Very similar to my "correct" patch. There's one other issue I found when I
was coming up with a patch which is that Darwin:getgrouplist also doesn't
return the number of groups. It sets the in/out parameter, but then just
returns 0.
The only thing I wasn't too sure about was how risky it is to just keep
doubling. If some bizzaro-world getgrouplist was just implemented to return
-1, well, the results wouldnt be great. But maybe that's not a significant
concern.
My patch:
diff --git a/gl/lib/mgetgroups.c b/gl/lib/mgetgroups.c
index e697013..8d08ae8 100644
--- a/gl/lib/mgetgroups.c
+++ b/gl/lib/mgetgroups.c
@@ -83,7 +83,17 @@ mgetgroups (char const *username, gid_t gid, GETGROUPS_T
**groups)
GETGROUPS_T *h;
/* getgrouplist updates max_n_groups to num required. */
+ /* Some old getgrouplist return -1 but don't update max_n_groups
*/
+ int previous_max_n_groups = max_n_groups;
ng = getgrouplist (username, gid, g, &max_n_groups);
+
+ if (ng < 0 && max_n_groups == previous_max_n_groups)
+ {
+ max_n_groups <<= 1;
+ } else {
+ /* some old getgrouplist don't return max_n_groups on success
*/
+ ng = max_n_groups;
+ }
if ((h = realloc_groupbuf (g, max_n_groups)) == NULL)
{
_______________________________________________
Bug-coreutils mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/bug-coreutils
