[dpdk-dev] [PATCH v7 0/8] implement packed virtqueues

[dpdk-dev] [PATCH v7 0/8] implement packed virtqueues

[Jens Freimann]

I'm sending this out to get some comments especially on the TX path code. 

I added support for mergeable rx buffers, out of order processing and
indirect. The receive path works well but the TX path sometimes locks up
after a random number of packets transmitted, so please review this
patch extra careful. This is also why I didn't add any new performance
numbers to the cover letter yet.

To support out-of-order processing I used the vq_desc_extra struct to
add a .next field and use it as a list for managing descriptors. This
seemed to add less complexity to the code than adding a new data
structure to use as a list for packed queue descriptors. 

I also took out the patch for supporting virtio-user as it turned out
more complex than expected. I will try to get it working for the next
version, but if I don't can we add it in a later pach set (and then
probably not in 18.11?]

This is a basic implementation of packed virtqueues as specified in the
Virtio 1.1 draft. A compiled version of the current draft is available
at https://github.com/oasis-tcs/virtio-docs.git (or as .pdf at
https://github.com/oasis-tcs/virtio-docs/blob/master/virtio-v1.1-packed-wd10.pdf

A packed virtqueue is different from a split virtqueue in that it
consists of only a single descriptor ring that replaces available and
used ring, index and descriptor pointers.

Each descriptor is readable and writable and has a flags field. These flags
will mark if a descriptor is available or used.  To detect new available descriptors
even after the ring has wrapped, device and driver each have a
single-bit wrap counter that is flipped from 0 to 1 and vice versa every time
the last descriptor in the ring is used/made available.


Jens Freimann (8):
  net/virtio: vring init for packed queues
  net/virtio: add packed virtqueue defines
  net/virtio: add packed virtqueue helpers
  net/virtio: dump packed virtqueue data
  net/virtio: implement transmit path for packed queues
  net/virtio: implement receive path for packed queues
  net/virtio: add virtio send command packed queue support
  net/virtio: enable packed virtqueues by default

 drivers/net/virtio/virtio_ethdev.c | 161 +++++++--
 drivers/net/virtio/virtio_ethdev.h |   5 +
 drivers/net/virtio/virtio_pci.h    |   8 +
 drivers/net/virtio/virtio_ring.h   |  96 ++++-
 drivers/net/virtio/virtio_rxtx.c   | 544 ++++++++++++++++++++++++++++-
 drivers/net/virtio/virtqueue.c     |  23 ++
 drivers/net/virtio/virtqueue.h     |  52 ++-
 7 files changed, 846 insertions(+), 43 deletions(-)

-- 
2.17.1

[dpdk-dev] [PATCH v7 0/8] implement packed virtqueues

[Jens Freimann]

I'm sending this out to get some comments especially on the TX path code. 

I added support for mergeable rx buffers, out of order processing and
indirect. The receive path works well but the TX path sometimes locks up
after a random number of packets transmitted, so please review this
patch extra careful. This is also why I didn't add any new performance
numbers to the cover letter yet.

To support out-of-order processing I used the vq_desc_extra struct to
add a .next field and use it as a list for managing descriptors. This
seemed to add less complexity to the code than adding a new data
structure to use as a list for packed queue descriptors. 

I also took out the patch for supporting virtio-user as it turned out
more complex than expected. I will try to get it working for the next
version, but if I don't can we add it in a later pach set (and then
probably not in 18.11?]

This is a basic implementation of packed virtqueues as specified in the
Virtio 1.1 draft. A compiled version of the current draft is available
at https://github.com/oasis-tcs/virtio-docs.git (or as .pdf at
https://github.com/oasis-tcs/virtio-docs/blob/master/virtio-v1.1-packed-wd10.pdf

A packed virtqueue is different from a split virtqueue in that it
consists of only a single descriptor ring that replaces available and
used ring, index and descriptor pointers.

Each descriptor is readable and writable and has a flags field. These flags
will mark if a descriptor is available or used.  To detect new available descriptors
even after the ring has wrapped, device and driver each have a
single-bit wrap counter that is flipped from 0 to 1 and vice versa every time
the last descriptor in the ring is used/made available.


Jens Freimann (8):
  net/virtio: vring init for packed queues
  net/virtio: add packed virtqueue defines
  net/virtio: add packed virtqueue helpers
  net/virtio: dump packed virtqueue data
  net/virtio: implement transmit path for packed queues
  net/virtio: implement receive path for packed queues
  net/virtio: add virtio send command packed queue support
  net/virtio: enable packed virtqueues by default

 drivers/net/virtio/virtio_ethdev.c | 161 +++++++--
 drivers/net/virtio/virtio_ethdev.h |   5 +
 drivers/net/virtio/virtio_pci.h    |   8 +
 drivers/net/virtio/virtio_ring.h   |  96 ++++-
 drivers/net/virtio/virtio_rxtx.c   | 544 ++++++++++++++++++++++++++++-
 drivers/net/virtio/virtqueue.c     |  23 ++
 drivers/net/virtio/virtqueue.h     |  52 ++-
 7 files changed, 846 insertions(+), 43 deletions(-)

-- 
2.17.1

Re: [dpdk-dev] [PATCH v7 0/8] implement packed virtqueues

[Jens Freimann]

On Wed, Oct 03, 2018 at 03:11:10PM +0200, Jens Freimann wrote:
>I'm sending this out to get some comments especially on the TX path code.

Tested with this QEMU: https://github.com/jensfr/qemu/pull/new/wexu-packed-ring-v1-fixed

Re: [dpdk-dev] [PATCH v7 0/8] implement packed virtqueues

[Maxime Coquelin]

Hi Jens,

On 10/03/2018 03:11 PM, Jens Freimann wrote:
> To support out-of-order processing I used the vq_desc_extra struct to
> add a .next field and use it as a list for managing descriptors. This
> seemed to add less complexity to the code than adding a new data
> structure to use as a list for packed queue descriptors.

Looking at the series, I don't see a specific path when in-order has
been negotiated. Is it intended?
Wouldn't we save some cache misses when assuming packets are processed
in-order?

Maxime

[dpdk-dev] [PATCH v7 0/8] implement packed virtqueues

[Jens Freimann]

I'm sending this out to get some comments especially on the TX path code. 

I added support for mergeable rx buffers, out of order processing and
indirect. The receive path works well but the TX path sometimes locks up
after a random number of packets transmitted, so please review this
patch extra careful. This is also why I didn't add any new performance
numbers to the cover letter yet.

To support out-of-order processing I used the vq_desc_extra struct to
add a .next field and use it as a list for managing descriptors. This
seemed to add less complexity to the code than adding a new data
structure to use as a list for packed queue descriptors. 

I also took out the patch for supporting virtio-user as it turned out
more complex than expected. I will try to get it working for the next
version, but if I don't can we add it in a later pach set (and then
probably not in 18.11?]

This is a basic implementation of packed virtqueues as specified in the
Virtio 1.1 draft. A compiled version of the current draft is available
at https://github.com/oasis-tcs/virtio-docs.git (or as .pdf at
https://github.com/oasis-tcs/virtio-docs/blob/master/virtio-v1.1-packed-wd10.pdf

A packed virtqueue is different from a split virtqueue in that it
consists of only a single descriptor ring that replaces available and
used ring, index and descriptor pointers.

Each descriptor is readable and writable and has a flags field. These flags
will mark if a descriptor is available or used.  To detect new available descriptors
even after the ring has wrapped, device and driver each have a
single-bit wrap counter that is flipped from 0 to 1 and vice versa every time
the last descriptor in the ring is used/made available.


Jens Freimann (8):
  net/virtio: vring init for packed queues
  net/virtio: add packed virtqueue defines
  net/virtio: add packed virtqueue helpers
  net/virtio: dump packed virtqueue data
  net/virtio: implement transmit path for packed queues
  net/virtio: implement receive path for packed queues
  net/virtio: add virtio send command packed queue support
  net/virtio: enable packed virtqueues by default

 drivers/net/virtio/virtio_ethdev.c | 161 +++++++--
 drivers/net/virtio/virtio_ethdev.h |   5 +
 drivers/net/virtio/virtio_pci.h    |   8 +
 drivers/net/virtio/virtio_ring.h   |  96 ++++-
 drivers/net/virtio/virtio_rxtx.c   | 544 ++++++++++++++++++++++++++++-
 drivers/net/virtio/virtqueue.c     |  23 ++
 drivers/net/virtio/virtqueue.h     |  52 ++-
 7 files changed, 846 insertions(+), 43 deletions(-)

-- 
2.17.1