| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
DMA sync functions should use the same direction that was used by DMA
mapping. Use DMA_BIDIRECTIONAL for XDP_TX from regular RQ, which reuses
the same mapping that was used for RX, and DMA_TO_DEVICE for XDP_TX from
XSK RQ and XDP_REDIRECT, which establish a new mapping in this
direction. On the RX side, use the same direction that was used when
setting up the mapping (DMA_BIDIRECTIONAL for XDP, DMA_FROM_DEVICE
otherwise).
Also don't skip sync for device when establishing a DMA_FROM_DEVICE
mapping for RX, as some architectures (ARM) may require invalidating
caches before the device can use the mapping. It doesn't break the
bugfix made in
commit 0b7cfa4082fb ("net/mlx5e: Fix page DMA map/unmap attributes"),
since the bug happened on unmap.
Fixes: 0b7cfa4082fb ("net/mlx5e: Fix page DMA map/unmap attributes")
Fixes: b5503b994ed5 ("net/mlx5e: XDP TX forwarding support")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Fix a typo in the function name: mpqwe -> mpwqe (stands for multi-packet
work queue element).
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When merged with Linus tree, the cited patch below will cause the
following build warning:
In function 'fortify_memset_chk',
inlined from 'mlx5e_xmit_xdp_frame' at drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c:438:3:
include/linux/fortify-string.h:242:25: error: call to '__write_overflow_field' declared with attribute warning: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Werror=attribute-warning]
242 | __write_overflow_field(p_size_field, size);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Fix that by grouping the fields to memeset in struct_group() to avoid
the false alarm.
Fixes: 9ded70fa1d81 ("net/mlx5e: Don't prefill WQEs in XDP SQ in the multi buffer mode")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Suggested-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Link: https://lore.kernel.org/r/20220322172224.31849-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
|
|
|
|
|
|
|
| |
This commit enables passing multi buffer XDP frames to the TX handlers
on XDP_TX. Fragments are DMA synchronized to the device and queued to
the xdpi_fifo for a subsequent unmapping.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
| |
The next commit will add more indentation levels to mlx5e_xmit_xdp_buff.
To keep indentation minimal, unindent the else-block of the if-statement
by doing an early return.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
| |
xmit_xdp_frame is extended to support sending fragmented XDP frames. The
next commit will start using this functionality.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When MPWQE is disabled, mlx5e_open_xdpsq() prefills the common fields of
WQEs in the XDP SQ to save time when sending packets.
mlx5e_xmit_xdp_frame() runs on the prefilled fields, however, sending
multi buffer XDP frames would require changing some of these fields on a
per-packet basis. Besides that, mlx5e_xmit_xdp_frame() will be used as a
fallback to send multi buffer XDP frames when MPWQE is enabled (MPWQE
can only handle linear packets).
In order to prepare for XDP multi buffer support, this commit introduces
a mode for mlx5e_xmit_xdp_frame() that fills all the fields itself.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When MPWQE is disabled, mlx5e_open_xdpsq prefills the common fields of
WQEs in the XDP SQ to save time when sending packets. One of such fields
is eseg->inline_hdr.sz, which can be either 0 or MLX5E_XDP_MIN_INLINE,
depending on the inline mode of the SQ.
The inline mode can't change during the lifetime of the SQ, so setting
this field again in mlx5e_xmit_xdp_frame is redundant. Moreover, the
xmit function only sets it to MLX5E_XDP_MIN_INLINE, but not to 0 in the
other case.
This commit removes the redundant assignment in mlx5e_xmit_xdp_frame.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The implementations of xmit_xdp_frame get the xdpi parameter of type
struct mlx5e_xdp_info for the sole purpose of calling
mlx5e_xdpi_fifo_push() on success.
This commit moves this call outside of xmit_xdp_frame, shifting this
responsibility to the caller. It will allow more fine-grained handling
of XDP info for cases when an xdp_frame is fragmented.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Use page_pool_set_dma_addr() to store the DMA address of a page inside
struct page, in order to avoid passing struct mlx5e_dma_info to XDP
handlers. Previously, struct mlx5e_dma_info was used to pass both the
DMA address and the page, and it worked well for the single-fragment
case.
When XDP multi buffer is in use, and a fragmented xdp_frame has to be
transmitted, the driver needs to know the DMA addresses of fragments,
however, the array of fragments in struct skb_shared_info doesn't
contain them. In order to pass the DMA addresses, the driver puts them
into struct page itself, which is accessible from the array of fragments
in struct skb_shared_info. The existing XDP handlers are modified to
remove the dependency on struct mlx5e_dma_info.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit adds XDP multi buffer support to the RX path in the
non-linear legacy RQ mode. mlx5e_xdp_handle is called from
mlx5e_skb_from_cqe_nonlinear.
XDP_TX action for fragmented XDP frames is not yet supported and
blocked.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The len parameter of mlx5e_xdp_handle is used to output the new packet
length after XDP has processed the packet and returned XDP_PASS.
However, this value can be calculated on the caller site, as the caller
knows if it was an XDP_PASS.
This commit drops the len parameter and moves the calculation to the
caller, reducing the number of parameters passed to the function and
preparing for XDP support in non-linear legacy RQ.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Instead of early return inside mlx5e_xdp_handle(), let the caller check
if an XDP program is loaded. This allows saving a few unnecessary
function calls and calculations in case !prog.
Performance test: single core, drop packets in iptables
Before: 3,872,504 pps
After: 3,975,628 pps (+2.66%)
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
| |
Calculate maximal count of MPW WQEBBs on SQ's creation and store it
there. Remove MLX5E_TX_MPW_MAX_NUM_DS and MLX5E_TX_MPW_MAX_WQEBBS.
Update mlx5e_tx_mpwqe_is_full() and mlx5e_xdp_mpqwe_is_full() .
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Prior to this patch the maximal value for max WQEBBs (WQE Basic Blocks,
where WQE is a Work Queue Element) on the TX side was assumed to be 16
(fixed value). All firmware versions till today comply to this. In order
to be more flexible and resilient, read from FW the corresponding:
max_wqe_sz_sq. This value describes the maximum WQE size given in bytes,
thus max WQEBBs is given by the division in WQEBB's byte size. The
driver uses the top between 16 and the division result. This ensures
synchronization between driver and firmware and avoids unexpected
behavior. Store this value on the different SQs (Send Queues) for easy
access.
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In preparation for FORTIFY_SOURCE performing compile-time and run-time
field bounds checking for memcpy(), memmove(), and memset(), avoid
intentionally writing across neighboring fields.
Use flexible arrays instead of zero-element arrays (which look like they
are always overflowing) and split the cross-field memcpy() into two halves
that can be appropriately bounds-checked by the compiler.
We were doing:
#define ETH_HLEN 14
#define VLAN_HLEN 4
...
#define MLX5E_XDP_MIN_INLINE (ETH_HLEN + VLAN_HLEN)
...
struct mlx5e_tx_wqe *wqe = mlx5_wq_cyc_get_wqe(wq, pi);
...
struct mlx5_wqe_eth_seg *eseg = &wqe->eth;
struct mlx5_wqe_data_seg *dseg = wqe->data;
...
memcpy(eseg->inline_hdr.start, xdptxd->data, MLX5E_XDP_MIN_INLINE);
target is wqe->eth.inline_hdr.start (which the compiler sees as being
2 bytes in size), but copying 18, intending to write across start
(really vlan_tci, 2 bytes). The remaining 16 bytes get written into
wqe->data[0], covering byte_count (4 bytes), lkey (4 bytes), and addr
(8 bytes).
struct mlx5e_tx_wqe {
struct mlx5_wqe_ctrl_seg ctrl; /* 0 16 */
struct mlx5_wqe_eth_seg eth; /* 16 16 */
struct mlx5_wqe_data_seg data[]; /* 32 0 */
/* size: 32, cachelines: 1, members: 3 */
/* last cacheline: 32 bytes */
};
struct mlx5_wqe_eth_seg {
u8 swp_outer_l4_offset; /* 0 1 */
u8 swp_outer_l3_offset; /* 1 1 */
u8 swp_inner_l4_offset; /* 2 1 */
u8 swp_inner_l3_offset; /* 3 1 */
u8 cs_flags; /* 4 1 */
u8 swp_flags; /* 5 1 */
__be16 mss; /* 6 2 */
__be32 flow_table_metadata; /* 8 4 */
union {
struct {
__be16 sz; /* 12 2 */
u8 start[2]; /* 14 2 */
} inline_hdr; /* 12 4 */
struct {
__be16 type; /* 12 2 */
__be16 vlan_tci; /* 14 2 */
} insert; /* 12 4 */
__be32 trailer; /* 12 4 */
}; /* 12 4 */
/* size: 16, cachelines: 1, members: 9 */
/* last cacheline: 16 bytes */
};
struct mlx5_wqe_data_seg {
__be32 byte_count; /* 0 4 */
__be32 lkey; /* 4 4 */
__be64 addr; /* 8 8 */
/* size: 16, cachelines: 1, members: 3 */
/* last cacheline: 16 bytes */
};
So, split the memcpy() so the compiler can reason about the buffer
sizes.
"pahole" shows no size nor member offset changes to struct mlx5e_tx_wqe
nor struct mlx5e_umr_wqe. "objdump -d" shows no meaningful object
code changes (i.e. only source line number induced differences and
optimizations).
Fixes: b5503b994ed5 ("net/mlx5e: XDP TX forwarding support")
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In non trivial scenarios, the action id alone is not sufficient to
identify the program causing the warning. Before the previous patch,
the generated stack-trace pointed out at least the involved device
driver.
Let's additionally include the program name and id, and the relevant
device name.
If the user needs additional infos, he can fetch them via a kernel
probe, leveraging the arguments added here.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Link: https://lore.kernel.org/bpf/ddb96bb975cbfddb1546cf5da60e77d5100b533c.1638189075.git.pabeni@redhat.com
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We want to change the current ndo_xdp_xmit drop semantics because it will
allow us to implement better queue overflow handling. This is working
towards the larger goal of a XDP TX queue-hook. Move XDP_REDIRECT error
path handling from each XDP ethernet driver to devmap code. According to
the new APIs, the driver running the ndo_xdp_xmit pointer, will break tx
loop whenever the hw reports a tx error and it will just return to devmap
caller the number of successfully transmitted frames. It will be devmap
responsibility to free dropped frames.
Move each XDP ndo_xdp_xmit capable driver to the new APIs:
- veth
- virtio-net
- mvneta
- mvpp2
- socionext
- amazon ena
- bnxt
- freescale (dpaa2, dpaa)
- xen-frontend
- qede
- ice
- igb
- ixgbe
- i40e
- mlx5
- ti (cpsw, cpsw-new)
- tun
- sfc
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Ioana Ciornei <ioana.ciornei@nxp.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Camelia Groza <camelia.groza@nxp.com>
Acked-by: Edward Cree <ecree.xilinx@gmail.com>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Shay Agroskin <shayagr@amazon.com>
Link: https://lore.kernel.org/bpf/ed670de24f951cfd77590decf0229a0ad7fd12f6.1615201152.git.lorenzo@kernel.org
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Convert mlx5 driver to xdp_return_frame_bulk APIs.
XDP_REDIRECT (upstream codepath): 8.9Mpps
XDP_REDIRECT (upstream codepath + bulking APIs): 10.2Mpps
Co-developed-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/250460319fd868b7b5668fc1deca74dd42813a90.1605267335.git.lorenzo@kernel.org
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux
Saeed Mahameed says:
====================
mlx5-updates-2020-09-21
Multi packet TX descriptor support for SKBs.
This series introduces some refactoring of the regular TX data path in
mlx5 and adds the Enhanced TX MPWQE feature support. MPWQE stands for
multi-packet work queue element, and it can serve multiple packets,
reducing the PCI bandwidth spent on control traffic. It should improve
performance in scenarios where PCI is the bottleneck, and xmit_more is
signaled by the kernel. The refactoring done in this series also
improves the packet rate on its own.
MPWQE is already implemented in the XDP tx path, this series adds the
support of MPWQE for regular kernel SKB tx path.
MPWQE is supported from ConnectX-5 and onward, for legacy devices we need
to keep backward compatibility for regular (Single packet) WQE descriptor.
MPWQE is not compatible with certain offloads and features, such as TLS
offload, TSO, nonlinear SKBs. If such incompatible features are in use,
the driver gracefully falls back to non-MPWQE per SKB.
Prior to the final patch "net/mlx5e: Enhanced TX MPWQE for SKBs" that adds
the actual support, Maxim did some refactoring to the tx data path to
split it into stages and smaller helper functions that can be utilized and
reused for both legacy and new MPWQE feature.
Performance testing:
UDP performance is improved in a single stream pktgen test:
Packet rate: 16.86 Mpps (±0.15 Mpps) -> 20.94 Mpps (±0.33 Mpps)
Instructions per packet: 434 -> 329
Cycles per packet: 158 -> 123
Instructions per cycle: 2.75 -> 2.67
TCP and XDP_TX single stream tests show no performance difference.
MPWQE can reduce PCI bandwidth:
PCI Gen2, pktgen at fixed rate of 36864000 pps on 24 CPU cores:
Inbound PCI utilization with MPWQE off: 80.3%
Inbound PCI utilization with MPWQE on: 59.0%
PCI Gen3, pktgen at fixed rate of 56064000 pps on 24 CPU cores:
Inbound PCI utilization with MPWQE off: 65.4%
Inbound PCI utilization with MPWQE on: 49.3%
MPWQE can also reduce CPU load, increasing the packet rate in case of
CPU bottleneck:
PCI Gen2, pktgen at full rate on 24 CPU cores:
Packet rate with MPWQE off: 37.5 Mpps
Packet rate with MPWQE on: 49.0 Mpps
PCI Gen3, pktgen at full rate on 24 CPU cores:
Packet rate with MPWQE off: 57.0 Mpps
Packet rate with MPWQE on: 66.8 Mpps
Burst size in all pktgen tests is 32.
CPU: Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz (x86_64)
NIC: Mellanox ConnectX-6 Dx
GCC 10.2.0
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This commit adds support for Enhanced TX MPWQE feature in the regular
(SKB) data path. A MPWQE (multi-packet work queue element) can serve
multiple packets, reducing the PCI bandwidth on control traffic.
Two new stats (tx*_mpwqe_blks and tx*_mpwqe_pkts) are added. The feature
is on by default and controlled by the skb_tx_mpwqe private flag.
In a MPWQE, eseg is shared among all packets, so eseg-based offloads
(IPSEC, GENEVE, checksum) run on a separate eseg that is compared to the
eseg of the current MPWQE session to decide if the new packet can be
added to the same session.
MPWQE is not compatible with certain offloads and features, such as TLS
offload, TSO, nonlinear SKBs. If such incompatible features are in use,
the driver gracefully falls back to non-MPWQE.
This change has no performance impact in TCP single stream test and
XDP_TX single stream test.
UDP pktgen, 64-byte packets, single stream, MPWQE off:
Packet rate: 16.96 Mpps (±0.12 Mpps) -> 17.01 Mpps (±0.20 Mpps)
Instructions per packet: 421 -> 429
Cycles per packet: 156 -> 161
Instructions per cycle: 2.70 -> 2.67
UDP pktgen, 64-byte packets, single stream, MPWQE on:
Packet rate: 16.96 Mpps (±0.12 Mpps) -> 20.94 Mpps (±0.33 Mpps)
Instructions per packet: 421 -> 329
Cycles per packet: 156 -> 123
Instructions per cycle: 2.70 -> 2.67
Enabling MPWQE can reduce PCI bandwidth:
PCI Gen2, pktgen at fixed rate of 36864000 pps on 24 CPU cores:
Inbound PCI utilization with MPWQE off: 80.3%
Inbound PCI utilization with MPWQE on: 59.0%
PCI Gen3, pktgen at fixed rate of 56064000 pps on 24 CPU cores:
Inbound PCI utilization with MPWQE off: 65.4%
Inbound PCI utilization with MPWQE on: 49.3%
Enabling MPWQE can also reduce CPU load, increasing the packet rate in
case of CPU bottleneck:
PCI Gen2, pktgen at full rate on 24 CPU cores:
Packet rate with MPWQE off: 37.5 Mpps
Packet rate with MPWQE on: 49.0 Mpps
PCI Gen3, pktgen at full rate on 24 CPU cores:
Packet rate with MPWQE off: 57.0 Mpps
Packet rate with MPWQE on: 66.8 Mpps
Burst size in all pktgen tests is 32.
CPU: Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz (x86_64)
NIC: Mellanox ConnectX-6 Dx
GCC 10.2.0
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
As preparation for the upcoming TX MPWQE support for SKBs, rename struct
mlx5e_xdp_mpwqe to mlx5e_tx_mpwqe and move it above struct mlx5e_txqsq.
This structure will be reused in the regular SQ and in the regular TX
data path. Also rename mlx5e_xdp_xmit_data to mlx5e_xmit_data - it will
be used in the upcoming TX MPWQE flow.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
As preparation for the upcoming TX MPWQE for SKBs, create a function
(mlx5e_tx_mpwqe_is_full) to check whether an MPWQE session is full. This
function will be shared by MPWQE code for XDP and for SKBs. Defines are
renamed and moved to make them not XDP-specific.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
A constant for the number of DS in an empty WQE (i.e. a WQE without data
segments) is needed in multiple places (normal TX data path, MPWQE in
XDP), but currently we have a constant for XDP and an inline formula in
normal TX. This patch introduces a common constant.
Additionally, mlx5e_xdp_mpwqe_session_start is converted to use struct
assignment, because the code nearby is touched.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Use MLX5E_XDP_MPW_MAX_WQEBBS to reserve space for a MPWQE, because it's
actually the maximal size a MPWQE can take.
Reorganize the logic that checks when to close the MPWQE session:
1. Put all checks into a single function.
2. When inline is on, make only one comparison - if it's false, the less
strict one will also be false. The compiler probably optimized it out
anyway, but it's clearer to also reflect it in the code.
The MLX5E_XDP_INLINE_WQE_* defines are also changed to make the
calculations more correct from the logical point of view. Though
MLX5E_XDP_INLINE_WQE_MAX_DS_CNT used to be 16 and didn't change its
value, the calculation used to be DIV_ROUND_UP(max inline packet size,
MLX5_SEND_WQE_DS), and the numerator should have included sizeof(struct
mlx5_wqe_inline_seg).
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
|
|\ \
| |/
|/|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Two minor conflicts:
1) net/ipv4/route.c, adding a new local variable while
moving another local variable and removing it's
initial assignment.
2) drivers/net/dsa/microchip/ksz9477.c, overlapping changes.
One pretty prints the port mode differently, whilst another
changes the driver to try and obtain the port mode from
the port node rather than the switch node.
Signed-off-by: David S. Miller <davem@davemloft.net>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Currently, the RQs are temporarily deactivated while hot-replacing the
XDP program, and napi_synchronize is used to make sure rq->xdp_prog is
not in use. However, napi_synchronize is not ideal: instead of waiting
till the end of a NAPI cycle, it polls and waits until NAPI is not
running, sleeping for 1ms between the periodic checks. Under heavy
workloads, this loop will never end, which may even lead to a kernel
panic if the kernel detects the hangup. Such workloads include XSK TX
and possibly also heavy RX (XSK or normal).
The fix is inspired by commit 326fe02d1ed6 ("net/mlx4_en: protect
ring->xdp_prog with rcu_read_lock"). As mlx5e_xdp_handle is already
protected by rcu_read_lock, and bpf_prog_put uses call_rcu to free the
program, there is no need for additional synchronization if proper RCU
functions are used to access the pointer. This patch converts all
accesses to rq->xdp_prog to use RCU functions.
Fixes: 86994156c736 ("net/mlx5e: XDP fast RX drop bpf programs support")
Fixes: db05815b36cb ("net/mlx5e: Add XSK zero-copy support")
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Daniel Borkmann says:
====================
pull-request: bpf-next 2020-09-01
The following pull-request contains BPF updates for your *net-next* tree.
There are two small conflicts when pulling, resolve as follows:
1) Merge conflict in tools/lib/bpf/libbpf.c between 88a82120282b ("libbpf: Factor
out common ELF operations and improve logging") in bpf-next and 1e891e513e16
("libbpf: Fix map index used in error message") in net-next. Resolve by taking
the hunk in bpf-next:
[...]
scn = elf_sec_by_idx(obj, obj->efile.btf_maps_shndx);
data = elf_sec_data(obj, scn);
if (!scn || !data) {
pr_warn("elf: failed to get %s map definitions for %s\n",
MAPS_ELF_SEC, obj->path);
return -EINVAL;
}
[...]
2) Merge conflict in drivers/net/ethernet/mellanox/mlx5/core/en/xsk/rx.c between
9647c57b11e5 ("xsk: i40e: ice: ixgbe: mlx5: Test for dma_need_sync earlier for
better performance") in bpf-next and e20f0dbf204f ("net/mlx5e: RX, Add a prefetch
command for small L1_CACHE_BYTES") in net-next. Resolve the two locations by retaining
net_prefetch() and taking xsk_buff_dma_sync_for_cpu() from bpf-next. Should look like:
[...]
xdp_set_data_meta_invalid(xdp);
xsk_buff_dma_sync_for_cpu(xdp, rq->xsk_pool);
net_prefetch(xdp->data);
[...]
We've added 133 non-merge commits during the last 14 day(s) which contain
a total of 246 files changed, 13832 insertions(+), 3105 deletions(-).
The main changes are:
1) Initial support for sleepable BPF programs along with bpf_copy_from_user() helper
for tracing to reliably access user memory, from Alexei Starovoitov.
2) Add BPF infra for writing and parsing TCP header options, from Martin KaFai Lau.
3) bpf_d_path() helper for returning full path for given 'struct path', from Jiri Olsa.
4) AF_XDP support for shared umems between devices and queues, from Magnus Karlsson.
5) Initial prep work for full BPF-to-BPF call support in libbpf, from Andrii Nakryiko.
6) Generalize bpf_sk_storage map & add local storage for inodes, from KP Singh.
7) Implement sockmap/hash updates from BPF context, from Lorenz Bauer.
8) BPF xor verification for scalar types & add BPF link iterator, from Yonghong Song.
9) Use target's prog type for BPF_PROG_TYPE_EXT prog verification, from Udip Pant.
10) Rework BPF tracing samples to use libbpf loader, from Daniel T. Lee.
11) Fix xdpsock sample to really cycle through all buffers, from Weqaar Janjua.
12) Improve type safety for tun/veth XDP frame handling, from Maciej Żenczykowski.
13) Various smaller cleanups and improvements all over the place.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Rename the AF_XDP zero-copy driver interface functions to better
reflect what they do after the replacement of umems with buffer
pools in the previous commit. Mostly it is about replacing the
umem name from the function names with xsk_buff and also have
them take the a buffer pool pointer instead of a umem. The
various ring functions have also been renamed in the process so
that they have the same naming convention as the internal
functions in xsk_queue.h. This so that it will be clearer what
they do and also for consistency.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Björn Töpel <bjorn.topel@intel.com>
Link: https://lore.kernel.org/bpf/1598603189-32145-3-git-send-email-magnus.karlsson@intel.com
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Replace the explicit umem reference passed to the driver in AF_XDP
zero-copy mode with the buffer pool instead. This in preparation for
extending the functionality of the zero-copy mode so that umems can be
shared between queues on the same netdev and also between netdevs. In
this commit, only an umem reference has been added to the buffer pool
struct. But later commits will add other entities to it. These are
going to be entities that are different between different queue ids
and netdevs even though the umem is shared between them.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Björn Töpel <bjorn.topel@intel.com>
Link: https://lore.kernel.org/bpf/1598603189-32145-2-git-send-email-magnus.karlsson@intel.com
|
|/
|
|
|
|
|
|
|
|
|
| |
A single cacheline might not contain the packet header for
small L1_CACHE_BYTES values.
Use net_prefetch() as it issues an additional prefetch
in this case.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
|
|
|
|
|
|
|
|
|
| |
Replace the existing /* fall through */ comments and its variants with
the new pseudo-keyword macro fallthrough[1]. Also, remove unnecessary
fall-through markings when it is the case.
[1] https://www.kernel.org/doc/html/v5.7/process/deprecated.html?highlight=fallthrough#implicit-switch-case-fall-through
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
| |
Use the indirect call wrapper API macros for declaration and scope
of the RX post WQEs functions.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
| |
Use INDIRECT_CALL_2() helper to avoid the cost of the indirect call
when/if CONFIG_RETPOLINE=y.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
|
|
| |
In order to use standard 'xdp' prefix, rename convert_to_xdp_frame
utility routine in xdp_convert_buff_to_frame and replace all the
occurrences
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Link: https://lore.kernel.org/bpf/6344f739be0d1a08ab2b9607584c4d5478c8c083.1590698295.git.lorenzo@kernel.org
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Daniel Borkmann says:
====================
pull-request: bpf-next 2020-05-23
The following pull-request contains BPF updates for your *net-next* tree.
We've added 50 non-merge commits during the last 8 day(s) which contain
a total of 109 files changed, 2776 insertions(+), 2887 deletions(-).
The main changes are:
1) Add a new AF_XDP buffer allocation API to the core in order to help
lowering the bar for drivers adopting AF_XDP support. i40e, ice, ixgbe
as well as mlx5 have been moved over to the new API and also gained a
small improvement in performance, from Björn Töpel and Magnus Karlsson.
2) Add getpeername()/getsockname() attach types for BPF sock_addr programs
in order to allow for e.g. reverse translation of load-balancer backend
to service address/port tuple from a connected peer, from Daniel Borkmann.
3) Improve the BPF verifier is_branch_taken() logic to evaluate pointers
being non-NULL, e.g. if after an initial test another non-NULL test on
that pointer follows in a given path, then it can be pruned right away,
from John Fastabend.
4) Larger rework of BPF sockmap selftests to make output easier to understand
and to reduce overall runtime as well as adding new BPF kTLS selftests
that run in combination with sockmap, also from John Fastabend.
5) Batch of misc updates to BPF selftests including fixing up test_align
to match verifier output again and moving it under test_progs, allowing
bpf_iter selftest to compile on machines with older vmlinux.h, and
updating config options for lirc and v6 segment routing helpers, from
Stanislav Fomichev, Andrii Nakryiko and Alan Maguire.
6) Conversion of BPF tracing samples outdated internal BPF loader to use
libbpf API instead, from Daniel T. Lee.
7) Follow-up to BPF kernel test infrastructure in order to fix a flake in
the XDP selftests, from Jesper Dangaard Brouer.
8) Minor improvements to libbpf's internal hashmap implementation, from
Ian Rogers.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Use the new MEM_TYPE_XSK_BUFF_POOL API in lieu of MEM_TYPE_ZERO_COPY in
mlx5e. It allows to drop a lot of code from the driver (which is now
common in AF_XDP core and was related to XSK RX frame allocation, DMA
mapping, etc.) and slightly improve performance (RX +0.8 Mpps, TX +0.4
Mpps).
rfc->v1: Put back the sanity check for XSK params, use XSK API to get
the total headroom size. (Maxim)
v1->v2: Fix DMA address handling, set XDP metadata to invalid. (Maxim)
v2->v3: Handle frame_sz, use xsk_buff_xdp_get_frame_dma, use xsk_buff
API for DMA sync on TX, add performance numbers. (Maxim)
v3->v4: Remove unused variable num_xsk_frames. (Jakub)
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200520192103.355233-12-bjorn.topel@gmail.com
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Move the AF_XDP zero-copy driver interface to its own include file
called xdp_sock_drv.h. This, hopefully, will make it more clear for
NIC driver implementors to know what functions to use for zero-copy
support.
v4->v5: Fix -Wmissing-prototypes by include header file. (Jakub)
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200520192103.355233-4-bjorn.topel@gmail.com
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently, different formulas are used to estimate the space that may be
taken by WQEs in the SQ during a single packet transmit. This space is
called stop room, and it's checked in the end of packet transmit to find
out if the next packet could overflow the SQ. If it could, the driver
tells the kernel to stop sending next packets.
Many factors affect the stop room:
1. Padding with NOPs to avoid WQEs spanning over page boundaries.
2. Enabled and disabled offloads (TLS, upcoming MPWQE).
3. The maximum size of a WQE.
The padding is performed before every WQE if it doesn't fit the current
page.
The current formula assumes that only one padding will be required per
packet, and it doesn't take into account that the WQEs posted during the
transmission of a single packet might exceed the page size in very rare
circumstances. For example, to hit this condition with 4096-byte pages,
TLS offload will have to interrupt an almost-full MPWQE session, be in
the resync flow and try to transmit a near to maximum amount of data.
To avoid SQ overflows in such rare cases after MPWQE is added, this
patch introduces a more robust formula to estimate the stop room. The
new formula uses the fact that a WQE of size X will not require more
than X-1 WQEBBs of padding. More exact estimations are possible, but
they result in much more complex and error-prone code for little gain.
Before this patch, the TLS stop room included space for both INNOVA and
ConnectX TLS offloads that couldn't run at the same time anyway, so this
patch accounts only for the active one.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The mlx5 driver have multiple memory models, which are also changed
according to whether a XDP bpf_prog is attached.
The 'rx_striding_rq' setting is adjusted via ethtool priv-flags e.g.:
# ethtool --set-priv-flags mlx5p2 rx_striding_rq off
On the general case with 4K page_size and regular MTU packet, then
the frame_sz is 2048 and 4096 when XDP is enabled, in both modes.
The info on the given frame size is stored differently depending on the
RQ-mode and encoded in a union in struct mlx5e_rq union wqe/mpwqe.
In rx striding mode rq->mpwqe.log_stride_sz is either 11 or 12, which
corresponds to 2048 or 4096 (MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ).
In non-striding mode (MLX5_WQ_TYPE_CYCLIC) the frag_stride is stored
in rq->wqe.info.arr[0].frag_stride, for the first fragment, which is
what the XDP case cares about.
To reduce effect on fast-path, this patch determine the frame_sz at
setup time, to avoid determining the memory model runtime. Variable
is named frame0_sz to make it clear that this is only the frame
size of the first fragment.
This mlx5 driver does a DMA-sync on XDP_TX action, but grow is safe
as it have done a DMA-map on the entire PAGE_SIZE. The driver also
already does a XDP length check against sq->hw_mtu on the possible
XDP xmit paths mlx5e_xmit_xdp_frame() + mlx5e_xmit_xdp_frame_mpwqe().
V3+4: Change variable name first_frame_sz to frame0_sz
V2: Fix that frag_size need to be recalc before creating SKB.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Tariq Toukan <tariqt@mellanox.com>
Cc: Saeed Mahameed <saeedm@mellanox.com>
Link: https://lore.kernel.org/bpf/158945348021.97035.12295039384250022883.stgit@firesoul
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In our fast-path design, a WQE (Work Queue Element) must not cross the
page boundary. To enforce that, for WQEs consisting of more than one BB
(Basic Block), the driver checks the available contiguous space in the
WQ in advance, and if it's not enough, it pads it with NOPs.
This patch modifies the code that calculates the position of next WQE,
considering the padding, and prepares the WQE. This code is common for
all SQ types. In this patch it's reorganized in a way that makes the
usage pattern unified for all SQ types, and makes the implementations
self-contained and look almost the same, preparing the repeating code to
further attempts to deduplicate it.
One place is left as is: mlx5e_sq_xmit and mlx5e_fill_sq_frag_edge call
inside, because it is special in a way that it may also copy WQE's cseg
and eseg when reserving space. This will be eliminated in one of the
following patches, and this place will be converted to the new approach,
too.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There are multiple functions mlx5{e,i}_*_fetch_wqe that contain the same
code, that is repeated, because they operate on different SQ struct
types. mlx5e_sq_fetch_wqe also returns void *, instead of the concrete
WQE type.
This commit generalizes the fetch WQE operation by putting this code
into a single function. To simplify calls of the generic function in
concrete use cases, macros are provided that substitute the right WQE
size and cast the return type.
Before this patch, fetch_wqe used to calculate pi itself, but the value
was often known to the caller. This calculation is moved outside to
eliminate this unnecessary step and prepare for the fill_frag_edge
refactoring in the next patch.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
| |
Upon an error completion on an XDP SQ, print the offending WQE
to ease the debug process.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Aya Levin <ayal@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
|
| |
Error CQE was dumped only for TXQ SQs.
Generalise the function, and add usage for error completions
on ICO SQs and XDP SQs.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Aya Levin <ayal@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
|
|
| |
With the addition of the unaligned chunks option, we need to make sure we
handle the offsets accordingly based on the mode we are currently running
in. This patch modifies the driver to appropriately mask the address for
each case.
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
|
|
|
|
|
|
|
|
|
|
| |
Instead of passing an output param, let function return the
WQE pointer.
In addition, pass &pi so it gets its value in the function,
and save the redundant assignment that comes after it.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In MPWQE mode, when transmitting packets with XDP, a packet that is smaller
than a certain size (set to 256 bytes) would be sent inline within its WQE
TX descriptor (mem-copied), in case the hardware tx queue is congested
beyond a pre-defined water-mark.
If a MPWQE cannot contain an additional inline packet, we close this
MPWQE session, and send the packet inlined within the next MPWQE.
To save some MPWQE session close+open operations, we don't open MPWQE
sessions that are contiguously smaller than certain size (set to the
HW MPWQE maximum size). If there isn't enough contiguous room in the
send queue, we fill it with NOPs and wrap the send queue index around.
This way, qualified packets are always sent inline.
Perf tests:
Tested packet rate for UDP 64Byte multi-stream
over two dual port ConnectX-5 100Gbps NICs.
CPU: Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz
XDP_TX:
With 24 channels:
| ------ | bounced packets | inlined packets | inline ratio |
| before | 113.6Mpps | 96.3Mpps | 84% |
| after | 115Mpps | 99.5Mpps | 86% |
With one channel:
| ------ | bounced packets | inlined packets | inline ratio |
| before | 6.7Mpps | 0pps | 0% |
| after | 6.8Mpps | 0pps | 0% |
As we can see, there is improvement in both inline ratio and overall
packet rate for 24 channels. Also, we see no degradation for the
one-channel case.
Signed-off-by: Shay Agroskin <shayag@mellanox.com>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit adds support for AF_XDP zero-copy RX and TX.
We create a dedicated XSK RQ inside the channel, it means that two
RQs are running simultaneously: one for non-XSK traffic and the other
for XSK traffic. The regular and XSK RQs use a single ID namespace split
into two halves: the lower half is regular RQs, and the upper half is
XSK RQs. When any zero-copy AF_XDP socket is active, changing the number
of channels is not allowed, because it would break to mapping between
XSK RQ IDs and channels.
XSK requires different page allocation and release routines. Such
functions as mlx5e_{alloc,free}_rx_mpwqe and mlx5e_{get,put}_rx_frag are
generic enough to be used for both regular and XSK RQs, and they use the
mlx5e_page_{alloc,release} wrappers around the real allocation
functions. Function pointers are not used to avoid losing the
performance with retpolines. Wherever it's certain that the regular
(non-XSK) page release function should be used, it's called directly.
Only the stats that could be meaningful for XSK are exposed to the
userspace. Those that don't take part in the XSK flow are not
considered.
Note that we don't wait for WQEs on the XSK RQ (unlike the regular RQ),
because the newer xdpsock sample doesn't provide any Fill Ring entries
at the setup stage.
We create a dedicated XSK SQ in the channel. This separation has its
advantages:
1. When the UMEM is closed, the XSK SQ can also be closed and stop
receiving completions. If an existing SQ was used for XSK, it would
continue receiving completions for the packets of the closed socket. If
a new UMEM was opened at that point, it would start getting completions
that don't belong to it.
2. Calculating statistics separately.
When the userspace kicks the TX, the driver triggers a hardware
interrupt by posting a NOP to a dedicated XSK ICO (internal control
operations) SQ, in order to trigger NAPI on the right CPU core. This XSK
ICO SQ is protected by a spinlock, as the userspace application may kick
the TX from any core.
Store the pointers to the UMEMs in the net device private context,
independently from the kernel. This way the driver can distinguish
between the zero-copy and non-zero-copy UMEMs. The kernel function
xdp_get_umem_from_qid does not care about this difference, but the
driver is only interested in zero-copy UMEMs, particularly, on the
cleanup it determines whether to close the XSK RQ and SQ or not by
looking at the presence of the UMEM. Use state_lock to protect the
access to this area of UMEM pointers.
LRO isn't compatible with XDP, but there may be active UMEMs while
XDP is off. If this is the case, don't allow LRO to ensure XDP can
be reenabled at any time.
The validation of XSK parameters typically happens when XSK queues
open. However, when the interface is down or the XDP program isn't
set, it's still possible to have active AF_XDP sockets and even to
open new, but the XSK queues will be closed. To cover these cases,
perform the validation also in these flows:
1. A new UMEM is registered, but the XSK queues aren't going to be
created due to missing XDP program or interface being down.
2. MTU changes while there are UMEMs registered.
Having this early check prevents mlx5e_open_channels from failing
at a later stage, where recovery is impossible and the application
has no chance to handle the error, because it got the successful
return value for an MTU change or XSK open operation.
The performance testing was performed on a machine with the following
configuration:
- 24 cores of Intel Xeon E5-2620 v3 @ 2.40 GHz
- Mellanox ConnectX-5 Ex with 100 Gbit/s link
The results with retpoline disabled, single stream:
txonly: 33.3 Mpps (21.5 Mpps with queue and app pinned to the same CPU)
rxdrop: 12.2 Mpps
l2fwd: 9.4 Mpps
The results with retpoline enabled, single stream:
txonly: 21.3 Mpps (14.1 Mpps with queue and app pinned to the same CPU)
rxdrop: 9.9 Mpps
l2fwd: 6.8 Mpps
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
|
|
|
|
|
|
|
|
|
|
|
| |
Use the existing mlx5e_get_linear_rq_headroom function to calculate the
headroom for mlx5e_xdp_max_mtu. This function takes the XSK headroom
into consideration, which will be used in the following patches.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When an XDP program returns XDP_TX, and the RQ is XSK-enabled, it
requires careful handling, because convert_to_xdp_frame creates a new
page and copies the data there, while our driver expects the xdp_frame
to point to the same memory as the xdp_buff. Handle this case
separately: map the page, and in the end unmap it and call
xdp_return_frame.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
|