From patchwork Thu Jun 10 22:59:40 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mat Martineau X-Patchwork-Id: 12314223 X-Patchwork-Delegate: mat@martineau.name Received: from mga06.intel.com (mga06.intel.com [134.134.136.31]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 3685C6D0D for ; Thu, 10 Jun 2021 22:59:53 +0000 (UTC) IronPort-SDR: UXDOsx6lvFnwMMOJCzM9I2t80gsLpe7Qxci8GYrtBwb/0BC8V3lf5O/VYPs8l4/xv8MslL6AXO v1Uo//PqBBcQ== X-IronPort-AV: E=McAfee;i="6200,9189,10011"; a="266575615" X-IronPort-AV: E=Sophos;i="5.83,264,1616482800"; d="scan'208";a="266575615" Received: from orsmga007.jf.intel.com ([10.7.209.58]) by orsmga104.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 10 Jun 2021 15:59:49 -0700 IronPort-SDR: 1C/FuNPpMYZlVHSvidCbR830ZYzFELsp3/ADTNfNkr98m8J582trb0r/MVRatuwzFHsGkr2TZo 7Z0LwaBGSuAQ== X-IronPort-AV: E=Sophos;i="5.83,264,1616482800"; d="scan'208";a="441387020" Received: from mjmartin-desk2.amr.corp.intel.com (HELO mjmartin-desk2.intel.com) ([10.209.70.185]) by orsmga007-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 10 Jun 2021 15:59:49 -0700 From: Mat Martineau To: netdev@vger.kernel.org Cc: Paolo Abeni , davem@davemloft.net, kuba@kernel.org, matthieu.baerts@tessares.net, mptcp@lists.linux.dev, Mat Martineau Subject: [PATCH net 1/5] mptcp: try harder to borrow memory from subflow under pressure Date: Thu, 10 Jun 2021 15:59:40 -0700 Message-Id: <20210610225944.351224-2-mathew.j.martineau@linux.intel.com> X-Mailer: git-send-email 2.32.0 In-Reply-To: <20210610225944.351224-1-mathew.j.martineau@linux.intel.com> References: <20210610225944.351224-1-mathew.j.martineau@linux.intel.com> X-Mailing-List: mptcp@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 From: Paolo Abeni If the host is under sever memory pressure, and RX forward memory allocation for the msk fails, we try to borrow the required memory from the ingress subflow. The current attempt is a bit flaky: if skb->truesize is less than SK_MEM_QUANTUM, the ssk will not release any memory, and the next schedule will fail again. Instead, directly move the required amount of pages from the ssk to the msk, if available Fixes: 9c3f94e1681b ("mptcp: add missing memory scheduling in the rx path") Signed-off-by: Paolo Abeni Signed-off-by: Mat Martineau --- net/mptcp/protocol.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index 5edc686faff1..534cf500521d 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -280,11 +280,13 @@ static bool __mptcp_move_skb(struct mptcp_sock *msk, struct sock *ssk, /* try to fetch required memory from subflow */ if (!sk_rmem_schedule(sk, skb, skb->truesize)) { - if (ssk->sk_forward_alloc < skb->truesize) - goto drop; - __sk_mem_reclaim(ssk, skb->truesize); - if (!sk_rmem_schedule(sk, skb, skb->truesize)) + int amount = sk_mem_pages(skb->truesize) << SK_MEM_QUANTUM_SHIFT; + + if (ssk->sk_forward_alloc < amount) goto drop; + + ssk->sk_forward_alloc -= amount; + sk->sk_forward_alloc += amount; } /* the skb map_seq accounts for the skb offset: