Created attachment 448761 [details] Test program Description of problem: This problem is very similar to bug #550968, yet it's something different. The attached test program shows that gettimeofday() exposes second increments earlier than time(). There is a fix for this in the mainline kernel, but it was never integrated in the stable 2.6.21 series or the RHEL kernel. The patch should be integrated in the RedHat kernel. I have several wrongly-named dump/backup files at a customer site because of this. Fix in mainline (between 2.6.22-rc2 and 2.6.22-rc3): http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=d0aff6e6f4e54f79f9c89d147d371bad384454e9 Version-Release number of selected component (if applicable): 2.6.18-194.el5
From BZ 461184: "jturner noted that performance testing on the new time() syscall resulted in a 2200% increase in the amount of time to call time(). This is unacceptable so I am reverting this small change which will leave time() and gettimeofday() unsynchronized in RHEL5." P.