tag 411059 + wontfix
severity 411059 wishlist
thanks
On Thu, Feb 15, 2007 at 12:51:12PM -0800, Steve Langasek wrote:
> On Thu, Feb 15, 2007 at 03:47:53PM +0000, Ian Jackson wrote:
> > Steve Langasek writes ("Re: recent etch upgrade... sashroot (uid=0) started to impersonate uid=0 (root)"):
> > > Again, I believe the behavior is not a bug because the behavior of
> > > getpwuid() when two users share the same uid is undefined.
>
> > Where is the format of /etc/passwd standardised, so that we can see
> > what that says about the meaning of entries with identical uids ?
>
> I'm not aware that it is. But getpwuid() certainly is standardized.
> http://www.opengroup.org/onlinepubs/009695399/functions/getpwuid.html, for
> example, doesn't say anything about how getpwuid() is required to resolve
> uid conflicts in the password database. That makes it "undefined" in my
> book.
>
> The behavior of most getpwuid() implementations that act on /etc/passwd is
> *predictable*, but AFAIK not standardized.
>
> > > This behavior is related to other nscd issues in the past that /were/ bugs
>
> > It seems to me that nscd is buggy in that it fails to preserve the
> > long-established de-facto standard behaviour for getpwuid.
>
> The de-facto standard that's violated by nss_ldap, nss_nisplus, nss_hesiod,
> nss_db, and any other backend that doesn't use an ordered flatfile for its
> data storage?
I totally agree with that. the _gnu libc_ getpwuid implementation is
nothing, even not a de facto standard. I'm almost sure it does not
behave the same on other OS'es.
--
·O· Pierre Habouzit
··O madcoder@debian.org
OOO http://www.madism.org
Attachment:
pgphgFtuxa3cx.pgp
Description: PGP signature