SVCRDMA: Check num_sge when setting LAST_CTXT bit

From: Tom Tucker <[email protected]>

The RDMACTXT_F_LAST_CTXT bit was getting set incorrectly when the last chunk in the read-list spanned multiple pages. This resulted in a kernel panic when the wrong context was used to build the RPC iovec page list.

RDMA_READ is used to fetch RPC data from the client for NFS_WRITE requests. A scatter-gather is used to map the advertised client side buffer to the server-side iovec and associated page list.

WR contexts are used to convey which scatter-gather entries are handled by each WR. When the write data is large, a single RPC may require multiple RDMA_READ requests so the contexts for a single RPC are chained together in a linked list. The last context in this list is marked with a bit RDMACTXT_F_LAST_CTXT so that when this WR completes, the CQ handler code can enqueue the RPC for processing.

The code in rdma_read_xdr was setting this bit on the last two contexts on this list when the last read-list chunk spanned multiple pages. This caused the svc_rdma_recvfrom logic to incorrectly build the RPC and caused the kernel to crash because the second-to-last context doesn't contain the iovec page list.

Modified the condition that sets this bit so that it correctly detects the last context for the RPC.

---
 net/sunrpc/xprtrdma/svc_rdma_recvfrom.c |   21 +++++++++++----------
 1 files changed, 11 insertions(+), 10 deletions(-)

One more server-side rdma patch for 2.6.25.

--b.

index 9712716..c22d6b6 100644
@@ -322,15 +322,6 @@ next_sge:
 		ctxt->direction = DMA_FROM_DEVICE;
 		clear_bit(RDMACTXT_F_READ_DONE, &ctxt->flags);
 		clear_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags);
-		if ((ch+1)->rc_discrim == 0) {
-			/*
-			 * Checked in sq_cq_reap to see if we need to
-			 * be enqueued
-			 */
-			set_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags);
-			ctxt->next = hdr_ctxt;
-			hdr_ctxt->next = head;
-		}
 
 		/* Prepare READ WR */
 		memset(&read_wr, 0, sizeof read_wr);

@@ -348,7 +339,17 @@ next_sge:
 		rdma_set_ctxt_sge(ctxt, &sge[ch_sge_ary[ch_no].start],
 				  &sgl_offset,
 				  read_wr.num_sge);
-
+		if (((ch+1)->rc_discrim == 0) &&
+		    (read_wr.num_sge == ch_sge_ary[ch_no].count)) {
+			/*
+			 * Mark the last RDMA_READ with a bit to
+			 * indicate all RPC data has been fetched from
+			 * the client and the RPC needs to be enqueued.
+			 */
+			set_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags);
+			ctxt->next = hdr_ctxt;
+			hdr_ctxt->next = head;
+		}
 		/* Post the read */
 		err = svc_rdma_send(xprt, &read_wr);
 		if (err) {

1.5.5.rc1