wtmp corruption (was Re: Is write(2) "atomic" ?)

[HO]

In article <11410006 at eecs.nwu.edu>, naim at eecs.nwu.edu (Naim Abdullah) writes:
> However, I was able to solve the original problem by using O_APPEND
> when I open(2)'ed the output file. 
>              the 4.3bsd login(1) opens wtmp using O_APPEND as login faces
> the same problem of multiple concurrent writers to the same wtmp
> file (when many people are logging in and out at the same time).
> I just checked our System V rel 3.1 sources and the system V login
> fseeks before fwriting to wtmp so I imagine it will suffer from this
> problem too.

The wtmp file corruption is a problem (for acct).
Especially so for some systems that I use:

1) over 200 users on the system,  
2) the system has more than 1 CPU,  
3) the system has more than one /etc/init (one is the real init,
	the other is a kind of init, but for mux interface).

Months ago,  I have asked Summit for the O_APPEND change in init/getty/login,
and I hope this will be in SVR4.  It is time to junk wtmpfix.

	Paul Ho