From: Bailey Forrest <bcf@xxxxxxxxxx>
The NIC requires each TSO segment to not span more than 10
descriptors. NIC further requires each descriptor to not exceed
16KB - 1 (GVE_TX_MAX_BUF_SIZE_DQO).
The descriptors for an skb are generated by
gve_tx_add_skb_no_copy_dqo() for DQO RDA queue format.
gve_tx_add_skb_no_copy_dqo() loops through each skb frag and
generates a descriptor for the entire frag if the frag size is
not greater than GVE_TX_MAX_BUF_SIZE_DQO. If the frag size is
greater than GVE_TX_MAX_BUF_SIZE_DQO, it is split into descriptor(s)
of size GVE_TX_MAX_BUF_SIZE_DQO and a descriptor is generated for
the remainder (frag size % GVE_TX_MAX_BUF_SIZE_DQO).
gve_can_send_tso() checks if the descriptors thus generated for an
skb would meet the requirement that each TSO-segment not span more
than 10 descriptors. However, the current code misses an edge case
when a TSO segment spans multiple descriptors within a large frag.
This change fixes the edge case.
gve_can_send_tso() relies on the assumption that max gso size (9728)
is less than GVE_TX_MAX_BUF_SIZE_DQO and therefore within an skb
fragment a TSO segment can never span more than 2 descriptors.
Fixes: a57e5de476be ("gve: DQO: Add TX path")
Signed-off-by: Praveen Kaligineedi <pkaligineedi@xxxxxxxxxx>
Signed-off-by: Bailey Forrest <bcf@xxxxxxxxxx>
Reviewed-by: Jeroen de Borst <jeroendb@xxxxxxxxxx>
Cc: stable@xxxxxxxxxxxxxxx
---
Changes from v1:
- Added 'stable tag'
- Added more explanation in the commit message
- Modified comments to clarify the changes made
- Changed variable names 'last_frag_size' to 'prev_frag_size' and
'last_frag_remain' to 'prev_frag_remain'
- Removed parentheses around single line statement
drivers/net/ethernet/google/gve/gve_tx_dqo.c | 22 +++++++++++++++++++-
1 file changed, 21 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/google/gve/gve_tx_dqo.c b/drivers/net/ethernet/google/gve/gve_tx_dqo.c
index 0b3cca3fc792..f879426cb552 100644
--- a/drivers/net/ethernet/google/gve/gve_tx_dqo.c
+++ b/drivers/net/ethernet/google/gve/gve_tx_dqo.c
@@ -866,22 +866,42 @@ static bool gve_can_send_tso(const struct sk_buff *skb)
const int header_len = skb_tcp_all_headers(skb);
const int gso_size = shinfo->gso_size;
int cur_seg_num_bufs;
+ int prev_frag_size;
int cur_seg_size;
int i;
cur_seg_size = skb_headlen(skb) - header_len;
+ prev_frag_size = skb_headlen(skb);
cur_seg_num_bufs = cur_seg_size > 0;
for (i = 0; i < shinfo->nr_frags; i++) {
if (cur_seg_size >= gso_size) {
cur_seg_size %= gso_size;
cur_seg_num_bufs = cur_seg_size > 0;
+
+ if (prev_frag_size > GVE_TX_MAX_BUF_SIZE_DQO) {
+ int prev_frag_remain = prev_frag_size %
+ GVE_TX_MAX_BUF_SIZE_DQO;
+
+ /* If the last descriptor of the previous frag
+ * is less than cur_seg_size, the segment will
+ * span two descriptors in the previous frag.
+ * Since max gso size (9728) is less than
+ * GVE_TX_MAX_BUF_SIZE_DQO, it is impossible
+ * for the segment to span more than two
+ * descriptors.
+ */
+ if (prev_frag_remain &&
+ cur_seg_size > prev_frag_remain)
+ cur_seg_num_bufs++;
+ }
}
if (unlikely(++cur_seg_num_bufs > max_bufs_per_seg))
return false;
- cur_seg_size += skb_frag_size(&shinfo->frags[i]);
+ prev_frag_size = skb_frag_size(&shinfo->frags[i]);
+ cur_seg_size += prev_frag_size;
}
return true;
--
2.45.2.1089.g2a221341d9-goog
