cvs commit: src/sys/sys _types.h resource.h

[Dag-Erling Smørgrav]

Colin Percival <cperciva at tarsnap.com> writes:
> Dag-Erling Smørgrav <des at des.no> writes:
> > [...] we might as well change the value of RLIM_INFINITY to (rlim_t)-1
> > to match other OSes, and we can do it without significant breakage.
> Should we get a ports experimental run for this?

Sure.  It might also be a good idea to ask portmgr@ to grep tarballs for
rlim_t so we can inspect source code.  I just realized that there is a
potential for non-benign errors if a userland program compares a number
to a limit that happens to be set to RLIM_INFINITY (an older binary will
perform a signed comparison, which will return a different result if
RLIM_INFINITY is negative when interpreted as a signed number).  But
userland programs shouldn't be comparing numbers to limits - that's the
kernel's job, and the kernel will always be consistent with itself.

DES
-- 
Dag-Erling Smørgrav - des at des.no