Re: [PATCH v6 00/45] use stdatomic API

On Fri, May 17, 2024 at 06:58:57PM +0200, David Marchand wrote: > On Tue, May 14, 2024 at 6:36 PM Tyler Retzlaff > wrote: > > > > This series converts all non-generic built atomics to use the rte_atomic > > macros that allow optional enablement of standard C11 atomics. > > > > Use of generic atomi

[PATCH v7 9/9] net/null: use generic SW stats

Use the new common code for statistics. This also fixes the bug that this driver was not accounting for bytes. Signed-off-by: Stephen Hemminger --- drivers/net/null/rte_eth_null.c | 80 +++-- 1 file changed, 17 insertions(+), 63 deletions(-) diff --git a/drivers/net/

[PATCH v7 8/9] net/tap: use generic SW stats

Use new common sw statistics. Signed-off-by: Stephen Hemminger --- drivers/net/tap/rte_eth_tap.c | 88 ++- drivers/net/tap/rte_eth_tap.h | 15 ++ 2 files changed, 18 insertions(+), 85 deletions(-) diff --git a/drivers/net/tap/rte_eth_tap.c b/drivers/net/tap/r

[PATCH v7 7/9] net/ring: use generic SW stats

Use generic per-queue infrastructure. This also fixes bug where ring code was not accounting for bytes. Signed-off-by: Stephen Hemminger --- drivers/net/ring/rte_eth_ring.c | 71 + 1 file changed, 28 insertions(+), 43 deletions(-) diff --git a/drivers/net/ring/rt

[PATCH v7 6/9] test/pmd_ring: initialize mbufs

Do not pass uninitialized data into the ring PMD. The mbufs should be initialized first so that length is zero. Signed-off-by: Stephen Hemminger --- app/test/test_pmd_ring.c | 26 +- 1 file changed, 21 insertions(+), 5 deletions(-) diff --git a/app/test/test_pmd_ring.c b

[PATCH v7 5/9] net/pcap: use generic SW stats

Use common statistics for SW drivers. Signed-off-by: Stephen Hemminger --- drivers/net/pcap/pcap_ethdev.c | 125 +++-- 1 file changed, 26 insertions(+), 99 deletions(-) diff --git a/drivers/net/pcap/pcap_ethdev.c b/drivers/net/pcap/pcap_ethdev.c index bfec085045..b1a

[PATCH v7 4/9] net/af_xdp: use generic SW stats

Use common code for all SW stats. Signed-off-by: Stephen Hemminger --- drivers/net/af_xdp/rte_eth_af_xdp.c | 97 +++-- 1 file changed, 24 insertions(+), 73 deletions(-) diff --git a/drivers/net/af_xdp/rte_eth_af_xdp.c b/drivers/net/af_xdp/rte_eth_af_xdp.c index 6ba455bb

[PATCH v7 2/9] ethdev: add common counters for statistics

Introduce common helper routines for keeping track of per-queue statistics in SW PMD's. The code in several drivers had copy/pasted the same code for this, but had common issues with 64 bit counters on 32 bit platforms. Signed-off-by: Stephen Hemminger --- lib/ethdev/ethdev_swstats.c | 101 +

[PATCH v7 3/9] net/af_packet: use generic SW stats

Use the new generic SW stats. Signed-off-by: Stephen Hemminger --- drivers/net/af_packet/rte_eth_af_packet.c | 82 --- 1 file changed, 14 insertions(+), 68 deletions(-) diff --git a/drivers/net/af_packet/rte_eth_af_packet.c b/drivers/net/af_packet/rte_eth_af_packet.c index

[PATCH v7 1/9] eal: generic 64 bit counter

This header implements 64 bit counters using atomic operations but with a weak memory ordering so that they are safe against load/store splits on 32 bit platforms. Signed-off-by: Stephen Hemminger Acked-by: Morten Brørup --- lib/eal/include/meson.build | 1 + lib/eal/include/rte_counter.h |

[PATCH v7 0/9] Use weak atomic operations for SW PMD counters

Adds common code for collecting basic statistics used by many SW based PMD's such as af_packet, af_xdp, null, tap and ring. This version uses std atomic on all platforms. The memory order is chosen to be the best performance but still providing the guarantees expected of PMD counters. By using co

Re: [PATCH v2] eal/unix: support ZSTD compression for firmware

On Mon, May 13, 2024 at 1:37 PM Bruce Richardson wrote: > On Mon, May 13, 2024 at 01:12:04PM +0200, David Marchand wrote: > > Ubuntu 24.04 started to compress firmware files with ZSTD compression. > > > > Bugzilla ID: 1437 > > Cc: sta...@dpdk.org > > > > Signed-off-by: David Marchand > Acked-by:

Re: [PATCH v6 00/45] use stdatomic API

On Tue, May 14, 2024 at 6:36 PM Tyler Retzlaff wrote: > > This series converts all non-generic built atomics to use the rte_atomic > macros that allow optional enablement of standard C11 atomics. > > Use of generic atomics for non-scalar types are not converted in this > change and will be evaluat

Re: [PATCH] maintainers: update for eal

17/05/2024 18:15, David Marchand: > On Tue, Apr 30, 2024 at 10:52 PM Tyler Retzlaff > wrote: > > > > Add myself as a maintainer for eal and tests. > > > > Signed-off-by: Tyler Retzlaff > > Acked-by: David Marchand Applied, thanks for volunteering, EAL is a big piece.

Re: [PATCH] maintainers: update for eal

On Tue, Apr 30, 2024 at 10:52 PM Tyler Retzlaff wrote: > > Add myself as a maintainer for eal and tests. > > Signed-off-by: Tyler Retzlaff Acked-by: David Marchand -- David Marchand

Re: [PATCH v6 4/9] net/af_xdp: use generic SW stats

On Fri, 17 May 2024 13:34:30 + "Loftus, Ciara" wrote: > > - tx_bytes += mbuf->pkt_len; > > + rte_eth_count_packet(&txq->stats, pkt_len); > > This change resolves the bugzilla you reported recently (1440 - use after > free in af_xdp). Should this be mentioned in the com

RE: [PATCH v6 4/9] net/af_xdp: use generic SW stats

> Subject: [PATCH v6 4/9] net/af_xdp: use generic SW stats > > Use common code for all SW stats. > > Signed-off-by: Stephen Hemminger > --- > drivers/net/af_xdp/rte_eth_af_xdp.c | 98 - > 1 file changed, 25 insertions(+), 73 deletions(-) > > diff --git a/drivers/net

RE: [PATCH v2] net/af_xdp: fix umem map size for zero copy

> > The current calculation assumes that the mbufs are contiguous. However, > this assumption is incorrect when the memory spans across a huge page. > Correct to directly read the size from the mempool memory chunks. > > Signed-off-by: Frank Du Hi Frank, Thanks for the patch. Before your patc

Re: [PATCH] doc: ensure sphinx output is reproducible

On Mon, 27 Nov 2023 at 17:04, Bruce Richardson wrote: > > On Mon, Nov 27, 2023 at 05:45:52PM +0100, Thomas Monjalon wrote: > > 06/07/2023 14:49, Christian Ehrhardt: > > > On Mon, Jul 3, 2023 at 5:29 PM Thomas Monjalon > > > wrote: > > > > > > > > 29/06/2023 14:58, christian.ehrha...@canonical.co

Minutes of Technical Board meeting 2024-05-15

Next meeting will be 29th May, with Hemant as Chair Attendees - Stephen Hemminger Bruce Richardson Aaron Conole Konstantin Ananyev Kevin Traynor Hemant Agrawal Jerin Jacob Maxime Coquelin Honnappa Nagarahalli Thomas Monjalon David Marchand Nathan Southern Tyler Retzlaff Robin Jarry Minute

[DPDK/ethdev Bug 1443] net/af_packet: defragmented packet will be truncated when exceed the frame size

https://bugs.dpdk.org/show_bug.cgi?id=1443 Bug ID: 1443 Summary: net/af_packet: defragmented packet will be truncated when exceed the frame size Product: DPDK Version: 23.03 Hardware: x86 OS: Linux

Re: [PATCH v5] devtools: add .clang-format file

On Thu, May 16, 2024 at 08:20:51AM +, Abdullah Ömer Yamaç wrote: > clang-format is a tool to format C/C++/Objective-C code. It can be used > to reformat code to match a given coding style, or to ensure that code > adheres to a specific coding style. It helps to maintain a consistent > coding st

[v1 2/2] app/testpmd: introduce VXLAN-GBP and VXLAN-GPE fields

After unification of VXLAN and all its extension protocols(VXLAN-GPE and VXLAN-GBP currently), all related RTE items should be spicified after 'vxlan'. And it's user's responsibility to specify correct UDP port for the protocol to be matched. Below are the examples to match standard VXLAN, VXLAN-G

[v1 1/2] net: extend VXLAN header to support more extensions

Currently, DPDK supports VXLAN and VXLAN-GPE with similar header structures and we are working on adding support for VXLAN-GBP which is another extension to VXLAN. More extension of VXLAN may be added in the future. VXLAN and VXLAN-GBP use the same UDP port(4789) while VXLAN-GPE uses a different o

[v1 0/2] extend VXLAN header to support more extensions

In this patch series, all the VXLAN extension header will be merged with VXLAN as union if the overlapped field has different format among protocols. The existing VXLAN-GPE will be marked as deprecated and new extensions of VXLAN should be added to VXLAN instead of a new RTE item. RFC https://patc

DPDK 23.11.1(LTS) released

Hi all, Here is a new stable release: https://fast.dpdk.org/rel/dpdk-23.11.1.tar.xz The git tree is at: https://dpdk.org/browse/dpdk-stable/?h=23.11 Thanks to the authors who helped with backports and to the following who helped with validation: Red Hat, Nvidia and Intel. Xuem

[PATCH 10/10] net/cnxk: define CPT HW result format for PMD API

From: Srujana Challa Defines CPT HW result format for PMD API, rte_pmd_cnxk_inl_ipsec_res(). Signed-off-by: Srujana Challa --- drivers/net/cnxk/cn10k_ethdev_sec.c | 4 ++-- drivers/net/cnxk/rte_pmd_cnxk.h | 28 ++-- 2 files changed, 28 insertions(+), 4 deletions(-)

[PATCH 09/10] net/cnxk: clear CGX stats during xstats reset

From: Sunil Kumar Kori Currently only NIX stats are cleared during xstats reset and CGX stats are left as it is. Clearing CGX stats too during xstats reset. Signed-off-by: Sunil Kumar Kori --- drivers/net/cnxk/cnxk_stats.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/net/cnxk

[PATCH 08/10] common/cnxk: add API to reset CGX stats

From: Sunil Kumar Kori Similar to NIX stats reset API, adding API to reset CGX stats. When user requests to reset the stats then it clears if nix_lf is PF otherwise silently discard the request. Signed-off-by: Sunil Kumar Kori --- drivers/common/cnxk/roc_nix.h | 1 + drivers/common/cnxk/

[PATCH 07/10] net/cnxk: fix issue with outbound security higher pkt burst

Fix issue with outbound security path while handling mixed traffic i.e both plain and inline outbound pkts being present as part of burst and burst size is > 32. The loop needs to be broken when we don't have space for 4 pkts in LMT lines for CPT considering both the full lmt lines and partial lmt

[PATCH 06/10] net/cnxk: add option to disable custom meta aura

Add option to explicitly disable custom meta aura. Currently custom meta aura is enabled automatically when inl_cpt_channel is set i.e inline dev is masking CHAN field in IPsec rules. Also decouple the custom meta aura feature from custom sa action so that the custom sa action can independently be

[PATCH 05/10] net/cnxk: update SA userdata and keep original cookie

Update SA userdata as part of session_update() and keep the original cookie that is used to identify SA. Fixes: 8efa348e8160 ("net/cnxk: support custom SA index") Cc: sta...@dpdk.org Signed-off-by: Nithin Dabilpuram --- drivers/net/cnxk/cn10k_ethdev_sec.c | 57 - 1 f

[PATCH 04/10] common/cnxk: override UDP encap ports with session data

Override UDP encap ports with session info when non-zero on cn10k. This makes the UDP encap ports configurable by user as needed. Default UDP source and destination ports will still be 4500. Signed-off-by: Nithin Dabilpuram --- drivers/common/cnxk/cnxk_security.c | 22 ++ 1 f

[PATCH 03/10] net/cnxk: fix extbuf handling for multisegment packet

From: Rakesh Kudurumalla Avoid double free of extbuf when during TX path in multisegmented packet with extbuf as one of segment. Fixes: dd9446991212 ("net/cnxk: add transmit completion handler") Cc: sta...@dpdk.org Signed-off-by: Rakesh Kudurumalla --- drivers/net/cnxk/cn9k_tx.h | 1 + 1 file

[PATCH 02/10] net/cnxk: set VF default root schedule weight

From: Satha Rao To get better performance on LBK or VF interfaces, set the default root schedule weight to known tested value. Signed-off-by: Satha Rao --- drivers/net/cnxk/cnxk_ethdev.c | 4 1 file changed, 4 insertions(+) diff --git a/drivers/net/cnxk/cnxk_ethdev.c b/drivers/net/cnxk/c

[PATCH 01/10] common/cnxk: sync VF root weight with kernel

From: Satha Rao By default set VF root scheduling weight same as other kernel configured VFs. Also fix a compilation issue when cflags has -Werror=shadow=compatible-local. Signed-off-by: Satha Rao --- drivers/common/cnxk/roc_nix.h | 1 + drivers/common/cnxk/roc_nix_queue.c | 3 ++- drive

RE: [PATCH] mempool: dump includes list of memory chunks

> From: Stephen Hemminger [mailto:step...@networkplumber.org] > Sent: Thursday, 16 May 2024 17.21 > > On Thu, 16 May 2024 10:59:40 +0200 > Morten Brørup wrote: > > > + fprintf(f, " mem_list:\n"); > > + n = 0; > > + STAILQ_FOREACH(memhdr, &mp->mem_list, next) { > > + fprintf(f, "