Skip to contentSkip to conversation

net: fix pskb_trim_rcsum_slow() with odd trim offset

PATCH In review Dimitris Michailidis 3 participants 3 messages Oct 20, 2018

Subsystems (2)
Networking Network drivers
lore
Share & tools
lore

Dimitris Michailidis

author
#0 lore raw

PATCH 1/1 net: fix pskb_trim_rcsum_slow() with odd trim offset

We've been getting checksum errors involving small UDP packets, usually 59B packets with 1 extra non-zero padding byte. netdev_rx_csum_fault() has been complaining that HW is providing bad checksums. Turns out the problem is in pskb_trim_rcsum_slow(), introduced in commit 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends").

The source of the problem is that when the bytes we are trimming start at an odd address, as in the case of the 1 padding byte above, skb_checksum() returns a byte-swapped value. We cannot just combine this with skb->csum using csum_sub(). We need to use csum_block_sub() here that takes into account the parity of the start address and handles the swapping.

Matches existing code in __skb_postpull_rcsum() and esp_remove_trailer().

Fixes 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
Signed-off-by Dimitris Michailidis <[email protected]>
---
 net/core/skbuff.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
net/core/skbuff.c +3 -2
index 428094b577fc..f817f336595d 100644
@@ -1846,8 +1846,9 @@ int pskb_trim_rcsum_slow(struct sk_buff *skb, unsigned int len)
 	if (skb->ip_summed == CHECKSUM_COMPLETE) {
 		int delta = skb->len - len;
 
-		skb->csum = csum_sub(skb->csum,
-				     skb_checksum(skb, len, delta, 0));
+		skb->csum = csum_block_sub(skb->csum,
+					   skb_checksum(skb, len, delta, 0),
+					   len);
 	}
 	return __pskb_trim(skb, len);
 }
Signature · 1 line
2.19.1.568.g152ad8e336-goog
Reviewed-by on v1 1/1
#1 lore raw
Quoted 13 lines · Dimitris Michailidis

We've been getting checksum errors involving small UDP packets, usually 59B packets with 1 extra non-zero padding byte. netdev_rx_csum_fault() has been complaining that HW is providing bad checksums. Turns out the problem is in pskb_trim_rcsum_slow(), introduced in commit 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends").

The source of the problem is that when the bytes we are trimming start at an odd address, as in the case of the 1 padding byte above, skb_checksum() returns a byte-swapped value. We cannot just combine this with skb->csum using csum_sub(). We need to use csum_block_sub() here that takes into account the parity of the start address and handles the swapping.

Quoted 4 lines

Matches existing code in __skb_postpull_rcsum() and esp_remove_trailer().

Fixes 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
Signed-off-by Dimitris Michailidis <[email protected]>

Thanks a lot Dimitris for finding this.

Reviewed-by Eric Dumazet <[email protected]>
maintainer
#2 lore raw
In reply to Dimitris Michailidis

From: Dimitris Michailidis <[email protected]>
Date: Fri, 19 Oct 2018 17:07:13 -0700

Quoted 13 lines

We've been getting checksum errors involving small UDP packets, usually 59B packets with 1 extra non-zero padding byte. netdev_rx_csum_fault() has been complaining that HW is providing bad checksums. Turns out the problem is in pskb_trim_rcsum_slow(), introduced in commit 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends").

The source of the problem is that when the bytes we are trimming start at an odd address, as in the case of the 1 padding byte above, skb_checksum() returns a byte-swapped value. We cannot just combine this with skb->csum using csum_sub(). We need to use csum_block_sub() here that takes into account the parity of the start address and handles the swapping.

Quoted 4 lines

Matches existing code in __skb_postpull_rcsum() and esp_remove_trailer().

Fixes 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
Signed-off-by Dimitris Michailidis <[email protected]>

Applied and queued up for -stable, thanks!