2010/6/21 Pádraig Brady <P(a)draigbrady.com>:
Sorry I haven't been following this really, but I loath
config options that aren't absolutely required, and this
seems like a place where we could just stat() (cache) always.
What's the problem with doing that?
I too would be really interested in any real-world application for
which the cost of the stat() was any kind of major performance hit.
p.s. not having looked at the code,
the atomic ops seems unusual in the
presence of the static last_... variables.
We are implicitly relying on aligned word size assignment atomicity
here it looks like, yes. I haven't looked if there are other places
in glibc doing this or if there's some infrastructure for it. To be
clear this patch is derived from the OpenSUSE one; I didn't change the
semantics there.