[dpdk-dev,v2,6/7] eal: pci: export pci_map_device

  Normally we could set RTE_PCI_DRV_NEED_MAPPING flag so that eal will
invoke pci_map_device internally for us. From that point view, there
is no need to export pci_map_device.

However, for virtio pmd driver, which is designed to work without
binding UIO (or something similar first), pci_map_device() will fail,
which ends up with virtio pmd driver being skipped. Therefore, we can
not set RTE_PCI_DRV_NEED_MAPPING blindly at virtio pmd driver.

Therefore, this patch exports pci_map_device, and let virtio pmd
call it when necessary.

Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
---
 lib/librte_eal/bsdapp/eal/eal_pci.c             |  2 +-
 lib/librte_eal/bsdapp/eal/rte_eal_version.map   |  6 ++++++
 lib/librte_eal/common/eal_common_pci.c          |  2 +-
 lib/librte_eal/common/eal_private.h             | 11 -----------
 lib/librte_eal/common/include/rte_pci.h         | 11 +++++++++++
 lib/librte_eal/linuxapp/eal/eal_pci.c           |  2 +-
 lib/librte_eal/linuxapp/eal/rte_eal_version.map |  6 ++++++
 7 files changed, 26 insertions(+), 14 deletions(-)
  

On Tue, Jan 12, 2016 at 7:59 AM, Yuanhan Liu <yuanhan.liu@linux.intel.com>
wrote:

> Normally we could set RTE_PCI_DRV_NEED_MAPPING flag so that eal will
> invoke pci_map_device internally for us. From that point view, there
> is no need to export pci_map_device.
>
> However, for virtio pmd driver, which is designed to work without
> binding UIO (or something similar first), pci_map_device() will fail,
> which ends up with virtio pmd driver being skipped. Therefore, we can
> not set RTE_PCI_DRV_NEED_MAPPING blindly at virtio pmd driver.
>
> Therefore, this patch exports pci_map_device, and let virtio pmd
> call it when necessary.
>

Well, if you introduce map function, I suppose, for hotplug, you would need
unmap.

[snip]

diff --git a/lib/librte_eal/common/include/rte_pci.h
> b/lib/librte_eal/common/include/rte_pci.h
> index 334c12e..e9e1725 100644
> --- a/lib/librte_eal/common/include/rte_pci.h
> +++ b/lib/librte_eal/common/include/rte_pci.h
> @@ -485,6 +485,17 @@ int rte_eal_pci_read_config(const struct
> rte_pci_device *device,
>   */
>  int rte_eal_pci_write_config(const struct rte_pci_device *device,
>                              const void *buf, size_t len, off_t offset);
> +/**
> + * Map this device
> + *
> + * This function is private to EAL.
> + *
> + * @return
> + *   0 on success, negative on error and positive if no driver
> + *   is found for the device.
> + */
> +int rte_eal_pci_map_device(struct rte_pci_device *dev);
> +
>

If you export it, then this can not be marked as private anymore.
Description could be better (I agree it was not that great before).
And a little comment on when to call: driver should not set
RTE_PCI_DRV_NEED_MAPPING flag if it wants to use it.

The rest looks good to me.
  

On Tue, Jan 12, 2016 at 09:31:05AM +0100, David Marchand wrote:
> On Tue, Jan 12, 2016 at 7:59 AM, Yuanhan Liu <yuanhan.liu@linux.intel.com>
> wrote:
> 
>     Normally we could set RTE_PCI_DRV_NEED_MAPPING flag so that eal will
>     invoke pci_map_device internally for us. From that point view, there
>     is no need to export pci_map_device.
> 
>     However, for virtio pmd driver, which is designed to work without
>     binding UIO (or something similar first), pci_map_device() will fail,
>     which ends up with virtio pmd driver being skipped. Therefore, we can
>     not set RTE_PCI_DRV_NEED_MAPPING blindly at virtio pmd driver.
> 
>     Therefore, this patch exports pci_map_device, and let virtio pmd
>     call it when necessary.
> 
> 
> Well, if you introduce map function, I suppose, for hotplug, you would need
> unmap.

Good remind. Thanks. I will export pci_unmap_device as well.

> [snip]
> 
> 
>     diff --git a/lib/librte_eal/common/include/rte_pci.h b/lib/librte_eal/
>     common/include/rte_pci.h
>     index 334c12e..e9e1725 100644
>     --- a/lib/librte_eal/common/include/rte_pci.h
>     +++ b/lib/librte_eal/common/include/rte_pci.h
>     @@ -485,6 +485,17 @@ int rte_eal_pci_read_config(const struct
>     rte_pci_device *device,
>       */
>      int rte_eal_pci_write_config(const struct rte_pci_device *device,
>                                  const void *buf, size_t len, off_t offset);
>     +/**
>     + * Map this device
>     + *
>     + * This function is private to EAL.
>     + *
>     + * @return
>     + *   0 on success, negative on error and positive if no driver
>     + *   is found for the device.
>     + */
>     +int rte_eal_pci_map_device(struct rte_pci_device *dev);
>     +
> 
> 
> If you export it, then this can not be marked as private anymore.

Oops, a silly C&P error. Will fix it.

> Description could be better (I agree it was not that great before).
> And a little comment on when to call: driver should not set
> RTE_PCI_DRV_NEED_MAPPING flag if it wants to use it.

Good suggestion.

> The rest looks good to me.

Thanks.

	--yliu