Re: Build error with GCC 10 in NetworkInterface.c and k_standard.c

[Koichi Sakata]

Hi Andrew,

Thank you for teaching me.
I made the patch that returns NaN. Could you please sponsor it?
In this case I think it is better to fix k_standard.c itself rather than 
ignoring the warning. However, I was able to learn there are some good 
ways to suppress warnings. So I'd like to thank Ioi and Yasumasa.

Thanks,
Koichi

===== PATCH =====
diff --git a/src/java.base/share/native/libfdlibm/k_standard.c 
b/src/java.base/share/native/libfdlibm/k_standard.c
--- a/src/java.base/share/native/libfdlibm/k_standard.c
+++ b/src/java.base/share/native/libfdlibm/k_standard.c
@@ -739,6 +739,10 @@
                         errno = EDOM;
                 }
                 break;
+            default:
+                exc.retval = zero / zero;
+                errno = EINVAL;
+                break;
         }
         return exc.retval;
 }
diff --git a/src/java.base/unix/native/libnet/NetworkInterface.c 
b/src/java.base/unix/native/libnet/NetworkInterface.c
--- a/src/java.base/unix/native/libnet/NetworkInterface.c
+++ b/src/java.base/unix/native/libnet/NetworkInterface.c
@@ -1296,7 +1296,8 @@
 static int getIndex(int sock, const char *name) {
     struct ifreq if2;
     memset((char *)&if2, 0, sizeof(if2));
-    strncpy(if2.ifr_name, name, sizeof(if2.ifr_name) - 1);
+    strncpy(if2.ifr_name, name, sizeof(if2.ifr_name));
+    if2.ifr_name[sizeof(if2.ifr_name) - 1] = 0;

     if (ioctl(sock, SIOCGIFINDEX, (char *)&if2) < 0) {
         return -1;
@@ -1359,7 +1360,8 @@
 static int getFlags(int sock, const char *ifname, int *flags) {
     struct ifreq if2;
     memset((char *)&if2, 0, sizeof(if2));
-    strncpy(if2.ifr_name, ifname, sizeof(if2.ifr_name) - 1);
+    strncpy(if2.ifr_name, ifname, sizeof(if2.ifr_name));
+    if2.ifr_name[sizeof(if2.ifr_name) - 1] = 0;

     if (ioctl(sock, SIOCGIFFLAGS, (char *)&if2) < 0) {
         return -1;


On 2020/07/22 17:24, Andrew Haley wrote:
On 17/07/2020 12:26, Koichi Sakata wrote:

  > You'll need to find a reviewer that understands what that
  > method is supposed to do in that case, that's not me ;-)

I understand. This ML is suitable for finding a reviewer, isn't it?
Or, there is another way. We can avoid the error by the accepting
maybe-uninitialized warning in libfdlibm instead of fixing k_standard.c.

I think it'd be better to fix it. Although the logic is such that all of
the cases are covered by the switch statement in k_standard.c, the default
case should return NaN in retval: use zero/zero.

I'll approve that.