Hi all, Well, after having written my nice NSS library, I started trying it out with some applications. All was going wonderfully until I noticed that pam_unix uses getpwnam() internally, which has the nasty side effect of overwriting the static buffer that was returned by that function previously. So my orignal username was blown away again. Is there some particular reason that pam_unix uses getpwnam over the reentrant version, getpwnam_r()? There seems to be some uncertainty re: getpwent_r(), but the #ifdefs would indicate that getpwnam_r is standardized. Would a patch substituting getpwnam_r for getpwnam be accepted? Thanks, Chris
