
> > index 4d2c3e2..460749b 100644
> > --- a/lib/librte_ether/rte_ethdev_vdev.h
> > +++ b/lib/librte_ether/rte_ethdev_vdev.h
> > @@ -58,25 +58,29 @@ rte_eth_vdev_allocate(struct rte_vdev_device
> *dev, size_t private_data_size)
> > struct rte_eth_dev *eth_dev;
> > const char *name = rte_vdev_device_name(dev);
> >
> > - eth_dev = rte_eth_dev_allocate(name);
> > - if (!eth_dev)
> > - return NULL;
> > -
> > - if (private_data_size) {
> > - eth_dev->data->dev_private = rte_zmalloc_socket(name,
> > - private_data_size, RTE_CACHE_LINE_SIZE,
> > - dev->device.numa_node);
> > - if (!eth_dev->data->dev_private) {
> > - rte_eth_dev_release_port(eth_dev);
> > + if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
> > + eth_dev = rte_eth_dev_allocate(name);
> > + if (!eth_dev)
> > return NULL;
> > +
> > + if (private_data_size) {
> > + eth_dev->data->dev_private = rte_zmalloc_socket(name,
> > + private_data_size, RTE_CACHE_LINE_SIZE,
> > + dev->device.numa_node);
> > + if (!eth_dev->data->dev_private) {
> > + rte_eth_dev_release_port(eth_dev);
> > + return NULL;
> > + }
> > }
> > + } else {
> > + eth_dev = rte_eth_dev_attach_secondary(name);
>
> I don't see the point why the secondary process should call
> rte_eth_vdev_allocate() in the first place. The driver needs to setup
> the IPC anyway so it should just call rte_eth_dev_attach_secondary()
> instead.
Hmm... make sense. Will fix it in the next version.
Thanks for the input.
Thanks,
Jianfeng
@@ -58,25 +58,29 @@ rte_eth_vdev_allocate(struct rte_vdev_device *dev, size_t private_data_size)
struct rte_eth_dev *eth_dev;
const char *name = rte_vdev_device_name(dev);
- eth_dev = rte_eth_dev_allocate(name);
- if (!eth_dev)
- return NULL;
-
- if (private_data_size) {
- eth_dev->data->dev_private = rte_zmalloc_socket(name,
- private_data_size, RTE_CACHE_LINE_SIZE,
- dev->device.numa_node);
- if (!eth_dev->data->dev_private) {
- rte_eth_dev_release_port(eth_dev);
+ if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
+ eth_dev = rte_eth_dev_allocate(name);
+ if (!eth_dev)
return NULL;
+
+ if (private_data_size) {
+ eth_dev->data->dev_private = rte_zmalloc_socket(name,
+ private_data_size, RTE_CACHE_LINE_SIZE,
+ dev->device.numa_node);
+ if (!eth_dev->data->dev_private) {
+ rte_eth_dev_release_port(eth_dev);
+ return NULL;
+ }
}
+ } else {
+ eth_dev = rte_eth_dev_attach_secondary(name);
}
eth_dev->device = &dev->device;
eth_dev->intr_handle = NULL;
-
eth_dev->data->kdrv = RTE_KDRV_NONE;
eth_dev->data->numa_node = dev->device.numa_node;
+
return eth_dev;
}