From: Julian Wiedmann <jwi@linux.ibm.com>
Subject: s390/qeth: fix race in used-buffer accounting
Patch-mainline: v4.19-rc1
Git-commit: a702349a4099cd5a7bab0904689d8e0bf8dcd622
References: FATE#326350, LTC#169511, bsc#1113509

Summary:     qeth: performance improvements
Description: This adds recent functional and performance improvements for the
             qeth network driver.
             Primarily this brings Scatter-Gather support for HiperSockets,
             reduced CPU consumption in the L3 IPv4 transmit path for OSA,
             improved Promiscuous Mode performance due to IFF_UNICAST_FLT,
             support for Scatter-Gather on z/VM virtual NICs, and
             support for delayed GRO flushing.

             For sanity & stability reasons, this effectively constitutes a
             backport of the qeth device driver from 4.19 mainline.
             

Upstream-Description:

             s390/qeth: fix race in used-buffer accounting

             By updating q->used_buffers only _after_ do_QDIO() has completed, there
             is a potential race against the buffer's TX completion. In the unlikely
             case that the TX completion path wins, qeth_qdio_output_handler() would
             decrement the counter before qeth_flush_buffers() even incremented it.

             Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
             Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Acked-by: Petr Tesarik <ptesarik@suse.com>
---
 drivers/s390/net/qeth_core_main.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

--- a/drivers/s390/net/qeth_core_main.c
+++ b/drivers/s390/net/qeth_core_main.c
@@ -3505,13 +3505,14 @@ static void qeth_flush_buffers(struct qe
 	qdio_flags = QDIO_FLAG_SYNC_OUTPUT;
 	if (atomic_read(&queue->set_pci_flags_count))
 		qdio_flags |= QDIO_FLAG_PCI_OUT;
+	atomic_add(count, &queue->used_buffers);
+
 	rc = do_QDIO(CARD_DDEV(queue->card), qdio_flags,
 		     queue->queue_no, index, count);
 	if (queue->card->options.performance_stats)
 		queue->card->perf_stats.outbound_do_qdio_time +=
 			qeth_get_micros() -
 			queue->card->perf_stats.outbound_do_qdio_start_time;
-	atomic_add(count, &queue->used_buffers);
 	if (rc) {
 		queue->card->stats.tx_errors += count;
 		/* ignore temporary SIGA errors without busy condition */