From: Chuck Lever chuck.lever@oracle.com
commit ca1c671302825182629d3c1a60363cee6f5455bb upstream.
The @nents value that was passed to ib_dma_map_sg() has to be passed to the matching ib_dma_unmap_sg() call. If ib_dma_map_sg() choses to concatenate sg entries, it will return a different nents value than it was passed.
The bug was exposed by recent changes to the AMD IOMMU driver, which enabled sg entry concatenation.
Looking all the way back to commit 4143f34e01e9 ("xprtrdma: Port to new memory registration API") and reviewing other kernel ULPs, it's not clear that the frwr_map() logic was ever correct for this case.
Reported-by: Andre Tomt andre@tomt.net Suggested-by: Robin Murphy robin.murphy@arm.com Signed-off-by: Chuck Lever chuck.lever@oracle.com Cc: stable@vger.kernel.org Reviewed-by: Jason Gunthorpe jgg@mellanox.com Signed-off-by: Anna Schumaker Anna.Schumaker@Netapp.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org
--- net/sunrpc/xprtrdma/frwr_ops.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-)
--- a/net/sunrpc/xprtrdma/frwr_ops.c +++ b/net/sunrpc/xprtrdma/frwr_ops.c @@ -298,8 +298,8 @@ struct rpcrdma_mr_seg *frwr_map(struct r { struct rpcrdma_ia *ia = &r_xprt->rx_ia; struct ib_reg_wr *reg_wr; + int i, n, dma_nents; struct ib_mr *ibmr; - int i, n; u8 key;
if (nsegs > ia->ri_max_frwr_depth) @@ -323,15 +323,16 @@ struct rpcrdma_mr_seg *frwr_map(struct r break; } mr->mr_dir = rpcrdma_data_dir(writing); + mr->mr_nents = i;
- mr->mr_nents = - ib_dma_map_sg(ia->ri_id->device, mr->mr_sg, i, mr->mr_dir); - if (!mr->mr_nents) + dma_nents = ib_dma_map_sg(ia->ri_id->device, mr->mr_sg, mr->mr_nents, + mr->mr_dir); + if (!dma_nents) goto out_dmamap_err;
ibmr = mr->frwr.fr_mr; - n = ib_map_mr_sg(ibmr, mr->mr_sg, mr->mr_nents, NULL, PAGE_SIZE); - if (unlikely(n != mr->mr_nents)) + n = ib_map_mr_sg(ibmr, mr->mr_sg, dma_nents, NULL, PAGE_SIZE); + if (n != dma_nents) goto out_mapmr_err;
ibmr->iova &= 0x00000000ffffffff;