[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20170511141212.870537195@linuxfoundation.org>
Date: Thu, 11 May 2017 16:11:57 +0200
From: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To: linux-kernel@...r.kernel.org
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
stable@...r.kernel.org, Xinming Hu <huxm@...vell.com>,
Amitkumar Karwar <akarwar@...vell.com>,
Kalle Valo <kvalo@...eaurora.org>
Subject: [PATCH 4.9 028/103] mwifiex: remove redundant dma padding in AMSDU
4.9-stable review patch. If anyone has any objections, please let me know.
------------------
From: Xinming Hu <huxm@...vell.com>
commit 5f0a221f59ad6b72202ef9c6e232086de8c336f2 upstream.
We already ensure 64 bytes alignment and add padding if required
during skb_aggr allocation.
Alignment and padding in mwifiex_11n_form_amsdu_txpd() is redundant.
We may end up accessing more data than allocated size with this.
This patch fixes following issue by removing redundant padding.
[ 370.241338] skbuff: skb_over_panic: text:ffffffffc046946a len:3550
put:72 head:ffff880000110000 data:ffff8800001100e4 tail:0xec2 end:0xec0 dev:<NULL>
[ 370.241374] ------------[ cut here ]------------
[ 370.241382] kernel BUG at net/core/skbuff.c:104!
370.244032] Call Trace:
[ 370.244041] [<ffffffff8c3df5ec>] skb_put+0x44/0x45
[ 370.244055] [<ffffffffc046946a>]
mwifiex_11n_aggregate_pkt+0x1e9/0xa50 [mwifiex]
[ 370.244067] [<ffffffffc0467c16>] mwifiex_wmm_process_tx+0x44a/0x6b7
[mwifiex]
[ 370.244074] [<ffffffffc0411eb8>] ? 0xffffffffc0411eb8
[ 370.244084] [<ffffffffc046116b>] mwifiex_main_process+0x476/0x5a5
[mwifiex]
[ 370.244098] [<ffffffffc0461298>] mwifiex_main_process+0x5a3/0x5a5
[mwifiex]
[ 370.244113] [<ffffffff8be7e9ff>] process_one_work+0x1a4/0x309
[ 370.244123] [<ffffffff8be7f4ca>] worker_thread+0x20c/0x2ee
[ 370.244130] [<ffffffff8be7f2be>] ? rescuer_thread+0x383/0x383
[ 370.244136] [<ffffffff8be7f2be>] ? rescuer_thread+0x383/0x383
[ 370.244143] [<ffffffff8be83742>] kthread+0x11c/0x124
[ 370.244150] [<ffffffff8be83626>] ? kthread_parkme+0x24/0x24
[ 370.244157] [<ffffffff8c4da1ef>] ret_from_fork+0x3f/0x70
[ 370.244168] [<ffffffff8be83626>] ? kthread_parkme+0x24/0x24
Fixes: 84b313b35f8158d ("mwifiex: make tx packet 64 byte DMA aligned")
Signed-off-by: Xinming Hu <huxm@...vell.com>
Signed-off-by: Amitkumar Karwar <akarwar@...vell.com>
Signed-off-by: Kalle Valo <kvalo@...eaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
---
drivers/net/wireless/marvell/mwifiex/11n_aggr.c | 19 +++++++------------
1 file changed, 7 insertions(+), 12 deletions(-)
--- a/drivers/net/wireless/marvell/mwifiex/11n_aggr.c
+++ b/drivers/net/wireless/marvell/mwifiex/11n_aggr.c
@@ -101,13 +101,6 @@ mwifiex_11n_form_amsdu_txpd(struct mwifi
{
struct txpd *local_tx_pd;
struct mwifiex_txinfo *tx_info = MWIFIEX_SKB_TXCB(skb);
- unsigned int pad;
- int headroom = (priv->adapter->iface_type ==
- MWIFIEX_USB) ? 0 : INTF_HEADER_LEN;
-
- pad = ((void *)skb->data - sizeof(*local_tx_pd) -
- headroom - NULL) & (MWIFIEX_DMA_ALIGN_SZ - 1);
- skb_push(skb, pad);
skb_push(skb, sizeof(*local_tx_pd));
@@ -121,12 +114,10 @@ mwifiex_11n_form_amsdu_txpd(struct mwifi
local_tx_pd->bss_num = priv->bss_num;
local_tx_pd->bss_type = priv->bss_type;
/* Always zero as the data is followed by struct txpd */
- local_tx_pd->tx_pkt_offset = cpu_to_le16(sizeof(struct txpd) +
- pad);
+ local_tx_pd->tx_pkt_offset = cpu_to_le16(sizeof(struct txpd));
local_tx_pd->tx_pkt_type = cpu_to_le16(PKT_TYPE_AMSDU);
local_tx_pd->tx_pkt_length = cpu_to_le16(skb->len -
- sizeof(*local_tx_pd) -
- pad);
+ sizeof(*local_tx_pd));
if (tx_info->flags & MWIFIEX_BUF_FLAG_TDLS_PKT)
local_tx_pd->flags |= MWIFIEX_TXPD_FLAGS_TDLS_PACKET;
@@ -190,7 +181,11 @@ mwifiex_11n_aggregate_pkt(struct mwifiex
ra_list_flags);
return -1;
}
- skb_reserve(skb_aggr, MWIFIEX_MIN_DATA_HEADER_LEN);
+
+ /* skb_aggr->data already 64 byte align, just reserve bus interface
+ * header and txpd.
+ */
+ skb_reserve(skb_aggr, headroom + sizeof(struct txpd));
tx_info_aggr = MWIFIEX_SKB_TXCB(skb_aggr);
memset(tx_info_aggr, 0, sizeof(*tx_info_aggr));
Powered by blists - more mailing lists