I hit a segfault in add_name with a mountd built with gcc-4.9.0. Some
NULL pointer checks got reordered such that a pointer was dereferenced
before checking to see whether it was NULL. The problem was due to
nfs-utils relying on undefined behavior, which tricked gcc into assuming
that the pointer would never be NULL.
At first I assumed that this was a compiler bug, but Jakub Jelinek and
Jeff Law pointed out:
"If old is NULL, then:
strncpy(new, old, cp-old);
is undefined behavior (even when cp == old == NULL in that case),
therefore gcc assumes that old is never NULL, as otherwise it would be
invalid.
Just guard
strncpy(new, old, cp-old);
new[cp-old] = 0;
with if (old) { ... }."
This patch does that. If old is NULL though, then we still need to
ensure that new is NULL terminated, lest the subsequent strcats walk off
the end of it.
Cc: Jeff Law <law@xxxxxxxxxx>
Cc: Jakub Jelinek <jakub@xxxxxxxxxx>
Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxxxxxxx>
---
support/export/client.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/support/export/client.c b/support/export/client.c
index dbf47b966522..f85e11c8b535 100644
--- a/support/export/client.c
+++ b/support/export/client.c
@@ -482,8 +482,12 @@ add_name(char *old, const char *add)
else
cp = cp + strlen(cp);
}
- strncpy(new, old, cp-old);
- new[cp-old] = 0;
+ if (old) {
+ strncpy(new, old, cp-old);
+ new[cp-old] = 0;
+ } else {
+ new[0] = 0;
+ }
if (cp != old && !*cp)
strcat(new, ",");
strcat(new, add);
--
1.9.0
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
