netdev - [PATCH net-next v3 2/2] net: mhi: Allow decoupled MTU/MRU

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 29 Mar 2021 17:39:32 +0200
From:   Loic Poulain <loic.poulain@...aro.org>
To:     kuba@...nel.org, davem@...emloft.net
Cc:     netdev@...r.kernel.org, Loic Poulain <loic.poulain@...aro.org>
Subject: [PATCH net-next v3 2/2] net: mhi: Allow decoupled MTU/MRU

MBIM protocol makes the mhi network interface asymmetric, ingress data
received from MHI is MBIM protocol, possibly containing multiple
aggregated IP packets, while egress data received from network stack is
IP protocol.

This changes allows a 'protocol' to specify its own MRU, that when
specified is used to allocate MHI RX buffers (skb).

For MBIM, Set the default MTU to 1500, which is the usual network MTU
for WWAN IP packets, and MRU to 3.5K (for allocation efficiency),
allowing skb to fit in an usual 4K page (including padding,
skb_shared_info, ...).

Signed-off-by: Loic Poulain <loic.poulain@...aro.org>
---
  v2: squashed 1/2 and 1/2 (decoupled mtu/mru + mbim mru)
     Reduced MRU from 32k to 3.5K
  v3: no change

 drivers/net/mhi/mhi.h        |  1 +
 drivers/net/mhi/net.c        |  4 +++-
 drivers/net/mhi/proto_mbim.c | 11 +++++++++++
 3 files changed, 15 insertions(+), 1 deletion(-)

diff --git a/drivers/net/mhi/mhi.h b/drivers/net/mhi/mhi.h
index 12e7407..1d0c499 100644
--- a/drivers/net/mhi/mhi.h
+++ b/drivers/net/mhi/mhi.h
@@ -29,6 +29,7 @@ struct mhi_net_dev {
 	struct mhi_net_stats stats;
 	u32 rx_queue_sz;
 	int msg_enable;
+	unsigned int mru;
 };
 
 struct mhi_net_proto {
diff --git a/drivers/net/mhi/net.c b/drivers/net/mhi/net.c
index b1769fb..4bef7fe 100644
--- a/drivers/net/mhi/net.c
+++ b/drivers/net/mhi/net.c
@@ -270,10 +270,12 @@ static void mhi_net_rx_refill_work(struct work_struct *work)
 						      rx_refill.work);
 	struct net_device *ndev = mhi_netdev->ndev;
 	struct mhi_device *mdev = mhi_netdev->mdev;
-	int size = READ_ONCE(ndev->mtu);
 	struct sk_buff *skb;
+	unsigned int size;
 	int err;
 
+	size = mhi_netdev->mru ? mhi_netdev->mru : READ_ONCE(ndev->mtu);
+
 	while (!mhi_queue_is_full(mdev, DMA_FROM_DEVICE)) {
 		skb = netdev_alloc_skb(ndev, size);
 		if (unlikely(!skb))
diff --git a/drivers/net/mhi/proto_mbim.c b/drivers/net/mhi/proto_mbim.c
index 5a176c3..fc72b3f 100644
--- a/drivers/net/mhi/proto_mbim.c
+++ b/drivers/net/mhi/proto_mbim.c
@@ -26,6 +26,15 @@
 
 #define MBIM_NDP16_SIGN_MASK 0x00ffffff
 
+/* Usual WWAN MTU */
+#define MHI_MBIM_DEFAULT_MTU 1500
+
+/* 3500 allows to optimize skb allocation, the skbs will basically fit in
+ * one 4K page. Large MBIM packets will simply be split over several MHI
+ * transfers and chained by the MHI net layer (zerocopy).
+ */
+#define MHI_MBIM_DEFAULT_MRU 3500
+
 struct mbim_context {
 	u16 rx_seq;
 	u16 tx_seq;
@@ -281,6 +290,8 @@ static int mbim_init(struct mhi_net_dev *mhi_netdev)
 		return -ENOMEM;
 
 	ndev->needed_headroom = sizeof(struct mbim_tx_hdr);
+	ndev->mtu = MHI_MBIM_DEFAULT_MTU;
+	mhi_netdev->mru = MHI_MBIM_DEFAULT_MRU;
 
 	return 0;
 }
-- 
2.7.4