[BusyBox] Minor problems

Erik Andersen andersen at lineo.com
Tue Jan 30 18:03:45 UTC 2001


On Tue Jan 30, 2001 at 08:31:24AM -0800, Matt Kraai wrote:
> On Tue, Jan 30, 2001 at 08:49:34AM -0700, Erik Andersen wrote:
> > On Mon Jan 29, 2001 at 07:15:25PM -0800, Matt Kraai wrote:
> > > On Mon, Jan 29, 2001 at 05:02:01PM -0800, Larry Doolittle wrote:
> > > > 2. I _think_ I built this copy with USE_SYSTEM_PWD_GRP = true,
> > > >   but ls on an NFS mounted directory gives me:
> > > > # ls -l /mnt/recycle 
> > > > -rwxrwxr-x   1 ls: unknown uid 501
> > > 
> > > ls.c uses the my_getpwuid functions from utility.c.  I guess it
> > > should be converted to the standard ones now that pwd_grp/ is
> > > included.  A patch to fix both 1 and 2 is attached.
> > 
> > If you look in utility.c, the my_* password and group functions
> > now use the standard interfaces internally.  And since they do 
> > the error checking and pulling the needed string out from struct 
> > passwd or struct group they do save some memory that way.
> 
> Should ls exit abnormally when it can't find the name associated
> with a file's uid?  GNU ls prints the uid and continues.

Hmm.  No, it should print the uid.  Yes, ok I see.  The current my_getpwuid()
and my_getgrgid() should be modified to string-ify the uid on error instead of
doing error_msg_and_die().  And my_getgrnam() and my_getpwuid should return -1
on failure, not simply exit.  This should fix the problem generally, not just
for ls.  I think the stuff I just checked in should make things behave.

 -Erik

--
Erik B. Andersen   email:  andersen at lineo.com
--This message was written using 73% post-consumer electrons--





More information about the busybox mailing list