On Thu, May 31, 2018 at 5:49 PM, Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote:
> Hello Selvin Xavier,
>
> The patch 1ac5a4047975: "RDMA/bnxt_re: Add bnxt_re RoCE driver" from
> Feb 10, 2017, leads to the following static checker warning:
>
> drivers/infiniband/hw/bnxt_re/qplib_fp.c:452 bnxt_qplib_enable_nq()
> error: we previously assumed 'nq->bar_reg_iomem' could be null (see line 438)
>
> drivers/infiniband/hw/bnxt_re/qplib_fp.c
> 430 nq->bar_reg = NQ_CONS_PCI_BAR_REGION;
> 431 nq->bar_reg_off = bar_reg_offset;
> 432 nq_base = pci_resource_start(pdev, nq->bar_reg);
> 433 if (!nq_base) {
> 434 rc = -ENOMEM;
> 435 goto fail;
> 436 }
> 437 nq->bar_reg_iomem = ioremap_nocache(nq_base + nq->bar_reg_off, 4);
> 438 if (!nq->bar_reg_iomem) {
> ^^^^^^^^^^^^^^^^^
> NULL here.
>
> 439 rc = -ENOMEM;
> 440 goto fail;
> 441 }
> 442
> 443 rc = bnxt_qplib_nq_start_irq(nq, nq_idx, msix_vector, true);
> 444 if (rc) {
> 445 dev_err(&nq->pdev->dev,
> 446 "QPLIB: Failed to request irq for nq-idx %d", nq_idx);
> 447 goto fail;
> 448 }
> 449
> 450 return 0;
> 451 fail:
> 452 bnxt_qplib_disable_nq(nq);
> ^^
> This function checks for NULL but only after it's already been
> dereferenced. Goto fail is always buggy.
>
Thanks Dan for reporting it. I will fix it.
> 453 return rc;
>
> regards,
> dan carpenter
--
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
