Fix an off-by-one error when deciding to request a tx interrupt

The canonical check for whether or not a ring is drainable is
TXQ_AVAIL() > MAX_TX_DESC() + 2.  Use this same construct here,
in order to avoid a potential off-by-one error where we might otherwise
fail to request an interrupt.

Reviewed by:	mmacy
Sponsored by:	Netflix
This commit is contained in:
Andrew Gallatin 2018-05-07 18:11:22 +00:00
parent f185a3dc33
commit 1f7ce05d1d
Notes: svn2git 2020-12-20 02:59:44 +00:00
svn path=/head/; revision=333329

View file

@ -3299,7 +3299,7 @@ iflib_encap(iflib_txq_t txq, struct mbuf **m_headp)
*/
txq->ift_rs_pending += nsegs + 1;
if (txq->ift_rs_pending > TXQ_MAX_RS_DEFERRED(txq) ||
iflib_no_tx_batch || (TXQ_AVAIL(txq) - nsegs - 1) <= MAX_TX_DESC(ctx)) {
iflib_no_tx_batch || (TXQ_AVAIL(txq) - nsegs) <= MAX_TX_DESC(ctx) + 2) {
pi.ipi_flags |= IPI_TX_INTR;
txq->ift_rs_pending = 0;
}