The manual states that the checksum cannot lie in the last DWORD of the
transmission, so add a basic check for this and fall back to software
checksumming the packet.
This only seems to trigger for ACK packets with no options or data to
return to the other end, and the use of the tx-alignment option makes
it more likely to happen.
Signed-off-by: Ben Dooks <ben.dooks@xxxxxxxxxxxxxxx>
---
Fixes for v2:
- Fix spelling of check at Sergei's suggestion
- Move skb->len check into smsc95xx_can_tx_checksum()
- Change name of smsc95xx_can_checksum to explicitly say it is tx-csum
---
drivers/net/usb/smsc95xx.c | 19 ++++++++++++++++++-
1 file changed, 18 insertions(+), 1 deletion(-)
diff --git a/drivers/net/usb/smsc95xx.c b/drivers/net/usb/smsc95xx.c
index 8f7c473f3260..cc78ef78cc93 100644
--- a/drivers/net/usb/smsc95xx.c
+++ b/drivers/net/usb/smsc95xx.c
@@ -1997,6 +1997,23 @@ static u32 smsc95xx_calc_csum_preamble(struct sk_buff *skb)
return (high_16 << 16) | low_16;
}
+/* The TX CSUM won't work if the checksum lies in the last 4 bytes of the
+ * transmission. This is fairly unlikely, only seems to trigger with some
+ * short TCP ACK packets sent.
+ *
+ * Note, this calculation should probably check for the alignment of the
+ * data as well, but a straight check for csum being in the last four bytes
+ * of the packet should be ok for now.
+*/
+static bool smsc95xx_can_tx_checksum(struct sk_buff *skb)
+{
+ unsigned int len = skb->len - skb_checksum_start_offset(skb);
+
+ if (skb->len <= 45)
+ return false;
+ return skb->csum_offset < (len - (4 + 1));
+}
+
static struct sk_buff *smsc95xx_tx_fixup(struct usbnet *dev,
struct sk_buff *skb, gfp_t flags)
{
@@ -2021,7 +2038,7 @@ static struct sk_buff *smsc95xx_tx_fixup(struct usbnet *dev,
tx_cmd_a = tx_cmd_b | TX_CMD_A_FIRST_SEG_ | TX_CMD_A_LAST_SEG_;
if (csum) {
- if (skb->len <= 45) {
+ if (!smsc95xx_can_tx_checksum(skb)) {
/* workaround - hardware tx checksum does not work
* properly with extremely small packets */
long csstart = skb_checksum_start_offset(skb);
--
2.19.1
