DPDK patches and discussions
 help / color / mirror / Atom feed
From: Stephen Hemminger <stephen@networkplumber.org>
To: scott.k.mitch1@gmail.com
Cc: dev@dpdk.org, mb@smartsharesystems.com
Subject: Re: [PATCH v9] net: optimize raw checksum computation
Date: Thu, 8 Jan 2026 21:34:43 -0800	[thread overview]
Message-ID: <20260108213443.0ab2f0c3@phoenix.local> (raw)
In-Reply-To: <20260108214713.52987-1-scott.k.mitch1@gmail.com>

On Thu,  8 Jan 2026 16:47:13 -0500
scott.k.mitch1@gmail.com wrote:

> diff --git a/lib/net/rte_ip4.h b/lib/net/rte_ip4.h
> index 822a660cfb..63852717c9 100644
> --- a/lib/net/rte_ip4.h
> +++ b/lib/net/rte_ip4.h
> @@ -223,21 +223,17 @@ rte_ipv4_phdr_cksum(const struct rte_ipv4_hdr *ipv4_hdr, uint64_t ol_flags)
>  		uint8_t  zero;     /* zero. */
>  		uint8_t  proto;    /* L4 protocol type. */
>  		uint16_t len;      /* L4 length. */
> -	} psd_hdr;
> -
> -	uint32_t l3_len;
> -
> -	psd_hdr.src_addr = ipv4_hdr->src_addr;
> -	psd_hdr.dst_addr = ipv4_hdr->dst_addr;
> -	psd_hdr.zero = 0;
> -	psd_hdr.proto = ipv4_hdr->next_proto_id;
> -	if (ol_flags & (RTE_MBUF_F_TX_TCP_SEG | RTE_MBUF_F_TX_UDP_SEG)) {
> -		psd_hdr.len = 0;
> -	} else {
> -		l3_len = rte_be_to_cpu_16(ipv4_hdr->total_length);
> -		psd_hdr.len = rte_cpu_to_be_16((uint16_t)(l3_len -
> -			rte_ipv4_hdr_len(ipv4_hdr)));
> -	}
> +	} psd_hdr = {
> +		.src_addr = ipv4_hdr->src_addr,
> +		.dst_addr = ipv4_hdr->dst_addr,
> +		.proto = ipv4_hdr->next_proto_id,
> +		.len = (ol_flags & (RTE_MBUF_F_TX_TCP_SEG | RTE_MBUF_F_TX_UDP_SEG))
> +			? (uint16_t)0
> +			: rte_cpu_to_be_16((uint16_t)(rte_be_to_cpu_16(ipv4_hdr->total_length) -
> +					rte_ipv4_hdr_len(ipv4_hdr)))
> +	};
> +	RTE_SUPPRESS_UNINITIALIZED_WARNING(psd_hdr);
> +
>  	return rte_raw_cksum(&psd_hdr, sizeof(psd_hdr));
>  }

Since this is all fixed size, why not hand unroll it?
That is what Linux kernel does for IPv4 checksum and this would be
another case. No loop is really necessary.

static inline uint16_t
rte_ipv4_phdr_cksum(const struct rte_ipv4_hdr *ipv4_hdr, uint64_t ol_flags)
{
	union {
		struct {
			uint32_t src_addr;
			uint32_t dst_addr;
			uint8_t  zero;
			uint8_t  proto;
			rte_be16_t len;
		} psd;
		uint16_t u16[6];
	} hdr = {
		.psd = {
			.src_addr = ipv4_hdr->src_addr,
			.dst_addr = ipv4_hdr->dst_addr,
			.proto = ipv4_hdr->next_proto_id,
		}
	};
	uint32_t sum;

	if (!(ol_flags & (RTE_MBUF_F_TX_TCP_SEG | RTE_MBUF_F_TX_UDP_SEG))) {
		hdr.psd.len = rte_cpu_to_be_16(
			(uint16_t)(rte_be_to_cpu_16(ipv4_hdr->total_length) -
				   rte_ipv4_hdr_len(ipv4_hdr)));
	}

	/* Unrolled sum of 6 uint16_t words */
	sum = hdr.u16[0] + hdr.u16[1] + hdr.u16[2] +
	      hdr.u16[3] + hdr.u16[4] + hdr.u16[5];

	sum = (sum & 0xffff) + (sum >> 16);
	sum = (sum & 0xffff) + (sum >> 16);
	return (uint16_t)sum;
}

I.e Don't use the rte_raw_cksum at all.

Same for IPv6.

  reply	other threads:[~2026-01-09  5:34 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-01-08 21:47 scott.k.mitch1
2026-01-09  5:34 ` Stephen Hemminger [this message]
2026-01-09  5:36 ` Stephen Hemminger
2026-01-09  5:39 ` Stephen Hemminger
2026-01-09 17:50   ` Scott Mitchell
2026-01-10  3:46     ` Scott Mitchell

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260108213443.0ab2f0c3@phoenix.local \
    --to=stephen@networkplumber.org \
    --cc=dev@dpdk.org \
    --cc=mb@smartsharesystems.com \
    --cc=scott.k.mitch1@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).