On Dec 17 19:05, Johannes Schindelin wrote: > We should not blindly set the home directory of the SYSTEM account to > /home/SYSTEM, especially not when that value disagrees with what is > configured via the `db_home` line in the `/etc/nsswitch.conf` file. > > This fixes https://github.com/git-for-windows/git/issues/435 > > Signed-off-by: Johannes Schindelin > --- > winsup/cygwin/uinfo.cc | 6 +++++- > 1 file changed, 5 insertions(+), 1 deletion(-) > > diff --git a/winsup/cygwin/uinfo.cc b/winsup/cygwin/uinfo.cc > index a5d6270..8c51b82 100644 > --- a/winsup/cygwin/uinfo.cc > +++ b/winsup/cygwin/uinfo.cc > @@ -2129,7 +2129,11 @@ pwdgrp::fetch_account_from_windows (fetch_user_arg_t &arg, cyg_ldap *pldap) > it to a well-known group here. */ > if (acc_type == SidTypeUser > && (sid_sub_auth_count (sid) <= 3 || sid_id_auth (sid) == 11)) > - acc_type = SidTypeWellKnownGroup; > + { > + acc_type = SidTypeWellKnownGroup; > + home = cygheap->pg.get_home (pldap, sid, dom, domain, name, > + fully_qualified_name); Uhm, that's a bit over the top, isn't it? It will affect all S-1-5-X accounts as well as the S-1-5-11 Windows account SIDs. Is that really what you want? Using pldap here may SEGV in cygheap_pwdgrp::get_home, btw, because it may be NULL. cygheap_pwdgrp::get_home doesn't check pldap for validity, it expects a valid pointer. You could either use cldap, or cygheap_pwdgrp::get_home would have to check pldap before using it. However, either way there's another problem: Independently of the configured db_home schemes, you don't want to ask the DC for info on these builtin accounts. The better approach might be to call the PUSER_INFO_3 variant of cygheap_pwdgrp::get_home with a NULL ui pointer and just check for ui in the NSS_SCHEME_DESC case. The other called functions fetch_windows_home and fetch_from_path both can live with both pointers, pldap and ui being NULL. Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Maintainer cygwin AT cygwin DOT com Red Hat