On Mon, Nov 06, 2017 at 09:43:05AM +0800, oulijun wrote: > 在 2017/11/4 18:08, Leon Romanovsky 写道: > > On Wed, Nov 01, 2017 at 10:13:28AM -0600, Jason Gunthorpe wrote: > >> On Wed, Nov 01, 2017 at 06:00:40PM +0800, Lijun Ou wrote: > >>> @@ -385,7 +385,7 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *cq, > >>> case HNS_ROCE_RECV_OP_RDMA_WRITE_IMM: > >>> wc->opcode = IBV_WC_RECV_RDMA_WITH_IMM; > >>> wc->wc_flags = IBV_WC_WITH_IMM; > >>> - wc->imm_data = cqe->rkey_immtdata; > >>> + wc->imm_data = be32toh(cqe->rkey_immtdata); > >>> break; > >> > >> This can't be right, wc->imm_data is marked be32. > > > > Lijun, > > > > Are you going to fix it? > > > > Thanks > > > Yes， I think that it is not fixed. I have tested the roce device > aginst hip08, D05(ARM64) and CX3, and I also have tested the D05 > aginst cx3 and D05. I also tried to run sparse and don't produce a > userful result. I also follow Jason's advice. Is it ok? The patch I sent fixes the bug properly, you need to review and validate it. You'll also need to get sparse to work, in the mean time use 'buildlib/cbuild pkg travis' which will run sparse on hns once you apply my patch. My patch fixes all the sparse problems. Jason -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
