* [dpdk-dev] [PATCH] ethdev: The users could get device types now
@ 2014-09-28 3:27 Ding Heng
2014-09-28 3:50 ` Zhang, Helin
2014-09-28 12:36 ` Neil Horman
0 siblings, 2 replies; 4+ messages in thread
From: Ding Heng @ 2014-09-28 3:27 UTC (permalink / raw)
To: dev
As different PMDs support different features, application may want
to know the NIC type of a port, then decide how to use that port.
Add a new field in struct rte_eth_dev, users are able to get
device type now.
Signed-off-by: Ding Heng <hengx.ding@intel.com>
---
lib/librte_ether/rte_ethdev.c | 32 ++++++++++++++++++++++++++++++++
lib/librte_ether/rte_ethdev.h | 20 ++++++++++++++++++++
2 files changed, 52 insertions(+)
diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
index fd1010a..490ea3a 100644
--- a/lib/librte_ether/rte_ethdev.c
+++ b/lib/librte_ether/rte_ethdev.c
@@ -235,6 +235,38 @@ rte_eth_dev_init(struct rte_pci_driver *pci_drv,
if (diag == 0)
return (0);
+ /* Define the device type */
+ if (strcmp(eth_drv->pci_drv.name, "rte_i40e_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_I40E;
+ else if (strcmp(eth_drv->pci_drv.name, "rte_i40evf_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_I40E_VF;
+ else if (strcmp(eth_drv->pci_drv.name, "rte_em_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_EM;
+ else if (strcmp(eth_drv->pci_drv.name, "rte_igb_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_IGB;
+ else if (strcmp(eth_drv->pci_drv.name, "rte_igbvf_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_IGB_VF;
+ else if (strcmp(eth_drv->pci_drv.name, "rte_ixgbe_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_IXGBE;
+ else if (strcmp(eth_drv->pci_drv.name, "rte_ixgbevf_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_IXGBE_VF;
+ else if (strcmp(eth_drv->pci_drv.name, "eth_bond") == 0)
+ eth_dev->dev_type = ETH_DEV_BOND;
+ else if (strcmp(eth_drv->pci_drv.name, "eth_ring") == 0)
+ eth_dev->dev_type = ETH_DEV_RING;
+ else if (strcmp(eth_drv->pci_drv.name, "rte_virtio_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_VIRTIO;
+ else if (strcmp(eth_drv->pci_drv.name, "rte_vmxnet3_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_VMXNET3;
+ else if (strcmp(eth_drv->pci_drv.name, "eth_xenvirt") == 0)
+ eth_dev->dev_type = ETH_DEV_XENVIRT;
+ else if (strcmp(eth_drv->pci_drv.name, "rte_max_pmd") == 0)
+ eth_dev->dev_type = ETH_DEV_MAX;
+ else if (strcmp(eth_drv->pci_drv.name, "eth_pcap") == 0)
+ eth_dev->dev_type = ETH_DEV_PCAP;
+ else
+ eth_dev->dev_type = ETH_DEV_UNKNOWN;
+
PMD_DEBUG_TRACE("driver %s: eth_dev_init(vendor_id=0x%u device_id=0x%x)"
" failed\n", pci_drv->name,
(unsigned) pci_dev->id.vendor_id,
diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index 50df654..83a7ea5 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -268,6 +268,25 @@ enum rte_eth_rx_mq_mode {
ETH_MQ_RX_VMDQ_DCB_RSS, /**< Enable both VMDQ and DCB in VMDq */
};
+/*Device type*/
+enum rte_eth_dev_type {
+ ETH_DEV_UNKNOWN = 0,
+ ETH_DEV_EM,
+ ETH_DEV_IGB,
+ ETH_DEV_IGB_VF,
+ ETH_DEV_IXGBE,
+ ETH_DEV_IXGBE_VF,
+ ETH_DEV_I40E,
+ ETH_DEV_I40E_VF,
+ ETH_DEV_BOND,
+ ETH_DEV_PCAP,
+ ETH_DEV_RING,
+ ETH_DEV_VIRTIO,
+ ETH_DEV_VMXNET3,
+ ETH_DEV_XENVIRT,
+ ETH_DEV_MAX
+};
+
/**
* for rx mq mode backward compatible
*/
@@ -1487,6 +1506,7 @@ struct rte_eth_dev {
struct eth_dev_ops *dev_ops; /**< Functions exported by PMD */
struct rte_pci_device *pci_dev; /**< PCI info. supplied by probing */
struct rte_eth_dev_cb_list callbacks; /**< User application callbacks */
+ enum rte_eth_dev_type dev_type; /**< Device type info. */
};
struct rte_eth_dev_sriov {
--
1.9.3
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [dpdk-dev] [PATCH] ethdev: The users could get device types now
2014-09-28 3:27 [dpdk-dev] [PATCH] ethdev: The users could get device types now Ding Heng
@ 2014-09-28 3:50 ` Zhang, Helin
2014-09-28 12:36 ` Neil Horman
1 sibling, 0 replies; 4+ messages in thread
From: Zhang, Helin @ 2014-09-28 3:50 UTC (permalink / raw)
To: Ding, HengX, dev
Acked-by: Helin Zhang <helin.zhang@intel.com>
> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Ding Heng
> Sent: Sunday, September 28, 2014 11:27 AM
> To: dev@dpdk.org
> Subject: [dpdk-dev] [PATCH] ethdev: The users could get device types now
>
> As different PMDs support different features, application may want to know
> the NIC type of a port, then decide how to use that port.
> Add a new field in struct rte_eth_dev, users are able to get device type now.
>
> Signed-off-by: Ding Heng <hengx.ding@intel.com>
> ---
> lib/librte_ether/rte_ethdev.c | 32 ++++++++++++++++++++++++++++++++
> lib/librte_ether/rte_ethdev.h | 20 ++++++++++++++++++++
> 2 files changed, 52 insertions(+)
>
> diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c index
> fd1010a..490ea3a 100644
> --- a/lib/librte_ether/rte_ethdev.c
> +++ b/lib/librte_ether/rte_ethdev.c
> @@ -235,6 +235,38 @@ rte_eth_dev_init(struct rte_pci_driver *pci_drv,
> if (diag == 0)
> return (0);
>
> + /* Define the device type */
> + if (strcmp(eth_drv->pci_drv.name, "rte_i40e_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_I40E;
> + else if (strcmp(eth_drv->pci_drv.name, "rte_i40evf_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_I40E_VF;
> + else if (strcmp(eth_drv->pci_drv.name, "rte_em_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_EM;
> + else if (strcmp(eth_drv->pci_drv.name, "rte_igb_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_IGB;
> + else if (strcmp(eth_drv->pci_drv.name, "rte_igbvf_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_IGB_VF;
> + else if (strcmp(eth_drv->pci_drv.name, "rte_ixgbe_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_IXGBE;
> + else if (strcmp(eth_drv->pci_drv.name, "rte_ixgbevf_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_IXGBE_VF;
> + else if (strcmp(eth_drv->pci_drv.name, "eth_bond") == 0)
> + eth_dev->dev_type = ETH_DEV_BOND;
> + else if (strcmp(eth_drv->pci_drv.name, "eth_ring") == 0)
> + eth_dev->dev_type = ETH_DEV_RING;
> + else if (strcmp(eth_drv->pci_drv.name, "rte_virtio_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_VIRTIO;
> + else if (strcmp(eth_drv->pci_drv.name, "rte_vmxnet3_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_VMXNET3;
> + else if (strcmp(eth_drv->pci_drv.name, "eth_xenvirt") == 0)
> + eth_dev->dev_type = ETH_DEV_XENVIRT;
> + else if (strcmp(eth_drv->pci_drv.name, "rte_max_pmd") == 0)
> + eth_dev->dev_type = ETH_DEV_MAX;
> + else if (strcmp(eth_drv->pci_drv.name, "eth_pcap") == 0)
> + eth_dev->dev_type = ETH_DEV_PCAP;
> + else
> + eth_dev->dev_type = ETH_DEV_UNKNOWN;
> +
> PMD_DEBUG_TRACE("driver %s: eth_dev_init(vendor_id=0x%u
> device_id=0x%x)"
> " failed\n", pci_drv->name,
> (unsigned) pci_dev->id.vendor_id,
> diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h index
> 50df654..83a7ea5 100644
> --- a/lib/librte_ether/rte_ethdev.h
> +++ b/lib/librte_ether/rte_ethdev.h
> @@ -268,6 +268,25 @@ enum rte_eth_rx_mq_mode {
> ETH_MQ_RX_VMDQ_DCB_RSS, /**< Enable both VMDQ and DCB in
> VMDq */ };
>
> +/*Device type*/
> +enum rte_eth_dev_type {
> + ETH_DEV_UNKNOWN = 0,
> + ETH_DEV_EM,
> + ETH_DEV_IGB,
> + ETH_DEV_IGB_VF,
> + ETH_DEV_IXGBE,
> + ETH_DEV_IXGBE_VF,
> + ETH_DEV_I40E,
> + ETH_DEV_I40E_VF,
> + ETH_DEV_BOND,
> + ETH_DEV_PCAP,
> + ETH_DEV_RING,
> + ETH_DEV_VIRTIO,
> + ETH_DEV_VMXNET3,
> + ETH_DEV_XENVIRT,
> + ETH_DEV_MAX
> +};
> +
> /**
> * for rx mq mode backward compatible
> */
> @@ -1487,6 +1506,7 @@ struct rte_eth_dev {
> struct eth_dev_ops *dev_ops; /**< Functions exported by PMD */
> struct rte_pci_device *pci_dev; /**< PCI info. supplied by probing */
> struct rte_eth_dev_cb_list callbacks; /**< User application callbacks */
> + enum rte_eth_dev_type dev_type; /**< Device type info. */
> };
>
> struct rte_eth_dev_sriov {
> --
> 1.9.3
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [dpdk-dev] [PATCH] ethdev: The users could get device types now
2014-09-28 3:27 [dpdk-dev] [PATCH] ethdev: The users could get device types now Ding Heng
2014-09-28 3:50 ` Zhang, Helin
@ 2014-09-28 12:36 ` Neil Horman
2014-09-29 3:06 ` Stephen Hemminger
1 sibling, 1 reply; 4+ messages in thread
From: Neil Horman @ 2014-09-28 12:36 UTC (permalink / raw)
To: Ding Heng; +Cc: dev
On Sun, Sep 28, 2014 at 11:27:16AM +0800, Ding Heng wrote:
> As different PMDs support different features, application may want
> to know the NIC type of a port, then decide how to use that port.
> Add a new field in struct rte_eth_dev, users are able to get
> device type now.
>
> Signed-off-by: Ding Heng <hengx.ding@intel.com>
Nack. This patch embodies PMD specific information in a common library, which
really isn't necessecary. It implies that developers who maintain PMDs outside
of the core dpdk still need to do some maintenence in the core for all the dpdk
features to work. Its also a layering violation. The core shouldn't have to
know any specifics about a driver to initalize it, even its name.
If an application wants to know what type of driver a NIC is,
the application can just interrogate the pci drivers name field directly.
Neil
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [dpdk-dev] [PATCH] ethdev: The users could get device types now
2014-09-28 12:36 ` Neil Horman
@ 2014-09-29 3:06 ` Stephen Hemminger
0 siblings, 0 replies; 4+ messages in thread
From: Stephen Hemminger @ 2014-09-29 3:06 UTC (permalink / raw)
To: Neil Horman; +Cc: dev
Agree with NACK. This type of string matching won't scale.
THe better way is to encapsulate required behavior in feature flags.
I started that with drv_flags, and that seems natural home for such device
specific stuff.
On Sun, Sep 28, 2014 at 5:36 AM, Neil Horman <nhorman@tuxdriver.com> wrote:
> On Sun, Sep 28, 2014 at 11:27:16AM +0800, Ding Heng wrote:
> > As different PMDs support different features, application may want
> > to know the NIC type of a port, then decide how to use that port.
> > Add a new field in struct rte_eth_dev, users are able to get
> > device type now.
> >
> > Signed-off-by: Ding Heng <hengx.ding@intel.com>
> Nack. This patch embodies PMD specific information in a common library,
> which
> really isn't necessecary. It implies that developers who maintain PMDs
> outside
> of the core dpdk still need to do some maintenence in the core for all the
> dpdk
> features to work. Its also a layering violation. The core shouldn't
> have to
> know any specifics about a driver to initalize it, even its name.
>
> If an application wants to know what type of driver a NIC is,
> the application can just interrogate the pci drivers name field directly.
>
> Neil
>
>
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2014-09-29 2:59 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-09-28 3:27 [dpdk-dev] [PATCH] ethdev: The users could get device types now Ding Heng
2014-09-28 3:50 ` Zhang, Helin
2014-09-28 12:36 ` Neil Horman
2014-09-29 3:06 ` Stephen Hemminger
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).