Re: [dpdk-dev] [PATCH] test: fix division by zero

[Min Hu (Connor)]

在 2021/5/13 0:53, Bruce Richardson 写道:

On Fri, Apr 23, 2021 at 03:35:30PM +0800, Min Hu (Connor) wrote:

Variable i is used as a denominator which may be zero, and
this may result in segmentation fault.

This patch fixed it.

Fixes: 948bc3d6d095 ("test: add reciprocal based division")
Cc: sta...@dpdk.org

Signed-off-by: Min Hu (Connor) <humi...@huawei.com>
---
  app/test/test_reciprocal_division_perf.c | 5 +++--
  1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/app/test/test_reciprocal_division_perf.c 
b/app/test/test_reciprocal_division_perf.c
index a7be8aa..2647308 100644
--- a/app/test/test_reciprocal_division_perf.c
+++ b/app/test/test_reciprocal_division_perf.c
@@ -143,7 +143,7 @@ test_reciprocal_division_perf(void)
                                        "result %"PRIu64"",
                                        nresult_u64, rresult_u64);
                        result = 1;
-                       break;
+                       goto err;
                }
        }

@@ -182,7 +182,7 @@ test_reciprocal_division_perf(void)

                                        dividend_u64, divisor_u64,
                                        nresult_u64, rresult_u64);
                        result = 1;
-                       break;
+                       goto err;
                }
        }
        printf("64bit Division results:\n");
@@ -195,6 +195,7 @@ test_reciprocal_division_perf(void)
        printf("Cycles per division(reciprocal) : %3.2f\n",
                        ((double)tot_cyc_r)/i);

+err:

        return result;
  }

This looks correct as far as it goes, but I believe the same fix is needed
at lines 66 and 106 too.

One other thing I note is that currently the test will move on to the
next test case on failure, due to break, but using the goto will change
that behaviour. Therefore, I wonder if a better fix is to skip the
printouts if i == 0 in each case?

Good point, fixed in v2, thanks Bruce.

/Bruce
.