[MPTCP] Re: [PATCH net-next] Squash-to: "mptcp: move page frag allocation in mptcp_sendmsg()"

[MPTCP] Re: [PATCH net-next] Squash-to: "mptcp: move page frag allocation in mptcp_sendmsg()"

[Mat Martineau]

[-- Attachment #1: Type: text/plain, Size: 1605 bytes --]


On Wed, 21 Oct 2020, Paolo Abeni wrote:

> better conflict resolution for a recent rebase
>
> Signed-off-by: Paolo Abeni <pabeni(a)redhat.com>
> ---
> net/mptcp/protocol.c | 12 +++++-------
> 1 file changed, 5 insertions(+), 7 deletions(-)
>
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index 897c0c1d2888..975a3243d55c 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -978,8 +978,8 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
> 	struct mptcp_ext *mpext = NULL;
> 	struct sk_buff *skb, *tail;
> 	bool can_collapse = false;
> -	int avail_size, ret;
> -	size_t psize;
> +	int avail_size;
> +	size_t ret;
>
> 	pr_debug("msk=%p ssk=%p sending dfrag at seq=%lld len=%d already sent=%d",
> 		 msk, ssk, dfrag->data_seq, dfrag->data_len, info->sent);
> @@ -1018,15 +1018,13 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
> 			 info->limit > dfrag->data_len))
> 		return 0;
>
> -	psize = min_t(size_t, info->limit - info->sent, avail_size);
> -
> -	tail = tcp_build_frag(ssk, psize, info->flags, dfrag->page,
> -			      dfrag->offset + info->sent, &psize);
> +	ret = info->limit - info->sent;
> +	tail = tcp_build_frag(ssk, avail_size, info->flags, dfrag->page,
> +			      dfrag->offset + info->sent, &ret);
> 	if (!tail) {
> 		tcp_remove_empty_skb(sk, tcp_write_queue_tail(ssk));
> 		return -ENOMEM;
> 	}
> -	ret = psize;
>
> 	/* if the tail skb is still the cached one, collapsing really happened.
> 	 */
> -- 
> 2.26.2

Agreed, looks ready to squash.

--
Mat Martineau
Intel

[MPTCP] Re: [PATCH net-next] Squash-to: "mptcp: move page frag allocation in mptcp_sendmsg()"

[Matthieu Baerts]

[-- Attachment #1: Type: text/plain, Size: 419 bytes --]

Hi Paolo, Mat,

On 21/10/2020 18:45, Paolo Abeni wrote:
> better conflict resolution for a recent rebase

Thanks for the patch fixing my mistakes and for the review!

- 587ee99bf694: "squashed" in "mptcp: move page frag allocation in 
mptcp_sendmsg()"
- Results: 66965b9869c7..a1e156132206

Tests + export are in progress!

Cheers,
Matt
-- 
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net

[MPTCP] Re: [PATCH net-next] Squash-to: mptcp: move page frag allocation in mptcp_sendmsg()

[Paolo Abeni]

[-- Attachment #1: Type: text/plain, Size: 583 bytes --]

On Tue, 2020-10-06 at 14:20 +0200, Matthieu Baerts wrote:
> Hi Paolo,
> 
> On 05/10/2020 11:00, Paolo Abeni wrote:
> > This, as suggested by Florian, avoid probe coalescing, which
> > will hit a later WARN_ON_ONCE() and will cause stream corruption.
> > 
> > Additionally bail early when the MPTCP win is closed an no
> > probe.
> 
> Thank you for the patch!
> 
> I guess it was for "mptcp: keep track of advertised windows right edge" 
> commit and not "mptcp: move page frag allocation in mptcp_sendmsg()".

Indeed! thanks for the double check && correction!

/P

[MPTCP] Re: [PATCH net-next] Squash-to: mptcp: move page frag allocation in mptcp_sendmsg()

[Matthieu Baerts]

[-- Attachment #1: Type: text/plain, Size: 697 bytes --]

Hi Paolo,

On 05/10/2020 11:00, Paolo Abeni wrote:
> This, as suggested by Florian, avoid probe coalescing, which
> will hit a later WARN_ON_ONCE() and will cause stream corruption.
> 
> Additionally bail early when the MPTCP win is closed an no
> probe.

Thank you for the patch!

I guess it was for "mptcp: keep track of advertised windows right edge" 
commit and not "mptcp: move page frag allocation in mptcp_sendmsg()".

- 04554e1b1dfa: "squashed" in "mptcp: keep track of advertised windows 
right edge"
- Results: 345ad770cca7..10c53117392d

tests + export are going to be started soon.

Cheers,
Matt
-- 
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net