This is a note to let you know that I've just added the patch titled
RDMA/siw: Always consume all skbuf data in sk_data_ready() upcall.
to the 5.10-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary
The filename of the patch is:
rdma-siw-always-consume-all-skbuf-data-in-sk_data_re.patch
and it can be found in the queue-5.10 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@xxxxxxxxxxxxxxx> know about it.
commit 5f89be1d18980d565b0f63415f09c4028e990073
Author: Bernard Metzler <bmt@xxxxxxxxxxxxxx>
Date: Tue Sep 20 10:12:02 2022 +0200
RDMA/siw: Always consume all skbuf data in sk_data_ready() upcall.
[ Upstream commit 754209850df8367c954ac1de7671c7430b1f342c ]
For header and trailer/padding processing, siw did not consume new
skb data until minimum amount present to fill current header or trailer
structure, including potential payload padding. Not consuming any
data during upcall may cause a receive stall, since tcp_read_sock()
is not upcalling again if no new data arrive.
A NFSoRDMA client got stuck at RDMA Write reception of unaligned
payload, if the current skb did contain only the expected 3 padding
bytes, but not the 4 bytes CRC trailer. Expecting 4 more bytes already
arrived in another skb, and not consuming those 3 bytes in the current
upcall left the Write incomplete, waiting for the CRC forever.
Fixes: 8b6a361b8c48 ("rdma/siw: receive path")
Reported-by: Olga Kornievskaia <kolga@xxxxxxxxxx>
Tested-by: Olga Kornievskaia <kolga@xxxxxxxxxx>
Signed-off-by: Bernard Metzler <bmt@xxxxxxxxxxxxxx>
Link: https://lore.kernel.org/r/20220920081202.223629-1-bmt@xxxxxxxxxxxxxx
Signed-off-by: Leon Romanovsky <leon@xxxxxxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
diff --git a/drivers/infiniband/sw/siw/siw_qp_rx.c b/drivers/infiniband/sw/siw/siw_qp_rx.c
index 875ea6f1b04a..fd721cc19682 100644
--- a/drivers/infiniband/sw/siw/siw_qp_rx.c
+++ b/drivers/infiniband/sw/siw/siw_qp_rx.c
@@ -961,27 +961,28 @@ int siw_proc_terminate(struct siw_qp *qp)
static int siw_get_trailer(struct siw_qp *qp, struct siw_rx_stream *srx)
{
struct sk_buff *skb = srx->skb;
+ int avail = min(srx->skb_new, srx->fpdu_part_rem);
u8 *tbuf = (u8 *)&srx->trailer.crc - srx->pad;
__wsum crc_in, crc_own = 0;
siw_dbg_qp(qp, "expected %d, available %d, pad %u\n",
srx->fpdu_part_rem, srx->skb_new, srx->pad);
- if (srx->skb_new < srx->fpdu_part_rem)
- return -EAGAIN;
-
- skb_copy_bits(skb, srx->skb_offset, tbuf, srx->fpdu_part_rem);
+ skb_copy_bits(skb, srx->skb_offset, tbuf, avail);
- if (srx->mpa_crc_hd && srx->pad)
- crypto_shash_update(srx->mpa_crc_hd, tbuf, srx->pad);
+ srx->skb_new -= avail;
+ srx->skb_offset += avail;
+ srx->skb_copied += avail;
+ srx->fpdu_part_rem -= avail;
- srx->skb_new -= srx->fpdu_part_rem;
- srx->skb_offset += srx->fpdu_part_rem;
- srx->skb_copied += srx->fpdu_part_rem;
+ if (srx->fpdu_part_rem)
+ return -EAGAIN;
if (!srx->mpa_crc_hd)
return 0;
+ if (srx->pad)
+ crypto_shash_update(srx->mpa_crc_hd, tbuf, srx->pad);
/*
* CRC32 is computed, transmitted and received directly in NBO,
* so there's never a reason to convert byte order.
@@ -1083,10 +1084,9 @@ static int siw_get_hdr(struct siw_rx_stream *srx)
* completely received.
*/
if (iwarp_pktinfo[opcode].hdr_len > sizeof(struct iwarp_ctrl_tagged)) {
- bytes = iwarp_pktinfo[opcode].hdr_len - MIN_DDP_HDR;
+ int hdrlen = iwarp_pktinfo[opcode].hdr_len;
- if (srx->skb_new < bytes)
- return -EAGAIN;
+ bytes = min_t(int, hdrlen - MIN_DDP_HDR, srx->skb_new);
skb_copy_bits(skb, srx->skb_offset,
(char *)c_hdr + srx->fpdu_part_rcvd, bytes);
@@ -1096,6 +1096,9 @@ static int siw_get_hdr(struct siw_rx_stream *srx)
srx->skb_new -= bytes;
srx->skb_offset += bytes;
srx->skb_copied += bytes;
+
+ if (srx->fpdu_part_rcvd < hdrlen)
+ return -EAGAIN;
}
/*
