[PATCH 3.2 13/94] ibmvscsi: Add memory barriers for send / receive

[Ben Hutchings]

3.2.62-rc1 review patch.  If anyone has any objections, please let me know.

------------------

From: Brian King <brking@xxxxxxxxxxxxxxxxxx>

commit 7114aae02742d6b5c5a0d39a41deb61d415d3717 upstream.

Add a memory barrier prior to sending a new command to the VIOS
to ensure the VIOS does not receive stale data in the command buffer.
Also add a memory barrier when processing the CRQ for completed commands.

Signed-off-by: Brian King <brking@xxxxxxxxxxxxxxxxxx>
Acked-by: Nathan Fontenot <nfont@xxxxxxxxxxxxxxxxxx>
Signed-off-by: Christoph Hellwig <hch@xxxxxx>
[bwh: Backported to 3.2: as the iSeries code is still present, these
 functions have different names and live in rpa_vscsi.c.]
Signed-off-by: Ben Hutchings <ben@xxxxxxxxxxxxxxx>
---
 drivers/scsi/ibmvscsi/ibmvscsi.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

--- a/drivers/scsi/ibmvscsi/rpa_vscsi.c
+++ b/drivers/scsi/ibmvscsi/rpa_vscsi.c
@@ -104,6 +104,11 @@ static struct viosrp_crq *crq_queue_next
 	if (crq->valid & 0x80) {
 		if (++queue->cur == queue->size)
 			queue->cur = 0;
+
+		/* Ensure the read of the valid bit occurs before reading any
+		 * other bits of the CRQ entry
+		 */
+		rmb();
 	} else
 		crq = NULL;
 	spin_unlock_irqrestore(&queue->lock, flags);
@@ -122,6 +127,11 @@ static int rpavscsi_send_crq(struct ibmv
 {
 	struct vio_dev *vdev = to_vio_dev(hostdata->dev);
 
+	/*
+	 * Ensure the command buffer is flushed to memory before handing it
+	 * over to the VIOS to prevent it from fetching any stale data.
+	 */
+	mb();
 	return plpar_hcall_norets(H_SEND_CRQ, vdev->unit_address, word1, word2);
 }
 

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/