bug#12632: file permissions checking mishandled when setuid

[Paul Eggert <eggert@cs.ucla.edu>]

On 10/14/2012 01:17 PM, Eli Zaretskii wrote:
> If euidaccess always agrees with stat, why does it make sense to
> switch check_existing to using euidaccess

euidaccess always agrees with 'stat' with respect to access checks,
but there are other reasons that 'stat' could fail, e.g.,
on a file that is too large for 'stat' to represent the size.
In cases such as these, we want euidaccess here, because
euidaccess shouldn't fail for these other reasons.

> You are talking about the directories leading to the file.  I was
> talking about the file itself.

OK, but in this context it doesn't make sense to talk about just
the file itself, since the file is being accessed via a name, and
one needs proper permissions to use the name.

For both access and euidaccess, the F_OK check means "I don't
care about the permissions about the file itself; all I care about
are the permissions on the file's name", as far as permissions
are concerned.