Age | Commit message (Collapse) | Author |
|
This one is just an old copy of linux-generic, let's use the current one. Later
we can start off our own version if necessary.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Although DPKD will still print its output to stdout.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Based on the linux-generic implementation.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Based on the linux-generic implementation.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Based on the linux-generic implementation.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
And two related functions. This is based on linux-generic implementation,
performance could be improved in the future, e.g. it's using odp_packet_offset
to get the offset for every segment, which unnecessarily iterates through all
the segments from the beginning, every time.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Otherwise the next patch would like quite unreadable.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
rte_mbuf is already aligned, odp_packet_hdr_t is right between it and the head
of the buffer.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Apart from calling rte_eth_tx_burst the only thing it does is to copy the
pkt_table[] array into an another one. As far as I can determine DPDK doesn't
touch this array of pointers, so it's not necessary to copy it. I've merged the
rest of it into odp_pktio_send()
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
- return the pointer to the correct segment
- refactor duplicate code into a common function
- fix offset setting error check
- set the offset to their default value during packet receive
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
For odp_packet_input() pktio has to be adjusted as well to store the
odp_pktio_t in the pool structure.
A few header file dependencies are also fixed.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
According to the relevan section of include/odp/api/packet.h.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Actually it just needed a new implementation of odp_hdr_to_buf() to be in an
own odp_buffer_inlines.h, and an extra include.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
- set the data length from the parameter
- check that parameter
- and the return value during allocation as well
- use headroom as well
- memset only the header, not the whole packet
- reset rte_mbuf and rte_pktmbuf values as well for each segment
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
DPDK doesn't support pool destroy at the moment. Instead we should improve
odp_pool_create() to try to reuse pools. See this thread:
http://thread.gmane.org/gmane.comp.networking.dpdk.devel/16498
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Packet buffers remain the same. Save the ODP type as well.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Also translate the error scenario correctly
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
These changes are almost the same as with linux-generic, except that we make
sure the headroom for packets is at leas RTE_PKTMBUF_HEADROOM. Debug logs are
updated accordingly, and timeout type is no longer unimplemented.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
ODP_CONFIG_PACKET_HEADROOM is a minimum, we should check that it's smaller or
equal to RTE_PKTMBUF_HEADROOM.
ODP_CONFIG_PACKET_TAILROOM has to be 0, DPDK doesn't explicitly support
mandating tailroom size for received packets.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Save the pool handle to pool and buffer metadata areas so we can return it in
odp_buffer_pool.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
That's what the callers expect.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
According to the DPDK API, this value must be equal or smaller than
RTE_MEMPOOL_CACHE_MAX_SIZE. Also, it is recommended to be a divisor of the
number of elements.
This algorithm tries to find the greatest divisor which is smaller than the
max allowed, if there isn't any, the max is used.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Re-use the stubs that were used to develop linux-generic
Changed the return behaviour to be a fault
Original commit was 5c7aed2c
Signed-off-by: Mike Holmes <mike.holmes@linaro.org>
|
|
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
This reverts commit 3925e6a28da232eecc9a82cf8fdf61847d6b4896.
Conflicts:
platform/linux-dpdk/odp_pool.c
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
To store the number of buffers requested.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Also add an overflow check, and make the timeout type implementation
unimplemented for the time being.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
This is no longer a valid type.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
|
|
Ran checkpatch on platform/linux-dpdk to fix up some accumulated code
style problems, mostly indentation. This patch doesn't contain any
functional changes.
Signed-off-by: Zoltan Kiss <zoltan.kiss@linaro.org>
Reviewed-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
|
|
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
- This function is currently unimplemented
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
- It internally implements,
queue_enq_dummy
queue_enq_multi_dummy and
queue_deq_multi_destroy
Port of 72011097
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Port of d9a7f995
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Port of b776ca47, de37ffa4, 6c0c7f72
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Port of 5e47a0cf
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
- currently this function is unimplemented
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Port of d3066cb4
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Port of e822279e
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Port of a49b7b64
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|
|
Port of 37514635
Signed-off-by: Venkatesh Vivekanandan <venkatesh.vivekanandan@linaro.org>
Acked-by: Zoltan Kiss <zoltan.kiss.org>
|